Modern Concepts of the Theory of the Firm

Springer-Verlag Berlin Heidelberg GmbH

Sponsored by
Deutsche Forschungsgemeinschaft, Bonn, Germany
Erich-Gutenberg-Arbeitsgemeinschaft, Cologne, Germany
FernUniversität in Hagen, Germany
Märkische Bank e.G., Hagen, Germany
G. Fandei · U. Backes-Gellner
M. Schlüter · J. E. Staufenbiel
Editors

Modern Concepts
of the Theory
ofthe Firm
Managing Enterprises
of the New Economy

In Collaboration with
H. Raubenheimer
and C. Stammen-Hegen er

With too Figures and 82 Tables

'Springer
Prof. Dr. Günter Fandei
FernUniversität Hagen
Fachbereich für Wirtschaftswissenschaft
Lehrstuhl für Produktions- und Investitionstheorie
Universitätsstraße 41 · 58084 Hagen, Germany
Prof. Dr. Uschi Backes-Gellner
Universität Zürich
Lehrstuhl für BWL und empirische Methodik
der Arbeitsbeziehungen und Personalökonomik
Plattenstraße 14 · 8032 Zürich, Switzerland
Dr. Manfred Schlüter
Bauernweg 3 · 25524 Itzehoe, Germany
Dipl.-Kfm. Joerg E. Staufenbiet
Staufenbiet Personalberatung
Konrad-Adenauer-Ufer 33 · 50668 Köln, Germany

ISBN 978-3-642-07349-6 ISBN 978-3-662-08799-2 (eBook)

DOI 10.1007/978-3-662-08799-2

Cataloging-in-Publication Data applied for

A catalog record for this book is available from the library of Congress.
Bibliographie information published by Die Deutsche Bibliothek
Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data available in the internetat http.lldnb.ddb.de
This work is subject to copyright. All rights are reserved, whether the whole or part of the
material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data
banks. Duplication of this publication or parts thereof is permitted only under the provisions
oftheGerman Copyright Law of September 9, 1965, in its current version, and permission
for use must always be obtained from Springer-Verlag Berlin Heidelberg GmbH.
Violations are liable for prosecution under the German Copyright Law.

http://www.springer.de
© Springer-Verlag Berlin Heidelberg 2004
Original1y publisbed by Springer-Verlag Berlin Heidelberg New York in 2004
Softcover reprint of the bardeover 1st edition 2004

The use of general descriptive names, registered names, trademarks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
Cover design: Erich Kirchner, Heidelberg
SPIN 10943502 42/3130- 5 4 3 2 1 0- Printedon acid-free paper
Preface

This volume contains the results of the International Conference on "Man-

aging Enterprises ofthe New Economy by Modem Concepts ofthe Theory
ofthe Firm", which took place in Hagen from 12- 14 December 2002. The
conference was organised jointly by the FernUniversität in Hagen and the
Erich-Gutenberg-Arbeitsgemeinschaft in Cologne. The Deutsche For-
schungsgemeinschaft, the FernUniversität and the Märkische Bank in
Hagen provided generous financial support for the conference, and in fact
enabled its implementation and the publication of this record of the pro-
ceedings. We would like to express our gratitude to the sponsors for their
support.
The aim of the conference was the exchange of academic experience
with regard to new approaches to, and extensions of, the theory of the firm
that can be used to solve the problems of New Economy companies. The
starting point was the practical experience that the owners and managers of
New Economy companies paid too little attention to functional company
controls in the general euphoria regarding future economic trends and that
because of this the enterprises got into difficulties or were faced with
threats to their survival. These control deficits were detected equally in the
areas of sales, production, personnel planning, organisation and finance
and in the arrangement of management and control structures. Deficiencies
in company evaluations and accounting led to investors experiencing se-
vere disappointment on the loss of their invested capital.
These practical experiences gave rise to the considerations that Guten-
berg's foundations of a function-oriented theory of the firm were most
likely to provide a suitable basis for solving the pending problems. It is a
fact that the majority of the scientific papers in this conference report take
their methodological starting point from Gutenberg's foundations of the
theory of the firm. It is gratifying to see that Erich Gutenberg's conceptual
approaches are being diffused in this way into international academic re-
search as well.
We wish to thank in particular the chairmen ofthe conference sections,
who were at the same time members of the Scientific Committee, for put-
ting together the papers for the conference, the results of which are pre-
sented here, and for providing a very attractive programme for those taking
part. Of course, our thanks go on the first place to the speakers for their
contributions, which led to lively responses.
Heike Raubenheimerand Cathrin Stammen-Hegener took over the work
of editing this volume. This book would not have been published without
VI

their untiring work in bringing the manuscripts into a uniform content and
size, while at the same time harmonising the wishes of the authors with the
requirements of the publisher and the editors. They helped to design this
book and to organise the conference and our personal thanks go to them.
We also wish to thank Van Loi Nguyen as well, who very patiently put the
manuscripts into the right technical shape, so that the text could be used di-
rectly for printing. They all made every effort to carry out the necessary
editorial and technical work accurately and in time for the book to appear
very soon after the conference. We want to express our gratitude to Wemer
Müller and his colleagues at Springer-Verlag for enabling the publication
of these proceedings and for providing intensive support.
The Organisation Committee, which was composed of the editors,
wishes to thank Daniela Doliwa for her support in the local organisation.

Günter Fandei
Contents

Section 1: Corporate Governance I

Corporate Governance - Large and Small Corporations,

Agents, Principals, Competitors 3
Horst Albach

A New Management under IT Revolution in Japan 19

Koji Okubayashi

Section 2: Marketing 31

Relative Advantages ofE-Business Start-Ups versus Integrated

Units ofBricks-and-Mortar Companies 33
Sönke Albers

E-Business Performance: A Latent Class Examination 58

Timothy M. Devinney, Tim R. Coltman and David F. Midgley

Success Factars oflnternet-Based Business Models 69

Wolfgang Fritz

From the Old Economy towards the New Economy: Managing

the Transformation .from the Marketing Point of View 85
Michaela Haase and Michael Kleinaltenkamp

Mining Product Gonfigurator Data 110

Rainer Paffrath
VIII

Multi-Channel Management and its Impact on Customers'

Purehase Behavior 122
Bernd Skiera and Sonja Gensler

Section 3: Production and Procurement 139

E-Business in Production and Procurement - Some

Theoretical Perspectives 141
Ronald Bogaschewsky

E-Business Strategies in the Mechanical Engineering

Industry
An Approach to Explanations of Current Trends
in E-Business Diffusion 157
Axel Braßler and Herfried Schneider

Elements of the Production ofServices 175

Günter Fandei and Steffen Blaga

Strategie Supply Chain Management: A New Approach to

Analyze Product Life Cycles 190
Günter Fandei and Markus Stammen

An Analysis ofService OutputBasedon Production Theory 210

Ralf Gössinger

E-Business and New Forms ofCollaboration along the

Supply Chain 222
Stefan Kayser
 IX

Capahilifies ofthe Firm and their Effect on Performance-

Production ofInformation and Communication Technology
Services as an Example 238
Olli Kuivalainen and Sanna Taalikka

The Transaction Costs of eProcurement 253

Joachim Reese and Bjöm Saggau

A Complexity-Based Approach to Production Management

in the New Economy 264
Michael Reiss

Advanced Planning Systems - Basics and beyond 285

Hartmut Stadtier

Hierarchical Planning Structures in Supply Chain Management 301

Marion Steven

Section 4: Human Resource Management

and Economic Organization 313

Firm Foundations and Human Capital Investments:

The 0-Ring Approach to Organizational Equilibrium in
an Ernerging Industry 315
Oliver Fabel

Training Strategies and Remuneration Systems ofEnterprises

of the New and Old Economy- Similarities and Disparifies 339
Rosemarie Kay and Caroline Demgenski

Training: A Strategie Enterprise Decision? 355

Thomas Zwick
X

Section 5: Finance 367

Some New Properfies ofRisk Measures 369

Luca Bonaccorso, Salvatore Greco,
Benedetto Matarazzo and Pietro Platania

Discovery in the New Economy- Why Entrepreneurs may not

Contract with Investors 386
Helmut M. Dietl, Egon Franck and Stefan Winter

Japan 's Venture Capital Market from an Institutional

Perspective 398
Wemer Pascha and Stephan Mocek

Structural Analysis of Multinational Network Organizations 414

Manfred Perlitz and OlafN. Rank

Portfolio Return Characteristics ofDifferent Industries 434

Igor Pouchkarev, Jaap Spronk and Pim van Vliet

Valuation of Growth Companies and Growth Options 449

Markus Rudolf

Shareholder Value at Risk as an Instrument ofCompany

Valuation 474
Lars Schiefher and Reinhart Schmidt

Valuation ofIntangible Assets for Financial Reporting 491

Martin Scholich, Andreas Mackenstedt and Markus Greinert

Are Stars Worth their Pay? 505

Franz Wirl
 XI

Section 6: Accounting 521

A Tale oftwo Bubbles: A Preliminary Look at the US

Internet and Biotechnology Bubbles 523
Elizabeth Demers and Philip Joos

Communicating Intangible Resources for the Capital Market 552

Thomas W. Guenther

Business Valuation in the New Economy- Back to the Basics 575

Dirk Hachmeister

New Financial Accounting Standards for the New Economy?

- Same Remarks on the Ongoing Debate - 590
Christoph Kuhner

Controlling the Assets ofthe New Economy ... and not only the
New Economy 604
Rainer Strack

The Valuation oflntangibles in New Economy Firms 615

Peter Witt

ValueReporting TM
New Trends in Corporate Reporting 633
Joachim Wolbert

List of Contributors 643

 Section 1

Corporate Governance
Corporate Governance - Large and Small
Corporations, Agents, Principals, Competitors

Horst Albach

1 Preliminary remarks

1.1 Definition

Corporate Govemance is the set of rules of behavior for all activities of the
corporation. Two subsets of the rules of behavior may be distinguished:
the principals and the agents.
Principals have the right to in the final end set the rules of behavior.
Agents have to follow the rules of behavior. One person may be an agent
and a principal at the same time. Corporate govemance in this case is iden-
tical with corporate hierarchy. Corporate hierarchy is then a multi-level
principal-agents-relationship. In such a system, there is one person (group
ofpersons) which is a principal only, called the "source", and one group of
persons which is agents only (lowest rank in the hierarchy).
The set of rules of behavior may be partly standardized by law. Two
types of such legally defined corporate govemance systems may be distin-
guished:
- the one-tier-system
- the two-tier-system
In the one-tier-system the shareholders are the principals. They elect a
Board of Directors that govems the activities of the corporation in the
short term as well as in the long term. The board is expected to act in the
interest of the shareholders.
In the two-tier-system the shareholders are the principals. They elect a
Supervisory Board. The members of the Supervisory Board elect the
members of the Management Board. The Management Board govems the
activities of the corporation in the short term in the interest of the corpora-
tion. This is interpreted as balancing the interests of the various groups of
stakeholders in the corporation The Supervisory Board controls the be-
havior of the Management Board.
4

1.2 Justification of corporate hierarchy

The discussion about the effectiveness of different corporate govemance

systems has its origin in the frauds in and the crashes of corporations dur-
ing the last few decades. The thesis "supervision has failed" led to legisla-
tion for better supervision of top managers 1• However, there is a deeper
root to the discussion: does organization in general and corporate govem-
ance in particular matter in maintaining competitiveness of the company
on global markets?
It seems, therefore, necessary to shortly describe the economic theories
ofhierarchy that explain why corporate govemance systems matter.
Max Weber in his theories of govemance2 analyzed bureaucratic organi-
zations. He came to the conclusion that the hierarchies of bureaucratic or-
ganizations were most cost-efficient and therefore superior to all other
forms of govemance.
Robert Merton3 and others studied the extemal and intemal side effects
of bureaucratic hierarchies and proved that the costs of control for these
side-effects may become so high as to destroy bureaucratic organizations.
Therefore, control for dysfunctional effects pfhierarchies is mandatory4 •
In his "Rank in Organizations" Martin Beckmann5 studied the opti-
mality conditions for hierarchical organizations and for the span of control
in such govemance systems. He concluded from introducing hierarchical
forms into a Cobb-Douglas-production function that the owner of a com-
pany will introduce hierarchy as an organizational form if the productivity
gains from superiors controlling and consulting with workers on the shop
floor as well as subordinate managers outweigh the salaries of the manag-
ers on the different ranks in the organization.

1 Gesetz zur weiteren Reform des Aktien- und Bilanzrechts, zu Transparenz und
Publizität vom 19.7.2002. In: BGBl part I, p 2681; Gesetz zur Kontrolle und
Transparenz im Unternehmensbereich vom 27.4.1998, in: BGBl part I, p 786
2 Weber M (1956) Wirtschaft und Gesellschaft, Grundriss der verstehenden So-
ziologie. 4th ed, p 551
3 Merton RK (1940) Bureaucratic structure and personality. Social Forces: 560-
568
4 March JG, Sirnon HA (1958) Organizations. New York, pp 36- 47, see also:
Albach H, Gabelin T (1977) Mitarbeiterfiihrung. Wiesbaden 2nd ed. 1983, pp 20
-26
5 Beckmann MJ (1978) Rank in Organizations. Lecture Notes in Operations Re-
search and Mathematical Systems Vol. 161
 5

Kenneth Arrow6 studied the effects of non-observability of individual

effort in production regimes where the results can only be attributed to the
work group as a whole. In such a situation each individual is tempted to
maximize his (or her) personal utility by shirking from work. However, the
other members of the work group suffer a loss if there is one (or more) free
rider on the team. Therefore, hierarchy is in their interest. Workers will opt
for a hierarchical govemance system if the gains from preventing shirking
exceed the salaries of the managers that control the behavior of the work-
ers in the group.
Shirking is a special case of asymmetric information between principals
and agents in an organization. Such asymmetric information arises not
only from hidden action but also from hidden information and hidden
characteristics. Jensen and Meckling7 pointed out that asymmetric informa-
tion induces utility maximizing individuals to behave opportunistically and
confronts the principals with moral hazard. Principals will have to control
for "agency costs". Consequently, hierarchical govemance systems are vi-
able to the extent only that agency costs can be efficiently controlled.

1.3 Structure of top management

The question whether the one-tier-system or the two-tier system is a corpo-

rate govemance system with a relatively higher competitive advantage and
with lower agency costs is just a part of the general question of how organ-
izational form and processes influence the performance of companies.
In societies with a liberal legal tradition the decision on the corporate
govemance structure lies with the owners of the company. This principle
applies to privately held companies without restriction. Companies that is-
sue stock to the general public and whose stock is traded on the stock ex-
change cannot possibly meet the information requirements of investors in
privately held companies. Information cost would be excessively high.
Therefore, the corporate govemance system has tobe standardized by law.
There are societies that offer only one standardized form, e.g., the Akti-
engesellschaft. When the so-called Meiji reforms were introduced in Japan

6 Arrow KJ (1985) The Economics of Agency, Chapter 2. In: Pratt JW, Zeck-
hauser RJ (eds) Principals and Agents: The Structure ofBusiness, esp. p 47; see
also: Arrow KJ (1974) The Limits of Organization. In chapter 4 of this book
Arrow discusses "the value of authority". For "shirking" see also: Kräkel M
(1999) Organisation und Management, pp 104- 118
7 Jensen MC, Meckling WH (1976) Theory of the Firm: Managerial Behavior,
Agency Costs and Ownership Structure. Journal ofFinancial Economics 1976:
305-360
6

in 1869, the companies were given a choice among various forms of gov-
emance. Since the firms felt that the govemment preferred the two-tier-
system, they all opted for the same system - the two-tier system. In
France, the law offers the choice between the one-tier-system and the two-
tier-system. A vast majority of the French firms has opted for the one-tier-
system because shareholders were convinced that this system offered Stra-
tegie competitive advantages8 •

2 Efficient corporate governance systems

2.1 Corporate governance in the individual corporation

Recent research on corporate govemance systems attempts to provide em-

pirical evidence for the optimality of such options. At the present moment
it is evident that German corporations operating under the German two-
tier-system with full as with limited Co-determination have survived in
global competition. And so have US corporations with a mandatory one-
tier-system and Japanese corporations under their specific govemance sys-
tem. The conclusion therefore is that all three legal regimes provide for ef-
ficient organizational forms 9 •
However, this is surface evidence. More in-depth research has shown
that the one-tier system cannot efficiently control for agency costs. The
one-tier-system does not provide for effective supervision of management
behavior. Incentive-compatible contracts between the shareholders and the
top managers lead to fully exploiting the reporting standards of the al-
legedly shareholder-friendly USGAAP in order to boost stock prices, then
to "creative" reporting and then to fraud under a remarkable coalition of
top managers and CPAs 10 •
Stock options, originally advanced as incentive-compatible by Jensen
and Meckling, have produced counter-productive effects. They have re-

8 Albach H (2000) Globalisierung und Organisationsstruktur mittelständischer

Unternehmen - Eine Analyse aus europäischer Sicht. In: Schneider UH, et al
(eds) Deutsches und europäisches Gesellschafts- Konzern- und Kapital-
marktrecht, pp 673 - 688
9 LaPorta R, Lopez-de-Silanes L, Shleifer A, Vishny RW (2001) Law and Fi-
nance In: Schwalbach, Joachim (ed) Corporate Governance, pp 26-68
10 for a more detailed discussion see Albach H (2003) Zur Ökonomie der Habgier.
The Berlin Science Center
 7

sulted in what John Cassidy calls the "greed cycle" 11 • Control by the capi-
tal markets in the US has proved inefficient in the light of misreporting
profits by corporations.
While in Germany there is a tendency to change the govemance system
in the direction of the one-tier Board System, as clearly expressed in the
German Code of Good Corporate Govemance, observers in the United
States voice the opinion that the German two-tier system in combination
with the German more conservative reporting law may prove to be supe-
rior to the corporate govemance system of the United States of America. lt
seems to be more efficient in "reining in" the greed of CEOs.
Under the German Corporation Act quoted corporations are required to
publish a "comply" or "explain" declaration with respect to the German
Corporate Govemance Code (§ 161 AktG). That leaves German corpora-
tion with the choice between several Codes of Good Corporate Govem-
ance. They may feel that a declaration stating "We do not comply with the
recommendations of the German Code because we are convinced that we
achieve lower cost of capital if we comply with the OECD Code - which
we do" 12 improves their competitiveness on the capital market.

2.2 Corporate governance in global groups of companies

The German Code of Corporate Govemance does not mention any rules
for good govemance in multinational companies. In general, govemance in
these groups is carried out through board membership. In the two-tier-sys-
tem there are many problems that have gone unsolved so far. The outside
board members in the mother company are inadequately informed about
the activities and the risks in major subsidiaries. Top managers of major
subsidiaries practically never report to the board of the mother company.
The auditors of the mother company rely on the audit reports of the CP As
that have audited the subsidiaries. The CP A report submitted to the super-
visory board of the mother company practically never reports on the major
subsidiaries and their risk management system.
The outside directors in affiliated companies are never put in a position
to approve decisions with respect to their companies. The decisions have
been taken on the level of the mother company, approved by the supervi-
sory board of the mother company, and major strategic decisions that af-
fect the subsidiary cannot properly be reported on in a "dependence report"

11 Cassidy J (2002) The Greed Cycle. The New Yorker: pp 64 - 77

12 Ad Hoc Task Force on Corporate Govemance: OECD Principles of Corporate
Govemance. SG/CG(99)5, April16, 1999
8

(Abhängigkeitsbericht). Therefore, outside members on the supervisory

board of the subsidiary insist on signing a "govemance contract" (Beherr-
schungsvertrag) including profit and loss accrual to the mother company.
After such a contract has been signed the govemance system of the sub-
sidiary does not serve any other but a purely legal purpose: Under the law
it remains legally necessary for the subsidiary to have a supervisory board
with workers' representatives on it.

2.3 Corporate governance in start-up firms

During the recent boom of venture business, particularly in the e-business

sector, it has become evident that start-up firms with high-growth expecta-
tions give rise to very interesting govemance problems. It is important to
distinguish between the start-up-phase and the !PO-phase.

2.3.1 Corporate governance in the start-up-phase

The founders of a start-up firm do not have the amount of capital necessary
to finance the young firm. Investment banks do not consider it their busi-
ness to fmance venture firms. Nor do the universal banks in Germany.
Therefore, the founders need venture capital. The founders want to remain
in control of their company. This means that the venture capitalists have to
supply money without gaining a dominant position in the firm. This Ieads
to govemance problems.
The founders maintain a dominant position in their firm by receiving
stock that amounts to a majority of the equity in the firm. The shares is-
sued to them are either fully in compensation for the business idea or the
founders pay in cash in addition to the valuation of their business idea that
guarantees a stake of at least 51 % in the stock of the company. In general,
the cash comes from a private loan that the founders take up with their
banks against personal property and against the promise to retire the debt
when the firm goes public. The founders usually try to attract also business
angels that supply some cash and provide advice and reputation to the
start-up firm.
In such a set-up, there is
- a group of dominant shareholders (the founders)
- one or a group of venture capitalists that exercise all the rights of a
minority shareholder and that in general have a contract with the
firm that grants them certain veto rights
- a group ofventure capitalists, called business angels, that hold mi-
nority positions in the company
 9

- one or several banks that have an indirect interest in the firm

through their credit contracts with the founders.
Venture capitalists and banks by contract reserve the right to force the
company to go public whenever they feel the time is appropriate. There-
fore, they insist that the company is set up under the legal form of a "Sniall
Corporation" (Kleine Aktiengesellschaft) 13 • This facilitates an IPO. They
also insist on the company to prepare financial reports on the basis of
USGAAP. Thus, investors from the United States can more easily be at-
tracted to purchase shares in the IPO of the company.
The Small Corporation is govemed by a two-tier-system. The composi-
tion of the supervisory board is not subject to Codetermination Law. The
Management Board and the Supervisory Board work closely together.
They work closely together in all strategic and operational decisions and,
therefore, come close to the one-tier-system. These are factors that help to
attract foreign capital.
The German Code of Corporate Conduct does not deal with the Small
Corporation nor does it mention the govemance problems of companies
quoted on what used tobe the "New Market". Since most of the start-up
companies prepared for an IPO on the NASDAQ rather than on a German
stock exchange, start-up companies could not care less about compliance
or non-compliance with the German Code of Corporate Govemance. They
comply with the code that gives them most favorable treatment on the
NASDAQ.
Problems of corporate govemance may arise between the venture capi-
talists and the founders. In a practical case the venture capitalist insisted on
an IPO as stipulated in the equity contract, while the founders wanted to
merge with a large US competitor. The founders threatened to quit the
company if the venture capitalist insisted. He bad to give in.
Problems arise also in start-up companies with a major shareholder and
several business angels. In a practical case the major shareholder, a state
bank in Germany, sold software to the company in which it owned 30% of
the stock, thus increasing its share to 75 %. A CPA bad valued the soft-
ware under the going-concem assumption at a time when the bum rate of
the company clearly showed that this assumption no Ionger bad a sound
basis. The bank used its majority to call an extraordinary shareholders'
meeting on Christmas Eve. The majority voted to close down the company

13 The origins of the idea of the Small Corporation may be found in: Albach H,
Corte C, Friedewald R, Lutter M, Richter W (1988) Deregulierung des Aktien-
rechts: Das Drei-Stufen-Modell
10

informing the opposing businessangelsthat they had put up venture capi-

tal and venture capital was by its very nature a loss proposition.
Finally, problems anse between the top managers and the business an-
gels. The founders and top managers of the start-up company get their
salanes accepted by the compensation committee of the supervisory board.
The founders need high salanes in order to pay back the private loans they
have taken up with their banks in order to be able to sign up for shares in
their company. They have the insider information that informs them about
impending danger of survival. They try to prolong the lifetime of their
company as long as possible. There is no jurisdiction yet on when this re-
sults in illegal bankruptcy deferral. When eventually the company declares
insolvency, the founders have paid back their bank loans, and the business
angels have lost their money.

2.3.2 Corporate governance in the /PO-phase

Valuation of companies for an IPO poses serious methodological prob-
lems. Therefore, the risk of error in the valuation should not be bome by
the old investors and by the new ones, but also by the financial interme-
diary that is involved in the IPO. However, the New Marketin Germany
was introduced with the financial intermedianes exempted from practically
all responsibility. The small investors suffered very severe losses when the
IPO euphoria of start-up firms particularly in the e-business sector came to
an end and the New Market collapsed. Studies of pre-IPO valuations of
companies have shown that start-up companies were valued extremely dif-
ferently by different intermedianes and that during the two-week-book-
building period the main bank in the consortium forced the valuation down
by as much as 40 % 14 • And yet the new investors suffered severe losses
shortly after the IPO.
Good corporate govemance is not only a question of the relationship be-
tween present shareholders and management. It involves potential inves-
tors during an IPO as well. While investment bankers are known to act in
the interest of the incumbent shareholders, universal banks with invest-
ment banking and private equity activities will get in a conflict of interests
as financial intermedianes. This conflict is part of the govemance problern
in an IPO. The company going public is not required under the present law
to make this conflict transparent by informing the potential investors about

14 Hanusch A, Albach H (2003) Preisbildung :fiir junge Aktien bei einem IPO, in:
Albach H, Pinkwart A (eds) (2002) Unternehmensgründung und Unternehmer-
tum. Zffi-Supplement 2: pp 115- 125, see also: RudolfM, Witt P (2002) Be-
wertung von Wachstumsunternehmen
 11

the methods used in the valuation of the shares and about the results of the
various expert valuations during the whole IPO-process.

3 Corporate governance as competition between

systems

3.1 Efficient corporate governance in bank-related systems

To the extent that companies arenot free to shape their individual corpo-
rate governance system because the legal system limits free choice to cer-
tain standard forms, competition between companies with different corpo-
rate governance on the global markets is competition between different
nationallegal systems.
Such competition does not necessitate to harmonize the legal systems as
is assumed by the harmonization school in the European integration dis-
cussion. Theoretical analyses have shown that national systems may differ
in certain points and may have to be harmonized in others. Harmonization
of the systems is not the necessary result of the competitive process. If it
happens, it is the outcome of the attempts of one dominant system to mo-
nopolize the world 15 •
The German system has several elements. They are the following: two-
tier-system, bank finance, creditors' view in financial reports. These ele-
ments are consistent. The banks monitor their loans to the corporation and
by extending credit to them signal to the shareholders, incumbent and po-
tential, that their investment is safe. This leads to a reduction of the cost of
equity by reducing information costs and risk premia even if debt-equity
ratios are much higher than in a capital market-oriented system. The risk
premium for bank credit is reduced by the creditor-orientation of the finan-
cial reporting system.
Thus the German system is a consistent legal system of credit finance,
house bank monitoring, high debt-equity-ratios, conservative financial re-
porting, dominance of institutional investors with professional interpre-
tation of signaling by banks and professional evaluation of company risks
and with the right to monitor the corporation through a representative on
the supervisory board of the two-tier-system. The important conclusion is
that the discussion of efficient corporate governance systems has to take all

15 Witt P (2003) Corporate Governance-Systeme im Wettbewerb

12

these elements into account. It is not correct to focus on the one-tier-sys-

tem and the two-tier-system alone.

3.2 Efficient corporate governance in standardized production

regimes

Corporate govemance systems are not only linked to elements from the fi-
nance side of society, but also to the production regime that characterizes
society.
A standardized production regime is typical of the United States. Such a
system does not require long periods of leaming. Short on-the-job training
periods are sufficient to make a worker perform adequat~ly in a work
place. Investment in human capital by the worker as well as by the com-
pany is low in such a system. Accordingly, moving between firms is typi-
cal of an American worker. Average stay in one job is significantly shorter
than in Germany. Emotional commitment to the firm is not in general a
strong element in the individual's utility function. The extemallabor mar-
ket is efficient. Incentives to look for a new job, once a worker is given no-
tice, are fairly high.
Fairly short periods of work for one company are not only typical of
workers on the shop floor, but also of management. The officers of a com-
pany change much faster than in the German system. So do the share-
holders. Average holding periods of shares of one particular company are
much shorter than used to be characteristic of German investors. Conse-
quently, asymmetric information between shareholders and top executives
of a company is a much greater problern in the United States than in Ger-
many. The shareholders do not expect loyalty to the company to be the
foremost consideration of top managers. Therefore, they are convinced that
without offering incentive-compatible contracts to top managers, the hired
managers will act in their own but not in the interest of the shareholders.
Therefore, the acquisition of stock of the company and the offer of stock
options are necessary elements of the govemance system. Controlling top
management is much more important than in a system with long periods
on the job and retirement pay contracts. The Board System gives outside
directors a very good opportunity not only to control for results but also to
observe behavior of the top executives. The frequency of Board meetings
is much higher than in the German system.
In the highly specialized and sophisticated production regime typical of
Germany, an apprenticeship system is important. In such a system, invest-
ment of the firm in the human capital of its apprentices is high. Conse-
quently, it is in the interest ofthe company that the apprentices after finish-
 13

ing their training stay on in the company. The workers will continue with
the company even though firms that do not train apprentices may offer
them higher wages if the offer to stay on is a credible offer for lifetime
employment. In fact, it used to be typical of the German industrial system
that generations of workers worked for the same company. The offer is
credible if the top managers that have made the offer stay with the com-
pany as long as the worker. The workers thus have an interest in long peri-
ods of employment of the top managers. Mutual trust and credible loyalty
to the firm are characteristic of such a production regime. Of course, no
top manager can offer the worker a contract, not even an implicit one, for
lifetime employment. But he can express bis care for maintaining life-time
employability of bis workers. Life-long education systems installed by top
managers of a company can be interpreted as such a credible offer. If the
top managers rise from the ranks of the company to their executive posi-
tions, workers have a good "feel" oftheir character. Shareholders know the
track record of the people that are elected by the supervisory board to the
top positions. They are weil informed of the commitment of top managers
to the shareholders' interests and oftheir loyalty to the company. They can
and will offer long-term contracts with retirement pay promises to their top
managers. There is no need for stock options. The members ofthe supervi-
sory board trust the persons that they have elected to top positions. There-
fore, the hidden action and hidden information problems are much less im-
portant in such a system than in the American system. The frequency of
meetings ofthe supervisory board is lower.
Both production regimes and their corresponding govemance systems
seem to be consistent. Which one tums out to be globally more efficient
cannot be forecast with certainty as yet. But it should be clear again that
changing just one element in either system may cause inconsistency of the
whole system. This would result in serious strategic disadvantages of the
firm.

4 Concluding remarks

Observers of developments in the corporate world have recognized a ten-

dency towards the system of corporate govemance in the United States.
Voices that warn of such a development have become stronger in recent
years. Henry Gates has asked the question: "Why has betraying everyone
become a national pastime?". He continued with the question: "What be-
comes of loyalty to basic institutions when the only basic institution left is
14

the self?" He concludes: "You attain power not by putting down roots but
by "networking" 16 •
A system without loyalty to institutions and without trust in partners is a
system with high transaction costs. Such a system is in the words of
Andres Veiel "doomed": "One cannot be honestand at the sametime suc-
cessful in this firm. One has to make one's fingers dirty. The firm func-
tions on the basis of a mephistophelic system of seduction through bribes.
The firm has developed a structure in which everyone takes what he can
get. Loyalty to the firm is no Ionger important. Just effectiveness which is
rewarded through a generous system of stock options. This system of stock
options stands for nothing but greed and self-interest to get rieb and richer
at any cost. Within the last few years basic salaries of the members of the
Management Board increased eight times from one million to eight mil-
lion. Social responsibility and transparency are outdated and outmoded
concepts. The system is corrupt to the roots. Everybody is aware of it, eve-
rybody despite this awareness carries on, with ever increasing speed. The
whole system is doomed. Everybody knows it but does not do anything to
stop it" 17
George Akerlof in bis 1983 article on loyalty filters has shown a way to
stop this development18 • He proved. that loyalty to an institution is eco-
nomically reasonable. It is, in bis view, the highest form of self-interest.
Therefore, the conclusion may be drawn that in the final end those govem-
ance systems will prove to be optimal that are based on mutual trust, on
loyalty and on faimess.

16 Gates HL Jr. (1998) The End ofLoyalty. The New Yorker: 34-44
17 Veie1 A (2002) BlackBox BRD. A1fred Herrhausen, die Deutsche Bank, die
RAF und Wolfgang Grams
18 AkerlofGA (1983) Loyalty Filters. The American Economic Review 1983: 54
-63
 15

References

AkerlofGA (1983) Loyalty Filters. The American Economic Review: 54-63

Albach H (2003) Zur Ökonomie der Habgier. Paper delivered to the WZB-Forum
on January 28. The Berlin Science Center, Berlin
Albach H (2000) Globalisierung und Organisationsstruktur mittelständischer Un-
ternehmen - Eine Analyse aus europäischer Sicht. In: Schneider UH,
Hammelhoff P, Schmidt K, Timm W, Grunewald B, Drygala T (eds) Deut-
sches und europäisches Gesellschafts-, Konzern- und Kapitalmarktrecht Es-
says in Honor ofMarcus Lutter on his70th Birthday, Köln, pp 3- 15
Albach H (1999) Corporate Governance. Jaakko Honko Lectures 1998, Helsinki
Albach H (1997) Strategische Unternehmensplanung und Aufsichtsrat. Zeitschrift
für Unternehmens- und Gesellschaftsrecht 32 - 40
Albach H (1993) Unfinished Business. In: BierichMet al. (eds) Unternehmen und
Unternehmensfiihrung im Recht. Berlin New York, pp 673 - 688
Albach H, Corte C, Friedewald R, Lutter M, Richter W (1988) Deregulierung des
Aktienrechts: Das Drei-Stufen-Modell. Gütersloh
Albach H, Gabetin T Mitarbeiterfiihrung. Wiesbaden 1977, 2nd ed.: Wiesbaden
1983
Arrow KJ (1985) The Economics of Agency. Chapter 2. In: Pratt JW, Zeckhauser
RJ (eds) Principals and Agents: The Structure ofBusiness. Boston
Arrow KJ (1974) The Limits ofOrganization. New York
BauerS (1994) Große Freiheit für kleine Firmen. TopBusiness 1994: 120- 122
Baums T (200 1) Bericht der Regierungskommission Corporate Governance. Köln
bb (2002) Corporate Govemance bei ABB. Neue Zürcher Zeitung, February 14,
2002: 19 and NZZ February 15,2002: 19
Beckmann MJ (1978) Rank in Organizations. Lecture Notes in Operations Re-
search and Mathematical Systems, Vol. 161. Berlin Heidelberg New York
Bernhard W (2002) Der Deutsche Corporate Governance Kodex: Zuwahl (com-
ply) oder Abwahl (explain)? Diskussionsbeitrag Nr. 28 der Wirtschafts-
wissenschaftlichen Fakultät der Universität Leipzig, June 2002
Bernhardt W (2002) Vorstand und Aufsichtsrat (unter Einschluss des Verhält-
nisses zum Abschlussprüfer). In: Hornmethoff P. et al. (eds) Corporate Go-
vernance. Beihefte der Zeitschrift für das gesamte Handelsrecht und Wirt-
schaftsrecht: 119 - 130
Bernhardt W (1994) Keine Aufsicht und schlechter Rat? Zum Meinungsstreit um
deutsche Aufsichtsräte. Zffi 64: 1341 - 1350
Bierich M, Hornmethoff P, Kropff B (eds) (1993) Unternehmen und Unterneh-
mensfiihrung im Recht. Berlin New York
Blanke G (1994) Private Aktiengesellschaft und Deregulierung des Aktienrechts.
Betriebs-Berater: 1505-1512
Cassidy J (2002) The Greed Cycle. The New Yorker September 23, 2002: 64 - 77
Dilger E (1994) Die kleine AG und die Neuregelung zum Bezugsrechtsausschluss.
Die Bank: 610- 615
16

Directorate for Financial, Fiscal and Enterprise Affairs: OECD Principles of Cor-
porate Governance
Gates HL Jr. (1998) The End ofLoyalty. The New Yorker March 9, 1998: 34-44
Goergen A (2002) Der große Bluff. Der Verhaltenskodex für Unternehmens-
kontrolle stößt auf Kritik bei Anlegern und Investoren. Wirtschaftswoche,
January 31,2002:66-70
Government Commission "German Corporate Governance Code": German Cor-
porate Governance Code
Grandori A (1997) Governance Structures, Coordination Mechanisms and Cogni-
tive Models. Journal ofManagement and Governance Band 1: 29- 47
Gutenberg E (1962) Unternehmensfiihrung. Organisation und Entscheidungen.
Wiesbaden
Hahn J (1994) "Kleine AG", eine rechtspolitische Idee zum unternehmerischen
Erfolg. Der Betrieb: 1659-1665
Hanusch A, Albach H (2003) Preisbildung für junge Aktien bei einem IPO. In:
Albach H, Pinkwart A. (eds) Unternehmensgründung und Unternehmertum.
Zffi-Supplement 2
Heilbroner R (1972) In the Name ofProfit. New York
Heinrich RP (200 1) Corporate Governance - Die Bedeutung komplementärer E-
lemente für konsistente Reformen. Die Weltwirtschaft 4, Tübingen
Heraeus J (2002) Braucht der ehrbare Kaufinann die Corporate Govemance? FAZ
June 11, 2002: 19
Herr J (2003) "Ich rechne damit, als aktionärsfeindlich kritisiert zu werden". Die
Vor- und Nachteile des Corporate Governance-Kodex. Mit dem Vorsitzenden
des Aufsichtsrates von Siemens (Karl-Hermann Baumann) sprach Joachim
Herr. FAZ August 3, 2002: 12
Hirschman AO (1970) Exit, Voice, and Loyalty. Cambridge, pp 1-43
Hoffmann-Becking M (1995) Gesetz zur "kleinen AG"- unwesentliche Randkor-
rekturen oder grundlegende Reform? Zeitschrift für Wirtschaftsrecht 1: 1 - 10
HommelhoffP (1994) Jetzt die "Kleine" und dann noch eine ,,Anleger AG". Zeit-
schrift für Unternehmens- und Gesellschaftsrecht, Special Issue 12: 65- 84
Hopt KJ (2002) Unternehmensfiihrung, Unternehmenskontrolle, Modernisierung
des Aktienrechts - Zum Bericht der Regierungskommission Corporate Gover-
nance. Beihefte der Zeitschrift für das gesamte Handelsrecht und Wirt-
schaftsrecht 71: 27-67
Hopt KJ (2002) Corporate Governance in Europa: Neue Regelungsaufgaben und
Soft Law. Der Gesellschafter GesRZ 2002: 4- 13
Hopt KJ, Wymeersch E (eds) (1997) Comparative Corporate Governance. Essays
and Materials. Berlin New Y ork
Institute of Directors in Southern Africa, The King Report on Corporate Govern-
ance, November 24, 1994 (King I Report) and March 26, 2002 (King n Re-
port) www.iodsa.co.za/corpgov.html
Jensen MC, Meckling WH (1976) Theory of the Firm: Managerial Behavior,
Agency Costs and Ownership Structure. Journal ofFinancial Economics: 305
-360
 17

Jong HW de (1997) The Governance Structure and Performance of Large Euro-

pean Corporations. Journal ofManagement and Governance Band 1: 5-27
Kalifass H (1992) Managementkontrolle in Publikumsaktiengesellschaften. Ham-
burger Jahrbuch für Wirtschafts- und Gesellschaftspolitik 37: 277- 290
Kobayashi K Company Mission and Philosophy- Corporate Culture in Japan and
the West. Journal of Japanese Trade & Industry, vol. 21, No. 6: 8 - 11
Kräk:el MJ (1999) Organisation und Management, Tübingen
LaPorta R. Lopez-de-Silanes L, Shleifer A, Vishny RW (2001) Law and Finance.
In: Schwalbach J (ed) Corporate Governance. Berlin Heidelberg New York
Lück W (2002) Corporate Governance und Aufsichtsrat. FAZ October, 21 2002:
24
Lutter M (2002) Der deutsche Corporate-Governance-Kodex. Der Gesellschafter
Special Issue: 19- 24
Lutter M (2002) Der Bericht der Regierungskommission "Corporate Governance".
In: VGR (ed) Gesellschaftsrecht in der Diskussion 2001. VGR. Vol. 5: 47 -
59
Lutter M (1994) Das neue "Gesetz für kleine Aktiengesellschaften und zur Dere-
gulierung des Aktienrechts". Die Aktiengesellschaft: 429- 438
March JG, Sirnon HA (1958) Organizations. New York
Mayer C (1998) Financial Systemsand Corporate Governance. A Review ofthe
International Evidence. Journal of Institutional and Theoretical Economics
154: 144- 165
Merton RK (1940) Bureaucratic structure and personality. Social Forces: 560 -
568
OECD (1998) Corporate Governance. Ein Bericht der Beratergruppe für die Wirt-
schaft in Corporate Governance-Fragen an die OECD. Paris
Otto S (1997) Corporate Control-Mechanismen und Stakeholder Relations in Ja-
pan im Wandel. ZtB-Supplement 4:43-74
Pratt JW, Zeckhauser RJ (eds) (1985) Principals and Agents: The Structure of
Business. Boston
Rechenberg W-G Freiherr von: Die Hauptversammlung als oberstes Organ der
Aktiengesellschaft. Abhandlungen zum Arbeits- und Wirtschaftsrecht 48.
Heidelberg
Regierungskommission Deutscher Corporate Governance Kodex (2002) Deut-
scher Corporate Governance Kodex. Düsseldorf, February 26, 2002
Rudolf M, Witt P (2002) Bewertung von Wachstumsunternehmen. Traditionelle
und innovative Methoden im Vergleich. Wiesbaden
Sakamoto T, Nobuo S (eds) (1998) Kigyou shuudan shihai to kouporeito gabanan-
su. Tokyo
Schilling F (2002) Corporate Governance- Form oder Inhalt? F AZ May 27, 2002:
25
Schwalbach J (ed) (200 I) Corporate Governance. In: Essays in Honor of Horst
Albach. Berlin Heidelberg New Y ork
Seiht CH (2002) Deutscher Corporate Governance Kodex und Entsprechens-
Erklärung (§ 161 AktG-E). Die Aktien-Gesellschaft 47: 249- 259
18

Shleifer A, Vishny R (1997) A Survey on Corporate Governance. The Journal of

Finance: 737- 783
Teramoto Y (ed) (1997) Nihon Kigyou no kouporeito gabanansu. Tokyo
Veiel A (2002) BlackBox BRD. Alfred Herrhausen, die Deutsche Bank, die RAF
und Wolfgang Grams. Stuttgart München
Volk G (2002) Entwurf des deutschen Corporate Governance-Kodex. NWB Janu-
ary 21, 2002: 199- 201 and NWB March 18, 2002: 853- 855
Weber M (1956) Wirtschaft und Gesellschaft, Grundriss der verstehenden Sozio-
logie. 4th ed, Tübingen
Weidenbaum M (2002) Ein Weckruffür die Corporate Governance. Die amerika-
nische Unternehmenswelt vor neuen Herausforderungen. NZZ, July 27/28,
2002: 14
Wels T, Scholz F (2002) Corporate Governance wird zum Schlüsselkriterium für
Investoren. FAZ, July 8, 2002: 18
Wetzling F (1978) Der Konzernbetriebsrat Köln
Witt P (2003) Corporate Governance-Systeme im Wettbewerb. Wiesbaden
Witt P (2000) Zur Bewertung schnell wachsender Gründungsunternehmen. In:
Klandt H, Nathusius K, Szyperski N, Heil H. (eds) G-Forum 1999. Lohmar
Köln 2000, pp 173- 191
www .nzz.ch/dossiers/2002/enron
A New Management under IT Revolution in Japan

Koji Okubayashi

1 lntroduction

The effects of Information Technology on social as weil as economic life

are sometimes called the "Third Industrial Revolution". In the business
world, it changed significantly the way of management, strategies of com-
panies, production system, human resource management, corporate
govemance, accounting system, distribution channel, marketing system
and so on. These trends are often reported in business joumals and
academic papers.
This paper intends to access the following research questions based on
the experience of Japaneselarge companies since 1990.
I. How much is the investments for IT in private sector among private
facility investment since 1990 in Japan?
2. What are the directions of such changes of Japanese management
under the effects of IT revolution, especially among production
system, human resource management and corporate govemance?
3. What kind ofnew concepts can we deduct from these experiences for
theories of new management?

2 Japanese trends of IT investment since 1990

Japanese industries made a Iot of investment for IT use especially since

1990 when the "bubble" economy finished its prosperity and Japanese
economy went down into deflation. Investment in IT as a whole has to
include investment in IT facilities, development of software and systems,
training and education of the usage of IT facilities and many intangible
activities related to IT. However, herewewill refer only to the hardware
aspect of IT investment.
According to the Japanese white paper on communication published in
2000, real amount ofiT investment in private sectorwas 6,101 billion yen
in 1990, but it increased to 10,387 billion yen in 1998. The ratio of IT
investment to GDP was 1.4% in 1990 but 2.2% in 1998. The ratio ofiT
investment to facility investment in private sector was 7.3 % in 1990 but
20

12.4 % in 1998. These figures indicate a large amount of IT investment

since 1990 and make us to expect drastic changes of industrial as weil as
social life due to the diffusion of IT. IT, in this case, means personal
computers in office and at home, many types of machines supported by
computers, intemet and many systems using the intemet and any kind of
systems connected by computers. Office automation and many systems
using IT in companies brought drastic transformation in production
systems and management style

(%)

-
14 14
.....,.. Ratio ot IT lnvestmont to
12 GDP

/
12

10
.... 10
Real civll inveotment into
IT

...... - //
".,.

/
-- ......
-+--Ratio of IT inveabnent to
8 8 pr,vate sec:tor facillty
/ jnva atment

...
~

6 6

4 4

-
2 2

0 0 (trillion yen)
~-#'0 ~-#'~ ~-#>"' ~-#'~ ~"'· ~"'~ ~<I' ,<JO'

Japan White Paper on Communication 2000, p . 1 08

Fig. 1. Trends ofiT Investment since 1990 in Japan

3 Changes of production system

3.1 Multi-product small-volume production system

The introduction of computers into production system brought a funda-

mental shift of production principle from single-product mass-production
to multi-product small-volume production system. In electronics industry,
for example, new products are quickly developed and sold within a short
period of time. The first comer of electronics products can get major parts
of the pro fit in that market. They have to produce a new product one after
another and continue that process to get a lion's share ofthat market.
Many other industries like automobile and chemieals also follow the new
 21

principle of production system. Even service industries have to apply the

new production principle to respond quickly to the changing demands of
their customers.

3.2 Shift from manufacturing to solution business

Many manufacturers in Japan started to shift their main activities from

manufacturing to research and development as well as marketing to get
higher share of added value. They thought that they can not obtain high
added value by manufacturing standardized products, which would be
transferred to Asian countries with low labor cost. They will concentrate
their energy on higher- added -value stage of production process such as
R&D and marketing. Sony and Matsushita, for example, invested their
efforts to develop a new business system or new field of business by intro-
ducing Electronics Manufacturing Service, which is often categorized into
solution business.

organization of headquarter

~.!I h ~~ i s·Ii~.
! lj ii
Q"ca
~ ~. [·
Dl

r
I 1 1 ,---L
Engineering, [ • AV(Wl~. aoo"•'~l.
Manufacturing IT(information technology)
• device
~~
and
Customer Services • semiconductor
contracting
'----
corporation

Fig. 2. Sony's R&D and Manufacturing System in 1999

3.3 Loosely-structured organization

The new production system will result in the transformation of traditional

structure of organization. The traditional structure of organization with
mechanistic work organization and hierarchical management organization
22

is named here "tightly-structured organization". On the contrary, a new

structure of organization with organic work organization and flat manage-
ment organization is named here "loosely-structured organization".
The tightly-structured organization fits to single-product mass-produc-
tion system to accomplish high performance of organization. This type of
organization is usually seen at the long belt conveyer line of automobile
assembly shop. On the other side, loosely-structured organization is quite
effective for organizational performance of multi-product small-volume
production system. That type of organization is often identified at the cell
production system of electronics appliance. Many advanced Japanese
manufacturing companies are shifting their organizational structure to
loosely-structured organization to adapt to turbulent management en-
vironment.

Management organization

Hierarchical Hierarchical
Mechanistic Organic
( Tightly-structured)
Organization

Work organization --------+----------

Flat Flat
Mechanistic Organic
( Loosely-structured1
Organization )

Fig. 3. Four Types ofürganizational structure

3.4 E-net purchasing of product parts

Keiretsu-shitauke was one of main mechanisms to connect buyer-supplier

relations especially in automobile and electronics industry. A supplier in
some shitauke keeps technical, management, financial and personnel
relations with its assembler to maintain mutual-aiding corporate group.
However, these assemblers nowadays try to enlarge the group of suppliers
by using e-net to discount the price of these parts and stimulate
 23

competition among suppliers for improving the quality and design of their
parts. Through the e-net, the assemblers can easily find out suppliers even
abroad who can offer high-quality parts with low price. Therefore, e-net
transforms drastically the traditional buyer-supplier relations in many
industries.

Assembler A I (Buyer)

1Choice of best company

Internet

Application by companies

,_____ ~ Q ~ ~<i ~Q~ ~ Q ~ Q ~ ~Q~ ~b~ ~Q ~ ____ J (Supplier)

Ken Nakayama, "Network Strategy of Medium and Small-sized Company",
Douyuukan, 2001, p.142.

Fig. 4. E-net Purchasing System

4 New trends of human resource management

4.1 Education and training based on employability

The life-time employment (syuusinkoyou) provided all training and edu-

cation of employees necessary for carrying out their jobs at their company.
However, after the economic depression in 1990' even Japanese big com-
panies could not keep the life-time employment for all of their employees.
They insist on supporting the employability of their employees by pro-
viding many types of training and education only for employees who will
be eager to improve their skills and knowledge for their carriers. Em-
ployees themselves sometimes have to pay some money to artend training
course prepared by their company. It indicates an establishment of new
company-employees relations different from the traditional life-time em-
24

ployment. Workers themselves can not trust in the life-time employment

under recent economic conditions.

4.2 Management by objective

Japanese companies introduced the practice ofmanagement by objective in

1990' so that they can simulate the performance of their employees for
their survival under severe market condition. They evaluated the contribu-
tion of the employees to their company by the ability accumulated by their
experience at the company, which is usually called ability-based wage
system. However, the ability itself would be useless if it can not result in
concrete performance of their company. They could not pay only for the
ability of their employees to keep their employment to the company. To
measure the concrete contribution of each employee to the total per-
formance ofthe company, they introduced MBO. According to the degree
of accomplishment of the objective, they could pay large or small amount
of rewards to each employee, quite independent from their age and ability.
This is a new principle of payment different from seniority wage system.

4.3 Annual salary for managers

Another payment system introduced in 1990' was annual salary for

managers. Japanese companies have to revise seniority wage system es-
pecially of managers who shared more than ten percent of all of the em-
ployees. In the rapid economic growth period of 1950'and 1960', the
percentage of managers among employees was 3-4 percent. But due to
aging of employees and shortage of management positions under low eco-
nomic growth, Japanese companies prepared unnecessary management
positions to keep the traditional incentives for all employees. This Iabor
cost of managers imposed heavy burden on the companies. Therefore, they
introduced at the first stage the MBO for managers and then introduced
annual salary based on MBO. The annual salary is determined not by the
ability of managers, but by the degree of the accomplishment of their
target. The amount of annual salary will vary depending on the result of
MBO. This flexible salary results in stimulating morale of managers and
reducing the fixed labor cost of managers.
 25

5 Corporate governance

5.1 Rebinding of kigyo-shuudann

Kigyo-shuudann (Corporate assemblage) was also one ofthe main charac-

teristics of Japanese corporate system. The main six kigyo-shuudann
(Sumitomo, Mitsui, Mitsubishi, Fuyou, Sanwa, Daiiti-kangyou-ginkou)
shared major part of economic activities in the rapid economic growth
period and exerted strong influence on Japanese economy. The corporate
assemblage intended to maintain their mutual economic benefits among
their members and to compete with other assemblage. This mutual benefits
and competition among assemblage members promoted positive economic
activities of Japanese companies after the Second World War. The main
banks of these assemblages played an important role of coordinating busi-
ness activities of their member corporations and distributing financial
resources to these activities.
However, under the long depression and bankruptcy of financial organi-
zations, many of these main banks have to merge or reorganize among
themselves beyond the boundary of kigyou-shuudann. Mitsui Bank and
Sumitomo Bank merged into Mitsui-Sumitomo bank. Mizuho bank is
composed of Fuji, Daiiti-kanngyou, and Nihonkougyou Banks. UFJ Bank
is a result of merger between Sanwa and Toukai Banks. These mergers of
banks will Iead to reorganization of kigyou-shuudann and industries as a
whole.
 N
0\

Table 1. The Main Corporate Assernblages 1989.10.1 (1)

Mitsui kei Mitsubishi kei Sumitomo kei !
I
Nikikai 24 Kinyoukai 29 Hakusuikai 20
ln Foundation 1961.10 Foundation 1955 Foundation 1951.4 I
~
Bank • lnsurance Mitsui Bank Mitsubishi Bank Sumitomo Bank
Mitsui Trust Finance Mitsubishi Trust Finance Sumitomo Trust Finance
*Mitsui Life lnsurance *Meiji Life lnsurance *Sumitomo Life lnsurance
Taisho Marine & Fire lnsurance Tokyo Marine & Fire Sumitomo Marine & Fire
lnsurance lnsurance

Trading company Mitsui & Co. Mitsubishi Sumitomo

Agriculture • Forestry • Mitsui Mining Sumitomo Forestry

Mining *Hokkaido Tanko Kisen Sumitomo Goal Mining

Construction Mitsui Construction Sanki *Mitsubishi Construction Sumitomo Construction

Foods Nippon Flour Mills Kirin Brewery

Textiles Toray lndustries Mitsubishi Rayon

Pulp and paper Oji Paper Mills Mitsubishi Paper Mills

Chemistry Mitsui Toatsu Chemieals Mitsubishi Kasei Sumitomo Chemical

Mitsui Petrochemical lndustry Mitsubishi Gas Chemical Sumitomo Bakelite
Mitsubishi Plastics lndustry
*Mitsubishi Monsant
Chemical

Oil Mitsubishi Oil

Rubber
Table 2. The Main Corporate Assernblages (2)
Huyou kel Sanwakei lkkan kei
~e Huyoukal29 Sannsuikal Sankinkai I
ln Foundataon 1966.1 Foundation 1967.2 Foundation 1978.1 I

Bank · lnsurance Fuji Bank Sanwa Bank Dai-ichi Kangyo Bank I

Yasuda Trust & Banking Toyo Trust & Banking •Asahl Ufe lnsurance
*Yasuda Ufe lnsurance •Nippon Ufe I nsurance •Fukoku Ufe lnsurance
Yasuda Fire & Marine lnsurance Nissan Fire & Marine Insuranes
Talsei Fire & Marine lnsurance
Tradong company Niehirnen ltochu
I
*Nissho lwai Kanematsu Kosho
I
lwatani International *Nissho lwal
I
Kawasho I
Agriculture • Forestry • Mining I

Constructlon Obayashi Shimizu I

Zanitaka
Toyo Construct:ion
Sekisui Hause
Foods ltoham Foods
•suntory
Textiles Unitika Asahi Chemicel lndustry
Teijm
Pulp and papar Housyu Paper
Chemistry Tokuyama Soda Denki Kagaku Kogyo
Sakisul Chemlcai Kyowa Hakko kogyo
Ube lndustries Nippon Zaon
Hitschi Chemical Asahi Danka Kogyo
Tanabe Pharmaceutical Sankyo
Kansai Paint Shiseldo
Lion
Oll COSMOOIL Showa Shell Sakiyu
Rubber Toyo Tlre & Rubber Yokohama Rubber
Glass • Cement -·-··-···-
Nihon Cement Osaka cement Tltibu cemant --

N
-..l
28

5.2 Decrease of mutual stock holding

Companies in a kigyoushuudann have stock each other to keep financial

relations and group identity. However, under the low stock price in the
depression and a new practice of accounting based on present value of
stock, many companies even among kigyou-shuudann started to reduce
their mutual stock holding. The ratio of mutually-hold stock among total
stock in the stock market in non-financial companies decreased from
around ten percent in 1987 to four percent in 1999. This reduction of
mutual stock holding means to the companies within kigyou-shuudann a
decrease oftheir stable stock owners. Topmanagement ofthese companies
has to take care of the interests of stock holders and to be keen about their
stock price to obtain financial resource from stock market. The executives
of these companies were moved into unstable position compared with the
former one.
(%)
2S r-----------------------------------------~

20

-·-
15

-~ .. -·- ... .
... ····~······· ..•·······•
10

5
•·······• . .
· ······• ······ • .......~ .....: ...-···: ··~:· :~· ·~~·- ......
1987 1991 1995 1999 (year)

- - Ratio ofmutually-hold stock among market stock

•·- Ratio ofrnutually-hold stot:k by non·financial companies
.. .. .. Ratio ofmutually-ho ld stock by finaneiul c::ompanies

Mitsuak.i Okabe, "Mutual Stock Holding and Japanese Economic System",

Keiougijukudaigak.u-shuppankai, 2002, p.34 .

Fig. 5. Trends ofMutual Stock Holding since 1987

5.3 lntroduction of managing directors and "company" system

Some of Japanese leading companies like NEC and Sony introduced a new
practice of managing directors and "company" system which means a
company within company. Active arguments about corporate govemance
were made by managers and academics on the occasion of corporate
 29

scandals and bankruptcy of banks and security companies. Some compa-

nies shifted their target from stable stock-holders to active stock-holders to
keep friendly investor relationship.
At the same time, restructuring of top management organization came
from the demand of speedy decision making at the board of directors and
clarification of responsibility of each executive. The "company" system
was an effective replyto these requests to top management organization.

6 Concluding remarks

After accessing new trends of management practices of Japanese large

companies under IT revolution, what kind of new concepts in management
can we distill for the development ofthe theory ofthe firm?
First, economy of speed is a very important concert to explain the new
behavior of Japanese firms. The economy of speed is a source of profit
under turbulent market. It is also a driving force to reform organizational
structure of hierarchical management. The shift from manufacturing
process to solution business intends to speed up the reproduction process
of new products to get the first-comer benefits. The economy of speed is as
an important concept as economy of scale under IT revolution.
Second, loosely-structured organization will be a key concept to explain
a new type of organization beside hierarchical organization. A network
organization is often used to explain a new type of organization. The
concept of network organization would be quite useful to explain the rela-
tions among organizations, but the loosely-structured organization is quite
useful to analyze manufacturing organization which demands rigid line of
communication.
Third, flexible Iabor cost is also a key concept to rationalize new prac-
tices of human resource management. The traditional seniority wage
system tends to fix or increase Iabor cost for companies. MBO and annual
salary for managers intend to change Iabor cost according to final result of
the company or concrete performance of employees. Labor cost of a com-
pany also should be quite flexible depending on the business activities.
This is a new principle different from seniority wage system.
These new concepts need to be elaborated further to be applied to
another countfies beside Japan. They should be accessed from other
theories of firm to identify the usefulness to explain new behavior of firms.
We need international collaboration in these research activities.
30

References

Okubayashi K (1995) Japanese Effects ofNew Technology on Organization and

work. Zffi, Ergänzungsheft 4
Okubayashi K (2000) Japanese Style ofTeam Working. Zffi, Ergänzungsheft 1
Okubayashi K (2001) Japanese Manufacturers without Factries. Vezestudomany,
Evf
 Section 2

Marketing
Relative Advantages of E-Business Start-Ups
versus lntegrated Units of Bricks-and-Mortar
Companies

Sönke Albers

1 lntroduction

Only two years after the e-commerce hype in March 2000 some firms have
lost more than 90 percent of their value on the stock exchange while many
others have become insolvent and have been Iiquidated (Mahajan et al.
2000). However, this does not mean a failure of e-business. Rather, the
Internet is still growing. The number of consumers buying in the Internet is
increasing. Many more companies work hard to enable electronic trans-
actions via the Internet. In fact, we observe a number of companies that al-
ready achieve a profit with their electronic business operations (Albers et
al. 2002). There is no doubt, electronic business (e-business) cannot be ne-
glected by companies. Given the enormaus di:fficulties to turn an e-busi-
ness into a profitable one and the necessity to adapt one's business to the
Internet it is of utmost importance to know more about the success factors
of e-business operations.
There is a Iot of folklore around that gives prescriptions how to achieve
success on the Internet. Much of this was written before the hype. A good
example is the book by Hageland Armstrang (1997) on virtual communi-
ties. Although they named their book "Net gain" and described in detail
how one can be successful with community building there are only very
few communities profitable five years later. With these Contradietory Ob-
servations in mind we, therefore, do not want to rely on prescriptions but
rather investigate empirically which factors may have led to better results
than others.
This can be done in different ways. One way is to investigate successful
companies in depth and generalize what they have in common. Such a case
study design has been carried out by Albers et al. (2002). They found that
profitable companies have used the potential of this new interactive me-
dium from the very beginning by directing their business to completely
digitalized products and affering added-value which consumers are willing
to pay for. They have also leveraged network effects to enhance word-of-
mouth in order to avoid cash buming marketing efforts. All their actions
34

were directed to the goal of achieving immediate profitability. Another

way is to ask a larger sample of companies in the same or across different
industries about their way of doing business. We follow this latter ap-
proach by conducting a survey among relevant companies because we find
only a limited amount of empirical results (exceptions are Chen and Hort
2002; Geyskens et al. 2002, Coltman et al. 2002).
Of course, there exist numerous factors that may affect success. It is
generally accepted that success depends on a sustainable competitive ad-
vantage. This advantage can be achieved either by a superior marketing
strategy and/or by the resources the company can leverage (Wernerfelt
1984). One finding ofthe so called resource-based view isthat it is not so
much the real resource base but rather the intangible resources which de-
cide on success. As a result, our questionnarre focuses on the leverage of
resources provided by the founders (network, experience, knowledge ), the
technological solution for the Internet, and the motivation of the employ-
ees. Competitive advantage may be achieved by marketing strategies that
are not easy to imitate and thus create barriers to entry especially if the
company is a pioneer. We focus on some aspects ofthe marketing strategy
by investigating the concept, the communication instruments for marketing
the product, and the revenue structure.
Before the hype it seemed as if only start-ups would be able to build up
an entire new business on the Internet. Established companies were as-
sumed to be too inflexible and too slow to take advantage of the various
business opportunities on the Internet. With the bankruptcy of many
dot.coms and more and more examples of established firms being success-
ful in the Internet the evaluation has changed. Since there are still many
companies that have to choose the right implementation of their e-business
operations we want to investigate in particular whether it is better for e-
business companies to begin as a start-up (with the opportunity tobe inte-
grated later) or as a more or less controlled unit of a bricks-and-mortar
company.

2 lnvestigation of success relationships

With the advent of the Internet numerous publications appeared on how to

achieve success in the Internet. Many of these studies provided simple pre-
scriptions like golden rules that tumed out to be very questionable (Rangan
and Adner 2001). However, one could also find serious books with success
factors derived from plausible reasoning about the peculiarities of the
 35

Internet. This was acceptable in the early days when nobody knew what
would really turn out to be successful in the very end.
The first stream of research investigated which products might be the
best candidates for selling via the Internet. Schwartz (1997) proposed to
concentrate on products that are information-rieb and hence can be digi-
tized to a maximum extent. He also pointed out that self-services might be
good candidates because they provide the highest Ievel of customer com-
fort. Since then many researchers have modified and enriched these propo-
sitions. Albers (2000) finally concludes from an investigation of the top
ten of products sold via the Internet that it is not so much the type of the
product but rather the value created for the customer. Moreover, the oppor-
tunities to create value are not restricted to certain product types but only
to the creativity ofthe entrepreneurs (Amit and Zott 2001).
The second stream of research discussed revenue models. In a report for
the European Community Council Zerdick et al. (2000) classify various
types of revenue streams on the Internet. Besides traditional forms of earn-
ing money from transactions, like margins from sold products or commis-
sions for having enabled transactions, they point out opportunities to profit
from one-time access fees or continuous subscription fees. In addition,
there are several other indirect forms such as revenue from online-
advertising, sponsorships and data mining. Hanson (2000) is another good
example of elaborating revenue models. Today, it is clear that companies
having relied on indirect revenue streams like online-advertising have suf-
fered the most. Therefore, we currently observe a hectic search for other
revenue sources (Albers et al. 2002).
It is by no way surprising that neither the type of product nor the reve-
nue model does explain actual performance differences of e-business Op-
erations. Consequently, a third stream of research focused on the specific
characteristics of the information economy that drives the Internet. Shapiro
and Varian (1999) describe information as an experience good that has to
be supported by specific means. They stress that information goods rely on
networks which Iead to demand side economies of scale and positive feed-
back. Those markets are characterized by lock-in effects and switching
costs. Therefore, such standards are important that change competition for
a market to competition within a market. Hanson (2000) and Rayport and
Jaworski (2001) devote most oftheir books to the specific conditions un-
der which a company should operate in the Internet. However, a general
understanding of these factors does not guarantee their proper implementa-
tion. Rather, Rangan and Adner (2001) show seven misconceptions about
achieving profits in the Internet.
We want to investigate the difference in performance of start-ups and
integrated units of bricks-and-mortar companies. We are not so much in-
36

terested in the success of certain business models but rather in their dif-
ferent strategic approaches. This requires searching for success factors be-
yond the specific business model, to assess their relative importance in or-
der to evaluate what kind of organization has done better. As a
consequence, we focus on success factors that are not necessarily unique
for e-business operations but generally valid for all types of traditional
businesses. Literature has provided numerous proposals from which we
can draw the appropriate ones for our study.
Nowadays, it is well accepted that the success of a company depends on
its strategic orientation (Porter 1985) and on the resources devoted to the
market as it has been proposed by the resource-based view school
(Wernerfelt 1984). While the resource-based view is rather general at first
glance, it nevertheless provides valuable insights because the proponents
have pointed out that not so much the real resources count but the intan-
gible ones that give a company a sustainable competitive advantage. Intan-
gible resources can be found in particular in the knowledge-base of the
company or its employees (Grant 1996). Therefore, this study investigates
knowledge represented by the founders, the employees and the web-tech-
nology.
The question of how to set up a new venture requires an analysis of in-
stitutional aspects. There is a stream of papers dealing with the question
whether venture capitalists are helpful (Hinkel 2001, Shane and Cable
2002). We want to concentrate on the issue whether start-ups are better
suited for setting up an Internet-based business than new units within es-
tablished bricks-and-mortar companies (Gulati and Garino 2000, Coltman
et al. 2001 ). This implies the question how an established company should
proceed: either allocate its resources to financially support start-ups and to
integrate them later or start with an internal solution. The proponents for
start-ups have argued that such a new business needs flexible employees
that would feel lost in established structures. On the contrary, the world
largest mail-order company Otto claims to be the most successful e-
commerce company that has integrated its e-business because it is viewed
only as another sales channel which should be tightly controlled by the
company. Since the decision on these more institutional questions influ-
ences the flexibility and the motivation of the employees and managers
this institutional fit is also considered as a success factor. It allows us to
give the important recommendation which institutional model suits best.
In order to investigate the success factors we have to relate input vari-
ables to success. During the last years, the performance of e-business com-
panies has been assessed very differently. Before the hype, the market
capitalization depended mostly on page impressions, number of sub-
scribers etc. Now, profitorpositive cash-flows are as important as they are
 37

for any other company. However, since only a few companies were profit-
able at the time of our investigation (midth of 2001) we replace the objec-
tive performance measures such as cash-flow (in Euro) or return on profit-
ability (in %) by the satisfaction of the companies with their profitability.
Others use the stock market return as a success measure (Geyskens et al.
2002).
Many success factor studies are mainly focused on testing the signifi-
cance of certain success factors. This very often Ieads to the confirmation
of factors that any manager would take for granted. Therefore, we neither
formulate hypotheses nor do we test them. Rather, we are conducting an
explorative study in which we estimate the relative importance of the vari-
ous success factors and how that does explain differences between start-
ups and integrated unit solutions.

3 Operationalization

In order to test empirically the relative importance of the success factors

and the superiority of either a start-up or an integrated e-business unit of a
bricks-and-mortar company, an operationalization of the various success
factors developed in chapter 2 is necessary.
Fora better understanding ofthe process leading to success it seems ap-
propriate to summarize certain management decisions under abstract con-
structs. In our model we define, for example, the marketing concept as a
construct that consists of elements such as branding, assortment and price.
All these elements or variables are success drivers that determine the Ievel
of the marketing concept instead of just being measures for the overall
construct. Thus, these variables are formative indicators and not reflective
ones. This implies that we cannot work with latent variables as constructs
as has been popular for instance in the market orientation Iiterature (Jawor-
ski and Kohli 1993). Rather, constructs must be understood as indices of
more detailed policies. Insofar, it is not necessary to Iook for multi-item
representations of the constructs and to test for construct validity and reli-
ability (a detailed discussion is given in Diamantopoulos and Winklhofer
2001). Rather, we focus on composing a comprehensive Iist of potential
success drivers to explain success as good as possible.
Every success study has to incorporate strategy elements that describe
the approach a company uses to serve its market. While there is a multi-
tude of variables by which the marketing strategy can be operationalized
we want to concentrate on variables whose values may vary depending on
observing a start-up or an integrated unit. We expect these differences not
38

so much in the design of the strategy (which can be implemented by both

types of companies) but in the use ofthe marketing instruments. In particu-
lar, we want to investigate the influence ofthe constructs:
• what kind of marketing concept has been applied,
• with which communication instruments have the customers been ad-
dressed,
• what kind of revenue structure has been designed and
• whether it was possible to achieve a sustainable competitive advantage
thereby.
A detailed list of variables comprising the four constructs is given in ta-
ble 1. As a scale we have chosen for these and the following questions (if
not noted otherwise) a seven-point Likert-scale which asks for the agree-
ment with a certain statement.

Table 1. Indicators of Constructs representing Marketing Strategy

Constructs Indicators
Marketing Concept We sell brandsvia the Internet
(7 -point Likert scale) Wehave built up a new brand for our Internetbusiness
We address a clearly defined target group
We acquire our customers through a wide assortment
We acquire our customers through an attractive price
Communication Instruments Print
(yes/no) TV
Radio
Online Advertising (Banners)
One-to-One Marketing
Publicrelations
Revenue Sources (yes/no) Commissions on Transactions
Online-Advertising
Fixed Access and Usage Fees
Margins on Sold Products
Non-imitability Our business model can hardly be imitated because we have re-
(7 -point Likert scale) alized a time-advantage over our competitors
Our business model can hardly be imitated because we have
built up a strong brand

We have included the communication instruments because we are all

aware of the case of Boo.com that dumped tens of millions of dollars into a
TV-advertising launch campaign - without actually having something to
sell (Malmsten et al. 2001 ). In general, the assumption is that mass media
like TV, radio and print are not cost-effective for e-businesses while
online-advertising, one-to-one-communication, and public relations are
better suited (Albers et al. 2002).
Many e-business companies failed because they did not have sufficient
revenue sources. In particular, companies focusing on revenue streams
 39

from advertising had a hard time to survive because it became more and
more difficult to achieve online-advertising revenues unless the company
belonged to the few high-traffic portals (http://www.acnielsen.de/news/-
2002/10_15.htm). It turned also outtobethat profitable companies gener-
ate revenue from more than just one revenue source (Albers et al. 2002).
Following Zerdicket al. (2000), we, therefore, included variables describ-
ing the revenue sources of the companies.
An investigation of German e-commerce sites shows that many business
models are imitated from the US or successful examples in the domestic
market (e.g. Alando and Buch.de imitating Ebay and Amazon). Therefore,
firms faced a stiff competition already after a short time of introducing
their offering. Insofar, the success of a marketing strategy heavily depends
on the imitability of the marketing concept (Day and Wensley 1988), so
that we included variables operationalizing non-imitability (Figueiredo
2000).
Marketing strategies can only be realized successfully if the company
employs the right resources for the process of offering products and ser-
vice in the Internet. From a resource-based view it is not so much the real
resources but the intangible ones that count (Srivastava, Shervani, and
Fahey 1998). In our context of start-up companies or new e-business units
one of the most important resources are the founders or founding manag-
ers. There is a body of Iiterature stressing the importance of the networks
that the founders can utilize (Aldrich and Zimmer 1986, Brüder! et al.
1992, Hinkel 2001). In addition, we know that many new companies fail
because they lack the necessary expertise and experience regarding man-
agerneut skills.
Founders and key managers can only be successful if they work with the
right team that is motivated to perform as good as possible. Therefore we
treat the human resource as one of the determining intangible resources
(Bamey 1991). Since the effectiveness ofthe human resources depend on
many factors which cannot be investigated in a single questionnaire we
concentrate on the aspects whether
• the job is attractive enough to attract and keep competent people,
• financial incentives motivate the employees to work as hard and smart
as possible,
• the company applies a corporate culture that is appropriate for such kind
ofbusiness, and
• the company relies on efficient Internet IT-system solutions developed
on their own rather than made available by others.
40

A detailed list of indicators by which the resource constructs have been

operationalized is given in table 2.

Table 2. Indicators ofConstructs representing the companies' resources

Constructs lndicators
Network ofFounders The network of the founders or founding managers
(7 -point Likert scale) ... enabled a comprehensive technology and know-how transfer
... allowed the collection of detailed data on the market & competitors
... provided access to important suppliers & potential customers
... enabled the built-up of acceptance and positive references for the
own offering
Prior Experience The founders or founding managers were able to make valuable experi-
(7-point Likert scale) ences prior to their current position
.. .in a consulting company
... as an entrepreneur
... in an industry close to the current one
... in marketing or sales
... in the IT-technology area
.. .in the areas offmance/accounting/administration.
Financial lncentives We offer higher salaries than on average
(7-point Likert scale) We offer generous stock options or shares
Job Attractiveness We offer very interestingjob tasks
(7-point Likert scale) We offer secure jobs
We offer attractive career opportunities
Corporate Culture We come up with decisions in a hierarchical way
(7-point Likert scale) All the employees can actively participate and bring in their own ideas
We support continuing education of our employees
We tolerate failures in the case ofrisky decisions (risk tolerance)
Our e-commerce unit acts autonomously from other company units
We run a complaint-management system
IT-Solution Our Internet IT-solution has been developed by the company itself
(7 -point Likert scale) Our Internet IT -solution represents a commercially available solution
that has been configured for our requirements
Our Internet IT-solution is solely based on open-source software or
freeware
Our Internet IT -solution has largely been influenced by one of our main
customers
Our Internet li-solution has been tested intensively up-front by our
customers

To singleout the success factors we have to relate the marketing strate-

gies and the intangible resources as potential success drivers to per-
formance. There is a debate whether it is better to base such an investiga-
tion on objective or subjective measures (ELMAR 2002). Objective
measures are, for example, profit or cash-flow in Euro or, for comparison
purposes, return on investment (ROI). Unfortunately, the companies inves-
tigated in this case are in different stages of their development. Many of
them did not reach profitability even at the time of distributing the ques-
tionnaire. Moreover, the majority of companies refused to provide such
data. We, therefore had to rely on subjective measures such as the satisfac-
 41

tion with certain successes. Given the recent manipulations of sales and
profit figures by such companies as Enron and Worldcom, some authors
argue that subjective measures must not be worse than objective ones
(Atuahene-Gima in ELMAR 2002). Moreover, in a Iiterature review on
market orientation research, Dawes (1999) concludes that subjective and
objective measures are strongly correlated.

All scales are

agreement on a
7-point-Likert scale

Fig. 1. Causal Relationships and Operationalizations of Success

In our study, we elicited measures that represent different stages of the

success chain: satisfaction with the achieved Ievel of market share and
with its development over the last 12 months and the same for revenue. Of
course, both are only a prerequisite of profitability which we opera-
tionalized by cash-flow and ROI. In addition, we asked for the company's
judgment of achieved customer satisfaction and employee fluctuation rate.
Finally, we were interested in the future assessment of their business and
asked whether the company would like to continue its e-business operation
and/or increase its investments in this sector. The assumed causal relation-
ships are visualized in Figure 1.
42

Start-up

Integrated
Unit

Fig. 2. Conceptual Framework

Given these operationalizations, we can detail our model and assume the
following relationships: Market share is the heart of all activities and there-
fore influenced by the marketing concept, the communication strategy, and
the non-imitability. In addition, the founders' network and experience, job
attractiveness, financial incentives, and corporate culture influence the ef-
fort of the employees to fight for market share. Market share determines
revenue which is also directly influenced by the revenue model. Revenue,
then, determines profitability. Profitability is also influenced by non-imita-
bility because the market position determines the margins that can be real-
ized. The IT-solution and the financial incentives have cost consequences
and hence a direct influence on profitability, too. Profitability finally influ-
ences the future assessment of the operations given by continuing the Op-
erations and/or increasing investments.
If we want to investigate data from a broad cross-section of companies it
is advisable to include covariates as additional explanatory variables in or-
der to absorb heterogeneity of the sample. Regarding the assumed large
differences between B2C and B2B operations we included a respective
dummy-variable. Companies have already reached their break-even when
they started earlier so we included the time since being online. In addition,
we take into account whether the operation was financed by outside capital
which may impose different expectations on profitability. Finally, we have
 43

included a dummy-variable which distinguishes between start-ups and in-

tegrated units to capture residual variance that can not be explained by the
constructs. The relationships described so far represent a structural equa-
tion modelthat is given in Figure 2.

4 Empirical study and data

In order to test empirically the relative importance of the various success

factors and to examine whether a start-up or an integrated unit within an
established company is the better organizational solution, we have dis-
tributed a questionnaire via email (in a few cases also by fax and surface
mail) to 590 companies. We obtained the addresses ofthe companies with
the help of an online-search for phrases like e-commerce, start-up, online-
shop etc. Moreover, we systematically searched in web-newspapers and
shopping indices. The questionnaire was sent out from May until July
2001. We addressed it to members of the board, chief executive officers as
weil as managers of the e-business operations. A total of 191 companies
responded which resulted in a responserate of 32%. Unfortunately, 21 out
of 191 companies refused to fill in the necessary information on business
success. In addition, we excluded another 23 companies because they
showed more than 4 missing values in the success factors. In the other
cases missing values have been replaced by mean values of the respective
variables. As a result, we can base our analysis on 147 complete question-
natres.
The sample of responses does not claim to be representative but appears
to be typical. In order to get an idea on the sample structure we provide
some characteristics of the respondents. Figure 3 provides the industries in
which they operate. More than 50% of the respondents were members of
the board, chief executive officer or founder. Another 25% responses came
from the responsible e-business or project manager and 10% from the
marketing manager. We purposely tried to match the number of start-ups
with the number of integrated e-business units of established companies so
that we received with 75 responses from integrated units and 72 responses
from start-ups an almost equal number. The majority ofrespondents (58%)
is represented by business-to-consumer (B2C) companies which nearly
equally split into integrated units and start-ups. Another 31% respondents
are business-to-business (B2B) companies which are a little bit over-repre-
sented by start-ups while the rest (11 %) are involved in B2C and B2B op-
erations at the same time.
44

Food
I I
PC and Consumer Electron ic
Financial Services
I )

Apparel and Assecoires

Hou eand Garden
Cosmetics, Bath, and Health
Office Products
'
Books, CD, DVD, Video
Automobile

0 2 4 6 8 10 12 14 16 18

Fig. 3. Industries of responding companies

Unfortunately, the companies did not respond very weil to information

conceming size indicators like employees, revenue, subscribers, page im-
pressions. Nevertheless, for a better understanding of the sample structure,
we report median values that characterize the responding companies by 15
employees, 5,000 user, 250,000 page impressions, and sales in the interval
of 100,000 - 1,000,000 €.

Margins on Sold
Products

Commissions on
Transactions

Fixed Access or Usage

llI
I
Fees

Onl ine Advertising

____J
'
1•Start-up
0 lntegrated Uni!
No Revenue intended
:::J
0% I 0% 20% 30% 40% 50% 60% 70% 80% 90%

Fig. 4.Revenue Sources ofResponding Companies

It is also important to know from which source the companies draw their
revenue. This is depicted in Figure 4 for start-ups and integrated e-business
units separately:
 45

5 Estimation of structural equation model

When the success of managerial policies is mediated by several intermedi-

ate constructs like job attractiveness or marketing concept and intermediate
success measures like market share and revenue, then the influence of the
various policies can only be assessed with the help of a structural equation
model. However, this is based on formative rather than on reflective indi-
cators (see chapter 3).

Fig. 5. Estimation Results of Structural Equation Model

As a consequence, the estimation of the structural equation model with

the help of LISREL, which has been the favorite method for this type of
research in the past, is not appropriate because LISREL assumes reflective
measures. In addition, we do not want to test whether the model is a valid
representation of the success creation process as it would be the main pur-
pose of a structural variance-covariance model. Rather, we have taken very
plausible managerial policies whose effect is not controversial in theory
and practice. Therefore, we do not focus on testing whether the model,
which cannot be complete anyway (because no respondent would give an-
swers on any aspect of their policy), is valid and whether the variables
(policies) have a significant effect. Rather, we want to find out what the
relative importance of the various policies on success is. Therefore, we ap-
46

ply path regression for estimating the structural equation model. Thus, we
estimate regression coefficients in a least square sense rather than factor
loadings as used for variance-covariance models like LISREL. In our
model, the policy variables may be correlated because some policies only
make sense if combined in a suitable way. This is, however, no indication
that we have a reflective construct with different measures. The indicators
of our constructs do not necessarily have to be correlated but they could. In
order to avoid multicollinearity problems we apply Partial Least Squares
as a tool for path regression as implemented in the software package PLS-
Graph by Chin (1998).
We ran PLS-Graph for the structural modelas given by Figure 1 and 2.
Since the model is too complex tobe visualized in all details we restriet
our results to the values of the standardized regression coefficients of the
constructs and the explained variance (R2) in the graphic (see Figure 5).
In addition, we report the regression coefficients of the indicators on the
constructs in Table 3.

Table 3. Regression Coefficients oflndicators with Respect to Construct

Weight Mean of sub- Standard Er- T-Statistic Significance

AIIData
samp1e samp1es ror

Marketing Concee.t
• Selling brands -0,1098 -0,0037 0,3513 -0,3125 0,3776
• Created new brand -0,2167 -0,0860 0,2924 -0,7411 0,2299
• Selling to target group 0,6217 0,4365 0,3160 1,9677 0,0255
• Wide assortment 0,6863 0,4635 0,3922 1,7499 0,0412
• Attractive ~rice 0,3860 0,1870 0,3980 0,9698 0,1669
Communication Instruments
• Print -0,3515 0,0999 0,4212 -0,8345 0,2027
• TV 0,6193 0,2211 0,4551 1,3608 0,0879
• Radio -0,1473 0,0357 0,3370 -0,4371 0,3314
• Online-Advertising 0,6293 0,2091 0,4314 1,4588 0,0734
• One-to-One Marketing 0,0038 0,1647 0,3751 0,0101 0,4960
• Public Relations -0,5692 -0,2832 0,4352 -1~078 0,0965
Revenue Sources
• Commissions 0,2887 0,2256 0,4329 0,6669 0,2530
• Online-Advertising -0,4161 -0,0935 0,5262 -0,7908 0,2152
• Fixed fees 0,1813 0,0997 0,4152 0,4366 0,3315
• Mar!l!ins on Sold Products -0,8463 -0,4432 0,5742 -1,4738 0,0714
Non-lmitabilif)!_
• Time Advantage 0,2439 0,1553 0,3075 0,7931 0,2145
• Brand 0,8910 0,8015 0,3540 2,5166 0,0065
 47

Networks o[_Founders
• Technology & Know-how -0,2977 0,0384 0,5480 -0,5432 0,2939
• Data on market & competition 1,2384 0,5360 0,6893 1,7967 0,0373
• Access to suppliers & custom-
ers -0,4303 -0,0045 0,5754 -0,7479 0,2279
• Acce12tance and references -0,3623 -0,2482 0,4523 -0,8010 0,2122
Exe.erience o[Founders
• Consulting 0,3880 0,2076 0,3493 1,1109 0,1343
• Entrepreneur -0,2819 -0,0019 0,3709 -0,7600 0,2243
• Same industry 1,0489 0,5125 0,4984 2,1045 0,0186
• Marketing and sales -0,0562 0,0685 0,3693 -0,1522 0,4396
• IT-technology -0,2600 -0,0616 0,3846 -0,6759 0,2501
• Finance, Accounting and Ad-
ministration -0,0966 -0,0927 0,3433 -0,2814 0,3894
Financial Incentives
• Above average salary 0,5579 0,1599 0,6081 0,9174 0,1803
• Generous stock options and
shares 0,6878 0,3679 0,5978 1,1505 0,1259
Job Attractiveness
• Interesting Tasks -0,1150 -0,0884 0,4704 -0,2445 0,4036
• Secure Jobs 0,6892 0,6223 0,3282 2,0997 0,0188
• Career OJ2J20rtunities 0,5702 0,3876 0,5112 1,1155 0,1333
Core.orate Culture
• Hierarchical decision making 0,4661 0,1580 0,4117 1,1321 0,1298
• Supply of own ideas -0,3306 0,1116 0,3706 -0,8920 0,1870
• Continuing education -0,3788 0,0064 0,4847 -0,7815 0,2179
• Autonomous unit 0,3392 0,1370 0,3775 0,8986 0,1852
• Com12laint management -0,2762 -0,1322 0,3526 -0,7833 0,2174
Internet IT-Solution
• Individually developed 1,0838 0,6983 0,4154 2,6093 0,0050
• Commercially available 0,5221 0,3378 0,407 1,2827 0,1009
• Open source Software or
Freeware 0,4522 0,3722 0,3364 1,3442 0,0905
• lnfluenced by Iead-user 0,3431 0,2556 0,2922 1,1742 0,1212
• Tested UJ2front b~ customers -0,3228 -0,1966 0,3345 -0,9649 0,1681

While PLS allows for a relatively unbiased estimation of regression co-

efficients it does not provide significance measures. Therefore, PLS-Graph
offers the option to create a bootstrap sample and to run the model for all
the models, thereby providing information on the standard errors of the es-
timation from which the t-values and significance Ievels are calculated in
Table 3.
48

6 Discussion

One important result of the estimation of a structural equation model with

PLS is the formation of indices that serve as more abstract constructs in the
success factor model. In our case all of the success drivers have been
treated as indicators that form these constructs. The constructs determine
market share and revenue on the one hand and due to their cost implica-
tions profitability on the other hand. As a result, all constructs show posi-
tive relationships with market share and can therefore be interpreted as
business enablers (see Table 3).
With respect to the marketing concept we realize a negative effect from
strategies of either selling brands or creating brands which is, however, in-
significant. More helpful are strategies of serving well-defined target
groups, offering a wide assortment or an attractive price. These have a
positive and significant effect on the marketing concept. With respect to
the communication channels we realize a negative but insignificant effect
of mass media like print and radio while TV significantly contributes to a
successful communication strategy. Banners have a positive effect while
public relations have an unexpected negative impact, both being signifi-
cant. The indicator of a one-to-one-marketing strategy does not signifi-
cantly affect success. The non-imitability of the marketing strategy is sig-
nificant only through its indicator of having created a strong brand while a
time advantage did not turn out to create a significant disadvantage for the
imitators. Quite different effects can be observed from the various revenue
models. While the reliance on revenue from online-advertising and mar-
gins of sold products stimulates revenue it does hurt profitability. On the
other side commissions on transactions and fixed access fees reduce reve-
nue while improving profitability. However, note that only the effect of
margins on sold products is significant.
If we consider the resources provided by the founders, the employees,
and the technology we realize that the founders have some influence
whereas the conditions for the employees and the technology do neither
really explain market share nor do they explicate profitability. Notahle ex-
ceptions are a few indicators. With respect to the founders network we dis-
cover that the network is only helpful for data collection on markets and
competitors while the transfer of technology and know-how, the access to
important suppliers and customers, and the opportunity of getting refer-
ences do not impact market share. F ounders only benefit from experience
in the consulting sector or from the same industry. Experience as an entre-
preneur, from marketing and sales, IT-technology or finance, accounting
and administration had a slight negative, but insignificant impact. It is in-
 49

teresting to note that frorn the indicators representing the rnotivation of the
ernployees and the corporate culture only generous stock options or shares
on the one side and hierarchical decision rnaking and secure jobs on the
other side contribute positively and significantly to business. This rnight
have changed now with the fall ofthe DAX. All the other indicators are in-
significant which irnplies that rnotivation was not a key driver for explain-
ing perforrnance differences. The necessity to survive rnay have led to this
destructive finding. Finally, we can see that the strategy of developing the
IT-solution by the cornpany itself has a highly significant and positive ef-
fect on profitability while the use of standard software had no significant
irnpact on profitability.
In order to better identify the rnost irnportant indicators, Table 4 sorts
thern according to the t-value because this value provides inforrnation how
rnuch a variable contributes to the prediction of an outcorne (Hansen 1987,
p. 523).
These constructs resulting :frorn the indicators and other one-itern con-
structs were investigated whether they had an influence on intermediate
success rneasures such as ernployee fluctuation, rnarket share, revenue,
profitability and finally on the future engagernent of the cornpany. The co-
efficients are displayed in Figure 2 (see chapter 5).
With respect to the fluctuation we observe a negative irnpact of job at-
tractiveness and rnotivation on fluctuation and a positive irnpact of risk
tolerance as expected. However, all these influences are not significant.
The only significant effect cornes :frorn integrated units. While these corn-
panies have rnore stable conditions in general, they rnay have provided less
attractive working conditions in that case and had to lay-off parts of the
ernployees to achieve profitability.
Satisfaction with rnarket share is positively influenced by nearly all con-
structs but with a varying degree. A significant irnpact is given by the
rnarketing concept, the non-irnitability of the strategy, the founders' net-
works, and the risk tolerance. It is interesting to note that the ernployees'
working conditions have alrnost nothing to do with success. Cornpanies
that went online early, address their offerings to consurners (B2C), and that
belong to the start-ups, were significantly rnore satisfied with their
achieved rnarket share than the other types of cornpanies.
50

Table 4. Importance of single indicators according to their t-value

T-Value Positive Negative

lndividually Developed IT-
' II solution 2,6093
!-<~q
";!!:::!..
Prior Experience in Same
> Industry 2,1045
Secure Jobs 2,0997
Selling to Target Group 1,9677 Margins on Sold Products -1,4738
Network for Data on Mar-
ket and Competition 1,7967 Public Relations -1,3078
Wide assortment 1,7499 Stock Options and Shares -1,1505
q Online Advertising Com-
<'I
V munication 1,4588
TV Communication 1,3608
=
G)

-; Open Source Software or

~ Freeware 1,3442 Tested Upfront by Customer -0,9649
!-<
Commercially Available IT-

-
V
q solution 1,2827 Above Average Salary -0,9174
- Influenced by Lead-User 1,1742 Supply of Own ldeas -0,8920
Hierarchical Decision Mak-
ing 1,1321 Print Communication -0,8345
Network for Acceptance and
Career Opportunities 1,ll55 References -0,8010
Prior Consulting Experience 1,ll09 Time Advantage -0,7931
Revenue from Online-
Attractive price 0,9698 Advertising -0,7908
Autonomaus Unit 0,8986 Complaint Management -0,7833
Commissions 0,6669 Continuous Education -0,7815
Revenue Fixed fees 0,4366 Prior Entrepreneur Experience -0,7600

-
q
-
V
One-to-One Marketing 0,0101
Network for Access to Suppli-
ers and Customers
Created New Brand
-0,7479
-0,74ll
=
G)

-; Prior Experience in IT-

Ä
V
technology
Network for Technology and
-0,6759

0 Know-how -0,5432
si Radio Communication -0,4371
Selling brands -0,3125
Prior Experience in Finance and
Accounting -0,2814
lnteresting Tasks -0,2445
Prior Experience in Marketing
and Sales -0,1522

Satisfaction with revenue is significantly influenced by market share.

This is not surprising because market share is a prerequisite of revenue. In-
sofar, we have only tested whether there is an additional effect by the cho-
sen revenue structure, perceived customer satisfaction and the type of
company. Figure 2 shows that a revenue structure depending on banners
and margins on sold products, running a B2C business model and be-
longing to the integrated units positively impact revenue. These effects are
also significant (almost in the case of B2C). This means that integrated
 51

units focus on more broad revenue sources rather than on achieving a

dominant market position (as given by market share):
Satisfaction with profitability in terms of cash-flow and ROI is posi-
tively influenced by a number of constructs. First of all, market share and
revenue as prerequisites of profitability show a highly significant and posi-
tive impact. This was supported by a revenue structure focusing on com-
missions on transactions and fixed access fees. The reason for this is that
these revenue streams can be achieved mostly through digital operations
and therefore generate high profit contribution margins, while retailing
margins did not suffice to pay for the costs of the logistic processes.
Higher profitability was also achieved by having built up a competitive
advantage through a strong brand because this allows for extracting higher
margins.
A nearly significant positive influence is coming from the founders' ex-
perience in consulting or the same industry as well as from high employee
fluctuation. This fluctuation can be explained by the positive profit impli-
cations of a lay-off of employees. While risk tolerance has a positive effect
on revenue this is not true for profitability. The latter can only be achieved
if the managers do not overly take risky decisions. If companies are work-
ing with their own IT-solution this has a significant positive impact on
profitability. While self generated solutions are generally considered tobe
very expensive this is not true on the Web. Here, companies can work with
solutions that have been developed by the founders requiring only small
amounts of money. Companies belonging to the group of integrated units
and to those working with outside capital were less satisfied with their
profitability which could partly be due to another aspiration Ievel. In addi-
tion, companies that went online early are more satisfied. This can be ex-
plained by the Ionger period in which they could reach break-even.
While the results given so far allow for an evaluation of what significant
effects are they do not say anything about the relative importance of the
various policy variables. This is due to the various paths over which a pol-
icy variable can affect success in the very end. Therefore, it is necessary to
calculate the total effect of a single policy variable. This is done by multi-
plying the standardized regression coefficients of each link over the whole
path and summing these values over all paths by which a single indicator
exerts influence on success. We take profitability as the final success
measure. The results are provided in Table 5.
52

Table 5. Total Effects ofSingle Indicators for CF-ROI

positive negative

-
Time Since Online 0,2456 Venture Capital Funded -0,1980
V)
Brand Advantage 0,1872
o·
II Prior Experience in Same
Industry 0,1634
Individually developed IT-
solution 0,1040

-
V)
Employee Fluctuation 0,0998
0 Network for Data on market
V)
' and competition 0,0983
0 Secure Jobs 0,0690
0
Wide assortment 0,0678
Selling to target group 0,0615
Prior Consulting Experience 0,0604
Career opportunities 0,0571 Margins on Sold Products -0,0562
Time Advantage 0,0512 Prior Entrepreneur Experience -0,0439
Commercially available IT-
solution 0,0501 Public Relations -0,0413
Prior Experience in IT-
Stock options and shares 0,0480 technology -0,0405
V)
0 Online Advertising Network for Access to suppliers
0 Communication 0,0457 and customers -0,0342
'
IT -solution tested upfront by
"'00 TV Communication 0,0449 customers -0,0310
Open source Software or
Freeware 0,0434
Above average salary 0,0389
Attractive price 0,0382
IT-solution Influenced by
Iead-user 0,0329
Network for Acceptance and
Customer Satisfaction 0,0251 References -0,0288
Revenue from Online-
Business Model (B2C) 0,0232 Advertising -0,0276
Commissions 0,0192 Print Communication -0,0255
Technology and Know-how
Revenue Fixed fees 0,0120 Network -0,0236
Hierarchical decision
making 0,0096 Created new brand -0,0214
"'00 Prior Experience in Finance and
V Autonomaus unit 0,0070 Accounting -0,0150
Risk Tolerance 0,0029 Interesting Tasks -0,0115
One-to-One Marketing 0,0003 Selling brands -0,0109
Radio Communication -0,0107
Prior Experience in Marketing
and Sales -0,0088
Continuous education -0,0078
Supply of own ideas -0,0068
Complaint management -0,0057
 53

7 Difference between start-ups and integrated units of

bricks-and-mortar companies

The estimation of the structural equation model has shown a direct effect
of integrated units versus start-ups on market share, revenue, and profit-
ability. Integrated units are less experienced and show a high employee
fluctuation because of lay-offs, a non-significant negative impact on mar-
ket share and counter running effects on revenue and profitability. While
integrated units are more satisfied with their revenue they are less satisfied
with their profitability than start-ups. Even if one calculates the total effect
on profitability the result is negative. Note that these direct effects can only
be interpreted as residual effects, which means that they do not represent
effects that can be explained with the help of the Ievels of their success
drivers. In order to assess the difference between start-ups and integrated
units we computed the mean values of all success drivers used in our struc-
tural equation model for the groups of integrated units and start-ups sepa-
rately and multiplied them with the total effects of the success drivers. The
resulting coefficients indicate where the two groups deviate and which ef-
fect on profitability that has. In Table 6 the success drivers are sorted with
respect to their importance by explaining differences between start-ups and
integrated units.
The results in Table 6 show that overall the integrated units have chosen
better Ievels for their success drivers. They explain a difference of 0,15
units of standardized satisfaction with profitability. This has tobe consoli-
dated with a residual negative effect of 0,37 units suchthat from an overall
perspective start-ups seem to be better organizations for such a risky and
volatile business.
In more detail, we realize that start-ups have chosen a slightly better
revenue structure, have worked with slightly worse communication in-
struments, have realized a better marketing concept, provided better net-
works, work experience, more attractive incentives at the expense of a
slightly less job attractiveness and worked with the better IT-solution. All
these effects were outperformed by less fluctuation, less time for reaching
break-even and the burden ofworking with outside capital.
54

Table 6. Differences between Start-ups and Integrated Units

positive negative
> 0,10 Venture Capital Funded -0,1781
Network for Data on Market
~ and Competition 0,0563 Time Since Online -0,0788
0 lndividually Developed IT-
' solution 0,0500 Secure Jobs -0,0479
"""
0
0 Prior Experience in Same
Industry 0,0475
Prior Consulting Experience 0,0394 Employee Fluctuation -0,0281
"""
o. Stock Options and Shares 0,0295 Public Relations -0,0249
0 Career Opportunities 0,0218 Prior Entrepreneur Experience -0,0221
N
' Prior Experience in IT-
0
o· Selling to Target Group 0,0212 technology -0,0217
Created New Brand -0,0209
Network for Technology and
Margins on Sold Products 0,0170 Know-how -0,0154
Time Advantage 0,0164 Brand Advantage -0,0134
Revenue from Online-
Wide assortment 0,0163 Advertising -0,0124
Commercially Available IT-
Attractive price 0,0155 solution -0,0104
Online Advertising Com- Prior Experience in Finance and
munication 0,0121 Accounting -0,0098
Network for Access to Suppliers
Commissions 0,0068 and Customers -0,0081
N Customer Satisfaction 0,0063 Above Average Salary -0,006
0
o· Selling brands 0,0056 Tested Upfront by Customers -0,0054
V Revenue Fixed fees 0,0044 Influenced by Lead-User -0,0051
Network for Acceptance and
Continuous Education 0,0026 References -0,0041
TV Communication 0,0026 Print Communication -0,0033
Autonomous Unit 0,0025 Interesting Tasks -0,0027
Open Source Software or Free-
Risk Tolerance 0,001 ware -0,0024
Radio Communication 0,0008 Business Model (B2C) -0,0014
Prior Experience in Market-
ing and sales 0,0007 Hierarchical Decision Making -0,0010
Complaint Management 0,0004 Supply ofOwn Ideas -0,0007
One-to-One Marketing 0,0001

8 Summary and managerial implications

Despite the current pessimistic expectations of the retums of e-business

operations, the Internet is growing with respect to the number of users,
customers and suppliers. Therefore, almost all companies have to deal with
the question how to organize and implement their own e-business opera-
tions. This requires a detailed knowledge about the success factors of such
operations. In addition, companies face the organizational question
 55

whether they should stimulate the formation of start-ups and later try to in-
tegrate this business into their firm or whether they should work with an
integrated unit from the very beginning.
Basedon a survey of72 start-ups and 75 integrated units we empirically
tested with the help of a structural equation model what the most important
success drivers are. Based on the total effect on profitability we find that
non-imitability due to a strong brand advantage, prior experience in the
same industry, an individually developed Internet IT-solution, a network
for access of data on the market and competition, and a marketing strategy
focusing on selling a wide assortment to a specific target group are the key
drivers for success.
A comparison of start-ups and integrated units reveals that start-ups are
the better organizational implementation of e-business operations. The
most important difference can be observed with respect to the network for
access of data on the market and competition, the individually developed
Internet IT-solution and the prior experience in the same industry or in the
consulting business where start-ups are better. On the other hand start-ups
showed a disadvantage because of being financed through venture capital-
ists, a shorter period being online and of not providing attractive secure
jobs.

References

Albers S (2000) Was verkauft sich im Internet? - Produkte und Leistungen. In:
Albers S, Clement M, Peters K, Skiera B (eds) eCommerce - Einführung,
Strategie und Umsetzung im Unternehmen, 2nd edn. F.A.Z-Institut, Frankfurt
am Main, pp 21-36
Albers S, Panten G, Schäfers B (2002) Strategies for E-Profits: Lessons from pro-
fitable e-commerce companies. Working paper, University ofKiel
Aldrich H, Zimmer C (1986) Entrepreneurship through Social networks. In: Al-
drich H (ed) Population Perspectives on Organizations. Uppsala, pp 13-28
Amit R, Zott C (2001) Value Creation in eBusiness. Strategie Management Jour-
nal 22: 493-520
Anderson P, Anderson E (2002) The New E-Commerce Intermediaries. Sloan
Management Review 43: 53-62
Bamey JB (1991) Firm resources and sustained competitive advantage. Journal of
Management 17:99-120
Brüderl J, Preisendörfer P, Ziegler R (1992) Survival Chances ofNewly Founded
Business Organizations. American Sociological Review 57: 227-242
Chen PY, Hott LM (2002) Measuring Switching Costs and the Determinants of
Customer Retention in Internet-Enabled Businesses: A Study of the Online
Brokerage Industry. Information Systems Research 13: 255-274
56

Chin WW (1998) Issues and Opinion on Structural Equation Modeling. MIS

Quarterly 22:7-16
Coltman T, Devinney TM, Latukefu A, Midgley DF (2001): E-Business: Revolu-
tion, Evolution, or Hype? California Management Review 44: 57-74
Coltman TR, Devinney TM, Midgley DF (2001): Strategie Driversand Organiza-
tional Impediments to eBusiness Performance: Theory and Evidence. Work-
ingpaper
Day G, Wensley R (1988) Assessing Advantage: A Framework for Diagnosing
Competitive Superiority. Journal ofMarketing 52: 1-20
Dawes J (1999) The Relationship between Subjective and Objective Company
Performance Measures in Market Orientation Research: Further Empirical
Evidence. Marketing Bulletin 10: 65-75
Diamantopoulos A, Winklhofer HM (2001) Index Construction with Formative
Indicators: An Alternative to Scale Development. Journal of Marketing Re-
search 38: 269-277
ELMAR (2002) Moderated electronic mail network for marketing academics.
http://www.columbia.edu/~pbpllelmar/, Digest 947
Figueiredo JM (2000) Finding Sustainable Profitability in Electronic Commerce.
Sloan Management Review Summer: 41-52
Geyskens I, Gielens K, Dekimpe MG (2002) The Market Valuation of Internet
Channel Additions. Journal ofMarketing 66: 102-119.
Grant RM (1996) Toward a knowledge-based theory ofthe firm. Strategie Man-
agement Journal Summer Special Issue 17: 109-122
Gulati R, Garino J (2000) Get the Right Mix ofBricks & Clicks. Harvard Business
Review May June: 107-114
Hagel III J, Armstrong AG (1997) Net Gain- Expanding Markets through Virtual
Communities. Boston
Hansen G (1987) Forecasting Error and Multicollinearity (in German), Jahrbuch
für Nationalökonomie und Statistik 203/5-6: 517-531
Hanson W (2000) Principles of Internet Marketing. South-Western College Pub-
lishing, Cincinnati
Hinkel K (2001) Erfolgsfaktoren von Frühphasenfinanzierungen durch Wagniska-
pitalgesellschaften. Deutscher Universitäts-Verlag, Wiesbaden
Jaworski BJ, Kohli AK (1993) Market Orientation: Antecedents and Consequen-
ces. Journal ofMarketing 57: 53-70
Mahajan V, Srinivasan R, Wind J (2001) The dot.com retail failures of2000: were
there any winners? Journal ofthe Academy ofMarketing Science 30
Malmsten E, Portanger E, Drazin C (2001) boo hoo: A Dot.com Story from Con-
cept to Catastrophe. Random House, London
Porter M (1985) Competitive Advantage- Creating and Sustaining Superior Per-
formance. The Free Press, New York
Rangan S, Adner R (2001) Profitsand the Internet: Seven Misconceptions. Sloan
Management Review, 42: 44-53
Rayport JF, Jaworski BJ (2001) e-Commerce. McGraw-Hill/Irwin, New York
Schwartz EI (1997) Webonomics. Nine Essential Principles for Growing Your
Business on the World Wide Web. New York
 57

Shane S, Cable D (2002) Network Ties, Reputation, and the Financing of New
Ventures. Management Science 48: 364-381
Shapiro C, Varian HR (1999) Information Ru1es. A Strategie Guide to the Net-
work Economy. Boston
Srivastava R K, Shervani TA, Fahey L (1998) Market-Based Assets and Share-
holder Value: A Framework of Analysis. Journal ofMarketing 62: 2-18
Wemerfelt B (1984) A Resource-Based View ofthe Firm. Strategie Management
JournalS: 171-181
Zerdick A, Picot A, Schrape K, Artope A, Goldhammer K, Lange UT, Vierkant,
E, Lopez-Escobar E, Silverstone R (2000) CS. Strategies for the Digital Mar-
ketplace (European Communication Council Report), Springer, Ber-
lin!Heidelberg, originally published for the European Communication Council
1999
E-Business Performance: A Latent Class
Examination

Timothy M. Devinney

Tim R. Coltman

David F. Midgley

1 Abstract

Recent seminars, conferences and articles have dramatically increased

awareness conceming the strategic potential in Internet and related tech-
nologies. What remains poorly understand is why certain firms are suc-
cessful in capturing the potential in sophisticated information technolo-
gies-in particular ebusiness systems-while others remain hesitant or
unable to change. Building on the major theoretical arguments in strategic
management, this paper investigates the complex drivers of ebusiness per-
formance. Latent class modeling techniques enable us to show that busi-
ness performance and these drivers are heavily influenced by the unob-
servable heterogeneity across firms and that four distinct types of firms
exist in our data. The implication is that a single model cannot explain the
relationship between environment, structure, feasibility, managerial beliefs
and performance.

2 lntroduction

Even after the dot.com bust, there is a strong beliefthat the intemet is an
important factor driving the future performance of many established firms
through the creation of new transactional value between firms, suppliers,
complementary product service providers and customers (Kalak:ota and
Robinson 1999). The impressive performance returns of companies like
Dell Computers, Cisco Systems and General Electric illustrate the impor-
tance oflinking ebusiness to firm strategy. Yet in spite ofthese successes
the failure by many others firms--essentially operating in the same line of
business-raises some important questions.
 59

• Why do certain firms seize upon the strategic potential in ebusiness,

while others remain hesitant or unable to change?
• Why does ebusiness performance vary between organizations that
operate within the same line ofbusiness?
• To what extent are competitive differences structural-i.e., driven by
firm resources and infrastructure-or cognitive-i.e., driven by the be-
liefs and commitment of managers to an ebusiness future?
These questions are the focus of our paper and the remaining sections
set about testing a general framework of ebusiness performance, which,
after accounting for the pressures facing the firm, explains why and
through which mechanisms the adoption of ebusiness should Iead to
operational and competitive advantage. We test the importance of these
measures using field interviews and a survey of 294 senior executives. Our
results indicate that diverse demands and conditions (heterogeneity)
characterize the business environment, creating very different pressures for
change and significant variance in the performance outcomes of such
change. Using mixture regression modeling techniques we identify four
distinct segments that explain significantly more variation in ebusiness
performance than a model with only aggregate measures. Our analytical
approach is further developed in subsequent sections and represents a rela-
tively new approach to understanding unobserved heterogeneity within the
firm.

3 Model structure

Within the strategy field researchers have reached a general consensus that
strategy making is reflected in a pattem of important actions and decisions.
Typically these decisions are directed at: (1) maintaining alignment with
the extemal environment and (2) pursuing opportunities within the con-
straints ofmajor intemal interdependencies.
This line of thinking allows us to develop a model that captures the set
of pressures driving ebusiness performance along two dimensions: (1) the
extemal pressure to move online and (2) the feasibility of implementation
in terms of financial, business and organizational constraints. However,
these alone do not determine the ultimate ability of the firm to perform. A
third set of factors serve to mediate the relationship between intemal and
extemal pressures, the beliefs of the managers as to the extent to which
moving on-line is the correct strategy for the firm.
60

3.1 Empirical data

A survey was mailed to 2,000 organizations and a final sample of 294

responseswas used in the analysis (15% response rate). Industry distribu-
tion is relatively even across six industry groups: services (n=117),
govemment (n=60), retail (n=33), manufacturing (69), agriculture and
mining (n=21). Firm size was also well distributed, with 46% small to
medium-sized firms (less than 500 employees) and 54% large firms (more
than 500 employees). The average and median sizes of these firms were
2,480 and 650 employees respectively. Tests ofthe distribution ofretumed
surveys indicate that no industry or size bias existed in the responses
received.

3.2 Instrument development and measures

Using the strategic business unit (SBU) as the Ievel of analysis, we de-
veloped all scales using an extensive and recursive pre-testing procedure.
Firmperformance has consistently been used as the definitive dependent
variable and we use the approach advocated by Venketraman and
Ramanujam (1986). Three categories ofperformance are measured: finan-
cial, operational and overall effectiveness. The financial category
comprised six measures (a = 0.91) and five items were used to derive an
operational and overall effectiveness measure (a = 0.80). These various
measures were then combined into one scale that measured overall
performance dimensions; the inter-item consistency measure was 0.82.
Empirical programs have consistently shown that extemal pressures
create strong drivers for change in organizations. In this study we chose
not to replicate these complex measures since our interest lay only in the
extent that extemal pressures are driving firms to adopt ebusiness. This
construct was measured using a single item-"To what extent are market,
technological and environmental pressures moving the firm towards more
or less online products and/or services". We see no problern with using a
single item measure where respondents clearly understand that there is
only one characteristic being referred to in the question.
A key feature of established firms is that they have an organizational
context that exerts selection pressures that account in large part, for dif-
ferences in strategic choice and performance. These readily identifiable
organizational capahilifies are both tangible (e.g., physical IT infrastruc-
ture) and intangible (e.g., reflected in human knowledge sets and know-
how). Because these measures are causal (i.e., they influence the construct)
we combine these items into a single formative index.
 61

Managerial belieft were measured by asking respondents to rate the

extent to which they believe that ebusiness systems will create new opera-
tional and strategic benefits. Five items were used to capture beliefs ( a =
0.72).
Three separate items were used to measure organizational and technical
feasibility: (1) financial constraints incurred in setting up new ebusiness
operations, (2) the organizational and political constraints incurred in
setting up and taking down complex IT systems, and (3) the operational
implementation issues (001) incurred in terms of security, reliability and
privacy considerations. The Cronbach's alpha for each ofthese multi-item
scales is 0.82, 0.70 and 0.69 respectively.
To ensure the validity of our measures we examined key-informant bias,
non-response bias, common-method bias, unidimensionality, convergent
and discriminant validity. For the sake of brevity we provide abrief sum-
mary only. Seniormanagers were targeted from three functional areas (IT,
marketing and strategy) reducing the impact of key-informant bias. Based
on responses obtained from a short web-based form sent to all non-respon-
dents the threat from non-response biaswas not considered serious. To test
for common-method bias, we applied Harmann's ex post one-factor test,
across the entire survey. Thirty eight (38) distinct factors where needed to
explain the 80 per cent of the variance in the measures used with the
largest factor only accounting for 11 per cent of the variance. Hence, there
was no 'general factor' in the data that would represent a common method
bias. A correlation matrix ofthe constructs is shown in Table l.
62

Table 1. Correlations ofLatent Constructs

'1::1
...
('1)
tTl
~
('1)

3
-s
...., 0
g.
s·
-
....,
~
(')"'
0 'Tl
s·
§ to§
~
e.
o(JQ

-
Ö' 1:1' ('1) :::1 :::1 § (') 0 ('1) lll

8 '1::1
öl
~
2..... >
(')
0
~I
'J>
~ s;
-· §.: 0 ~~
til>S
lll
:::1 'J> (')
:;t.
<
::r s· =-· (') ('1)
g
(')
('1) s=...
'J>

('1) ~ q-· 0
~
.....
'J>
0
§. 0
~
'J>

Performance 1.00
Extemal
Pressures 0.23** 1.00
IT Infrastructure 0.13* 0.06 1.00
Online Activity 0.29** 0.15* 0.00 1.00
ITKnow-how 0.26** 0.15* 0.06 0.23** 1.00
Organizational
Constraints --0.13* --0.03 --0.05 -0.13* 0.07 1.00
Financial
Constraints --0.01 0.00 --0.08 --0.04 --0.02 0.22** 1.00
Oll 0.06 0.05 -0.05 0.09 -0.03 0.17** 0.33 1.00
Management
Beliefs 0.39** 0.33* 0.03 0.16** 0.23** -0.03 0.02 0.17** 1.00
State of
Implementation 0.26** 0.17* 0.06 0.15** 0.12* --0.13* --0.11 -0.04 0.28**

* p < 0.10, ** p <0.05

4 Results

Rather than use the traditional OLS approach we apply mixture regression
modeling to analyze our sample. Mixture models have been particularly
popular in marketing, and although rarely used, can be applied across all
social science disciplines where decomposition of a population into classes
or segments in useful (see Wedel and Kamakura (2000) for a general
explanation). The segments and latent class estimates offer considerable
improvement, accounting for a 60 percent increase (from 30 percent to 90
percent) in the amount ofvariance explained by the model.
Before we proceed further it is important to reflect on why the segments
identified in our analysis are real conditions and not simply statistical arti-
facts that may have capitalized on random numerical variations in our data.
Numerous criteria have been proposed to assist researchers in the choice of
segments. Ketchen and Shook (1996) advocate that segmentation tech-
 63

niques consider: (1) selection ofvariables, (2) whether variables have been
standardized or not, (3) multicollinearity among variables, (4) the number
and validity of segments selected. Firstly, in accordance with mostrelevant
methodological research, we use deductive theory to initially guide selec-
tion ofvariables. Secondly, we take a conservative approach and constrain
segment error variance to reduce the impact of outliers. Thirdly, we
address multicollinearity by using the variance inflation factor statistic. All
values were relatively small (< 1.1) and well below the cause for concem
value of 3. Fourthly, since there is no single criterion to guide segment
number selection, we follow the general rule that variants of the informa-
tion likelihood ratio test, such as Akaike (1974) information criteria (AIC),
MAIC and the consistent Akaike information criterion (CAIC), be used in
conjunction with more sophisticated approaches (Deb and Trivedi 1997).
The four-segment solution provided the clearest set of between segment
distinctions based on a combination of the lowest AIC, MAIC and CAIC
measures, the highest entropy measure and the theoretical meaning of the
results. Table 2 provides these statistics.

Table 2. Measures ofModel Fit and Parsimony by Segment

Number of Segments

2 3 4 5

Likelihood -224.7 -211.8 -201.6 -169.9 -139.7

AIC 475.4 471.3 467.2 424.4 431.8

CAIC 522.9 577.1 637.9 633.4 673.9
MAIC 486.5 494.2 502.5 471.6 490.4
NEC(S) 0.021 0.016 0.013 0.008
Entropy 1.00 0.289 0.359 0.717 0.660
Rz 0.13 0.41 0.64 0.70 0.76

DF 11 23 35 47 59
Minimum information criteria is shown in bold as is maximum entropy criteria
To increase the interpretability of our results, we compute "effect size"
estimates as a means of capturing the differences between the clusters
more effectively (coefficient estimates and significance measures are
available from the authors ). The effect size estimates are determined by
64

computing the value of the estimated coefficient for each cluster multiplied
by the mean for each variable. This provides a more accurate picture of the
contribution of that variable to the dependent variable and allows for
aggregation so that direct, mediated and total effects can be distinguished
more clearly. All statistics are shown in Table 3. The results indicate that
strong effect sizes exist on two variables in particular, extemal pressures to
move online and managerial beliefs. But this is only part of the narrative;
each cluster reveals its own story, providing additional insights that go
beyond single group estimates.
In the case of cluster 1 (N = 88), the highest performing group, firms are
most affected by the pressure to move online and the managerial beliefs
related to this. The total effect ofthese two variables is 5.16 (=3.47+4.47-
2. 78) meaning that it is the direct and mediating effects of the constraints
that are dragging down performance. In summary we can attribute high
performance in this segment to strong extemal pressures, high managerial
beliefs and an ability to overcome organizational constraints that arise
from competing business options.
In cluster 2 (N =52) the story is vastly different as these firms perform
to a different drumbeat. These firms are driven almost exclusively by the
overcoming the impact of operational implementation issues-network
performance, information security, brand protection and customer privacy.
This is mediated by both managerial beliefs (-2.82) and extemal pressures
(-11.90) but negatively. Similarly, this group is the most effected by finan-
cial constraints (-1.93).
Performance in cluster 3 (N = 65) is lower than any of the others.
Although, the direct effect of pressure to move on line is insignificant
(-0.24) and that of managerial beliefs is strong (3.53) the performance
effect is driven by the mediating effect of these two variables. If we
compare mean performance (2.39) with the total effect of both managerial
beliefs and extemal pressures, we see that nearly all of this is driven by
these two variables (2.78=1.28+2.06-0.56) leaving a slight negative
impact to be picked up by the direct effect of organizational conditions and
feasibility constraints.
Lastly, cluster 4 (N = 88) the second best performing group, faces sig-
nificant pressures to move online (4.13) with slightly less overall impact
from managerial beliefs. Firms in this segment are clearly sophisticated
operators with a large impact from high IT know-how (0.60) and as such
do not spend time arguing about the way in which IT decisions are made
(as evident in low organizational, financial and business constraints). In
summary, while these firms are supporters of ebusiness, they also appre-
ciate the limits to technology-based solutions and appear to place more
importance on complementary activities and know how.
 65

Another method to understand the underlying heterogeneity in these

results is to look at the different correlation structures evident in each
duster. To simplify this we have indicated which correlations between the
latent constructs are significant at the p < 0.05 Ievel and presented them in
Table 4. What this indicates is that each duster has a distinctive pattem.
66

Table 3. Effect Size Estimates*

Single
Number of Segments = 1 2 3 4 group

Externat Pressure
External Pressure to Move Online 6.82 3.46 --0.24 4.13 2.48
Organizational Conditions
ITKnow-How 0.07 --0.14 --0.11 0.60 0.55
IT Infrastructure 0.31 0.53 0.44 0.04 0.58
Online Activity 0.04 0.85 --0.03 -0.15 0.54
Managerial Cognitions
Managerial Beliefs 3.25 -9.52 3.53 1.20 0.48
Feasibility Constraints
Organizational 1.48 -8.18 -1.72 1.52 0.63
Financial 1.56 --6.51 3.99 -1.48 0.04
Operational Implementation Issues (Oll) --0.36 15.30 -2.13 0.67 0.20
State of Implementation 0.05 0.21 0.61 --0.05 0.63
Mediafing Effects (Individual Effects)
External Pressure*Managerial Beliefs -2.78 -4.90 --0.56 -2.22 0.23
External Pressure*Organizational Constraints -0.97 5.15 4.68 -2.61 -2.49
External Pressure*Financial Costs -1.53 --0.69 --6.06 -0.63 --6.06
External Pressure*OII -0.84 -11.90 2.90 1.82 -9.84
Managerial Beliefs*Organizational Constraints 0.52 3.26 -2.98 0.51 4.67
Managerial Beliefs*Financial Costs --0.33 5.27 2.20 2.82 1.14
Managerial Beliefs*OII 1.03 -2.82 -0.69 -2.53 9.56
Mediating Effects (Grouped)
Mediating Effects ofExternal Pressure§ -3.35 -7.44 1.52 -1.42 -11.23
Mediating Effects ofManagerial Beliefs§ 1.22 5.70 -1.48 0.79 5.33
Overall Impact ofExternal Pressuret 3.47 -3.98 1.28 2.70 -8.75
Overall Impact ofManagerial Beliefst 4.47 -3.82 2.06 2.00 5.81
Overall Impact ofOrganizational ConstraintsX 1.03 0.23 --0.02 --0.58 -4.29
Overall Impact ofFinancial ConstraintsX --0.30 -1.93 0.13 0.71 --0.24
___ Q-y~r~Ul!l.P.~C::.~.«?.f.QI.I_)(_____________ . ·-·----··------···············. -:-:9.17 ----~~_?_____ .9.:Q_S. __::():_Q_'t____:-:9.:~~
Estimated Mean Performance ofGroup 3.21 2.60 2.39 2.90 2.86

* Effect Sizes based on significant effects (from Table 2) represented in bold.

§ Excluding the joint effect of extemal pressures and managerial beliefs
t Excluding the joint effect of extemal pressures and managerial beliefs but including the direct effect
of the variable in question
X Including the joint effects of extemal pressures and managerial beliefs
 67

Table 4. Significant Correlations Amongst Latent Constructs (by Cluster)

"'0
0..., =l 0
:::3 =l ()""1
0 .."
:;· ~
Ö'
..., 5' :;· ;;<: 0~
~~ § w§
3
""
:::3 "'t=
V>
...:::3~ [ 0
:>
:::3
0
~
"'~ bl-·
:::3 :::3
0
~
oo""
=~~
0 ...
20 ~ :,.. :;· ::!'. <il'3
0
0
V> ""
-
2 <' 0 vr g ()
0
0
;a
@ ~· ~ eo. ~

Externat Pre sures 1, 3, 4

IT Infrastructure
Online Activity 1, 2, 3 4
IT Know-how I, 4 3,4
Organizational (3) (2)
Constraints
Financia l Constraints (2), 4 3,4
001 4 3 3 1, 3
Management Be1iefs I, 2, 3, 4 12 , 4 1, 2, 3
State of I, 2, 3 I, 3 1, 4
Implementation

Numbers indicate the cluster in which the correlation is significant at the p < 0.05
Ievel. Numbers in parentheses indicate that the correlation is negative.
The common denominator here is that online activity and managerial
beliefs are strongly correlated with performance in all the groups but that
differences appear in terms of how financial and organizational constraints
and operational implementation issues influence performance. Groups 2
and 3 are negatively affected by financial and organizational constraints
respectively while group 4's performance is positively correlated with
operational issues and financial constraints.
Looking at the relationship amongst the independent variables we see a
complex and differential pattern. Group 1 shows that managerial beliefs is
correlated with the state of implementation, external pressures to go online,
and the Ievel of online activity. Group 2' s managerial beliefs is related to
IT know-how and external pressures to go online with organizational con-
strains negatively related to online activity. Group 3 has the most complex
mixture of correlations with online activity being related to IT know-how,
operational implementation issues and the state of implementation. All the
feasibility constraints are related-financial, organizational and Opera-
tional-and management beliefs are related to IT know-how and external
pressures to go online.
68

5 Discussion

We began this paper with an important question: Why do certain firms

seize upon the strategic potential in ebusiness, while others remain hesitant
or unable to change? While prior research has examined the factors driving
adoption of IS functions (e.g. relational databases, CASE and object-ori-
ented technologies) or administrative processes (e.g. office, groupware or
decision support tools) there has been a paucity of research regarding the
drivers of more complex ebusiness systems. That IT be aligned with the
extemal and intemal subsystems is a given, but what is poorly understood
is the contingent nature of organizational impediments and the influence of
managerial cognitions. Our empirical results tell us the relationship are
complex and there is no single set of rules that apply to all firms.

References

Deb P, Trivedi PK (1997) "The Demand for Medical Care by the E1derly: A Finite
Mixture Approach" Journal of Applied Econometrics 12(3): 313-336
Kalakota R, Robinson M (1999) e-Business: Roadmap for Success, Boston:
Addison-Wesley.
Ketchen DJ, Shook CL (1996) "The Application of Cluster Analysis in Strategie
Management Research: An Analysis and Critique" Strategie Management
Joumal17: 441-456.
Venkatraman N, Ramanujam V (1986) "Measurement ofBusiness Performance in
Strategy Research: A Comparison of Approaches" Academy of Management
Review 11(4): 801-814.
Wedel M, Kamakura W (2000) Market Segmentation, London:
Kluwer Publishers
Success Factors of lnternet-Based Business
Models

Wolfgang Fritz

1 Abstract

The NASDAQ crash in April 2000 and the widespread stock market
upheavals seem to question the success of the Internet economy. Along-
side spectacular failures of dotcoms such as boo.com and webvan.com,
however, there are also very successful e-businesses such as eBay. In this
paper, key results of empirical studies on critical success factors of Inter-
net-based business models are presented and discussed. Because of several
research limitations and the premature stage of development of e-business,
much more sophisticated studies are needed in this new field of empirical
research.

2 lntroduction

In November 1999, analysts of Gartner Group presented to the public a

lifecycle model of e-business that outlined the future development of such
"new economy" businesses quite realistically. These analysts predicted
that many e-companies would tumble into a period of e-business disillu-
sionment by 2001, with 75% of projects failing to deliver on their
promises (Gartner Group 1999).
This prediction has become reality to a high degree. Since 2000, a heavy
dotcom shakeout has taken place in the U.S. as well as in Europe. More-
over, many brick-and-mortar companies had to face the failure of their e-
business projects because of immature technology, unready market, and
poor e-business strategies. Events like the NASDAQ crash in April 2000
and the closure of the German New Market in 2003 seem to indicate the
beginning of the end of e-business.
But Gartner Group's lifecycle model of e-business predicts not only a
"Trough ofDisillusionment", but also a "Slope ofEnlightenment" with the
e"mergence of "true" and sustainable e-business models in the long run.
The surviving business models would have made a transition, most likely
to a brick-and-click mix, and pure e-business itself would cease to exist
(Gartner Group 1999).
70

This last prediction seems to be questionable. Besides spectacular

failures of pure dotcoms (e.g., boo.com, webvan.com) and failures of e-
businesses of brick-and-mortar companies (e.g. Karstadt's myworld.de,
and Bertelsmann's bol.com), very successful pure e-businesses can be
found that should survive in the long run (e.g., eBay). These "e-commerce
winners" have been described in detail recently (see Albers, Panten, and
Schäfers 2002; Fischermann 2002; Mahajan, Srinivasan, and Wind2002).
This paper presents a review and meta-analysis of empirical studies on
key success factors of Intemet-based business models. These key success
factors should separate e-commerce winners from e-commerce losers and
should characterize "true" e-business models. It is also shown that,
because of several research limitations and the premature stage of de-
velopment of the "new economy", much more sophisticated studies are
needed in this new field of empirical research.

3 Conceptual background

3.1 The success factors approach

The empirical research on key or critical success factors (KSF or CSF) of

old economy businesses has a long tradition. The idea that there are a few
factors that are decisive for the success of a business was first discussed by
Danie1 (1961) and e1aborated 1ater main1y by Rockart (1979) in the context
of designing management information systems (Leidecker and Bruno
1984, p. 23; Grunert and Ellegaard 1993, p. 246). Later on, the concept of
critica1 success factors was transferred to the fie1ds of business strategy
research, where it was used in different ways. In strategic marketing and
management, the Profit Impact of Market Strategies project (PIMS) initi-
ated by Harvard Business School has stimu1ated a wide range of research
primarily in the field of industrial firms (Buzzell and Gale 1987). The suc-
cess factors approach has influenced empirical research in many other
areas such as retailing (e.g., Hildebrandt 1988) and even in accounting
(e.g., Hinterhuber 2002). Even a few decades after the success factor
research commenced, methodological questions of the success factors
research are usually discussed by making references to the PIMS data
(Hildebrandt and Buzzel11998; Annacker 2001).
Although the success factors approach is recognized in many different
areas of business studies for over two decades, no coherent scientific
research program has emerged until today. Many different and specific
 71

approaches can be found instead: confirmatory vs. exploratory research

designs; sturlies focusing on financial success only vs. sturlies using a more
comprehensive, multiple indicators set of success including non-financial
and even perceived measures of success; sturlies based on single cases vs.
sturlies based on data from big, representative and international samples
analyzed by sophisticated multivariate techniques; etc. (see in detail Fritz
1990, 1992, 1995, and 1997).
It is not surprising that the success factors research has spawned a con-
siderable variety of results, and many of these results are controversial
even after two decades (see e.g. Hildebrandt and Annacker 1998;
Annacker 2001). A meta-analysis of 40 empirical sturlies has shown, how-
ever, that "quality of human resources", "closeness to the customer",
"innovation potential", "quality of products", and "pattem of leadership"
are the most frequently mentioned key factors of corporate success (Fritz
1990 and 1997). Moreover, empirical research in the U.S. andin Germany
has proven correspondingly that the "market orientation" of a firm must be
regarded as a fundamental key success factor (Kohli and Jaworski 1990;
Narver and Slater 1990; Fritz 1992 and 1996; Jaworski and Kohli 1993;
Hornburg 1995). Against this background, recent criticism by March and
Sutton (1997) or Nicolai and Kieser (2002), after which the success factors
research must be regarded as completely unsuccessful, cannot be taken as
serious.
Although a variety of conceptual views of key success factors (KSF) or
critical success factors (CSF) can be discemed in the Iiterature (Grunert
and Ellegaard 1993, pp. 246), most KSF-CSF approaches share a number
of crucial aspects. First, it is postulated that success and failure of a firm or
a business can be traced back to a limited or small number of key factors.
Second, these key variables establish a causal relationship with the firm's
or business' s success and therefore explain a major part of the variance in
the success indicators. Third, these KSF or CSF can be shaped or managed
and therefore represent skills or resources a business should invest in (see
e.g. Leidecker and Bruno 1984, p. 24; Hildebrandt 1988, p. 92; Grunert
and Ellegaard 1993, p. 264; Fritz 1995, p. 594).

3.2 Business models on the internet

Several different definitions of the term "business model" exist. Timmers

(1999, p. 31) uses a comprehensive conception and defines a business
model " ... as the organization (or 'architecture') of product, service and
information flows, and the sources of revenues and benefits for suppliers
72

and customers." According to Elliot (2002, p. 7), "Business models specify

the relationships between different participants in a commercial venture,
the benefits and costs to each and the flows of revenues". Thus, the reve-
nue model can be considered as one of the core concepts within a business
model (Elliot 2002, p. 8). On the whole, a business model comprises - in
addition to the revenue (or better yet, the capital) sub-model - a market
sub-model, a supply sub-model, a production sub-model, an affering sub-
model, and a distribution sub-model (Wirtz 2001, p. 211 ).
Attempts have been made to distinguish several different types of business
models on the Internet (e.g. Choi and Whinston 2000, p. 104; Shaw 2000,
pp. 10; Turban, Lee, King, and Chung 2000, pp. 202; Timmers 1999). For
business-to-consumer (B2C) e-commerce, the focus of this article, a
typology by Wirtz seems to be widely accepted. According to Wirtz, at
least four different types of business models can be distinguished (Wirtz
2001, pp. 217; Wirtz and Kleineicken 2000):
• Content (e-information; e-entertainment; e-education);
• Commerce (attraction; bargaining/negotiation; transaction);
• Context (search engines; web catalogues);
• Connection (virtual communities; online networks).
In most cases, the core focus of an online business could be charac-
terized by one of the four basic types of business models. F or example, e-
zines like HotWired belang to the "content" category, online bookstores
like Amazon are examples of "commerce", search engines like Alta Vista
are directed to "context", and online networks like AOL provide "connec-
tion". The focus of an online business, however, could change over time or
could be augmented by elements of other businesses. This may lead to
hybrid and multifunctional business models. One farnaus example is
Yahoo!, which extended its core "context" business in multiple ways,
assimilating elements of most of the other three basic e-business models
(Wirtz and Kleineicken 2000, pp. 634).
Furthermore, as predicted by Gartner Group, the evolution of business
models may create a type of hybrid e-commerce that integrates online and
offline businesses. One example is Gateway 2000, a direct seller of
computers that uses the telephone and the Internetas its sales channels. To
overcome some disadvantages of being only a virtual organization, the
company opened Gateway Country Stores in key markets and began to
advertise that customers could "call, click, or come in" (Dholakia and
Dholakia 2002, p. 25). It is therefore useful to extend Wirtz's typology of
business models in the way that is suggested in Table 1.
 73

Table 1. A Typology of Business Models on the Internet

~
s
D Content Commerce Context Connection
V

e.g. Genios
Pure-click-Business e.g eBay e.g. Yahoo! e.g. AOL
Web Search

Brick-and-click e.g. Stiftung e.g. Deutsche

e.g. Time Inc. e.g. Otto
Business Warentest Telekom

4 Success factors of internet-based business models:

empirical findings

Empirical research on key success factors for e-businesses should be

directed towards each of the eight basic types of business models for the
Internet profiled in Table 1. With the exception of two Mc Kinsey sturlies
(Agrawal et al. 2001; Kemmler et al. 2001), no comprehensive empirical
study exists aiming at each of the eight fields simultaneously and trying to
discover the specific success factors within each field on a large-scale
base. Therefore, this article distinguishes primarily only between sturlies
aiming at the success factors of pure-click businesses on the one hand and
brick-and-click businesses on the other hand.

4.1 Success factors of pure-click companies

The years after 2000 have been crucial for most pure-click companies be-
cause a heavy dotcom shakeout took place during 1999-2000. While spec-
tacular failures grabbed the headlines, in this process the wheat was sepa-
rated from the chaff. Some very successful e-businesses emerged and
demonstrated that e-commerce could even be profitable against the back-
drop of an overall decline ofthe dotcom and Internet-driven economy. But
what are the reasons why businesses such as Google, eBay, DoubleClick,
Overture, Expedia, Yahoo!, PayPal (now acquired by eBay), and Webex
were more successful than others (Fischermann 2002)? Table 2 compares
the results of several studies that have tried to answer this question and to
identify the key success factors of e-businesses on an empirical basis in
2001 and 2002.
 --.]
+>.
Table 2. Some Empirical Studies ofE-business Key Success Factors (2001-2002)
Study Analyzed Type of Re- Sampie Definition of Key Success Factcrs
Business search Success
Models
A. Pure-click Businesses
Albers, Content Exploratory 10 cases of successful • Profit • High degree of digitization
Panten, Commerce Qualitative e-companies in • Positive cash • Matehing of supply and demand and creating high trans-
and Gontext Germany and Austria flow parency
Schäfers Connection • Transaction-based revenues
(2002) • Network effects
• Core business furthered by technological advantage
• Marketing at low expenses
• Outsourcing of functions to customers
• Reducing and monitaring costs
• Little venture capital
• Focus on core business

Elliot Commerce Exploratory 30 cases of success- Not given • A viable business model for the whole organization
(2002) Qualitative ful Internet retailers in • Clear priorities
6 countries on 4 • Adequate funds for expansion
continents and addi- • Understanding of target markets and learning from
tional interviews with customers and competitors
executives • Product type
• Branding
• Reliable suppliers
• Website and fulfillment capability
• First-mover advantage
• Generating sales and profits
-~
----------- ---- -----------··-·. -----~~
- ' -
Mahajan, Commerce Confir- 48 Internet retailers in Change in stock Profile of the winner:
Sriniva- matory the U.S. price since the A firm that offers
san,and Qualitative IPO • Search goods
Wind Stock options • Existing products
(2002) underwater o With offline expertise.

B. Brick-and-click Businesses
Agrawal, Content Exploratory 650 million visitors to E-performance • Focus on core product or service proposition that fit the
Arjona, and Commerce Qualitative web sites of 224 firms scorecard needs of well-defined consumer segments
Lemmens Context in North America, comprising 21 o Control of extensions of product lines and business

(2001 ); Connection Europe, and Latin indicators of models

Kemmleret America customer o Avoidance of bleeding-edge technology

al. (2001) attraction, o Commerce sites were more successful than content and

(McKinsey) conversion, community sites

and retention o Clothing is the most profitable e-tailing category

• Best e-tailers outperform offline competitors

o lncumbents' offspring outperform pure plays in e-tailing

• Pure plays outperform incumbents' offspring in content

Böing Commerce Confirma- 135 firms in various Perceived Basic orientation:
(2001 ); tory German industries attainment of e- • Technology and innovation orientation
Meffert and Quanti- and 93 additional commerce • Market orientation
Böing tative expert interviews goals Strategy and organization:
(2001) (primarily) • Detailed market entry planning
• Conflict and cooperative strategy towards dealers
• Building strong brands
• High autonomy of e-commerce department
Measures:
• Online communications
• Web site design (added value; transaction)
• Short period of delivery
• Strang monitaring -..1
Ul
 -..J
0'1

Geyskens, Content Confirma- 93 announcements of Firm's stock Most Internet Channel additions were successful,
Gielens, and tory Quanti- adding an Internet market return especially for
Dekimpe tative channel to the tradi- caused by the • Powerful firms with a few direct channels
(2002) tional business by 22 announcement • Early followers
publishers in 4 Euro- • lntroductions supported by a high publicity
pean countries I

Strauss Various Exploratory 1308 interviews with Perceived • Elaborated e-business strategy
and Qualitative e-business managers attainment of e- • Differentiation of hybrid strategy
Schoder (primarily) of various industries commerce • Realistic assessment of obstacles and opportunities
(2002) in Germany, Austria goals • Effective channel management
and Switzerland • One-to-One marketing and CRM
• Process Grientation
~ -~-- ~~ ~----
• Autonom~ of e-business dej)artment
 77

Part A of Table 2 summarizes the findings of three empirical studies of

pure-dick companies. The study by Albers, Panten, and Schäfers (2002)
analyzes 10 pure-clicks in Germany and Austria that are successful in
terms of profit and cash flow. These e-commerce winners can be found
nearly in each industry. It follows that the economical success of e-busi-
ness is not restricted to a single or specific business model. In terms of key
success factors, these e-commerce winners offer products and services
exhibiting an extremely high degree of digitization and benefit from
network effects. Furthermore, these e-commerce winners generate revenue
by participating in e-commerce transactions (sales commissions) and make
profits by monitoring and reducing costs systematically.
One overall good example of e-commerce success is eBay, the world-
wide leading Internet-based B2C auction site. As Lührig and Dholakia
(2002) have pointed out, the company has reached such a magnitude that it
will be very difficult and expensive for competitors to overtake its market-
share leadership. Because of this high market share, eBay can enjoy the
biggest scale effects and benefits from ongoing market growth. Further-
more, eBay takes full advantage of network effects and economies of
scale, because the more people use its auctions, the better the site becomes.
"The rising numbers of eBay users generate a growing tide of content for
the site. This growing content, in turn, attracts more people to the site"
(Lührig and Dholakia 2002, p. 117).
The international study by Elliot (2002) covering six countries on four
continents (focusing on Australia, Denmark, Greece, Hong Kong/China,
United Kingdom, and United States) was originally designed to explain the
adoption of and implementations of Internet-based retail e-commerce (p.
301). In addition, executives were asked to identify the overall determi-
nants of success for online retailers (p. 333). A1though no definition of
success was given, the author claims that he has identified a broad range of
major success factors in Internet retailing (pp. 333, and Table 2). Some of
these success factors correspond to those discovered by Albers, Schäfers,
and Panten (2002), e.g., "clear priorities/focus on core business", "produce
profits/outsourcing of functions, reducing and monitoring costs" and "first-
mover advantage/network effects". Other findings of Elliot (2002) and
Albers, Schäfers, and Panten (2002) do not support each other and seem
Contradietory in parts. According to Albers, Schäfers, and Panten (2002,
pp. 216, 223), the products most suited to e-commerce aredigital products,
and e-commerce winners did not make the detour of branding to attain a
high degree of awareness. In the Elliot study, however, not only digital but
also standardized physical products (e.g. books, computers, consumer
electronics) are regarded as well suited for e-commerce, and branding is
presented as a major success factor (Elliot 2002, p. 334).
78

Both studies focus on successful e-businesses solely. A comparative

sample of non-successful e-firms was not included. Thus, both studies
cannot demonstrate that the factors analyzed are characteristics of success-
ful businesses exclusively and do not exist for non-successful firms. The
next study will show that this objection may be relevant.
Mahajan, Srinivasan, and Wind (2002) conducted a study of 48 dot.com
retailers in December 2000. They identified 1-800contacts.com as the sole
winner, using two performance indicators: percentage change in stock
price since the initial public offering (IPO) and stock options "under-
water". A stock option is said to be underwater when the price of the stock
drops below the price at which the stock option was issued to an employee
and thus the employee has lost the investment in the stock (p. 476). Based
on a conceptual framework, the authors derived a specific profile for the
hypothesized winner (see Table 3). They argued that the hypothesized
winner should offer an existing (not new) and customized digital product
with a search quality, and that the winner retailer should have offline
experience (e.g. traditional offline stores) and a high number of alliances
{pp. 477). 1-800contacts.com, the only dotcom retailer whose stock
options were not underwater and whose stock prices showed an increase
since the IPO (105%), however supports only three ofthese six assertions.
As shown in Table 3, contrary to the hypothesized profile of an "e-com-
merce winner", l-800contacts.com offered a physical product (contact
lenses) without customization and without having alliances. This finding
cannot be generalized, however, because the six firms in the sample that
had filed for bankruptcy showed nearly the sameprofilein most cases (see
Table 3). The only remaining difference was that the winner had offline
experience, as hypothesized.
These fmdings demonstrate very clearly that one cannot identify key
success factors by analyzing a sample ofwinners solely, because the losers
may show some of the same characteristics as the winners. The Mahajan,
Srinivasan, and Wind (2002) study suggests the only distinct overall key
success factor to be the offline experience of online retailers. To under-
stand the traditional retail business seems to be critical to the success of
online retailing. In these authors' opinion, global retailers such as Wal-
Mart, Carrefour, and Metro are clearly well positioned to take advantage
of the market opportunities offered by the Internet (Mahajan, Srinivasan,
and Wind 2002, p. 484).
 79

Table 3. The Profile of the Winner and of Other Dotcom Retailers

Product and Firm Hypothesized Actual Winner: Bankrupt
Characteristics Winner 1-800contacts.com Firms (n=6)
Product characteristics
Product type Digital product Physical product 83 percent
physical
Product properties Search good Search good 67 percent
search
Product newness Existing product Existing product 100 percent
existing
Product customization Yes No 83 percent no

Firm characteristics
Offline experience Yes Y es 100 percent no
Number of alliances High None 5 (average)
(Source: Mahajan, Srinivasan, and Wind 2002, p. 482)

4.2 Success factors of brick-and-click companies

In Part B, Table 2 shows some major findings of several studies conducted

in the field of brick-and-click business. The study by Geyskens, Gielens,
and Dekimpe (2002) differs from the other studies because it addresses the
question of whether it is beneficial for a traditional firm to invest in an
additional new Internet channel. By using event-study methodology, the
authors show that in most cases adding an Internet channel to the tradi-
tional channels increases the firm's stock market retum in the newspaper
industry (pp. 112). The probability of successful channel addition is espe-
cially high for powerful firms with only a few direct channels, for early
followers, and for introductions being supported by a high level of pub-
licity (pp. 114). The authors show that Internet channel investments are
positive net-present-value investments especially for these firms. But the
study fails to address the question of whether such Internet channel addi-
tions are profitable in the Ionger run, years after their introduction. Recent
experience from the German newspaper industry creates some doubts in
this regard.
The first of the two McKinsey studies reported here shows that three
principles of the old economy are beneficial even for pure-dick businesses
(Agrawal et al. 2001, pp. 38): (1) Matehing the value proposition of
products and services to the needs of well-defined consumer segments, (2)
extending of product lines and business models not too far beyond the core
80

business, and (3) avoiding an overemphasizing of technology. But the

authors of the study that was in parts already conducted in 1999, do not
show empirically the causal relationships in detail.
According to the findings of the second study (Kemmler et al. 2001 ),
commerce sites are more successful than content and community sites.
Within commerce business models, clothing is the most profitable e-tailing
category. E-tailers launched by incumbent offline firms are doing better
than pure-click retailers because they benefit e.g. of existing brands and
order-fulfillment systems.
The sturlies by Böing (2001), Meffert and Böing (2001), and Strauss and
Schoder (2002) show correspondingly that an elaborated and comprehen-
sive e-business strategy and a high organizational autonomy of the e-
commerce department are important prerequisites for the success of brick-
and-click companies in e-commerce (Böing 2001, pp. 157, 199; Meffert
and Böing 2001, pp. 462; Strauss and Schoder 2002, pp. 28). These
findings seem to contradict the widespread conviction that only the inte-
gration of online and offline activities, in the form of amalgamated con-
cepts of multi-channel management, would lead to success.
It is also becoming evident that a balanced technology and market
orientation is needed to prevent firms from overemphasizing the techno-
logical aspects of e-business - a disastrous mistake that often occurred in
the past (see Böing 2001, pp. 152). In line with a strong market orienta-
tion, the building of strong brands and a short delivery cycle for goods are
regarded as important key success factors of brick-and-click businesses in
e-commerce. With the exceptions of online communications, website
design, one-to-one marketing and eCRM, many of the reported key suc-
cess factors of brick-and-click companies do not appear to differ very
much from those oftraditional firms beyond the Internet economy.

5 Discussion

The current state of research reported in this article delivers no clear pic-
ture of the key success factors of Intemet-based business models. Reasons
can be found first on the methodologicallevel.
The empirical success factors research in e-commerce is still a relatively
young and immature in methodological terms. Thus, no dominant research
design has yet emerged. Exploratory sturlies with very small sample sizes
can be found just as confirmatory sturlies based on large-scale surveys. A
wide range of non-comparable success indicators is used in the different
studies, ranging from perceived success measures to profit indicators and
 81

to stock market-based measures. In many cases, perceived success

measures as weil as stock prices do not reflect realized operating per-
formance of a firm and should be replaced by profit and cash flow (see
Geyskens et al. 2002, p. 117). However, profit and cash flow figures are of
course hard to come by for many e-commerce firms.
Moreover, each reported study here offers only a snapshot of the firms
at a specific point in time. Since the Internet economy changes very
rapidly (and hence the firms' success may change just as fast), longitudinal
analyses are needed (see Mahajan, Srinivasan and Wind 2002, p. 485). As
already mentioned, another serious problern may be the intemal validity of
the findings of those sturlies that analyze samples of successful firms
solely. In addition, most sturlies do not have a sufficient theoretical
grounding (see Böing 2001, p. 33).
Against this backdrop, the prevalence of multiple, different, and in parts
contradictory findings is not surprising. Perhaps only a few very general
overail principles of success in e-commerce can be derived from some of
the sturlies today. For example, a balance between technology and market
orientation appears to be an important prerequisite for e-business success.
More specific and indisputable success factors, however, can hardly be
identified in the cited studies, because many findings seem to be contra-
dictory. Some examples of such contradictions are worth noting:
• 1t seems that digital products are most suited for e-commerce. But the
winner firm in the Mahajan, Srinivasan and Wind (2002) study sold
physical products successfuily online - but so also did six firms that
filed for bankruptcy. According to McKinsey, clothing is the most
profitable e-tailing category.
• It seems that search goods are most suited for e-commerce. The growing
online demand for travel services, however, shows that goods whose
quality must be experienced can also be successfuily sold online.
• It seems that network effects are very often responsible for success in e-
commerce. While this can be shown for some business models (e.g.,
eBay), for others it is not so clear (e.g., Deil).
• It seems that branding is an important success factor in e-commerce.
But the e-commerce winners in the Albers, Panten and Schäfers (2002)
study were successful without a branding strategy.
• 1t seems that a multi-channel management is weil suited for brick-and-
click companies. Three sturlies show, however, that it is not the brick-
and-click integration, but instead the organizational separation and
autonomy of the e-business activities that must be regarded as a key
success factor.
82

Under the prevailing state of research, the approaches and findings of

the empirical key success factors (KFS) research in e-business are very
heterogeneous and sometimes conflicting and confusing. The studies
reviewed and compared in this article, however, characterize only the first
steps into a fascinating new field of empirical research. Many more steps
are needed and should be undertaken in order to overcome the short-
comings and limitations of these early-stage KFS-CFS studies on e-busi-
ness.

References

Agrawal V, Arjona LD, Lemmens R (2001) E-performance: The Path to Rational

Exuberance. In: The McKinsey Quarterly, 2001, no 1, pp 31-43
Albers S; Panten G, Schäfers B (eds) (2002) Die eCommerce-Gewinner, Frankfurt
amMain
Annacker D (2002) Unbeobachtbare Einflussgrößen in der strategischen Er-
folgsfaktorenforschung, Wiesbaden
Böing C (2001) Erfolgsfaktoren im Business-to-Consumer E-Commerce,
Dissertation, Universität Münster
Buzzell RD, Gale BT (1987) The PIMS Principles, New York
Choi SY, Whinston AB (2000) The Internet Economy, Austin, TX
Daniel DR (1961), "Management Information Crisis," Harvard Business Review,
September- October: 111-116
Dholakia N, Dholakia RR (2002) Markets and Marketing in the Information Age.
In: Dholakia N, Fritz W, Dholakia RR, Mundorf N (eds): Global E-
Commerce and Online Marketing, Westport, CT: 17-29
Elliot S (2002) Electronic Commerce, Chichester
FischermannT (2002) Die Rückkehr der Dotcoms. Die Zeit Nr. 43: 21
Fritz W (1990) Marketing - ein Schlüsselfaktor des Unternehmenserfolgs? In:
Marketing- Zeitschrift für Forschung und Praxis, vol12, no 2: 91-110
Fritz W (1992) Marktorientierte Unternehmensführung und Unternehmenserfolg,
Stuttgart (2nd. edn 1995)
Fritz W (1995) Erfolgsfaktoren im Marketing. In: Tietz B, Köhler R, Zentes J
(eds): Handwörterbuch des Marketing, 2nd. edn, Stuttgart, pp 594-607
Fritz W (1996) Market Grientation and Corporate Success: Findings from
Germany. In: European Journal ofMarketing, vol30, no 8: 59-74
Fritz W (1997) Erfolgsursache Marketing, Stuttgart
Gartner Group (1999): Traveling at the Speed of Hype, URL:
htpp://gartner11.gartnerweb.com/public/static/aboutgg/pressrel/hype 110299 .h
tml (download Nov 7, 2001)
Geyskens J, Gielens K, Dekimpe, MG (2002) The Market Valuation oflnternet
Channel Additions. Journal ofMarketing, vol66, April: 102-119
 83

Grunert KG, Ellegaard C (1993) The Concept ofKey Success Factors: Theory and
Method. In: Baker MJ (ed): Perspectives on Marketing Management, vol 3,
New York, pp 245-274
Hildebrandt L (1988) Store Image and the Prediction ofPerformance in Retailing.
Journal ofBusiness Research, voll7: 91-100
Hildebrandt L, Annacker D (1998) Panelanalysen zur Kontrolle "unbe-
obachtbarer" Einflußgrößen. In: Hildebrandt L, Hornburg C (eds): Die
Kausalanalyse, Stuttgart, pp 209-235
Hildebrandt L, Buzzell RD (1998) Ein Kausalmodell zu den Bestimmungsgründen
der Profitabilität. In: Hildebrandt L, Hornburg C (eds): Die Kausalanalyse,
Stuttgart, pp 209-235
Hinterhuber A (2002) Strategische Erfolgsfaktoren bei der Unternehmensbe-
wertung, 2nd. edn, Wiesbaden 2002
Hornburg C (1995) Kundennähe von Industriegüterunternehmen, Wiesbaden 1995
Jaworski BJ, and Kohli AK (1993) Market Orientation: Antecedents and
Consequences. Journal ofMarketing, vol 57, July: 53-70
Kemmler T, Kubicova M, Musslewhite R, Prezean R (2001) E-performance II:
The Good, the Bad, and the Merely Average. The McKinsey Quarterly, no 3
(Web exclusive)
Kohli AK, Jaworski B (1990) Market Orientation: The Construct, Research
Propositions, and Managerial Implications. Journal of Marketing, vol 54,
April: 1-8
Leidecker JK, Bruno AV (1984) Identif)ring and Using Critical Success Factors.
Long Range Planning, voll7, no l: 23-32
Lührig T, Dholakia N (2002) Online Auctions: The Ernerging New Electronic
Agora. In: Dholakia N, Fritz W, Dholakia RR, Mundorf N (eds) Global E-
Commerce and Online Marketing, Westport, CT, pp lll-121
Mahajan V, Srinivasan R, Wind J (2002) The Dot.com Retail Failures of 2000:
Where are the Winners? Journal of the Academy of Marketing Science, vol
30, no 4: 474-486
March JG, Sutton RI (1997) Organizational Performance as aDependent Variable.
Organization Science, vol 6: 689-706
Meffert H, Böing C (200 1) Erfolgsfaktoren und Eintrittsvoraussetzungen im
Business-to-Consumer E-Commerce - ausgewählte Ergebnisse einer empi-
rischen Analyse. In: Fritz W (ed): Internet Marketing, 2nd. edn, Stuttgart, pp
453-477
Narver JC, Slater SF (1990) The Effect of a Market Orientation on Business
Profitability. Journal ofMarketing, vol54, October: 20-35
Nicolai A, Kieser A (2002) Trotz eklatanter Erfolgslosigkeit: Die Erfolgs-
faktorenforschung weiter auf Erfolgskurs. Die Betriebswirtschaft, vol 62, no
6: 579-596
Rockart JF (1979) Chief Executives Define Their Own Data Needs. Harvard
Business Review, vol57, March-April: 81-92
Shaw MJ (2000) Electronic Commerce: State of the Art. In: Shaw MJ, Blanning
R, Strader T, Whinston AB (eds): Handbook on Electronic Commerce, Berlin
et al., pp 3-24
84

Strauss R, Schoder D (2002) eReality, Frankfurt am Mai

Timmers P (1999) Electronic Commerce, Chichester
Turban E, Lee J, King D, Chung MH (2000) Electronic Commerce, Upper Saddle
River, NJ
Wirtz BW (2001) Electronic Business, 2nd. edn, Wiesbaden
Wirtz BW, Kleineicken A (2000) Geschäftsmodelltypologien im Internet. WiSt -
Wirtschaftswissenschaftliches Studium, vol 29, no 11: 628-635
From the Old Economy towards the New
Economy: Managing the Transformation from the
Marketing Point of View

Michaela Haase

Michael Kleinaltenkamp

1 lntroduction

The customer and his needs are the origin of all value creation. From peo-
ple no Ionger producing and consuming in one act arises the division of
Iabor and, thereafter, the requirement of combination of those many single
activities undertaken by many different people or organizations. The
economy's stage of development, the creation and distribution of knowl-
edge within the economy, as well as cultural and other historical factors in-
fluence the processes of value creation, capture and protection. The
transformation from the assumed old economy towards the assumed new
economy is a typical example ofthat
Our paper argues that the new economy's ernerging characteristics re-
sult from new opportunities to solve problems that are already known
within the old economy. These are related to the ends of organizing an ef-
ficient division of Iabor, meeting the demands of customers and suppliers,
and economizing on both production and transaction costs, respectively.
Because of the development of Information and Communication Tech-
nology (ICT), special traits of the new economy emerge, one the one hand,
from ICT's potential to affect economic organization and, one the other
hand, from the extent actors are able to realize ICT's potential gains.
From our point of view, in order to analyze the new economy, no new
economics is necessary. 1 Our paper draws mainly on the new institutional
economics, Austrian market-process theories, and the "Leistungs" ap-
proach in order to analyze the new economy from the marketing point of
view. The new institutional economics provides the transaction cost con-
cept as well as the idea that transactions are governed by structures and in-
stitutions. As Foss (2002) has pointed out, questions concerning the crea-

1 This does not imply that we do not see a need for improvement of given theo-
ries, however.
86

tion, capture, and protection of value are not relevant in a world without
transaction costs. Herewe refer to the setting provided by Coase (1937,
1960), who distinguishes between a world without transaction costs and
one with transaction costs. In addition, we interpret the economic world,
no matter whether "new" or "old" - in terms of a transaction-cost and
-benefit world (Zajac/Olsen 1993; Dyer 1997). Market-process theories
contribute to the analysis of knowledge distribution in and over markets
but arerather silent on organizations. 2 On the contrary, the "Leistungs" ap-
proach (Kleinaltenkamp/Jacob 2002) merges the analysis of markets and
organizations. In particular, it meets the demand for theories suited to
analyze BtB customer integration. Against the background of BtB-BtB-
BtC chains, our paper focuses on BtB transactions, that is, on transactions
between supplying or buying organizations. These market transactions are
undertaken in order to produce output - that is, goods, services or bundles
composed of either- for other organizations that operate at higher or lower
market tiers.
The reminder of the paper is divided into three additional sections. Sec-
tion 2 describes old economy's attributes ofinterest. In subsection 2.1, we
discuss two cases of interfirm interaction from the "Leistungs" approach's
vantage point. The following subsection 2.2 presents a distinction between
bargaining and managerial transactions. Both kinds of transaction costs
matter for intra- as well as interorganizational relationships. Subsection 2.3
provides a description of tasks and consequences evolving from division of
Iabor as well as of problems of coordination and cooperation subsequent to
the prior. In 2.4 we illustrate the major consequences for the configuration
of value chains against the background of customer integration. The third
section of the paper deals with the new economy's special opportunities
with regard to cost effects ofiCT (3.1) and economic organization (3.2 and
3.3) of interorganizational relationships, respectively. In reference to the
new economy's potentialities and challenges, subsection 3.4 accentuates
value-chain networks and their intellectualleadership (Pathak/Zeng 2001 ).
We argue- from the business-to-business (BtB) marketing point ofview-
in favour of business strategy (Porter 2001) from which pertinent govem-
ance structures with respect to both integrative and speculative production
have to be derived. Section four concludes the presentation.

2 For an attempt to identifY an Austrian theory of the firm, see Lewin/Phelan

(2000).
 87

2 Old economy, old economics

As an academic discipline, BtB marketing is located at the interface of or-

ganization and market theory. As management task, BtB marketing con-
nects a firm's infrastructure (and the classes of activities related herewith)
with market demand. From the viewpoint of the "Leistungs" approach, the
firm's information system (communication, information transfer and proc-
essing, knowledge generation), production system, 3 and "transaction tech-
nology" (Foss 2002: 26) are to be linked with respect to intra- as well as
interorganizational tasks.

2.1 Interfirm interaction

As stated above, every demand in BtB markets has its origin in the needs
of final consumers. Therefore, final demand can only be met by means of
complex interrelationships among firms operating at the whole array of
market tiers, as depicted in figure 1:

3 We rely on a broad concept of production that encompasses all kinds of

combinations of production factors. In this regard, an attomey's preparation of
his office in expectation of a possible but still unknown dient subsequent to the
before-mentioned preparative activity, is also Iabelied "production", albeit
"speculative production."
88

----
Procurement

----
Tumover

----
----
Tumover
Procurement

----
----
Tumover
Procurement

Fig. 1. Derived Demand 4

----
Tumover
End Consumers

The concrete expression of the market demand that a high er market-tier

supplier has to meet, however, is articulated by its own customers. At first
glance, and with respect to a single transaction, the supplier has to meet
only his own customers' demand. At second glance, however, he has also
to participate in the fulfilment of his customers' buyers needs. From the
theory of the firm, that is, the "Leistungs" approach, the concept of inte-
grative production (Kleinaltenkamp 1997, Haase 2000) is derived and ap-
plied on phenomena of supplier-buyer interaction, especially on their co-
production of those output bundles that the buyers demand for. Co-pro-
duction requires the provision of those production factors as weil as capa-
bilities of the customer, which are complementary to the supplier's both
tangible and intangible assets and, thus, necessary for the joint production
process. It is the task of a firm's (marketing) management to introduce
governance structures suited to back up both the transfer of a customer's
production factors and the run ofthe joint production process.

4 In this figure, we refer to the value-chain model (Porter 1985) that depicts a
firm as a collection of activities. Due to this model, primary activities - encom-
passing production, logistics, marketing and distribution and after-sales services
- are the main "value drivers" (Amit/Zott 2001). A firm's gains result from an
excess of the activities' market value over their costs. Secondary activities
comprise the provision of a firm's infrastructure, procurement, and both man-
agerneut and development of human and technological resources. The model,
however, does not only depict a certain mode ofvalue creation: It is also a well-
confirmed tool for strategic planning. Notwithstanding, Porter's model is only
one of many instantiations of value-chain models (Esser/Ringlstetter 1991 ).
 89

The most important factor of production is information. From the per-

spective of the "Leistungs" approach, integrative production is the main
source of the flow of information from markets, that is, market partici-
pants, into organizations. Hayek's famous works on knowledge never ex-
plored how knowledge actually spreads within a market or from market to
market. In addition, he neither held interest in organizational knowledge
nor in the interfaces of knowledge transfer between markets and organiza-
tions. From our point ofview, it is one ofthe main tasks ofmarketing the-
ory- within its own potentialities and limitations- to explore and to out-
line the flow of knowledge within and between markets and organizations.
However, the flow of information does not appear automatically in perti-
nent amount and shape; rather, there is a need for steering and managing it
via appropriate govemance structures. To those, the evolving e-economy
adds the need for a pertinent ICT govemance structure5 that has to be
blended into existing structures or aimed at transaction arrangements con-
stituting the govemance structure of firms or other kinds of economic or-
ganization.

2.2 Bargaining and managerial transactions

From the above-sketched supplier-customer relationship follows that each

single transaction between firms - each bargaining or market transaction in
the meaning given by Commons (1931) or Williamson (1985)- accounts
for managerial transactions that shape and execute the interfirm co-pro-
duction. Transaction costs are interpreted as costs of social interaction that
are - within our setting - related to both bargaining and management.
Within a transaction economy, from the market theories' vantage point,
bargaining transactions displace or complement the working of the price
mechanisms. Managerial transactions are labelled "managerial" because
they are designed, triggered and controlled by the firm's management. In
general, transaction costs arise due to (inter-)actions that aim at the estab-
lishment of govemance structures, or are carried out within those already
established6 • With respect to our setting, govemance structures are a strat-

5 In our paper, we do not go into the details of an adequate ITC-govemance

structure. Rather, we refer to the work of Broadbent/Weill (1997) and WeilV
Woodham (2002). WeilVWoodham (2002: 1) define the concept ofiT-govem-
ance "as specifying the decision rights and accountability framework to encour-
age desirable behavior in the use ofiT."
6 Throughout the paper, we do not deal with the distinction between transaction
costs of establishing and running a firm, a bargaining relationship, or an inter-
firm network, respectively.
90

egy-guided means to the end of speculative and integrative production's

social engineering. The production-cost concept, one the one hand, applies
to costs both of factors and their combination and, one the other hand, to
the- non-interactive- steering and direction of combinations.
With respect to Commons, we hint at two differences between his and
our understanding of "managerial transaction." Commons only applies the
concept on the characterization of authority relationships within a firm,
that is, on superior-inferior relationships. In our setting, this would imply
that the concept is reserved to the analysis of speculative production
(Schneider 1997: 28), that is, intrafirm production processes that are car-
ried out without any supplier-buyer interactions. In the following, we use
the term in a broader sense of meaning that is not restricted to authority
relationships. To be precise, we enhance the meaning of the term in order
to cover also interfirm interactions pertaining to the realm of speculative
production.
Figure 2 shows how an interfirm bargaining transaction, cut into the
phases initiation, agreement, processing, control, and adjustment, 7 fimc-
tions as a clasp especially for those managerial transactions caused by the
bargaining transaction.

7 Integrative production starts with any kind of interaction in order to achieve a

subsequent exchange of output bundles plus property rights between supplier
and buyer.
 91

Buyer
-.....
....... Speculative

......................................... .
" c
Production Line oforder penetration
Q) ..
Lineof l>ll"
Support Activities
(intercompany) interaction "= "'= - Q)
>
1':1
0 ----------::,-------:--------:----
""fi Line of internal interaction
""
·~

~~ 5b .g Backstage Activities
-21':1 ....o----~--------

1
...... ~ Onstage Activities

=-.;=-..:...a.=.-=-=-=-=-"'-~--=--==.:=--r~'-=-.,.,1="""00 ! ~ -..,l '-t..

> = Onstage Activities
Q) o _ _ _ _ _ _ _ _ _ _ _ __

Lineof -" "'= - -~ ] Backstage Activities

........
c
(intercompany) interaction 2 8 Line o[internal interaction
- ~ ~-----~~~~~~~~

l>ll" Support Activities

"=
= "' ••••••••••••••••••••••••i~;~j~;Je~~~;;~ti~~
..
" " Production
~!-<
Speculative

Supplier

Fig. 2. Interconnection ofbargaining transaction and managerial transactions

Managerial transactions express the division of Iabor between single

suppliers and single buyers as weil as their respective interconnectedness
with diverse other market participants. Division of Iabor between supplier
and customer tak.es place because oftheir joint integrative production. Fig-
ure 2, a simplified version of a general model of firms' activities, 8 repre-
sents - in regard to integrative production - three classes of interrelated
activities, namely onstage, hackstage and support activities. Onstage ac-
tivities are, from the point of view of the respective organization who op-
erates them, such activities that allow for the transaction partner's obser-
vation of actions, procedures, or characteristics of objects of potential
interest. If we merged onstage and hackstage activities into one set, then
hackstage activities would be just those which are complementary to on-
stage activities, because the customer is not allowed for inspecting them.
Both are - as all activities caused by integrative production - immediately
related to a production process between a supplier and its single customer.
Both may be backed up by supportive actions undertaken by additional

8 The model presented is a generalized version of a so-called blueprint. See Fließ

(200 1) for a detailed description of service blueprints, their line of development
and references.
92

employees of one firm or both firms, which herewith had to cross a line of
internal interaction ( cp. Fig. 2).
In Figure 2, the line of order penetration separates integrative and in-
trafirm speculative production. Activities beyond that line deal with
speculative production that targets at the search for and procurement of
valuable resources, and, herewith, at the preparation of integrative produc-
tion. Speculative production also gives rise to interfirm interactions carried
out in order to create and manage those value chains that are in charge of
actuation subsequent to (more or less) different demands of single cus-
tomers. The "Leistungs" approach therefore distinguishes between two
kinds of managerial transactions due to their betonging either to resources,
preparation and pre-production, or to value creation by supplier-customer
interactions.
Managerial transactions related to integrative production are due to in-
trafirm interactions among supplier's and buyer's immediate contact per-
sons,9 as well as among employees within the one, the other or either firm.
Interactions among those contact persons are also true interfirm interac-
tions that may be undertaken within the boundaries of either firm. They are
the result of but are no part of, that class of activities usually subsumed
under the Iabel "bargaining transaction" because they are not concerned
with activities leading to the establishment of a bargaining-transaction
governance-structure like, for example, the search for transaction objects
or partners, bargaining for, and enforcing of contractual terms.
Managerial transactions arise because the "organization ... has a life of
its own" (Williamson 1999: 4). Managerial transactions, especially those
cases to which "output plasticity" (Alchian/Woodward 1988) applies, en-
compass: the design of output bundles; the organization, ruling and di-
recting of joint production activities, and the design of communication
structures that bring forward the necessary information flow. They also
provide a scaffold and leeway for weathering through conflict situations.
Managerial transactions always rest on decisions concerning design,
structure, execution and consequences of the joint value creation process,
on the one hand, and its preparation, on the other.
Because they arise before the transaction partners establish contact, ex
ante bargaining transaction costs are those that are most clearly separable
from managerial transaction costs. Our discussion shows, as market and
organization theory become more and more connected, the boundary
between the class of activities related to market transactions and that re-

9 In the given context, we refer to individuals only; but, with respect to different
possible Ievels of analysis, interactions of groups within firms or subsections of
firms may also be regarded.
 93

lated to organizational transactions, may no Ionger exist or at least begins

to blur:
Interfrrm Intrafirm
Bargaining Transactions
Managerial Transactions Managerial Transsetions

I
Supplier-buyer interactions 1 Supplier-buyer interactions Supplier-buyer interactions
during the phases of initia- 1 with respect to the transfer with respect to the transfer
lntegrative
tion, agreement, processing, 1 of the buyer' s production of the buyer's production
Production
control and adjustment of a 1 factors and the organiza- factors and the organiza-
contract and its fulfillment 1 tion of the factor combination tion of the factor combina-
I tion

I
The supplier' s interfirm The supplier' s intrafmn
I
interactions carried out preparation and execution
I
Speculative in order to create and of speculative production
I
Production manage value chains processes
I
I
I

Fig. 3. Typology of transactions with respect to integrative and speculative

production

The main advantage of Figure 2 is its suitability for modeling dyadic

interaction structures with respect to integrative production. Conceming
speculative production, the model is limited to the depiction of intrafirm
interaction. That notwithstanding, interfirm interaction systems that evolve
into value-chain networks, also arise due to speculative production. We Ia-
bel this kind of interfirm interactions "interfirm managerial transactions,"
or, more specifically with respect to those value-chain networks which we
will discuss below, "network transactions," whereas the classes of in-
trafirm managerial and interfirm managerial transactions overlap each
other.

2.3 Division of Iabor and costs of coordination

The main strategic goal of actors aims at the creation, capture and protec-
tion of value' 0 , whereas "value" is defined as difference between revenue
and value of all inputs (including opportunity costs). Each supplier has to
decide how to segregate speculative and integrative production. From the

10 These goals are interrelated. Amit and Zott (200 1: 498) argue, for example, that
"the prospect of value preservation or sustainability is an important incentive
for value creation."
94

supplier's point ofview, hisaffered amount of customer individualization

is based on strategy considerations. Marketing can help the firm to find a
pertinent segregation of markets as weil as of customer segments, which
offers to customers those output bundles with the highest utility measure.
The more activities are undertaken autonomously by the supplier, that
is, without the customer' s intervening support, the more the production as
weil as transaction costs are kept under the supplier' s control. The cus-
tomer' s influence on both cost categories hinges on, inter alia: first, the
suitability of his production factors; second, his expertise conceming com-
plementary productive activities; and third, the smoothness of communi-
cation and information flow among those actors involved in integrative
production. lf the customer is an expert in production, his involvement
therein may even reduce the measure of production costs compared to the
magnitude that might have arisen from the supplier' s being the single - or
comparatively more involved- producer. That notwithstanding, if activi-
ties formerly undertaken in co-production are now carried out by the sup-
plier himself, he might make use of standardized production processes that
may result in ultimate cost reductions.
Additionaily, if the supplier bought standardized components on mar-
kets, then he would have to compare the savings in production costs with
the measure of additional transaction costs that arise due to his vertical
disintegration. Keeping this argument at the back of our mind, we can
claim that production costs would decrease if the supplier reduced the
amount of integrative production in favour of speculative production un-
dertaken by and bought from other firms. Ceteris paribus, the supplier will
do better if, and only if, the increase in transaction costs caused by the
additional sourcing activities, will remain below the production costs' de-
crease:
(1) PC'~ + TC'siB) + [(P(xt) + TCsiSI)) + ... + (P(xn) + TCsiSn))]:::; PC~
+ TC 8j(B) 11
whereby
PC~ Production costs of supplier ~
PC'~ Production costs of supplier ~. expressing, in comparison
with PCsj, a higher share of speculative production
TC~(B) Transaction costs of supplier ~ due to interactions with
Buyer B

11 The term "PC~ + TC~(B)" expresses the costs of coordination bases on a

certain degree of integrative production, that is, of division of Iabor between
supplier ~ and buyer B.
 95

TC'sj{B) Transaction costs of supplier S1 due to interactions with

Buyer B, expressing, in comparison with TC~lB), a lower
degree of integrative production among S1 and B
Transaction costs of supplier S1 due to interactions with sup-
plier S; (i = I, ... , n; i :;t j)
Market price of supplier S;'s services
Transaction costs accruing from the supplier-buyer relationship will
probably increase (decrease ), if the customer acts as co-producer to a
higher (lower) degree than before. This is described as an increase in cus-
tomer individualization; ceteris paribus, it would shift the managerial
transaction cost curve, that expresses the costs of coordination, to the left.
From the common cost perspective of added production and transaction
costs, both supplier and buyer ofthe integrative production' s output bundle
produce best at the minimum 12 of their respective total-cost curves:

Costs (C) Total Costs

Managerial
Transaction Costs

c•

Production costs

DoL* Division of Labor (DoL)

Fig. 4. Division ofLabor's Effects on Costs

In words ofZajac/Olsen (1993 : 134), the choice ofboth firms (S1 andB)
to minimize their total costs is "a single-firm decision." In other words,
both firms are not engaged in a joint cooperative interorganizational strat-
egy. Even our sketchy statement of the problern of value creation by the
means of integrative production has illustrated how deep intra- and inter-
organizational transaction- and production-cost analyses are entangled. In
addition, the accruement of costs and value is equally interconnected.
Concerning integrative production, supplier and customer are not enforced
to economize exclusively on the costs of their joint activities. In contrast,

12 We do not equate the optimal measures emanating from calculative activities

within a transaction world with those that would result from calculation within
an idealized neoclassical setting.
96

they could attempt to work for a maximum of net value in the exchange
relationship. As Zajac/Olsen (1993: 132, the authors' accentuation) note,
"transaction cost and transaction value may often be correlated such that
the pursuit of greater joint value requires the use of govemance structures
that are less efficient from the transaction cost perspective."
Besides economizing on transaction as weil as on production costs, a fur-
ther enhancement of the usual transaction-cost setring is necessary. Be-
cause profits are the ultimate expression of competitive advantages (Porter
2001 ), it does not make sense to focus exclusively on costs. Profits are
both the supplier's and buyer's ultimate goal, and, profits result from the
difference between value creation, respectively revenue, and costs. Both
transaction and production are sources ofvalue, too.

2.4 The interconnectedness of value chains

In our attempt to clarify the different issues and their interconnectedness,

we are now going to deal separately with integrative and speculative pro-
duction, on the one hand, and with costs and benefits of transactions and
production, on the other hand. With respect to the previously mentioned
distinction, this subsection focuses only on speculative production. Com-
mercialization processes on BtB markets can be characterized through an
entanglement of the value-adding processes of supplier and buyer. By the
entanglement of firms' value chains, costs as weil as values of primary and
secondary activities 13 may change.

13 lf we recorded all activities within a firm carried out at a given point of time
and sorted them first by Porter's modeland second by ours', then the respective
sets of activities would not be exclusive to each other. Primary as well as
secondary activities may be executed during integrative or speculative produc-
tion.
 97

Supplier Buyer

Fig. 5. The interconnectedness of suppliers and buyers' value chains

Figure 5 symbolizes suppliers' opportunities to exert influence on buy-

ers' value cbains. This is not without impact on tbe buyers' competitive-
ness in their own markets. The more a supplier is able to offer pertinent
problern solutions for buyers wbicb will belp to gain advantages in terms
of competition, the bigher are tbe benefits offered to buyers, the bigher is
bis effectiveness, and, the bigher are tbe net gains of those activities en-
closed in bis own value cbain.
Witb respect to a single, isolated BtB relationsbip, the net value of inte-
grative, that is, co-operative, production (based on production and transac-
tions costs and value, respectively) needs to be compared witb tbe net
value of speculative production (based on production and transactions
costs and value, respectively). Witb respect to speculative production, the
same idea applies: The excbange relationsbip(s) between ~ and B - or
between Band (S1 or S2 or ... or Sn)- can be viewed with respect either
solely to cost minimization or to net value maximization in the relation-
ship.
We are now going to summarize our arguments before beginning with
our discussion of the new economy. The paper started with tbe matter of
fact tbat all BtB demand is derived demand. Then, we made two distinc-
tions: First, between integrative and speculative production; second,
between managerial and bargaining transaction costs. Although we stated a
transaction-cost setting, we abolisbed the ceteris paribus clause usually
imposed on production costs. In the immediately preceding subsection, we
illustrated bow the division of labor between supplier and buyer, on the
one band, and supplier and bis sub-contractor(s), on the otber band, exerts
influence on tbe interplay of production and transaction costs as well as
benefits. Regardless of the glaring similarity among tbe tasks of economic
organization with respect to integrative and speculative production, the
following became evident: each organization, engaged in integrative and
speculative production as well, bas to find a solution to tbis task with re-
98

spect to two different problern definitions. Both give rise to genuine tasks
of coordination and cooperation.
Besides derived demand as a matter of fact, integrative production is
another source, as weil as the means of conveyance, for the customer's
immediate influence on suppliers' production processes and output bun-
dles. With respect to output, Alchian and Woodward (1988) have coined
the expression "plasticity." The new economy enhances the applicability
of that concept from products and services to value chains, that is, to
"value chain plasticity." The new economy exchanges the procurement of
raw products for the "procurement" of value-chain segments. Out of the
technological opportunities provided by the new economy to interrelate
and manage value chains, firms may gain advantages in terms of competi-
tion. Effectiveness and efficiency of suppliers might increase, if they were
able to create additional customer value and to communicate this to mar-
kets, on the one hand, and to capture and protect a share ofthat value, on
the other hand.

3 The new economy's potential for value creation

In this section, we rely on the argument that the new and the old economy
are govemed by the same "laws" (ShapiroNarian 1999; Porter 2001 ).
Nevertheless, ICT development ailows for a more immediate permeation
ofBtB-BtB-BtC chains by customers' needs. As Porter (2001: 65) with re-
spect to the recent decline of many dot.coms emphasizes, the "creation of
the true economic value once again become the final arbiter of business
success." Whether firms can make use ofiCT's opportunities and whether
they are able to realize the potential gains of the new economy is therefore
an important question. Porter (200 1: 66) further notes that "(t)he great
paradox of the Internet (is) ... that its very benefits ... also make it more
difficult for companies to capture those benefits as profits." With respect
to new organizational forms pertaining to speculative production and
evolving from the new economy's path of development, the capture as
weil as protection of that value, created by interconnected value chains,
attracts particular attention.

3.1 Costs effects of ICT

Tobegin with, firms' use of ICT can reduce operation costs to a remark-
able extent. Operation costs may accrue from both integrative and specu-
lative production activities, on the one hand, and both managerial and bar-
 99

gaining transaction activities, on the other. Second, costs related to the

flow and direction of information are lowered substantially. ICT directly
reduces costs of transactions before a transaction, as well as during and
after it (Lucking-Reiley/Spulber 2001: 57; Amit/Zott 2001: 495). Search
costs and measurement costs 14 arising from quality or price camparisans
that are carried out in the phase before a bargaining transaction arrange-
ment is established may provide an example for the category of ex ante
costs. Costs of communication can be reduced before, during and even af-
ter transactions. These communications might be related to the search for
and inspection of alternatives, to the arranging of technical details, to
transfers of production factors, or to after-sales operations. Time and cost
savings may also emerge from physical meetings or physical moves of pa-
pers and documents made unnecessary by ICT. During the transaction,
costs of monitaring and control may be lowered. Information and commu-
nication costs conceming future delivery can be saved, too. 15
With respect to our distinction between managerial and bargaining
transaction costs, we conclude that ICT helps saving costs within both
categories of transaction-cost origin. Y et, cost reductions are not the only
factor that shapes the characteristics and scope of feasible transactions
within the new economy.

3.2 Market intermediation and market-making

According to Lucking-Reiley and Spulher (2001: 141), intermediation and

market-making are central activities in a market economy which bring

14 Nevertheless, the intemet is still a poorly suited medium for the conveyance of
sense impressions necessary for the inspection of several goods.
15 As Jankowski (2001: 1) puts it, with respect to Covisint, an Intemet-based e-
commerce platform for the autornative industry, "Covisint won't totally elimi-
nate the need to get tagether face-to-face and work collaboratively in person,
but it will dramatically reduce the need for that. And when Covisint's goals are
accomplished, it brings time and money savings to the organizations that are
involved."
100

buyers and sellers together. 16 Intermediaries can tak:e over both the design-
ing of govemment structures and their operation. Spulher (1996: 136)
identifies "four of the most important actions of economic intermediaries:
setting prices and clearing markets; providing liquidity and immediacy;
coordinating buyers and sellers; and guaranteeing quality and monitoring
performance."
The most important offering of ICT to integrative as weil as speculative
production processes is the reduction of transaction costs in those realms
that allow for automation. In reaction to that, producers can search for op-
portunities to outsource processes of speculative production to other pro-
ducers who - because of economies of scale - can deliver value-chain
segments below self-costs of the former producer (Amit/Zott 200 I: 495).
This is important if, as Porter stresses for example (200 1), the customer is
the main winner 17 of power in the new economy: The so empowered cus-
tomer tak:es home rents that have arisen especially due to many dot.com's
misguided price policies and business strategies. Many firms that are to-
tally unknown to today's final customers work for firms at- or close to-
the BtC end of the value chain, providing electronic manufacturing ser-
vices and functioning rather as extended work benches than value-creation
partners. From a pure cost perspective, it can be stated that, if intermedi-
aries can tak:e over coordinative and inspective activities and if they can
enable their clients to realize more gains than before, then their services
will be made use of. A company substituting a firm's n exchange relation-
ships with his vendors for one may provide an example. Intermediaries
transform transaction costs of their customers into production costs em-
bodied in output sold back to their customers. A firm B will mak:e use of
an intermediary's services if, ceteris paribus, the sum ofthe services' mar-
ket price and those of B's transaction costs, which arise from B's interac-
tion with the intermediary, does not exceed the sum of those transaction
and production costs - saved by B - that would have been accrued from
his exchange activities with n different suppliers:

16 Whereas some authors forecast the elimination of at least "some" intermediar-

ies in the new economy's path of development (see Litan/R.ivlin 2001: 315),
others, f. e., Vorst et al. (2002) conjecture that new intermediafies will afise
who are not simply going to lengthen the supply chain. In our view, the survival
of extant intermediafies as well as the appearance of new types of intercompany
intermediation depends on a compafison of costs and value of their services for
the buyer. See Amit/Zott (2001: 495) for kinds ofre-intermediation.
17 Other authors, for example, (Litan/Rivlin 2001: 315) point to the vafiety of end
users' benefits - like wider choice, added convenience, and customization that
never show up in productivity statistics; seealso Fraumeni (2001: 320).
 101

whereby

P(xi)19 Market price ofthe intermediary rs Services XI

PCJ(xi) The intermediary Fs production COStS ofhis Services XI
TC.B(x1) Transaction costs TC of x/s buyer B (arising from his
exchange with the intermediary I)
Transaction costs TC ofxi's buyer B
(arising from his exchange with S;), i = 1, ... , n
Market price20 of supp1ier S;' s output bundle xi
(i = 1, ... , n)
Market-makers or "hubs" (Kaplan/Sawhney 2000) themselves create
markets. Garicano and Kaplan (2000: 10) distinguish between neutraland
biased market-makers. Neutral market-makers are "true" market-makers
because they bring together both suppliers and buyers. Biased market-
makers either work for the supp1y or for the demand side of the market.
Market-making is the exact opposite of market-taking: Intermediaries are
market-makers; firmsthat take as given prices, signals, govemance struc-
tures and institutions, respectively, are market-takers (Spulber 1996: 137).

3.3 Value-chain management by e-hubs

E-hubs are electronic platforms especially suited for value-chain collabo-
ration and coordination. According to Pathak and Zenk (200 1: 6), e-hubs
are intermedianes "that focus on specific industry verticals or specific
business processes, host electronic marketplaces, and use various market-
making mechanisms to mediate any-to-any transactions among busi-
nesses." In the following, the authors cited above draw on Berryman and
Heck21 who cut the development of BtB e-hubs into three phases. First,
independent on-line market-makers opened up exchanges on the web and
matched buyers and sellers by charging a transaction fee. The second
phase is characterized by activities undertaken by large firms like GM,
Ford and Chrysler. These firms like Covisint, for example focus on cost

18 The right hand side of the inequality can be abbreviated to: LP(xi) + LTC.B(xi)
with i = 1, ... , n.
19 P(xi) 2: PCJ(xi) + TCJ(x1) with TCJ(xi) = the intermediary's total transaction
costs ofhis services x1.
20 This price has to cover both the supplier's production and transaction costs.

21 Berryman, K./Heck, S. (2001): Is the Third Time the Charm for BtB?, in: The

McKinsey Quarterly, No. 2: On-Line Tactics, 18-22.

 102

reduction and do not pay any special attention to small-scale participants. 22

They are thus not perceived to be neutral market-makers. Expectations
conceming the third phase center on the "integration of information for
various Stakeholders ( ... ) and integration of business functions"
(Pathak/Zenk 2001: 8).
E-hubs are the center for everything about demand and supply of a cer-
tain BtB-BtB-BtC chain to pass through. They provide the technological
facilities, cost reduction, and transaction automation that allows for an op-
timization of supply-chain Operations from end to end (McKelvie/
Simmonds 2001). 23 They provide transparency in all stages of a customer
transaction and optimize the flow of inventories across the supply chain.
According to McKelvie and Simmonds (2001: 3), they are "all about
taking the most efficient action upon a customer commitment."
From an idealized point of view at SCM or value chain management,
customer commitment means that the customer' s needs trigger those
value-chain operations necessary to produce exactly that final output bun-
dle the customer demands for. This would require a remarkable amount of
coordination with respect to both the flow of information and the execution
of Operations from the one to the other end of the supply chain. The infor-
mation flow starts with ( a more or less high share of) integrative produc-
tion at the BtC end ofthe BtB-BtB-BtC chain where the customer transfers
information conceming the desired output bundle to the supplier. It then
has to find all its way across the actuated composition of value-chain seg-
ments of different firms until the whole value-creation procedure has been
accomplished.
In this sense, customer commitment implies a need for the restructuring
of value chains in such a way that both the requirements of effectiveness
and efficiency are fulfilled. The restructuring of value chains is an instance
of "creative destruction" (Schumpeter 1934); and, taken together, both
requirements are a source of innovation and value creation (cp. Amit/Zott
2001: 496 ff.).

22 Market power undermining posttlve welfare effects expected of the new

economy is not only ascribed to consumers, but also to large firms.
Pathak/Zeng (200 1: 7) summarize: "Although it is too early to give judgement,
both first and second wave e-hubs have not produced the results as expected."
23 McKelvie/Simmonds (200 1) and also Pathak/Zeng (200 1) refer especially to
supply chains and their potentialities for development ernerging from ICT. We
draw on them, however, with respect to the broader concept ''va1ue chain" that
we do not restriet in its meaning to that advocated by Porter. That is, we use the
term "value chain" without special reference to Porter's model of the value-
creation process.
 103

3.4 lntellectual or cognitive leadership in value-chain

networks

Whereas, at present, e-hubs work on a common understanding of concems

like managing, controlling, monitoring, and executing activities across the
value chain, they do not provide "intellectual leadership" (Pathak/Zeng
2001) or "cognitive leadership" (Foss 2001a). 24 Intellectual leadership is
necessary in order to step out of line of transaction automation that may
suffice with respect to cost savings. lf, in addition to cost reductions, the e-
hub is regarded by those, who participate in the value-chain network, as a
means that should provide value-adding services for its members, then the
value-chain network should evolve a kind of identity that helps to distin-
guish it from other possible value-chain networks providing comparable
services. Hence, with respect to its business strategy, each firm has to de-
cide which "value-chain community," or "value-chain network" either, if
extant, fits best to its strategic decisions conceming both speculative and
integrative production or needs to be developed towards that direction.
Following our line of argument that draws on both cost savings as well as
gains of cooperation, firms will rather join such networks where, compared
to alternatives,
• the network contributes to the firm's strategic means and ends,
• knowledge may be better grown, utilized and transferred,
• their (tangible as well as intangible) resources and capabilities are com-
plementary to those of other participants,
• the networks focusing exclusively on the realm of speculative produc-
tion give rise to yields above those gainable by comparable networks,
• common gains are captured, protected by and shared within the value-
chain network.

24 According to Foss (2001a: 3), cognitive leadership aims at the coordination of

many peoples' complementary actions through the creation of common knowl-
edge. Foss refers to the concept of common knowledge as something that is
known from coordination games. That concept effects the communication costs
- and herewith transaction costs - ernerging from the need of coordination.
 104

"Intellectualleadership" can not be developed and carried out solely by

making use of ICT. With respect to the above-listed requirements, trans-
action automation and automized communication will quickly reach its
natural limit.25 Thus, e-hubs that express the solution to the task of eco-
nomic organization of co-operative speculative production across BtB-
BtB-BtC chains, "will focus on information and knowledge sharing among
various partners with personalized facility (Pathak/Zenk 2001: 7)." Within
such value-chain networks, issues of trust, on the one band, or oppor-
tunism and the need for (as weil as ability of) foresight (Williamson 1999),
on the other band, will noterode because the transaction-cost (and -bene-
fit) world is a world of bounded rationality, information incompleteness,
information asymmetry, or even uncertainty.
Admittedly, at this point of time it is rather unclear what "intellectual
leadership" of a value-chain network exactly is and where it emerges from.
From our problern statement, we can single out at least two sources: First,
it has its origin in integrative production, that is, in the flow of information
and the accruement of knowledge evolving throughout supplier-buyer in-
teractions. Second, it results from the flow of information across the BtB-
BtB-BtC chain. Depending on the actual requirement, this flow may be
technologically supported, entangled, or disentangled. But, as already
recognized with respect to the old economy, if there is nobody who picks
up these several strands of information, integrates them into new or dif-
ferent models of what-is-going-on, and compares these models with those
underlying firms or value communities' strategies, then no knowledge will
be created, developed, reinterpreted, or even destroyed.
All members of the value-chain community exert influence on specula-
tive production respectively on the net value accruing from division and
coordination of labor within the network. Alienability and quality of output
bundles26 depend therefore, first, on supplier-buyer interactions; second, on
cooperative speculative production among the producers in charge of those
value-chain segments that contribute to value creation of each participant
in the value chain. Value chain plasticity paves the way for those able and
apt to participate in the creation and preservation of intellectualleadership.
Referring back to Figure 4, we can summarize: In the new economy -
not unlike in an ICT enabled old economy- the use of pertinent govem-

25 Still, value-chain networks are one possible answer to Hayek's problern of

utilization ofknowledge that is not given to anyone in totality (cp. Foss 200lb).
26 Compare Dyer (1997: 538) whose Iist of value creation behavior for
'noncontractibles' encompasses innovation, quality, and responsiveness.
 105

ance structures with respect to transaction27 and ICT may shift the transac-
tion cost curve to the right, therefore allowing for a higher degree of divi-
sion of Iabor. That notwithstanding, a higher degree of customer
individualization may trigger higher costs of coordination for each single
supplier committed to it, and, hence, lead to a shift of the transaction cost
curve to the left. This tendency may be at least partially compensated by a
higher degree of division of Iabor in the area of speculative production,
which in turn gives rise to the emergence of a kind of network-related
managerial transaction costs. Which tendency will finally prevail is still
open. Not lawlike economic mechanisms pave the way towards the new
economy but bounded rational actors in their attempt to make sense of the
economic world. From the marketing point of view, not ICT is the main
driver of economic growth and business organization (O'Donell/Henriksen
2002: 89) but the economic actors' ability to make use of it. Conceming
strategic maxims, the marketing point of view singles out the interlocking
of markets and organizations' matters among which are, inter alia, the
creation, capture and protection of value from
• collecting and utilizing ofknowledge,
• designing and managing of transaction arrangements, that is, govem-
ance structures that cover the bargaining transactions as well as the
managerial transactions triggered by them,
• shaping and managing intrafirm speculative production, that is, procure-
ment, development and preparation of alienable as well as non-alienable
resources,
• shaping and managing interfirm speculative network-production, which
implies participation in value-chain networks that add value to their
members28 and provide that intellectual leadership necessary for the
achievement of competitive advantages for the whole network and its
single members as well.

5 Conclusions

The analysis of the new economy provides no striking new insights. lt

rather reinforces and strengthens insights already pointed out by several of
today's economic approaches. Conceming our analysis of economic or-

27 Production costs are also influenced by the implementation of ICT. We do not

go into the details, however, throughout our paper.
28 Unfortunately, within value-chain networks, the classical free-rider problern
arises, too.
106

ganization, our paper remains clearly within both our theory of the firm,
the "Leistungs" approach, and the tradition oftransaction costs economics,
as recently once more characterized by Coase (2002: 5): "( ... ) I think the
key to the development of a sensible analysis is the comparison between
the additional production resulting from the rearrangement of activities and
the costs of the transaction needed to bring the rearrangement about."
From the analysis carried out in our paper, this statement has not lost in
importance.
What are the effects of lowered intrafirm transaction costs on economic
organization? Coase predicts that understanding the new economy's trans-
action costs will remind people of Adam Smith: "lt enables you to have
more specialization and greater production, because you're more efficient.
... You will get more small firms as a result, but large firms will also get
larger, because they can concentrate on core activities and contract out
what they can't do weil" (Tedeschi 2000). Because transaction costs go
down, firms' boundaries will change if transactions, formerly undertaken
within the firm, are carried out in the market.
The reduction of transaction costs is indeed an important source of value
creation. ICT lowers transaction costs at remarkable amount; herewith, it
paves the way for new forms of economic organization. This does not im-
ply, however, that transaction costs should be reduced to zero. Like pro-
duction, transaction is a source of both value creation and resource con-
sumption (Dyer 1997). Transaction costs reflect the instalment of
governance structures and consumption of resources that evolve due to the
need of direct, non-parametric interaction (Johansen 1981) between eco-
nomic actors within and between organizations and markets.
With respect to economic organization, Coase argues that there are lots
of interdependencies which have to be taken into account as, for example,
the consequences of less costly intercompany transactions because the
Internet lowers the costs of information and communication between as
weil as within firms. Concerning possible effects of the Internet on the
economy's organization, Coase (2002: 6) summarizes: "So you really can't
say whether firms are going to get larger or smaller." From our point of
view, the diversity of both new and old organizational forms that emerge
from the evolved technological opportunities for value creation is the main
issue of interest.
The shape ofthe path leading towards the new economy, as weil as the
speed and success of those performing on it, is not only dependent on tech-
nology but also on knowledge - encompassing that knowledge necessary
to form and carry through a successful business strategy. With respect to
advantages on the field of competition, we transferred Porter' s (200 1)
dieturn of distinctiveness from firms to value-chain networks. Of course,
 107

we do not agree with his position that partnering erodes company dis-
tinctiveness and just increases price competition (Porter 2001: 69). That
notwithstanding, our analysis gives further underpinning to the resource-
based view's position29 that strategy is first and foremost about sustainable
competitive advantages. Yet, resources are not the only source of advan-
tages as regards competition. Besides knowledge about value and use of
complementary resources, economic organization is a source of value, too.
Despite the wide array of existing forms of organizational arrangements,
we ultimately focused on integrative production, on the one hand, and,
speculative production, on the other hand, and claimed that both produc-
tion and transaction and their economic organization are sources of value.
As advocated by Foss (2001b), coordination mechanisms may duster in
predictable ways in govemance structures. Value-chain networks are, if
successful, true market-makers: They work on prices, signals, govemance
structures, and institutions as well. The value-chain network, enabled with
intellectual or cognitive leadership, is a new organizational form that is
even today still more fiction than reality. Yet, starting with the potentiali-
ties unleashed by ICT and effecting both transaction and production, then
taking into account that the customers' needs are the ultimate source of
value creation, and assuming that firms' managements are able to trans-
form their knowledge about these preconditions into business strategies,
value-chain communities and their value-adding services are a real option.

References

Alchian AA, Woodward S (1988) The Firm is dead: Long live the Firm-A Re-
view of Oliver E. Williamson's The Institutions of Capitalism. Journal of
Economic Literature, 26: 65 -79
Amit R, Zott, C (2001) Value Creation in E-Business. Strategie Management
Journal, 22: 493 - 520
Horenstein S, Sa1oner G (2001) Economics and Electronic Commerce. Journal of
Economic Perspectives, 15 (1): 3- 12
Broadbent M, Weill P (1997) Management by Maxim: How Business and IT
Managers Can Create IT Infrastructures. Sloan Management Review, Spring
1997: 77-92
Coase RH (1937) The Nature ofthe Firm. Economica, 2: 386-405

29 As Foss (2002) notes, the resource-based view has an unnecessary starting

point in Chicago's industrial organization theory. Because the resource-based
view draws on the perfect equilibrium model, whereas we started with a trans-
action-cost world, both approaches arenot compatible to each other.
108

Coase RH (1960) The Problem of Social Cost. The Journal of Law and Eco-
nomics, III: 1 - 44
Coase RH (2002) Why Economics Will Change. ISNIE Newsletter, 4 (1), Summer
2002: 1 and4-7
Commons JR (1931) Institutional Economics. American Economic Review,
21:648-657
Dyer JH (1997) Effective Interfirm Collaboration: How Firms Minimize Transac-
tion Costs and Maximize Transaction Value. Strategie Management Journal,
18 (7): 535-556
Esser WM, Ring1stetter M (1991) Die Rolle der Wertschöpfungskette in der stra-
tegischen Planung. In: Kirsch W (ed): Beiträge zum strategischen Ma-
nagement, München: Verlag Barbara Kirsch, pp 515 - 53 7
Fließ S (2001) Die Steuerung von Kundenintegrationsprozessen: Effizienz in
Dienstleistungsunternehmen, Wiesbaden: Gabler
Foss NJ (2001a) Influencing Beliefs: A Crucial Capability for Value Creation in
the Network Economy, http://www.cbs.dk/link/papers/files/LINKwpO 1-22.
doc
Foss NJ (2001b) Whither Economic Organization? http://www.cbs.dk/link/papers/
files/LINKwpO 1-23 .doc
Foss NJ (2002) The Strategy and Transaction Cost Nexus: Past Debates, Central
Questions, and Future Research Possibilities, http://www.cbs.dk/ivs/link/
papers/files/LINKWP02-09 .doc
Fraumeni B (2001) E-Commerce: Measurement and Measurement Issues.
American Economic Association (AEA), Papersand Proceedings, 91 (2): 318
-322
Garicano L, Kaplan SN (2000) The Effects of Business-to-Business E-Commerce
on Transaction Costs, Working Paper 8017, NBER Working Paper Series.
http://www.nber.org/papers/w8017, November 2000
Haase M (2000) Institutionenökonomische Betriebswirtschaftstheorie: Allgemeine
Betriebswirtschaftslehre auf sozial- und institutionentheoretischer Grundlage,
Wiesbaden: Gabler
Jankowski D (2001) Enabling the Auto Industry: A Covisint Interview,
http://www.ascet/com/white.asp?a_id=517
Jobansen L (1981) Interaction in Economic Theory. Economie Appliquee, 34: 229
-267
Kaplan SN, Sawhney M (2000) E-Hubs: The New B2B Marketplaces. Harvard
Business Review, May-June 2000
Kleinaltenkamp M (1997) Integrativität als Kern einer umfassenden Leis-
tungslehre In: Backhaus K, Günter B, Kleinaltenkamp M, Plinke W, Raffee
H. (eds): Marktleistung und Wettbewerb - Strategische und operative
Perspektiven der marktorientierten Leistungsgestaltung, Wemer H.
Engelhardt zum 65. Geburtstag, Wiesbaden: Gabler, pp 83-114
Kleinaltenkamp M, Jacob F (2002) German Approaches to Business-to-Business
Marketing Theory: Origins and Structure. Journal of Business Research, 55:
149- 155
 109

Lewin P, Phelan SE (2000) An Austrian Theory ofthe Firm. Review of Austrian

Economics, 13: 59-79
Litan RE, Rivlin AM (2001) Projecting the Economic Impact of the Internet.
American Economic Association (AEA), Papers and Proceedings, 91 (2): 313
-317
Lucking-Reiley D, Spulher DF (2001) Business-to-Business Electronic Com-
merce. Journal ofEconomic Perspectives, 15 (1): 55-68
McKelvie KM, Simmonds M (2001) E-hubs: The New Web-Enabled Technology
Driving True Supply Chain Collaboration, (4/15/01) Ascet Volume 3.
http://www.ascet.com/documents.asp?d_ID=505
O'Donell D, Henriksen LB (2002) Philosophical Foundations for a Critical
Evaluation of the Social Impact of ICT. Journal of Information Technology,
17:89-99
Pathak BK, Zeng, AZ (2001) Achieving Information Integration in Supply Change
Management through E-Hubs: Concepts and Analyses, Paper submitted to the
First International Conference on Electronic Business, December 19-21, 2001,
Hong Kong. http://www.mouse2mailbox.com/scm.com
Porter ME (1985) Competitive Advantage: Creating and Sustaining Superior
Performance, New York: Free Press
Porter ME (2001) Strategy and the Internet. Harvard Business Review, March
2001: 63-78
Schumpeter JA (1934) The Theory of Economic Development: An Inquiry into
Profits, Capital, Credit, Interest, and the Business Cycle, Cambridge (Mass.):
Harvard University Press
Shapiro C, Varian HR (1999) Information Rules: A Strategie Guide to the
Network Economy, Boston (Mass.): Harvard Business School Press
Spulher DF (1996) Market Microstructure and Intermediation. Journal of
Economic Perspectives, 10 (3): 135- 152
Tedeschi B (2000) E-Commerce Report: Divining the Nature of Business,
http://www.nytimes.com/2000/1 0/02/technology/02ECOMMERCE.html.
Vorst JGAJ van der, Dongen S van, Nougier S, Hithorst R (2002) E-business
Initiatives in Food Supply Chains; Defmitions and Typology of Electronic
Business Models. International Journal of Logistics: Research and
Applications, 5 (2): 119- 138
WeiH P, Woodham R (2002) Don't Just Lead, Govern: Implementing Effective IT
Governance, Center For Information Systems Research, MIT Sloan School of
Management, CISR Working Paper No. 326. http://ssrn.com/
absract id=31 7319
Williamson OE (1985) The Economic Institutions ofCapitalism- Firms, Markets,
Relational Contracting. New York: The Free Press
Williamson OE (1999) Human Actors and Economic Organization, http://www.
econ-pol.unisi.it/quaderni/247.pdf
Zajac EJ. Olsen CP (1993) From Transaction Cost to Transaction Value Analysis:
Implications for the Study of Interorganizational Strategies. Journal of
Management Studies, 30 (1): 131- 145
Mining Product Configurator Data

Rainer Paffrath 1

1 lntroduction

Up to now product configurators have been referred to as software to

simplify the configuration process for complex products with various
features (including services). Configurators particularly include con-
sistency checks in order to permit only valid configurations. Invalid
configurations result, for example, from technical reasons as complex
products regularly have mutually exclusive features (Lackes and Schnoedt
1999, p 212). Primary users are, for example, salespersons in the business-
to-business context that need assistance with complex products. Another
application is in the business-to-consumer sector where customers con-
figure a product using their home computer. In some cases a configurator
is the constituent component of an order process.
This view just takes into account the user's (salesperson's, customer's)
benefit. But market researchers might benefit from product configurators
as well. A market researcher might leam a lot from interpreting the trans-
action data that is produced during the various configuration sessions.
Recording and analyzing the data from a configuration session is like
observing customers in a showroom.
As we have only little knowledge on what can really be leamed from
these data the objective of the research presented in this article is explora-
tory. Prior to applying or adapting specific methods in order to analyze the
user behavior and the relation to real buying we need to develop software
modules that enable configurators to record the needed data. This article
suggests an experimental configurator that is equipped with the necessary
functions.
The article also reports on a first study conducted with the experimental
configurator. With the data gained from the configurator we will get an
impression of the Ievel of revealed information and its usefulness. This
information might be used in the case of an anonymous production setting
in order to determine concrete product specifications within a medium-

1 The author thanks Lars Tiedemann for his help with programming issues and for
being an excellent sparring partner.
 111

ranged planning. Or will we be able to calculate segment-wise/individual

preferences in order to derive subtly differentiated measures within an
explicit customer relationship management? At least we will be able to put
tagether a concise schedule for future research.

2 Terminology

There are several reasons for the emergence of product configurators

(Lackes and Schnoedt 1999). On the one hand measures of individualiza-
tion are needed in order to survive in highly competitive markets. On the
other hand we observe a broad product differentiation due to advanced
manufacturing systems. The relative costs of producing variants have been
reduced drastically within the last time. The sales department is challenged
in a twofold way. lt needs to acquire new customers and keep up-to-date
on a broad range of complex products. In this situation product configura-
tors play an important role. They support the configuration process for
complex products with various features.
A product configurator is software in its broadest sense that dismandes a
product into customer-relevant attributes (Mertens et al. 1993). A user
chooses among the customer-relevant attribute levels and composes a
product. The configuration software considers the production-relevant
interdependences and hides the complexity. The origin ofproduct configu-
rators is in the business-to-business sector (Kurbel 1992). The emergence
of the Internet results in product configurators being adapted for lang
lasting consumer goods. This article refers to configurators in the business-
to-consumer context only. We assume that the configurator is accessible
through the Internet.
A very broad understanding of data mining will be pursued within this
article. According to this understanding data mining is a process of
exploring patterns2 in data samples using specific methods (Bensberg
2001, pp 63)3 • This definition emphasizes that data mining composes
different phases (Fayyad et al. 1996, p 10). The process starts with
selecting and extracting the required data. The experimental product con-
figurator presented in Chap. 3 will perform this step. Another important
step with regard to high data quality is data cleaning. This step includes the
identification and correction of input errors, missing values and outliers. In

2 A pattem is a type of data constellation that is found in large data sample

(Hagedorn et al. 1997).
3 Actually mining product configurator data is a sub-discipline in web usage
mining (Bensberg 2001, p 131 ff.).
 112

order to be able to apply specific data mining methods the data needs to be
coded or transformed. There exists a core list of typical data mining
methods (Hagedorn et al. 1997) but other related data analysis techniques
might be applied as well. In any case a data mining method needs to be
capable of analyzing !arge data samples. A final step in the data mining
process is evaluating the identified patterns with regard to their relevance
for managerial decisions.

3 An experimental product configurator

3.1 Requirements

Among other requirements those related to data play the most important
role, as the main task of an experimental product configurator is to record
the user's data. A product configurator should be capable ofrecording the
following data:
1. the initial configuration,
2. the final configuration,
3. the ordered product (if applicable),
4. interaction data4,
5. the sequences in which features were chosen,
6. the time elapsed during the various choice tasks and
7. termination data5 •
With data oftype 1. to 7. it is possible to re-tell the user's configuration
process. Given this data we will be able to derive the maximum level of
information.

3.2 Architecture

The architecture is taken from a simple Internet (client/server) -application.

The user interface is performed by a standard WWW-browser coded in
HTML (Paffrath 2002, pp 251 ). An example of the user interface is shown
in Fig. 1. The user chooses among the features shown on the left side of

4 A user gradua11y determines the final product by adding, exchanging or de1eting

features (data on "playing" with the configurator).
5 Termination data covers a "photo" of the termination setting. Users should be
asked to give reasons for termination.
 113

the screen. There is a drop-down list box for each attribute. When the price
calculation button is pressed, the data is sent to the hosting WWW-server.
A script-program running on an application-server takes care of storing the
data in a data file and calculates the price. The program sends back a
HTML-page including price information.
In order to record interaction and time data the application logic is
supplemented by Java-scripts running on the /oca/ machine. Java-scripts
translate event-oriented programming into action. They take each mause
click as an event and retum the event setting and the time data 6 •

HanfDrive
L~!:o.n.~~~·g~_G_~ a
Dcfault Pricc:

1277,00 €
Software

= Pricc/ = Havcn't found

CDfDCJ).Drive Performance an acccptablc :
O.K. configuration '
-------------------------------------------- .. .......!

Fig. 1. Product Configurator User Interface

Fig. 2 shows an activity diagram representing the configuration process

when working with the experimental product configurator. The configura-
tor suggests a default configuration and the price that goes with iC. First
the user inspects the default configuration. He may change the default
configuration and adapt it to bis individual preferences. Once the user has
found a "good" configuration he requests a price calculation8 • If the user
agrees with the configuration's price/performance ratio he will finish the
configuration process and quit the program. If he does not agree he will

6 The time measurement is implemented on the client machine in order to avoid

web lags (Janetzko 1999, pp 127).
7 The price of the default configuration is 1277 €. The attributes in the default
configuration occur in randomized order.
8 Prices range from 1038 € to 2504 €.
114

exchange features and calculate the price again. It is optional to change

several features at a time. He will repeat exchanging until he has found an
optimal configuration. If he cannot find an adequate configuration he will
cancel the configuration process.

yes
performance ok?

price/performance no
ok?

yes
atlribute Ievel ok?

Fig. 2. Configuration process 9

3.3 Attributes and Ieveis

As can be deduced from Figure 1 the product configurator deals with note-
book computers 10 • A user can choose among the levels of six attributes.
The complete list of attributes and levels is put together in Table 1. The
table also reports the components of the default configuration (marked
with bullets). A notebook price can be calculated by simply adding the
component prices plus a basic price (170 €). Users are asked to put
themselves in a fictive buying decision. They are supposed to have a
budget available. The brand is not specified. Users should assume a well-
accepted good brand.

9 The notation of this diagram conforms to the Unified Mode1ing Language. An

ellipse represents a (user) activity. A rhomb symbo1izes a decision.
10 The study is 1imited to 1ong-1asting consumer goods (Paffrath 2002, pp 64).
 115

Table 1. Attributes and Levels

Attribute Level
Memory (I) 256MB (128 €) •
(2) 512MB (170 €)
Display (I) 14,1" TFT (425 €) •
(2) 15" TFT (510 €)
Hard Drive (1) 20GBcapacity (94 €) •
(2) 30 GB capacity ( 179 €)
(3) 40GBcapacity (264 €)
CD/DVD-Drive (1) 8x DVD ROM-Drive (136 €) •
(2) 8x4x24 CD RW-Drive (200 €)
(3) 24xl0x24 CD RW-Drive (264 €)
(4) Combo-Drive (327 €)
Processor (I) Intel Pentium III, 1.2 GHz (128 €) •
(2) Intel Celeron, 1.2 GHz (85 €)
(3) Intel Pentium IV, 1.6 GHz (187 €)
(4) Intel Pentium IV, 2.2 GHz (255 €)
Software (l) Windows XP Horne (196 €) •
(2) Nosoftware (0 €)
(3) Windows XP Professional (306 €)
(4) Windows XP Home/Works 7.0 (238 €)
(5) Windows XP Prof./Office XP Standard (808 €)

3.4 A sample data output

The data output is divided into two files. Table 2 contains the contents of
the first file. Each row contains a configuration that is part of a user's
session. A price calculation has been requested for each configuration
contained in a single row.
The user (IP-address 62.157.83.15) changes the default configuration
with regard to the hard drive, the CD/DVD-drive and the processor. After
changing these features he requests a price calculation. As he is not satis-
fied with the price/performance ratio of this configuration he continues
"playing" with the software. First he chooses a slower processor and
second he leaves out any software. The next step is undertaken in order to
inspect the additional expenses of a 512MB memory.
116

Table 2. Session Protocol

IP-address Memory Display Hard CD/DVD Processor Software
drive -drive
121.39.510.59
121.39.510.59 Accepted
62.157.83.15 1 ~ ~ ~
62.157.83.15 2 4 ~
62.157.83.15 2 4 3 2
62.157.83.15 ~ 2 4 3 2
62.157.83.15 ! ! 4 1 2
62.157.83.15 1 ~ 4 4 2
62.157.83.15 ! 4 4 2
62.157.83.15 Accepted
217.84.15.15
Changes to a preceding configuration are marked bold and underlined.

The following row reveals that the user prefers a faster processor to
}arger memory and to a hard drive with higher capacity. He has now
reached his final configuration but checks again if a higher capacity hard
drive is affordable to him. Finally, he chooses the lower capacity alterna-
tive.
Table 2 does not contain any time data. That is why the following
extract of the second file contains the complete session history:
&history=PROCESSOR:2154 PROCESSOR:l582 HARD
DISC : 3104 HARD DISC:ll82 CD/DVD-DRIVE:2943
CD/DVD-DRIVE:l922 DISPLAY:l813 SUBMIT TIME:3445
PROCESSOR:2393 PROCESSOR:l712 SUBMIT TIME:931
SOFTWARE : 3525 SOFTWARE:l292 SUBMIT TIME:3285
MEMORY:3375 MEMORY:l352 SUBMIT TIME : l472

HARD DRIVE:3515 HARD DRIVE:2263 SUBMIT TIME:l392

&date=l5/10/2002&start_t=23 : 26:37
The output is read as follows : 2.154 seconds after arriving at the site the
user decides to choose the attribute processor. 1.582 seconds later he
makes a change to this attribute. Another 3.104 seconds later he chooses
the hard drive attribute. lt takes him 1.182 seconds to make a change. He
also chooses among the different CD/DVD-drives. Then, after 2.943
seconds he takes a Iook at the different Ievels of the attribute display but
does not make a change. As he now arrives at a point where he wants to
see the price that goes with the configuration he presses the submit button.
He submits the configuration 3.445 seconds after considering the different
 117

displays. The expression SUBMIT TIME indicates the start of a price

calculation. As the history documents the session shown in Table 2. it is
easy to pursue to the end of the data file. After the last activity the file
stores the exact session start date and time. The total session duration can
be easily calculated by summing up the individual time values.

4 A study with the experimental configurator

4.1 Descriptive results

The configurator has been placed in the Internet on October 15, 2002 11 •
The protocol file used in this article recorded data until November 21,
2002. In order to obtain a widespread use students of the business ad-
ministration department have been informed to attend a configuration
session. Extra-university user data has been recorded as weil, as the soft-
ware was accessible worldwide. 240 users started a configuration session.
Among those, 31 users terminated the session without having found a
satisfying configuration. A final configuration was stored for 209 users.
The number of price calculations per session averages 3.4. The average
number of changes to individuallevels per session equals 7.5. The number
of considered attributes within one session amounts to 9.2 (median = 7).
Consequently 1. 7 attributes are considered to be changed but are left
unchanged. The average session duration is 83.2 seconds. The time it takes
a user to change an individual attribute level averages 5.7 seconds.

11 The configurator is accessib1e through https://prowi.uni-1ueneburg.de/mischer-

start.html. A pilot study preceded the online study. With the results ofthe pilot
study the configurator has been developed into a much more user-friendly
version.
118

4.2 Analysis of preferences 12 -simple counts

A simple way of analyzing preferences is by counting choices. For

example, we can count the occurrence of each attribute Ievel in each user's
first configuration. The hypothesis behind this is that a first product repre-
sents a kind of "dream product". At that time prices are unknown. The
final configuration 13 instead is much closer to the real buying behavior as
the price system is well known.
Table 3 lists the percentages of users choosing an attribute Ievel in the
first and in the final configuration. F or example, in the first configuration
52% of users preferred 256 MB memory to 512 MB memory. This
percentage changed in the final configuration where only 37% of users
preferred the smaller memory size to the bigger one.
A comparison of the first and final configurations shows an upgrading
behavior with regard to most of the attributes except for software. The
Ievel "No software" gained 12%. The reason is clear as this choice means
cutting down expenses of at least 196 €. There is only a slight price
difference between the average prices of the first and final configurations.
The average price of the first configuration is 1701 € whereas the average
price ofthe final configuration amounts to 1685 €.

4.3 Towards segment-wise/individual results

So far the results are on the aggregate Ievel. The marketing approach asks
for a more differentiated view though. For example, one of the core
requirements in an explicit customer relationship management is an exten-
sive segmentation. That is why a classification analysis (hierarchical
duster analysis) has been conducted, a specific method within data mining
that serves to identify homogenous groups in data samples (Hagedorn et al.
1997, p 609). In order to calculate distances between objects, squared

12 Generally we use conjoint analysis in order to decompose the importance of

individual attribute Ievels from concept-related judgments (Carroll/Green
1995). Working with a configurator is a very similar task to answering pairs
questions in conjoint analysis experiments as the user is confronted with several
trade-offs. Unfortunately conjoint analysis is not an appropriate method to
analyze preferences in this context. The amount of revealed information per
configuration session is very small. Even on an aggregate Ievel it is impossible
to apply conjoint analysis methods as the assumption of attribute independency
is violated. In the study presented in this article the price is linear dependent on
the choice ofthe remaining attributes.
13 Accepted configurations are permitted only.
 119

Eudidian distances served as a similarity measure. The complete linkage

method has been applied to duster objects (Hammann and Erichson 2000).

Table 3. Analysis by counting choices

Attribute Level First Config. Final Config. Change
256MB 52% 37% - 14%
512MB 48% 63% + 14%
14,1" TFT 51% 44% -7%
15"TFT 49% 56% +7%
20GB 42% 30% - 12%
30GB 20% 25% +5%
40GB 38% 45% +8%
8x DVD ROM-Drive 30% 16% -15%
8x4x24 CD RW-Drive 8% 11% +4%
8x4x24 CD RW-Drive 13% 14% +1%
Combo-Drive 49% 59% +10%
Intel Pentium III, 1.2 GHz 31% 21% - 10%
Intel Celeron, 1.2 GHz 7% 9% +2%
Intel Pentium IV, 1.6 GHz 34% 36% +2%
Intel Pentium IV, 2.2 GHz 28% 34% +6%
WinXPHome 43% 37% -6%
o software 18% 31% + 12%
Win XP Professional 13% 14% +2%
Win XP Home/Works 7.0 4% 5% + 1%
Win XP Pro/Office XP Stand. 22% 13% -9%

The specialty of the preferred configuration in segment 1 is a low price

that results from choosing budget-priced hardware components. Segment 2
and 3 reside in a medium price region. Segment 2 succeeds in keeping the
price in a medium range because medium-priced hardware components
have been chosen. This segment distinguishes itself by equipping the note-
book with an operating system though. Segment 3 chooses high-priced
hardware components. As this segment dispenses without any software the
price is kept at a medium Ievel. Segment 4 is an attractive duster as the
willingness to pay of its members is high. The preferred product in this
segment is equipped with high-end components.
As a resume, the 209 users proved to be a homogenous group
(students?). There exist remarkable segments though. The management
may decide to establish a "premium" product, for example. The mining
product configurator yields in a first useful result.
120

Table 4. Cluster Analysis Results

Segment Segm. 1 Segm. 2 Segm. 3 Segm.4
Attribute Level ~Segment Size2 (68) {242 {282 {892
256MB +27%
512MB +1 6% + 15%
14,1"TFT + 15% +23% +1 0%
15" TFT +20%
20GB +26%
30GB +33%
40GB +26% + 15%
8x DVD ROM-Drive +6%
8x4x24 CD RW-Drive +6%
8x4x24 CD RW-Drive
Combo-Drive + 13% +3%
Intel Pentiurn III, 1.2 GHz +10%
Intel Celeron, 1.2 GHz
Intel Pentiurn IV, 1.6 GHz + 10%
Intel Pentium IV, 2.2 GHz + 14%
Win XPHome +9%
Nosoftware + 16%
Win XP Professional +2%
Win XP Home/Works 7.0 +1%
Win XP Pro/Office XP Stand. +2%
Price Range 1250- 1550- 1550- 1700-
1400 € 1700€ 1700€ 1850 €

5 Evaluation and future research

Within the research presented in this article an experimental product con-

figurator has been developed. The configurator is capable of recording
interaction data in the business-to-consumer context. This is a first step
towards exploring the configuration behavior. In order to turn the configu-
rator data into valuable information further research has to be done.
The next step within this project is the application of specific data
mining methods. The objective is to explore patterns in the configurator
usage, for example users' strategies in order to arrive at the final configu-
ration. In order to prove the correctness of the derived results it is neces-
sary to counter check configuration data with real buying data. For
examp1e, it is important to inspect the relation between the final
configurations given in the configurator data and real orders. In a further
 121

step the functionality suggested in Chap. 3 needs to be implemented in a

real configurator.

References

Bensberg F (2001) Web Log Mining als Instrument der Marketingforschung.

Deutscher Universitäts-Verlag, Wiesbaden
Carroll J, Green PE (1995) Psychometrie Methods in Marketing Research: Part I,
Conjoint Analysis. Journal ofMarketing 32: 385-391
Fayyad UM, Piatetsky-Shapiro G, Smyth P (1996) From Data Mining to
Knowledge Discovery: An Overview. In: Fayyad UM et al. (eds) Advances in
Knowledge Discovery and Data Mining. Menlo Park, Cambridge London, pp
1-33
Hagedom J, Bissantz N, Mertens P (1997) Data Mining (Datenmustererkennung):
Stand der Forschung und Entwicklung. Wirtschaftsinformatik 39: 601-612
Harnmann P, Erichson B (2000) Marktforschung. 4th edn. Lucius & Lucius,
Stuttgart
Janetzko D (1999) Statistische Anwendungen im Internet. Addison-Wesley,
München et al.
Kurbel K (1992) Entwicklung und Einsatz von Expertensystemen. 2nd edn.
Springer, Berlin Heidelberg
Lackes R, Schnoedt G (1999) Wissensbasierte Entscheidungsunterstützung bei der
Konfiguration und Angebotserstellung von Produktvarianten. In: Hippner H,
Meyer M, Wilde KD, Computer Based Marketing. 2nd edn. Friedr. Vieweg &
Sohn Verlagsgesellschaft, Braunschweig Wiesbaden
Mertens P, Borkowski V, Geis W Betriebliche Expertensystem-Anwendungen.
3rd edn. Springer, Berlin et al.
Paffrath R (2002) Marktorientierte Planung des Produktsystems: Entwicklung
eines objektorientierten Referenzmodells. Deutscher Universitäts-Verlag,
Wiesbaden
Paffrath R (2002) Ein Referenz-Datenmodell fiir Online-Fragebögen im World
Wide Web. In: Fischer C, Nissen D, Ott I, Schöning S (eds) Fokus
Mittelstand. Peter Lang, Frankfurt a.M., pp 247-265
Multi-Channel Management and its Impact on
Customers' Purehase Behavior

Bernd Skiera

Sonja Gensler

1 Abstract

The aim of this paper is to analyze the effects which additional sales
channels have on customers' purchase behavior. Therefore, a method is
proposed to decompose the impact of additional sales channels on reve-
nues into an up-selling effect, a cross-selling effect and a loyalty effect.
Furthermore, the use of intervention analysis is proposed to separate cus-
tomers' self-selection effects from the effects additional sales channels
have on customers' purchase behavior. The conducted empirical study
demonstrates that a simple comparison of customers' average purchase
behavior does not allow to properly analyze the effects that different sales
channels have on customers' purchase behavior because they do not con-
sider systematic differences of customers that are independent of the sales
channel.

2 lntroduction

Multi-channel management is rapidly gammg importance because the

Internet enables almost every company to establish an additional sales
channel. This additional sales channel might enable companies to save
costs or to realize cross- and up-selling opportunities as well as to increase
customer loyalty. Cost savings have been especially realized by the
banking industry as the Internet allows to integrate the customer in the
production process by, e.g., enabling them to get appropriate information
concerning banking products without costly human interaction. Literature
has labeled this effect the "prosumer effect" because the consumer is inte-
grated into the production process (Vishwanath and Mulvin 2001). Reve-
nue opportunities have been especially outlined in the early times of the
Internet. Knott et al. 2002 and Sonnenberg 1988, for example, highlighted
the opportunities ofthe Internet for up- and cross-selling, i.e., selling either
 123

products with a higher profit margin or additional products via a

sophisticated web-presence. In addition, Garczorz and Krafft 2000, among
others, outlined ways to increase customer loyalty via an appropriate use
of the Internet.
Nevertheless, apart from anecdotal evidence that suggests an increase in
customers' purchase behavior if customers face the Internet as an addi-
tional sales channel (e.g., Sonnenberg 1988), only Hitt and Frei 2002 ana-
lyzed empirically in more detail the revenue generating effects of the
Internet as an additional sales channel in the context of the online banking
industry. They already pointed out that an examination of the effects of an
additional sales channel such as the Internet requires to control for
customer differences that are independent of the sales channels, but might
lead to an observation of different purchasing behavior across sales chan-
nels. Thus, the question is whether higher average revenues of customers
who use online- and offline-channels instead of only offline-channels are
due to the additional online-channel or whether those customers using
online- and offline-channels have always generated higher revenues, even
before starting to use the additional online-channel. The first effect might
be considered a channel effect and the second one a self-selection effect.
Comparing average profits or revenues per customer across the two groups
of customers (e.g., customers using online- and offline-channels versus
customers who use offline-channels only) without controlling for sys-
tematic differences across the customers and attributing differences in
means to the effect the additional online-channel has would show a spu-
rious effect in case ofthe latter.
As a result, the first aim of this paper is to outline a method to decom-
pose the effects additional sales channels have on revenues into an up-
selling effect, a cross-selling effect and a loyalty effect. Second, the use of
intervention analysis is proposed to separate customers' self-selection
effects from the effects additional sales channels have on customers' pur-
chase behavior. Third, the differences that might occur if self-selection
effects of customers are not properly taken into account are analyzed by an
empirical study. Therefore, the rest of the paper is organized as follows:
Section 3 outlines a decomposition method to separate up- and cross-
selling effects from loyalty effects. Section 4 describes methods to take
customer self-selection effects into account and outlines in more detail
how intervention analysis works. Section 5 outlines the aim of the empiri-
cal study and Section 6 describes the results. Section 7 summarizes the
main results ofthis paper.
124

3 Decomposition of revenue effects

The idea of decomposing the revenue into several components, i.e., effects,
is to analyze in more depth why changes in revenues occur. Therefore, the
profit contribution of the i-th customer is defined as the product of the
average pro:fit margin per purchased item of the i-th customer, the number
of items per purchase of the i-th customer, the number of purchases of the
i-th customer and the number of periods the i-th customer is active:
(iEf) (1)

where:
1t;: Profit contribution ofthe i-th customer,
iii;: Averageprofit margin per purchased item ofthe i-th customer,
q; : Averagenumber of items per purchase of the i-th customer,
n;: Number of purchases of the i-th customer per period,
1;: Number ofperiods the i-th customer is active (customer lifetime).
The profit contribution of a customer can be increased by selling items
with a higher profit margin (iii; increases), by selling more items per
purchase ( q; increases), by increasing the number of purchases per period
(n; increases), or, stated differently, by reducing the interpurchase time,
and by increasing the number of periods a customer is active, thus cus-
tomer lifetime (/; increases).
The ability to sell items with a higher profit margin is reflected by the
value of m; and is considered to tak:e up-selling effects into account. The
average number of items per purchase mirrors the cross-selling effect
whereas the number of purchases per period and the customer lifetime
reflect loyalty effects. Knowledge conceming changes in this variables due
to an additional sales channel allows to derive strategic and operational
best practices regarding the channel structure, allowing to determine, e.g.,
if customers should be actively pushed to an additional sales channel or
not.

4 Separation of self-selecting effects and channel

effects

Hitt and Frei 2002 are the frrst toseparate self-selection effects from chan-
nel effects in their study to analyze if more valuable customers utilize
 125

electronic sales channels in the case of PC banking. Their idea is to com-

pare the average value of PC banking customers with those of regular
customers after accounting for observable customer characteristics ( e.g.,
age, income, marital status, home ownership, relationship duration with the
institution). Basically, they consider two different approaches to account
for an influence of these observable characteristics on customer value in
their study using cross-sectional data. The first approach models the
influence of these observable characteristics on the customer value of the
two different groups via linear regression (for explaining profitability of
customers) or logistic regression (for determining the probability of asset
adoption and liability adoption). The second approachmatchesindividual
PC banking customers with individual regular customers on the basis of
their observable characteristics, e.g., same age (nearest 10 years), marital
status, income, home ownership, and relationship duration. This approach
avoids the problern that the functional form chosen for the influence of the
observable characteristics on the dependent variable is misspecified.
However, it is not always possible to find a corresponding regular
customer for all PC banking customers. Based on these two approaches,
Hitt and Frei 2002 show that not accounting for self-selection effects tends
to overstate the incremental value of online banking, so that e-commerce
profitability might be skewed upwardly.
The approach taken by Hitt and Frei 2002 is appropriate as long as there
are only cross-sectional data available containing customer-specific
information such as age, income and marital status. Sometimes, however,
pooled data containing information concerning several customers along
different time periods are available at the cost that customer-specific
information (e.g., age, marital status) is not available. Reasons might be
that customer-specific information is not recorded properly or channels
such as the Internet allows customers to approach companies almost
anonymously, while the Internet also enables companies to collect longi-
tudinal data at lower costs. In such instances, intervention analysis allows
to still disentangle the above-mentioned effects (Shao 1997; Tiao and Box
1981; Box and Tiao 1975; Hanssenset al. 2001, pp. 293; Deleersnyder et
al. 2001). Such an intervention analysis allows to analyze the effects that
an intervention, which occurred at time TBi, has on adependent variable:

(2)

where:
126

Yit: Dependent variable, e.g. sales ofthe i-th customer for the t-th pur-
chase,
DU;i Step dummyvariable which is responsible for a level shift of the
dependent variable of the i-th customer for the t-th and all coming
purchases,
DT; 1: Growth rate dummyvariable which is responsible for a shift in the
growth rate ofthedependent variable ofthe i-th customer for the t-
th and all coming purchases,
DP; 1: Pulsedummy variable which is responsible for a one-time shift for
the dependent variable ofthe i-th customer for the t-th purchase,
eu: Error term ofthe i-th customer for the t-th purchase,
I: Index set of customers,
T;: Indexset ofpurchases ofthe i-th customer,
TB;: Potentialbreakdate ofthe i-th customer.

The lagged first differences are added to ensure that the residual time-
sefies is indeed white noise. The parameter value of b allows to identify a
stationary time-series: If b < 1, the unit-root null-hypothesis is rejected,
thus the time-series is stationary. Otherwise, it is evolving (Deleersnyder et
al. 2001, p. 5, Leeflang et al. 2000, p. 463). In case of a stationary time-
series, the hypotheses tests conceming the other parameters can be carried
out using conventional t-tests (Holden and Perman 1994, Perron 1994).
The dummy variables that capture the different effects of the interven-
tion are defined as follows:
_ {t;-TB;+1 if t ~TB; (iel, teT) (3)
DT;,r- O else

_{1
DU;,r- O l
if t; ~TB; (iel, teT) (4)
ese

1 if t;= TB;
{ (iel, teT) (5)
DP;,r = 0 eise

Due to these definitions of the dummy variables, an intervention analy-

sis is able to capture a permanent Ievel shift in the dependent variable (via
the step dummy variable DU;,t and a significant value for the parameter
c1), a permanent shift in the growth rate ofthedependent variable (via the
 127

growth rate shift dummy variable DTit and a significant value for the
parameter c2), and a temporary shift of the dependent variable (via the
pulse dummy variable DPi,t and a significant value for the parameter c3 )
(Hanssens et al. 2001, p. 293).
Figure 1 demonstrates these effects that are captured by positive values
ofthe parameters cl> c2 and c3 .

y step intervention y pulse intervention y growth rate intervention

L . . . __ _ _ . __ _ _ time
' - - - - - - ' - - - - time '----.......1...--- time
TB TB TB

Fig. 1. Examples for different types of interventions

5 Aim of the empirical study

The aim of the empirical study is to analyze the effects an additional sales
channel has on customers' purchase behavior and to compare the results
that are observed by taking and not taking self-selection effects into
account. Therefore, data from an European retailer who sells products via
call-center and the Internet are considered. Historically, this retailer only
advertised his products via television and sold his products via call-center.
Only recently, this retailer started to use the Internetas an additional sales
channel. Therefore, this retailer has two different groups of customers: One
group of customers who buys products only via call-center and another
group of customers who buys products via call-center and the Internet. 1
Knowledge concerning the different effects of the additional Internet sales
channel is of interest to this retailer because it would give an indication
whether the retailer should aggressively move the call-center customers to
the Internet (e.g., via advertising this sales channel on television or during
customers' waiting time in the call-center) or not.
Data are available for all purchased items during a period of 15 months,
covering several million purchased items from several hundred product

1 A third group, those customers that only buy via the Internet, was too small to
be considered here.
128

categories that were bought by more than one million customers. Unfortu-
nately, information concerning customer-specific characteristics as well as
profit margins of the purchased items is not available. Hence, only the
effects of the additional sales channel on cross-selling and loyalty are con-
sidered, but not the effects on up-selling. Self-selection effects are taken
into account by using intervention analysis and are compared to the results
that are observed by simply comparing the average purchase behavior of
the two groups of customers. Hence, intervention analysis focuses on the
change in behavior a particular customer shows from the point in time the
Internet is used as an additional sales channel (intervention) and can thus
be considered an intra-customer analysis. In contrast, a simple comparison
of an average purchase behavior across two groups compares behavior
across customers and is called an inter-customer analysis.

6 Results of the empirical study

6.1 Evolution of the share of internet purchases

lt is interesting to know which sales channels customers choose after they

have started to use a second sales channel (here the Internet). A dimin-
ishing share over time of the purchases via Internet in comparison to the
total number of purchases via both sales channels would indicate that the
customers are not satisfied with the sales channel Internet and an in-
creasing share would indicate that this sales channel is going to even fur-
ther substitute the existing sales channel over time. Therefore, the evolu-
tion of the share of purchases via the Internet is analyzed in comparison to
the total number of purchases over time after the first month customers
decided to use the Internet as an additional sales channel for all cohorts of
customers that started to use the Internet within the same month. This
approach leads to the share oflnternet purchases shown in Figure 2 for 13
cohorts. A linear regression with the share of Internet purchases as the
dependent variable and time as the independent variable:
sk,g = a+b·monthk,g +ck,g (kEK, gEGk\1) (6)

where:
sk,g: share of Internet purchases of the k-th cohort in the g-th
month after starting to use the Internet,
 129

monthk.g: Number indicating how many months are between the g-

th month and the first month the k-th cohort started to use
the Internet as an additional sales channel,
&k,g: Error term of the k-th cohort for the g-th month after starting
to use the Internet
K: index set of cohorts,
Gk: index set of months the k-th cohort used the Internet as an
additional sales channel.
yields the following results: a = 0.332 (p-value = 0.000), b = 0.001 (p-
value = 0.81), R2 = 0.001, F-value = 0.58, indicating that the share of
Internet purchases is not evolving over time.

0,6
<ll
0)
<ll 0,5
"'...,
..c
<.>
0,4
..,"'
0..

E 0,3
c
0)

....0 0,2
....
0)

"'
..c
(I)
0,1

0
2 3 4 5 6 7 8 9 10 II 12 13 14

month

--+-- cohort_ l --+-cohon_2 cohon_J cohon_4 _._ cohon_5 --+- cohon_6 -+- oohon _7
- cohon_8 - cohon_9 - - cohon_ IO - • cohon_ ll cohon_ l2 -'K-cohon_ l3

Fig. 2. Monthly share of Internet purchases with respect to first month using the
Internet as a sales channel
130

6.2 lnter-customer analysis

To analyze the cross-selling effect of an additional sales channel, a

Student's t-test is used to compare the two groups of customers. Two
samples are drawn from all customers whose first observed purchase took
place during the third month of the observation period. One sample con-
sists of customers who only use the offline-channel during their lifetime (N
= 1631) and the other sample uses the offline-channel as well as the
online-channel (N = 2007).
The following measures are considered to evaluate cross-selling effects:
average number of items per purchase,
number of categories.
The average number of items per purchase provides evidence for the
size of the shopping basket and is a component of equation ( 1). The num-
ber of different categories used by one customer is also of interest to
evaluate the share-of-wallet and, thus, the cross-selling effect. Therefore,
both measures are considered and provide information about the validity of
the results.
Comparing the two groups shows that the customers who use both sales
channels order more items per purchase (see Table 1). Moreover, cus-
tomers who use both channels also purchase from a greater number of
categories, providing additional support for the validity of the results.
Hence, multi-channel management seems to have a positive impact on
cross-selling. One reason might be that multiple channels allow to better
inform customers about the retailer's assortment and therefore, provide the
customers with more opportunities to fulfill their needs.

Table 1. Impact of muJti-channel management on cross-selling (inter-customer

analysis)

Group Average number of Number of categories

items per purchase
Customers using the offline- 1.576 5.220
channel only (N = 1631)
Customers using the offline- 1.890 8.060
and online-channel (N = 2007)
p-value 0.000 0.000
 131

As mentioned above, the profit contribution of one customer cannot

only be increased by increasing the number of purchases but also by
reducing the interpurchase time. Therefore, to evaluate the impact of multi-
channel management on loyalty the number of purchases as weil as the
average interpurchase time are considered, whereas the latter serves as a
measure to validate the results.

Table 2. Impact of multi-channel management on loyalty (inter-customer analy-

sis)

Group Number of purchase Average interpurchase

time

Customers using the offiine- 6.340 50.781

channel only (N = 1631)
Customers using the offline- 8.750 62.139
and on1ine-channel (N = 2007)
p-value 0.000 0.000

Results show that customers who use both sales channels have a higher
number of purchases than customers who only use the offline-channel (see
Table 2). On the other hand, customers who use both sales channels have a
Ionger interpurchase time. These Contradietory results are due to differ-
ences in the active time defined by the time between the last and first
observed purchase of one customer (average active time of customers
using only the offline-channel = 239.65 days, average active time of
customers using the offline- and online-channel = 274.80 da.ys (p-value =
0.000)). Nevertheless, these ambiguous results do not allow a clear state-
ment conceming the impact ofmulti-channel management on loyalty.
To demonstrate the usefulness ofthe decomposition method proposed in
Section 3, the inter-customer analysis is also computed for the number of
items purchased. This variable presents a combination of the average
number of items per purchase and the number of purchases, hence a
combination of the cross-selling effect and the loyalty effect. Table 3
shows the result of this analysis which indicates a positive impact of
offering multiple channels. However, it does not indicate whether this
increase in revenues is driven by a higher loyalty of the customers or by
higher cross-selling opportunities. Multiplying the corresponding results of
 132

Table 1 and Table 2 Ieads approximately to the results of Table 3: 1.576 ·

6.340 = 9.992 ~ 10.044 and 1.890 · 8.750 = 16.538 ~ 16.749).2

Table 3. Comparison regarding the number of items purchased (inter-customer

analysis)

Group Nurober of items purchased

Customers using the offline-channel only 10.044
(N = 1631)
Customers using the offline- and online-channel 16.749
(N = 2007)
p-value 0.000

Summing up the results of the inter-customer analysis, it can be stated

that multi-channel management has a positive impact on cross-selling and
loyalty measured by the number of purchases.

6.3 lntra-customer analysis

To account for self-selection effects requires an intra-customer analysis.

Here the sample consists of 9,790 customers who use both sales channels
and make more than two purchases. An intervention analysis is conducted
as an intra-customer comparison to measure the impact that multiple
channels have on customers' purchase behavior. Intervention analysis
requires data at the individual Ievel. Hence, the number of items per
purchase is considered to evaluate the cross-selling effect of multi-channel
management and the following model is calibrated:
NOIPO,,t = a + b · NOIPO,,t-1 + Ct • DU,,t + cz · DT,,t + CJ · DP,,1 + 8;,1

(iel, teTi\1) (7)

where:
NOIPOi.t: Number of items per purchase of the i-th customer for the t-th
purchase,

2 The differences occur because multiplying with the mean of the average
nurober of items per purchase determines a greater weighting of customers with
few purchases than of customers with many purchases.
 133

DU; 1: Step dummy variable which is responsible for a Ievel shift of

the dependent variable of the i-th customer for the t-th and all
coming purchases,
DT i,t: Growth rate dummy variable which is responsible for a shift in
the growth rate of the dependent variable of the i-th customer
for the t-th and all coming purchases,
DP; 1: Pulsedummy variable which is responsible for a one-time shift
for the dependent variable of the i-th customer for the t-th
purchase,
&;, 1: Error term ofthe i-th customer for the t-th purchase,
1: Index set of customers,
T;: Indexset ofpurchases ofthe i-th customer,
TB;: Potentialbreakdate (here: firsttime oflnternet usage) ofthe i-
th customer.
The results presented in Table 4 show that the time-series is stationary,
thus allowing to use t-tests for testing the significance of the parameters.
The number of items per purchase decreases slightly when multiple chan-
nels are used (-0.043 for the parameter of the step dummy variable, indi-
cating that the customers purchase fewer products). However, the number
of items for the first purchase via Internet is significantly higher than the
average number of items (0.209 for the parameter of the pulse dummy
variable). The growth rate dummy variable has no impact on the number of
items per purchase and is not significant. Hence, in the short-run, the
additional online-channel shows a positive impact on the number of items
per purchase that is reflected in the one-time effect captured by the pulse
dummy variable. In the long run, however, the decrease in the number of
items per purchase, captured by the step dummy variable, offsets this one-
time positive effect (an overallnegative effect should be observed after the
twelfth purchase: 12 · l-0.0431 = 0.516>0.494), indicating a negative
impact of the additional online-channel on cross-selling.
134

Table 4. Results of intra-customer analysis regarding " umber of Items per

Purchase"

Variable Parameter Value p-value

Constant (a) 1.530 0.000
Lag number of items per purchase (b) 0.209 0.000
Step dummyvariable Internet channel (c 1) -0.043 0.000
Growth rate dummy variable ( c2) 0.000 0.070
Pulsedummy variable Internet channel (c3) 0.494 0.000
R 2 = 0.050, F = 1683.192, = 136,596

The interpurchase time at the individual Ievel is used to analyze the

impact of multi-channel management on customer loyalty so that the
following model is calibrated:
IPL,t = a + b · IPTi,t-1 + c, · DU;,r + c2 · DT;,t + C3 · D(TB;,t) + &;,t

where:
IPTi 1: Interpurehase Time of the i-th customer for the t-th purchase.
Results displayed in Table 5 show that the observed time-series is
stationary (b < 1) and the growth rate dummy variable has nearly no
impact on loyalty. The parameter ofthe step dummyvariable indicates that
from the point in time the Internet is used the interpurchase time decreases,
whereas the pulse dummy variable has a one-time positive impact on the
interpurchase time. The parameter of the pulse dummy variable indicates
that the interpurchase time before the Internet is used the first time is
higher than on average. This one-time effect is offset in the long-run by the
effect of the step dummy variable, indicating that multi-channel
management has a positive impact on loyalty measured by the
interpurchase time at the individual Ievel in the long-run.
 135

Table 5. Results of intra-customer analysis regarding "Interpurehase Time"

Variable Parameter Value p-value

Constant (a) 14.025 0.000
Lag interpurchase time (b) 0.186 0.000
Step dummyvariable Internet channel (c 1) -6.225 0.000
Growth rate dummy variable ( c2 ) 0.008 0.000
Pulsedummy variable Internet channel (c 3 ) 33.211 0.000
R2 = 0.111 , F = 3649.75, N = 136,596

By summarizing the results of the intra-customer analysis, it can be

stated that the impact of multi-channel management on customers' pur-
chase behavior seems to be ambiguous: in the long run the impact of multi-
channel management on cross-selling is negative, whereas the impact on
loyalty is positive.

6.4 Comparison of results

A comparison of the results of the inter-customer and the intra-customer

analysis shows that an inter-customer analysis (i.e., a comparison of means
across groups) Ieads to different results for the effects of multiple sales
channels on customers' purchase behavior than an intra-customer analysis
(i.e., intervention analysis).
The inter-customer analysis for the retailer under consideration comes
up with the result that an additional sales channel has a positive impact on
cross-selling, whereas the intra-customer analysis shows that the impact is
negative in the long run (see Table 6). Regarding the loyalty effect, the
inter-customer analysis is ambiguous about the impact, because the impact
measured by the interpurchase time is negative, whereas the impact meas-
ured by the number of purchases is positive. The intra-customer analysis
indicates a positive impact of an additional sales channel on loyalty meas-
ured by the interpurchase time in the long run.
136

Table 6. Comparison of inter-customer analysis and intra-customer analysis

Effect lnter-Customer Analysis Intra-Customer Analysis

Cross- . Multi-channel management has a · In the long run multi-channel

Selling positive impact on cross-selling management has a negative
Effect - Customers who use both channels impact on cross-selling
order more items per purchase and
purchase from a greater number of
product categories
Loyalty- - Impact of multi-channel manage- - In the long run multi-channel
Effect ment on loyalty is ambiguous management has a positive
- Customers who use both channels impact on loyalty
have a Ionger interpurchase time,
but make a higher number ofpur-
chases

7 Summary

This paper proposed intervention analysis to analyze the impact of multi-

channel management on customers' purchase behavior. This intervention
analysis allows to separate channel effects from customers' self-selection
effects. In addition, a decomposition method was proposed to attribute
changes in sales to up-selling effects, cross-selling effects or loyalty
effects. The empirical study showed that an inter-customer analysis that
compared purchase behavior across two groups of customers (one group
who only uses the offline-channel and another group who uses the offline-
and online-channel) led to different results than an intra-customer analysis
on the basis of intervention analysis. In accordance to the results of the
study from Hitt and Frei 2002, it was shown that an inter-customer analy-
sis tended to overstate the incremental value of an additional online-chan-
nel. Hence, managers run the risk of overstating e-commerce profitability
if they do not take customers' self-selection effects into account.
 137

References

Box GEP, Tiao GC (1975) Intervention Analysis with Applications to Economic

and Environmental Problems. Journal of the American Statistical Association
70:70-79
Deleersnyder B, Geyskens I, Gielens K, Dekimpe MG (2001) How Cannibalistic
is the Internet Channel? ERIM Report Series (ERS-2002-22-MKT). Erasmus
Universiteit Rotterdam (NL)
Garczorz I, Krafft M (2000) Wie binde ich den Kunden? - Kundenbindung.
eCommerce. Einstieg, Strategie und Umsetzung im Unternehmen. In: S.
Albers, M. Clement, K. Peters and B. Skiera (eds) eCommerce. Einstieg,
Strategie und Umsetzung im Unternehmen. F.A.Z.-Institut, Frankfurt am
Main, pp 147-160
Hanssens DM, Parsons LJ, Schultz RL (2001) Market Response Models: Econo-
metric and Time Series Analysis. Kluwer Academic, Boston et al.
Ritt LM, Frei FX (2002) Do Better Customers Utilize Electronic Distribution
Channels? The Case ofPC Banking. Management Science 48:732-748
Holden D, Perman R (1994) Unit Roots and Cointegration for the Economist.
Cointegration for the Applied Economist. In: B. Bhaskara Rao (eds) Cointe-
gration forthe Applied Economist. New York, pp 47-112
Knott A, Hayes A, Neslin SA (2002) Next-Product-to-Buy Models for Cross-
Selling Applications. Journal of Interactive Marketing 16:59-75
Leeflang PSH, Wittink DR, Wedel M, Naert PA (2000) Building Models for Mar-
keting Decisions. Kluwer, Dordrecht
Perron P (1994) Trend, Unit Root and Structural Change in Macroeconomic Time
Series. Cointegration for the Applied Economist. In: B. Bhaskara Rao (eds)
Cointegration for the Applied Economist. New York, pp 113-146
Shao YE (1997) Multiple Intervention Analysis with Application to Sales Promo-
tion Data. Journal of Applied Statistics 24:181-191
Sonnenberg F (1988) The Power of Cross-Selling. The Journal of Business
Strategy 9:56-59
Tiao GC, Box GEP (1981) Modeling Multiple Time Series with Applications.
Journal ofthe American Statistical Association 76:802-816
Vishwanath V, Mulvin G (2001) Multi-Channels: The Real Winner in the B2C
Internet Wars. Business Strategy Review 12:25-33
 Section 3

Production and Procurement

E-Business in Production and Procurement -
Some Theoretical Perspectives

Ronald Bogaschewsky

1 New economy - macroeconomic considerations

The term New Economy was coined by the business press grasping two
broad trends in world economy (Shephard 1997, cited in Pohjola 2002a, p.
134): First, globalization ofbusiness and second, revolution in information
and communication technology {ICT). A more specific definition is given
by the US Department of Commerce: The New Economy is "an economy
in which IT and related investments drive higher rates of productivity
growth" {Temple 2002, p. 242).
In order to evaluate, if IT investments and IT use led to higher rates of
productivity growth in the last years, it is helpful to take a look on
statistical data of productivity growth accounting. As Temple (2002, p.
243) shows, the productivity growth was rather slow over the last decade
as a whole. At least the rate was unremarkable by the standards of the
1980s and lower than in the early 1960s. We can detect faster growth in
the second half of the 1990s, where the average growth rate was 2.4%
(1995:4th quarter - 2001:4). This figure is about one percentage point
higher than in the preceding 20 years, but less than during 1950-1972
(2.7%).
Table 1 shows both, productivity growth and real GDP growth over dif-
ferent periods of time for the U.S.A. When analyzing the data, it can be
stated that neither productivity growth nor GDP growth was extraordinary
even in the second half of the 1990s when compared to the figures of ear-
lier decades, especially the 1950s and 1960s. Furthermore, it has to be
taken into account that the U.S. uses hedonic pricing, where quality
aspects of products accrue to growth rates. This adds about 1% to the
growth figures in the 1990s (Welfens 2002).
142

Table 1. Real GDP growth and productivity growth during different time periods
Real GDP growth (%) Productivity growth (%)
1950:2 - 1972:2 3.9 2.7
1972 :2 - 1995 :4 2.9 1.4
1995:4 - 2001:4 3.5 2.4
(Temple 2002, p. 243)

In addition to this data, it has to be mentioned that output volatility was

declining since the mid-1980s due to consumption smoothing, higher sta-
bility in capital goods production, and better inventory management
(Temple 2002, pp. 252ff.). Therefore, effects measured by growth
accounting might partly accrue to these factors.
To grasp the role ICT played for both, GDP and productivity growth we
have to take a more detailed look at the field of interest: GDP growth
boosted from 2.5% at the start of the last decade to 4.5% at its end. This
could possibly be explained by the fact that it takes some time before ICT
investments (hardware, software, communications equipment) are getting
into effective use. Furthermore, ICT is a general purpose technology which
shows a rather long Iasting impact on the potential for economic growth.
Over time ICT capital deepening rose and does accrue to 6.3% of nominal
gross income today (Jalava and Pohjola 2002).
The role of ICT becomes clearer when perceiving that 2/3 of the 1%
step-up in labor productivity growth is due to growing use ofiCT and effi-
ciency improvements in computer production (Oliner and Sichel 2000).
Jalava and Pohjola (2002) point out that the benefits from ICT use out-
weigh those ofiCT production.
An interesting share of growth is usually contributed by Multi-Factor
Productivity (MFP), also known as Total Factor Productivity (TFP). The
growth of this figure shows the extent to which inputs are being combined
more efficiently. The growth rate can be calculated by subtracting the
growth contributions of the input factors from the total output growth. This
rate was rarely higher than 1% in the U.S.A. Many authors agree to
Jorgenson (2001), who claims that half oftbis is due to efficiency gains in
ICT production over 1995-9.
Table 2 shows the output growth contributors in the U.S.A. related to
different periods oftime (numbers may not add to totals due to rounding).
 143

Table 2. Output growth contributors in the U.S.A.

1974- 1990 1991- 1995 1996 - 1999
Output growth 3.1 2.8 4.8
Contributions from:
ICT capital 0.5 0.6 1.1
Hardware 0.3 0.3 0.6
Software 0.1 0.3 0.3
Comm. quip. 0.1 0.1 0.2
Other capital 0.9 0.4 0.8
Labor hours 1.2 0.8 1.5
Labor quality 0.2 0.4 0.3
MFP/TFP 0.3 0.5 1.2
(Jalava and Pohjola 2002, p. 194 based on Oliner and Sichel 2000)

As Table 2 shows, ICT capital and TFP are responsible for 2.3 per-
centage points ofthe 4.8% output growth in 1996-9. As mentioned above,
more than half of TFP growth is due to efficiency gains in ICT production
and because of ICT use.
Table 3 focuses on the contributions of different factors to Iabor
productivity growth (numbers may not add to totals due to rounding).

Table 3. Labor productivity growth contributors in the U.S.A.

1973- 1990 1990- 1995 1995- 1999
Labour roductivi rowth 1.26 J.l9 2.11
Aggregate contributions of:
ICT capital deepening 0.35 0.43 0.89
on-ICT capital deepening 0.44 0.21 0.35
ICT production TFP 0.19 0.25 0.50
on-ICT production TFP 0.06 -0.01 0.25
Labor g,uali!X 0.22 0.32 0.12
(Temple 2002, p. 248 based on Jorgenson 200 I)

To the rise from 1.19% in 1990-5 to 2.11% in 1995-9 ICT capital deep-
ening accrues 0.89% and ICT production TFP 0.5%. Therefore, the total
effect of ICT investment and ICT use of 1.39% dominates the contribu-
tions of the other factors.
Despite the not extraordinary growth figures, the U.S.A. enjoyed a
remarkable economic success in the 1990s compared to the rest of the
world. This success included a faster productivity growth, stability of
inflation, very low unemployment and a reduction in output volatility.
From the analysis above it can be said that this success can be partly
attributed to improvements in ICT production and more efficient ICT use,
heavy ICT investments (capital deepening) combined with even faster
144

decrease of relative ICT prices (from 3- to 2-year lifecycle in semicon-

ductors from 1996 in addition to Moore's Law). Nevertheless, we see a
current slow-down with a labor productivity growth rate which fell from
3.3% in 2000 to 1.9% in 2001, adding quite a bit water to the wine.
When looking at the European Union we see a very different picture. As
Daveri (2002, p. 355) states, there is no correlation between ICT invest-
ment share of GDP and the rate of labor productivity growth (1996-2001
with respect to 1990-5). Slow adopters of ICT such as Ireland or Greece
enjoy rather high labor productivity growth rates despite unremarkable or
even low ICT investments, while early adopters such as Sweden or Finland
have negative growth rates for labor productivity.
It can be stated that the EU as a whole is lagging behind the U.S.A. in
terms of adoption of ICT. Labor productivity halved from 2.5% (1990-5)
to 1.3% (1995-2000) (van Ark 2002) and is today near stagnation (2001:
0.4%) (Daveri 2002). However, hedonic pricing which is not applied in the
EU would add ~ 1% to the growth figures (Welfens 2002) resulting in a
slightly better picture of the economic situation.
The saying ofRobert Solow (1987) might still be true, especially for the
EU: "We see the computer age everywhere but in the productivity statis-
tics." When looking for possible reasons why there is no (sustainable)
effect we should consider that (OECD 2002, p. 58; Temple 2002, pp.
257ff.; Pohjola 2002b, pp. 380ff.):
- ICT capital share of total capital stock might be still too small to show
larger effects,
- there is a time lag between adoption of new technology and its efficient
use,
- we have market imperfections (EU: competition, ICT prices, labor), and
- heavy users of ICT are not leading producers and vice versa (except
U.S.A.).
Due to these arguments it can be stated that " ... most of the old economy
rules are still valid." (van Ark 2002, p. 1). If this might be frustrating or
not, it is essential to analyze how worldwide competition is affected by the
different growth rates in the U.S. andin the EU. An additional explanation
for the comparably late adoption of ICT in the EU can be seen in the fact
that medium- to low-tech producers may adopt high-tech ICT later without
loosing competitive advantage, thus avoiding leaming costs of first
movers. In this light the EU could be better off than the statistics say, but
this could be just wishful thinking.
 145

2 New economy - the firm's view

2.1 Definitions of e-commerce and e-business

For private companies the New Economy has a quite different meaning
compared to the arguments of macroeconomists. Before discussing this
issue we should make clear how we define E-Commerce and E-Business.
These two terms are anything but well defined in theory and practice. Our
working defmition of E-Commerce is separated in a more aggregated and a
business view, respectively.
In the aggregated view of E-Commerce entities of the economic net-
work (Electronic Businesses, governmental authorities, and consumers)
interact and process transactions facilitated by electronic means.
The firm as a specific single entity in this network views E-Commerce
as a part of its connections to and transactions with external entities. Once
again, these connections and transactions are facilitated by electronic
means, especially by modern ICT and more specific by internet tech-
nology. We can differentiate types of partner-specific relations: B2B -
Business-ta-Business, B2C- Business-to-Customer, and B2G- Business-
to-Government.
The term Electronic Business is used in this article to describe the entire
company, its functional areas, its processes, its organization and its
management systems, whereas the processes are facilitated by ICT, espe-
cially by internet technology. This view includes the (more strategic) E-
Management, E-Administration, E-Finance, E-Logistics, E-Marketing/Sal-
es, E-Production, and E-Procurement. It should be noted that this defini-
tion partly implies a functional view of the firm that can be questioned but
is of no relevance here.
Major challenges for a company to transform into an E-Business on the
one band and to participate in E-Commerce on the other are as follows:
1. Transforming to an E-Business: Reorganizing internal processes by
using (modern) ICT for more efficiency and more customer orientation;
qualification and motivation of human resources to support new goals
and strategies.
2. Parficipate in E-Commerce: Reorganizing external processes and
connecting them with internal processes by using (modern) ICT.
Develop and pursue new market strategies for new market opportunities.
3. Taking advantage of globalization: Internationalization strategies in
marketing and procurement.
146

4. Outsourcing of non-core processes: Redueed transaetion eosts and

availability of qualified suppliers may spur outsoureing aetivities.
For the reorganized "lean" and eustomer-oriented firm new opportuni-
ties may arise:
1. Faster, morefrequent product innovations: Shorter develop-to-market
times due to both, earlier pereeption of market I eustomer needs and
more efficient development proeesses. First mover advantages
eompared to "old eeonomy" eompanies.
2. More customer-tailored products and customer-driven processes:
Knowing eustomers needs better beeause of eonneeting to the market by
eleetronie means. Make-to-order proeesses and mass eustomization due
to more flexible proeesses.
3. Reduced manufacturing and inventory costs: Optimized intra- and inter-
organizational proeesses, espeeially strategie allianees along the value
ehain and eollaborative eommeree.
4. Positive network effects due to open and widely aeeepted standards.
Even though the potentials of the E-Business and E-Commeree are not
easy to realize and despite the faet that internet-based solutions are not
very weil aeeepted today, quite a few eompanies are very sueeessful by
applying this teehnology. However, when looking on the behavior of the
single businesses there seems tobe no entirely new eeonomy, so far. From
looking at the firms with sueeessful applieations of modern IT there might
be signs for the foundation of a new eeonomy. But it will take at least one
deeade until the majority of eompanies will adopt these new applieations,
leading to more effieient and more effeetive businesses.

2.2 E-procurement

Looking at Eleetronie Proeurement, three dominating areas where new IT

is applied today ean be identified (Bogasehewsky 2002):
- Desktop Procurement Systems and eCatalogs: Reorganizing proeure-
ment and transaetion proeesses for standard eatalog-based produets (e.g.
MRO-goods- Maintenanee, Repair, and Operations).
- Electronic market places: Automatie and standardized distribution of
Requests for Information (RFI), for Proposal (RFP), for Quotation
(RFQ), for Bids (RFB), standardized distribution of produet offers,
 147

information and communication platform, Reverse and Forward

Auctions for capital goods, general agreements, direct material.
- Virtual platforms and electronic networks: Collaborative applications
such as Supply Chain/Network Management (SCM/SNM), Simulta-
neaus Engineering (SE), Vendor Managed Inventory (VMI), Collabora-
tive Planning, Forecasting, and Replenishment (CPFR).
The major effects of these applications are expected in:
- a sharp reduction in process I transaction costs (main objective)
- enhanced competition (higher level of market transparency and more
open markets)
- process optimization in networks (reduced costs and inventories due
to collaboration)
However, today some major obstacles have to be taken into account:
- Reductions in process costs don't necessarily Iead to lower total costs
The cost of capacities (especially labor in this case) are fixed in the mid-
term (remanence). Only if personnel can be either used in other value
adding activities or be laid off, costs can be cut down. Therefore, produc-
tivity gains often cannot be transferred into lower costs. In addition to that
many potentials cannot be realized, because the internal (non-IT) precon-
ditions are not fulfilled. This includes company-wide spend analysis for
pooling activities, structural reorganization, and motivation of the staff.
Obviously, a lot of "homework" has tobe done first.
- Electronic market places are ignored by many
It is a fact, that many suppliers fear market transparency and the enhanced
competition followed by this. Therefore, these companies refuse to partici-
pate in competition enhancing market activities. This places a major
obstacle to the successful implementation and usage of electronic markets.
It is also a proof for the hypothesis that a lot of markets have imperfections
or are not really competitive. Another problern is that many procurement
professionals fear to disclose bad practices in the past when having big
successes by using virtual market places. On top of that they are not
willing- or don't have the budget- to pay additional intermediaries.
- Collaborative networks don't work (weil)
lt must be stated that collaborative networks are very complex to plan,
organize, and implement. Different strategies in management and IT have
to be harmonized. IT -systems have to work together or must be connected
to a new platform. Furthermore, collaboration on the operationallevel can
148

only be reached when the companies implement a strategic partnership.

The latter is not easily done and may take several years to develop.
Summarizing the potentials and obstacles as seen today of E-Procure-
ment it can be stated that this part of the E-Business concept is partly day-
to-day business in some - especially in big - companies, whereas it has not
been adopted to a !arger extent by the majority of small and medium sized
companies (SMEs). The great potentials of collaborative networks are still
at the concept level and only a few realizations, most of them in a proto-
type stage, can be seen today.

2.3 E-production

In the field of Electronic Production a number of new ideas exist that are
partly in the development state of prototyping. The most promising appli-
cations today are:
- Virtual platforms and electronic networks: Applications such as
SCM/SNM, SE, VMI, and CPFR do not only address procurement issues
as stated above, but change the way companies act in production and
logistical planning and scheduling.
- Enbanced ERP-Systems/Advanced Planning and Scbeduling-Systems
(APS): The further development ofERP-Systems and of APS, respectively
will give new opportunities to optimize the value chain. These applications
are closely related to the concepts of Supply Chain/Network Management.
On the concept level a closer integration of Customer Resource Manage-
ment (CRM), Production Planning and Scheduling and Supplier Relation-
ship Management (SRM) can be seen but is still to be implemented in
practice.
- Agent-based planning, scbeduling and control: New and innovative
applications more and more use "intelligent" software agents for a wide
range of tasks, e.g. negotiations for jobs or capacities between machines
can be facilitated by this technique.
- Macbine-to-Macbine and Macbine-to-Business Communication: The
intemet makes it easier and even less costly to control, configure or main-
tain machines and facilities from remote. A very interesting development
is the concept of self-controlled machine networks, possibly applying
agent technology. Furthermore, it can be thought ofthe vertical integration
of the firm from business planning to the shop floor facilitated by
electronic means (Kracke 2002).
 149

The expected effects of these applications are:

- Sharp reduction in process I transaction costs in the network due to
process optimization
- Customer oriented lean processes due to implementation of the
customer pull principle that minimizes cycle time, assumed that flexible
capacities (equipment and staff) is available. More transparency of
marginal costs because of real time knowledge of the actual capacity
situation, better planning and application of agent technology.
- Sound planning and control due to availability of reliable data on a
real time basis. Transponder technology might be applied to track work-
in-process and inventories. Hierarchical planning and other advanced
techniques are more widely used instead of sequential planning.
The potentials of E-Production seem to be very promising. However,
only very few applications can be seen today and it will take several years
before noticeable advances will be made in practice.

3 Modern theories of the firm and their imperfections to

explain the "New Economy"

The traditional theory ofthe firm is based in economics and market theory.
Business Management as a scientific discipline puts a focus on both, the -
more general - nature of the firm and the specific questions that arise
inside a company (Schmidt 2000). "Modem" theories of the firm as we
define it include transaction cost economics, agency theory, and property
rights theory as well as more specific approaches. A - not comprehensive -
set of popular theories is addressed in brief below. Each theory- as fasci-
nating as it might be - has substantial weaknesses when trying to explain
the "New Economy".
Transaction Cost Economics (Williamson 1985) became very popular
for discussing problems in and between companies. Taken transaction
costs into account in addition to production costs gives new insights on
how to determine best solutions about the govemance structure, including
make-or-buy decisions, Co-operations, etc. However, the theory cannot
fully explain what happens in practice due to the complexity of processes.
It falls short in considering restrictions such as organizational barriers, it
ignores sunk costs and it is, as Ossadnik et al. (200 1) point out, not always
compatible to empirical findings.
Agency Theory (Jensen and Meckling 1976) is another approach that
has been very popular in management theory in the last two decades. Since
150

the principal cannot be sure that his agent actually does what he wants him
to do and the agent may hide information and his intentions from the
principal, agency costs arise for setting up institutional arrangements and
for the results of opportunistic behavior of the agent. These aspects hold
true in any economy and are of great importance. Nevertheless, they just
address "one general" aspect.
Property Rights Theory (Alchian and Demsetz 1972) is related to
Agency Theory. It argues that ownership and control of a company could
be in one hand when considering economic efficiency. This is due to the
fact that a central monitor for all business activities is needed, who will not
shirk his responsibilities when he receives the residual profit.
Both theories, Agency and Property Rights, are excellent for explaining
the behavior of owner and non-owner or principal and agent. Therefore,
the theories give assistance when looking for basic assumptions and hints
on how to organize the firm. Furthermore, they put special attention on the
importance of leadership and motivation of personnel (without going into
details on how to do that). For explaining complex problems such as how
businesses ofthe "New Economy" should act they are far too basic.
The general Theory of Incomplete Contracts (Grossman and Hart
1986; Hart 1995) has been applied to specific problems such as deter-
mining the optimal nurober of suppliers or to buyer-supplier relationships
(Bakos and Brynjolfsson 1993a, 1993b). The key issue is that buyers may
maximize profits by limiting their options and reducing their bargaining
power, while suppliers invest in "non-contractibles" (innovation, respon-
siveness, etc. ). This may lead to partnerships or a "move-to-the middle" in
the hierarchy-market continuum. Obviously, this theory is helpful in
explaining specific issues regarding the selection of optimal institutional
arrangements, thus expanding the analysis based on transaction cost eco-
nomics. But still, the approach is far to general in order to explain the
"New Economy" entirely.
Network Theory (Sydow 1992; Klein 1996) gives insight into the
organizational and strategic aspects of a set of cooperating companies. lt is
rather broad and eclectically, describing the different aspects of networks
without going into details on how to decide and how to implement specific
measures. It has no detailed process focus. Furthermore, the theory usually
does not take into account that companies are part of different networks at
a time (Fleisch 2001, p. 97). Both, the more detailed view and the process
focus are necessary when looking at E-Businesses and E-Commerce. In
addition to that, E-Businesses are typically part of more than one network.
Therefore, Network Theory may deliver just another part of a comprehen-
sive theory of"New Economy" firms.
 151

Network Economics (Kelly 1998; Arthur 1994) focus on microeco-

nomic considerations such as increasing returns, path dependence, lock-in
effects, multiple equilibrium, and network extemalities. According to
Arthur (1990) increasing returns to scale can be seen in economies that
rely on knowledge-based products and he coins these "new" or "positive
feedback economies" as a counterpart to the conventional economies with
decreasing returns that are based on (natural) resources. Increasing returns
to scale on the producers side can be seen in connection with digital
products that have constant and very low marginal costs. On the users side
we see (positive) network extemalities (Shapiro and Varian 1999) when
each additional user adds benefit to all users already taking part in the net-
work. Path dependence means that "a minor or fleeting advantage or a
seemingly inconsequentiallead for some technology, product or standard
can have important irreversible influences on the ultimate market alloca-
tion of resources" (Liebowitz and Margolis 2002, p. 1). Lock-in effects are
situations when the costs of switching to a different, better technology are
higher than sticking to the inferior (old, but more profitable) technology.
These considerations are very helpful when trying to explain what
happens in the "New Economy". After all, they are more basic models to
explain empirical observations and as Liebowitz and Margolis (2002, p.
14) point out, Network Economics "suggests the importance of communi-
cation, planning, property and other market institutions ... which are essen-
tial elements in any explanation of the actual workings of the economy".
Coordination Theory (Malone et al. 1987, 1999) delivers a "set of
principles" to describe and solve coordination problems by considering
objectives, agents, activities, and resources (Malone 1987) and takes a
process-oriented view (Crowston 1994b, p. 5), explaining organizational
change facilitated by (new) electronic means (Crowston 1994a, p. 6).
Coordination is seen as the management of dependencies, e.g. between
agents that have the same objective or between activities relying on the
same resources (Malone and Crowston 1994). Even though Coordination
Theory combines different aspects of several theories that are relevant for
describing and solving the problern of how to explain firms in the "New
Economy", it still has a restricted focus on organizational and IT issues
(Fleisch 2001).
Behaviorism (Simon 1959; Cyert and March 1963) has been a widely
accepted theory in a certain domain of business science. The high
relevance of people and of personnel as human assets and owners of
crucial knowledge is unquestionable. Furthermore, the organization as a
whole has to be seen as a system influenced by human beings and acting as
an organic entity that e.g. learns as a whole. Behavioral theories might be
an important building block for explaining the complexity of the "New
152

Econorny", but it falls short to explain the entire problern all alone. As
Dibbem et al. (200 1, p. 681) point out, behavioral theories give no recorn-
rnendations on how and when to adapt to new environmental situations.
The Resource-based Theory (Penrose 1959) focuses the intemal physi-
cal, human, and organizational resources as a crucial source for econornic
success. In a way it is opposed to rnarket-oriented theories (Porter 1980,
1985) that have a rnore extemal focus (suppliers, custorners, poten-
tial/actual cornpetitors, potential substitute products). Resources that are
highly relevant for the cornpetitive position of the firm have been coined
core cornpetencies (Prahalad and Hamel 1990). The theory helps to iden-
tify rnarket strategies that are backed by the firm's cornpetencies. The
latter should be anchored in the organizational structure and the rnanage-
rnent systern, so that they cannot be easily copied by cornpetitors. If
intemal cornpetencies and rnarket opportunities do not fit, the resources
have to be developed to rnatch rnarket needs. Due to the high irnportance
of cornpetencies this approach has close relationships to the knowledge-
based view of the firm and to knowledge rnanagernent (Probst et al. 1998)
which focuses the non strategic level as weil. Beside other shortcornings,
these theories have no explicit focus on processes and behavioral assurnp-
tions and fall short on explaining network-related questions as can be seen
in the "New Econorny".
Information Economics (Hirshleifer 1971; Kaas 1995) is based on the
problern of information asymmetries that are a constitutiona1 elernent of
economic relationships. In order to overcome this problem, screening -
e.g. of the rnarket - and signalling - e.g. frorn vendor to buyer and vice
versa- can be used (Weiber 2002). As Hummel (2002, p. 723) pointsout
the theory ignores quality aspects of information and is therefore, beside
other shortcornings, not suitable to answer the questions of the "New
Econorny" alone.
It can be stated that other theories not rnentioned here - such as the
traditional theory of the firm that sees the firm as a production function,
neo-classical approaches or evolutionary theories (Knudsen 1995) - that
give no recommendations for strategic rnake-or-buy decisions (Winter
1987)- arealso not able to explain the "New Econorny" in its entirety.

4 Conclusions

lt can be sumrnarized that none of the theories rnentioned above can be

seen as a cornprehensive Theory of the Firm when looking at the E-Busi-
ness and E-Commerce as defined above. Each theory either takes a rather
 153

specific perspective or is very general and therefore basic. General con-

siderations and specific views have to be combined. Therefore, the
different Ievels of aggregation that are chosen by the theories have to be
connected. Empirical findings show that companies do not always act
according to theories. This is due to the fact that reality consists of (very)
heterogeneaus entities. A single theory either has to concentrate on the
"average" case, thus missing to grasp "standard deviation" or has to
choose a narrow focus, thus ignoring relevant empirical facts.
From experience it can be said that the following aspects are crucial in
the modern E-Business:
- time-based strategies,
- knowledge (management) and competencies,
- individual and organizationalleaming,
- process management,
- project management,
- trust and incentives,
- specificity of cases.
Modem theories of the firm should lay a special focus on these prob-
lems. This gives a hint that multi-theoretical approaches might be most
promising in explaining what happens in practice. What we see in practice
today is quite possibly different from what we will see when companies
made some steps on the leaming curve of applying e-technologies. This
Ieads to the question if it is the right time today in trying to derive new
theories or verify existing theories from empirical data. According to the
insights the author gained from empirical investigations, companies today
are still in a trial-and-error process regarding the usage of intemet tech-
nology and the transformation to an E-Business. Therefore, theories about
the "New Economy" on a micro-level can only be developed by rational
thinking, by combining pieces of different economic theories and - last,
not least- by a big chunk of common sense.

References

Alchian A, Demsetz H (1972) Production, Information Costs, and Economic

Organization. American Economic Review 62: 777-795
Arthur B (1990) Positive Feedbacks in the Economy. Scientific American 262: 92-
99
Arthur WB (1994) Increasing Returnsand Path Dependence in the New Economy,
AnnArbor
154

Bakos JY, Brynjolfsson E (1993) Information Technology, Incentives and the

Optimal Number of Suppliers. Journal of Management Information Systems,
Fall1993 (http://ccs.mit.edu/papers/CCSWP135.html accessed on 4.12.2002)
Bakos JY, Brynjolfsson E (1993b) From Vendors to Partners: Information
Technology and Incomplete Contracts in Buyer-Supplier Relationships. In:
Journal of Organizational Computing, (http://ccs.mit.edu/papers/-
CCSWP15.html accessed on 4.12.2002)
Bogaschewsky R (2002) Electronic Procurement - Katalog-basierte Beschaffung,
Marktplätze, B2B-Netzwerke. In: Hoppe U and Gabriel R (eds): Electronic
Business, Berlin et al., pp 23-43
Crowston K (1994a) A Taxonomy of Organizational Dependencies and
Coordination Mechanisms, MIT, Center for Coordination Science, Technical
Report 174 (http://ccs.mit.edu/papers/CCSWP174.html, accessed on
14.01.2003)
Crowston K (1994b) Electronic communication and new organizational forms: A
coordination theory approach, MIT, Center for Coordination Science,
Technical Report 175 (http://ccs.mit.edu/papers/CCSWP175.html accessed on
14.01.2003)
Cyert RM, March JG (1963) A Behavioral Theory ofthe Firm, Englewood Cliffs,
N.J.
Daveri F (2002) The new Economy in Europe, 1992-2001. In: Oxford Review of
Economic Policy 18: 345-362
Dibbern J et al. (200 1) Die Theorie der Unternehmung als Erklärungsansatz fiir
das selektive Outsourcing der Informationsverarbeitung. Entwicklung eines
theoretischen Bezugsrahmens. Zeitschrift fur Betriebswirtschaft 71: 675-700
FleischE (2001) Das Netzwerkunternehmen, Berlin et al.
Grassman S, Hart 0 (1986) The Costs and Benefits of Ownership: A Theory of
Vertical and Lateral Integration. Journal ofPolitical Economy 94: 691-719
Hart 0 (1995) Firms, Contracts, and Financial Structure, Oxford
Hirshleifer J (1971) The private and social value of information and the reward to
inventive activity. In: American Economic Review 61: 561-574
Hummel J (2002) Auswahl und Gestaltung transaktionsorientierter Geschäfts-
modelle im Internet - Eine Betrachtung aus Sicht der neuen Institutionen-
ökonomie. Zeitschrift fur Betriebswirtschaft 72: 713-733
Jalava J, Pohjola M (2002) Economic Growth in the New Economy: evidence
from advanced economies. Information Economics and Policy 14: 189-210
Jensen M, Meckling W (1976) Theory ofthe Firm: Managerial Behavior, Agency
Costs and Ownership Structure. Journal ofFinancial Economics 3: 305-360
Jorgenson DW (2001) Information Technology and the US Economy. American
Economic Review 91: 1-32
Kaas KP (1995) Informationsökonomik In: Tietz B, Köhler R, Zentes J (eds):
Handwörterbuch des Marketing, 2nd ed., Stuttgart, pp 971-981
Kelly K (1998) New Rules for the New Economy, New York
Klein S ( 1996) Interorganisationssysteme und Unternehmensnetzwerke,
Wiesbaden
 155

Knudsen C (1995) Theories of the Firm, Strategie Management, and Leadership,

in: Montgomery, C.A. (ed): Resource-Based and Evolutionary Theories ofthe
Firm: Towards a Synthesis, Boston-Dordrecht-London
Kracke U (2002): Business-to-Machine Communication- weltweite Vemetzung
von Automatisierungstechnik fiir Electronic Business, in: Hoppe U and
Gabriel R (eds): Electronic Business, Berlin et al., pp 45-66
Liebowitz SJ, Margolis SE (2002): Path Dependence, Lock-In, and History
(http://www. utdallas.edu/-liebowit/paths.html accessed on 6.12.2002)
Malone TW (1987) Modeling coordination in organizations and markets.
Management Science 33: 1317-1332
Malone TW et al. (1987) Electronic Markets and Electronic Hierarchies: Effects
of Information Technology on Market Structure and Corporate Strategies.
Communications ofthe ACM 30:484-497
Malone TW et al. (1999) Tools for inventing organizations. Management Science
45:425-443
Malone TW, Crowston K (1994) The Interdisciplinary Study of Coordination.
ACM Computing Surveys 26: 87-119
OECD (2002) OECD Economic Survey: Netherlands
Oliner SD, Sichel DE (2000) The resurgence of Growth in the Late 1990s: Is
Information Technology the Story?. Journal of Economic Perspectives 14: 3-
22
Ossadnik W et al. (2001) Transaktionskosten bei Zulieferbeziehungen in der
Automobilindustrie. Zeitschrift fiir Betriebswirtschaft 71: 869-891
Penrose ET (1959) The Theory ofthe Growth ofthe Firm, New York
Pohjola M (2002a) The New Economy: facts, impacts and policies. Information
Economics and Policy 14: 133-144
Pohjola M (2002b) The new Economy in Growth and Development. Oxford
Review ofEconomic Policy 18: 380-396
Porter ME (1980) Competitive Strategy, New York-London
Porter ME (1985) Competitive Advantage, New York-London
Prahalad CK, Hamel G (1990) The Core Competence ofthe Corporation. Harvard
Business Review 68: 79-91
Probst et al. (1998) Wissen managen. Wie Unternehmen ihre wertvollste
Ressource optimal nutzen, 2nd ed., Frankfurt/M.
Schmidt RH (2000) Erich Gutenberg and the Theory ofthe Firm, in: Albach, H. et
al. (eds): Theory ofthe Firm, Berlin et al., pp 3-39
Shapiro, C.Narian, H.R. (1999): Information Rules: A Strategie Guide to the
Network Economy, Boston, MA
Shephard SB (1997) The New Economy: what it really means. Business Week,
17.11.1997
Sirnon HA (1959) Theories of Decision Making in Economics and Behavioral
Science. American Economic Review 49: 253-283
Sydow J (1992) Strategische Netzwerke, WiesbadenTemple J (2002) The
Assessment: The new Economy. Oxford Review ofEconomic Policy 18: 241-
264
156

van Ark B (2002) Measuring the New Economy: An International Comparative

Perspective. Review oflncome and Wealth 48: 1-14
Weiber R (2002) Herausforderung Electronic Business -Mit dem Informations-
Dreisprung zu Wettbewerbsvorteilen auf den Märkten der Zukunft, in:
Weiber, R. (eds): HandbuchE-Business, Wiesbaden, pp 1-37
Welfens PJJ (2002) Intemet-economics.net- Macroeconomics, Deregulation, and
Innovation, Berlin et al.
Williamson, OE (1985): The Economic Institutions of Capitalism, New York-
London
Winter SG (1987) Knowledge and Competencies as Strategie Assets, in: Teece,
D.J. (Ed.): The Competitive Challenge: Strategies for Iudustrial Innovation
and Renewal, Cambridge, pp 159-184
E-Business Strategies in the Mechanical
Engineering lndustry

An Approach to Explanations of Current Trends in

E-Business Diffusion

Axel Braßler

Herfried Schneider

1 Abstract

An exploratory study on "E-business systems: current position and trends"

was carried out during the summer of 2001 in conjunction with VDMA,
the German Association for the mechanical engineering industry. The
main focus was on 23 e-business methods which are of intrinsic interest to
the mechanical engineering industry. The 195 usable returned question-
naires revealed a differentiated picture of the current situation and the
expected developments.
The results show, that the realization of e-business application in the
field of mechanical engineering is still in early stages. Extensive planning
activities for most application possibilities are nevertheless visible. In
addition, by using cluster analysis, significant variations between compa-
nies in their attitude and in the use of e-business methodology to which the
attitudes led were derived from the data obtained. The spectrum ranges
from complete refusal to a high degree of acceptance.
The authors try to give an economic explanation for such a heteroge-
neous situation. Issues from production economics and organizational
theory are tak:en into consideration as well as approaches from innovation
science and strategic management.
158

2 lntroduction

As a result of a considerable amount of success stories it is common belief

that enterprises in a future information and knowledge society should be
prepared for the use of e-business in all its facets. From this point of view
enterprises should be encouraged to dedicate themselves to the adoption of
e-business technology to stay competitive.
On the other hand the implementation of e-business solutions requires
substantial investment in hard- and software, in the training of employees
and, rather frequently, in the change of organization. Thus the questions
arise whether a reasonable pay-back can be expected and which the best
strategy for the adoption would be: Should a company be amongst the first
movers, just keep pace with the general development or avoid expenditures
in a technology which still have a long way to mature?
The answers to these questions may also explain why some companies
refuse e-business and why others believe that e-business is or can be de-
veloped into a competitive factor and whether characteristic differences
were recognizable among companies related to any kind of success factors.
These questions seem even more interesting regarding enterprises of the
so-called "Old economy" with its in most cases conservative organization
and on the background of the widespread discussion on the existence of an
"Old" and "New Economy".
Therefore the research reported on here was focused strictly on the
mechanical engineering brauch. The main objective was to fill the
explanatory gap between the current situation and the expected trends of
development by plausible considerations based on an empirical study.

3 Expected advantages of e-business

Profit, Costs, Time and Quality are the basic categories to judge and to
measure the performance of any value chain. From that it seems to be clear
that the expected advantages of e-business solutions have to have some
positive impacts on these factors. Hypothetically the following beneficial
effects can be accomplished by applying e-business solutions in a com-
pany:
• Increased profit by
- Additionalsales channels (e-sales, e-marketplaces)
- Expansion market (e-sales, e-marketplaces)
 159

- Creation services (e-product catalogues, e-product configuration, e-

order release)
- Taking advantage of temporary market opportunities (e-coopera-
tion)
• Reduced costs by
- Saving searching costs (e-procurement, e-logistics)
- Decreasing coordination costs (e-engineering, e-cooperation)
- Lowering transaction costs (e-order release, e-design, e-order
tracing and tracking)
• Saved time by
- Getting real time information (e-logistics, e-order tracing and
tracking)
- Reducing process time (e-procurement, e-reporting, e-workflows)
• Raised quality by
- Increasing flexibility (e-trade ofmachine capacity)
- Launehing compatible standards (e-marketplaces)
Despite these assumed advantages and the prognoses of the diffusion of
e-business applications, derived from these expectations, not too many
affirmative cases regarding the e-business reality have been provided up to
now.

4 Results of the empirical study

4.1 Research method and general data

As exploratory method a standardized questionnaire was used. This

method offers, in contrast to oral interviews, the advantage of reaching a
wider range of participants at a reasonable effort and the anonymity of a
questionnaire raises the willingness of sample enterprises to submit
reliable data.
With few exceptions closed questions were asked, to increase the objec-
tivity and comparability of the results as well as to minimize the time-con-
suming categorizing and coding of the answers.
In collaboration with the VDMA in mid-2001 a questionnaire was
mailed to 573 enterprises. 195 evaluable questionnaires were returned,
amounting to a response rate of 34%.
Twenty-three functions of electronically supported business processes
specific to Mechanical Engineering were defined and mentioned in the
160

questionnaire to analyze the level of e-business application, plans and

trends of development. The application of the e-business solutions cor-
responding to these business process functions were to be identified by the
distinguishing criteria "already realized", "planned within the next two
years", "not intended" and "unknown."
Since one of the objectives of the study was to identify (more or less
homogeneous) groups in respect to their attitude toward e-business, a
duster analysis was used to interpret the extracted information.
To describe the level a company is using e-business, a tuple was defmed
for the returned questionnaires. A company with 9 "realized", 8 "planned",
6 "not intended", 0 "unknown" e-business solutions would, for example,
be described in the tuple as (9,8,6,0).
With the helpoftbis formalization and using a Ward duster test it was
possible to group the enterprises depending on their tuple properties into
the previously mentioned archetypal groups. The Ward duster method was
used to obtain the lowest variance within the groups.
Questions about success-related characteristics as well as information
such as company size and means of production were also induded in the
questionnaire. The answers to these should later help to generate explana-
tions for different e-business situations and approaches.

Table 1. Statistics ofthe monetary and success-oriented indices.

~
~ ~ ~
~ -~ -~ s::
~ a e § § 0
Cl)
Cl) Cl)
"' 1ä -~
<
·ae
~ ....
0
~ -~
.s"' ::s .E!0
Cl) Cl) Cl)
Cl)

:g ~
~ ~
.2
::s
I I

<ii
"'
~ .,..~ .,..~ ~
> ~ N t-

0 Increase in I64 3I 8.04% 5% -3% 200 3% IO% I6.5%

turnover within the %
last 5 years
Turnover in Mi!. 176 19 210.584 54.9 I 5,000 20 I50.75 588.979
DM
Cash flow in 93 102 I4.153 3 -14 379 I 9.221 50.056
Mi!. DM
Research& 162 33 4.79% 4% 0% 40% 2% 6.63% 4.45%
development
expenditures in % of
turnover

The indices for the average increase in sales for the last 5 years, for the
monetary sales index, for cash flow and for R&D expenditures of the last
 161

fiscal year are given in Table 1 with the corresponding statistical

parameters.
In addition to the monetary and success-oriented indices shown in Table
1, information conceming company size and production methods is given
in Table 2.

Table 2. Company characteristics

"'
Cl) I) s=
"'
Cl)
"§ ~ "'
Cl) "0 "0
!ä .:a
"§ "§ -Cl:! 0
1-<
~ s= ;fl s=
g
S'~ .g ...0 ~t) 0
S's o V S'g ..... s=
·~ 0
GJ ·;::: ·~ ;:I s= .....
Cl) ...

-:g"' se"'
oon t) ~ oon > 0
"' t)
CI)"C oou
uY._, ~ V 1\ ..... •.j;j =.ä 0.. o.ä
s
t)
Cl:! 0 Cl)

-.g "' i~
Cl) ] t)
~ 000 1-<

s
CZl
.-6~ ~
.....:l e0..
CZIO.. .....
~
Cl:!
:I:
~
Absolute 37 104 47 21 32 61 70
frequency
Relative 19.68% 55.32% 25% 11.41% 17.39% 33.15% 38.05%
frequency

4.2 Data analysis and interpretation of results

The term e-business so far has not been consistently defined in publica-
tions. Most commonly it has been used as a general term to describe the
various possibilities to utilize electronic media in business processes.
When used in concrete terms regarding a subject area, terms such as e-
commerce, e-productions, e-procurement, e-finance or e-collaboration are
coined, all of which are used to Iabel an electronic business process. We
have included these various terms into the terminology of our study in
order to poll all possible application aspects of e-business in compromised
form, even though no definitive distinction has been made.
In Table 3, twenty-three e-business applications are included that appear
to have relevance to mechanical engineering. The selection was based
upon an analysis of literature, upon intemet research and expert dis-
cussions.
To avoid misinterpretation of questions, a short explanation was given
in the questionnaire for each of the defined e-business applications.
Systemizing dominant application areas, the following categories were
defined:
162

E-Business solutions, which predominately:

• Supportsalesand distribution processes,
• Support market oriented coordination ofbusiness processes,
• Support the coordination of networking processes,
• Supportservice processes,
• Support employee potential as well as infrastructure,
• Support manufacturing processes.
Table 3 summarizes the current position of the realization of e-business
applications. lt indicates the planned intentions and gives an overview
regarding the refusal rate. Conclusions conceming attitudes toward e-busi-
ness can be drawn from this data.

Table 3. Overview of the current position and development trends of individual e-

business applications.

l)
>.
.Q..S "'~
(J.)
~;::: ;:::
.....
0 "S
;s .s (J.)
>.
;:::
0
..s
"'(;j (J.)

.ß ~ u~
(J.)

"' "0 0
·p ~ N
:.::::"'0
"'(J.) ~ ~ ;::: (J.)
>- <IS (J.)
eo_~
(J.) N ;:I 0 0
8;:I :.:<IS:::
•...-! •'1""'1

ta..So
~i z
til
oo e <IS
Q.. ~ ·~ ~ Zs::l.
0

positive attitude negative attitude

towards e-business towards e-business
E-Business applications supportin!!: sales and distribution processes
Electronic Product Catalogues 40% 40% 19% - 1%
80% 20%
Electronic Product Configurators 9% 27% 55% 7% 2%
36% 64%
Electronic Order Release 10% 38% 46% 3% 3%
48% 52%
E-Business applications supporting market oriented coordination of
business processes
Electronic Sales 18% 56% 23% - 3%
74% 26%
Electronic Procurement 9% 55% 34% 1% 1%
64% 36%
Electronic Trade ofMachine 2% 11% 76% 11% -
Capacity 13% 87%
Electronic Engineering 2% 14% 73% 8% 3%
16% 84%
Electronic Logistics 2% 18% 68% 9% -
20% 80%
 163

E-Business applications supporting tbe coordination of networking

processes
Electronic Design 9% 114% 62% 19% 16%
23% 77%
Electronic Cooperation 4% 123% 63% 17% 13%
(Collaborative Planning) 27% 73%
Electronic Business Communities 5% 114% 66% 113% 12%
19% 81%
E-Business applications supportin!! service processes
Electronic order Tracing and 4% 126% 63% 14% 13%
Tracking 30% 70%
Electronic Service 12% 141% 45% 11% ll%
53% 47%
Electronic Hotline 31% J 35% 31% 11,5% !1,5%
66% 34%
Electronic Customer Relation 9% 143% 43% 13% 12%
Management 52% 48%
Electronic Documentation 17% 143% 38% 12% 1-
60% 40%
Products as Information Carriers 4% 117% 62% 113% 14%
21% 79%
E-Business applications supportin ~ employee_potential and infrastructure
Electronic Reporting 10% 120% 59% 18% 13%
30% 70%
Electronic Learning 5% 123% 64% 17% 11%
28% 72%
Virtual Information Pools 30% 136% 30% 13% 11%
66% 34%
Electronic Workforce Optimization 1% 119% 65% 113% 12%
20% 80%
Electronic Workflows 4% 125% 60% 19% 12%
29% 71%
E-Business applications supportin!! the manufacturing processes
Internet Technologies in the 10% 120% 59% 19% 12%
Control and Automation 30% 70%
Technology

Within the framework of e-business applications, which are predomi-

nately used to support sales processes, electronic product catalogues take a
domineering position. This e-business solutions has, in comparison to all
other applications, the highest utilization level at 40%. Another 40% of the
companies are planning to implement this application within the next two
years. In sum, 80% see beneficial solutions for their companies. Only 20%
164

of the companies questioned weren't currently considering electronic

product catalogues as a means of promoting sales.
Despite the obvious advantages for the considered industrial branch, the
electronic product configurators are only beginning to be introduced into
the sector. With 36%, more than a third ofthe companies see an advantage
in this application, but more than half of the companies haven't reached a
decision about realization.
Nearly half of the sample is considering electronic order release as an
effective sales support. However only 10% of the companies have already
implemented it.
The majority of mechanical engineering companies expect advantages
for procurement as well as sales by using electronic market places.
Nevertheless, the discrepancy between planned and implemented solu-
tions is extremely high. E-business solutions, mainly used to coordinate
networking processes have a realization level between 4 and 9%. These
applications arestill in early stages.
The skepticism of companies in these areas is also visible in the low
level of implementation plans. Presently, almost 2/3 of the companies
refuse this form ofbusiness support completely.
A heterogeneous view was achieved regarding e-business solutions with
distinct service character or which are used to support service processes.
While well over half of the companies considered e-hotlines and elec-
tronic documentation useful, applications such as e-order tracing and
tracking, with 35%, and products as information carriers, with 21%,
experienced a substantially lower approval rate.
E-services and e-CRM solutions, with 52 and 53%, were rated by over
half of the companies as useful. However, the level of implementation of
these solutions remains low with only 9 to 12%. E-business applications,
with the exception of virtual information pools, predominately serving to
support employee potential tended tobe judged skeptically. Their applica-
tion lies between 1 and 10%. Furthermore, less than 113 of the companies
have included this form of support in future planning objectives.

4.3 Compiling typical groups of e-business supporters and

rejecters

After analyzing the current situation and trends of development of e-busi-

ness applications, subsequent investigation is required to determine
whether or not the attitude of particular companies to e-business and the
resulting level of realization correlates and can be classified into charac-
teristic groups.
 165

By defining three different Ievels each for the parameters "Acceptance"

and "Realization" a matrix of nine fields was formed, into which the com-
panies were sorted by the help of a Wardclusteranalysis (see Table 4).
The first line consists of companies that hold a positive attitude and,
therefore, a "High Acceptance" toward such technology. These companies
scored lowly under the heading "No Realization". However, they differ
significantly regarding actual and planned implementation.
The groups 1, 2 and 3 were formed as a result of this divergence. The
group 3 enterprises see a decisively competitive success factor in realizing
e-business solutions. They have already realized nearly 10 applications,
while none of the companies in this group had implemented less than 8
solutions. In total, 12 companies could be categorized as e-business
experts.
In contrast, 12 companies fell into groupl. These companies, like those
in group 3 see a great opportunity for profit in e-business technology, but
are far from realization. In this group, less than two solutions have been
implemented in average, however a minimum of 14 concrete e-business
solutions has been planned. While group 1 and 3 enterprises stand in stark
cantrast to one another, the group 2 enterprises (22 companies) can be seen
as the link between the two.
This group has surpassed group 1 cluster and the enterprises belanging
Though they don't yet meet the criteria for the group 3 category, they are
on their way with an additional 11 planned applications.
In cantrast to groups 1, 2 and 3 e-business applications were met with
minimal acceptance by groups 7, 8 and 9. Companies classified at this
Ievel scored highly under "no realization."
Group 7 (16 companies) doesn't see a benefit in this technology. Up to
this point, the group 7 companies have implemented less than one solution
in average and have planned merely one application.
Consequently 22 of 23 possibilities were declined. Group 8, less
extremely positioned (32 companies), has also realized less than one appli-
cation in average. Nevertheless, the attitude toward e-business is more
positive than that of group 7 since group 8 has 3 or 4 planned applications.
Group 9 holds a similar position with 13 companies. This position can
be more or less distinguished from group 8 in that the applications have
not only been planned, but have also been partially realized.
Group 4 (49), 5 (36) and 6 (3) are placed between the two previously
described categories. In these groups, "No Realization" is planned on
average for 14 solutions.
166

Table 4. Results ofthe cluster analysis

acceptance average
low realization high realization
realization

c c c
0 0 0

i ~
·;:: i
. 1.
Cll Cll Cll
l.!!
. 1. I!
Cii > > Cii >
. 1.
> > Cii
.!! CD 'ECll Cll
1:11
'ECll .!! CD 'ECll
1:11 1:11
I! 'g 'g
I! 'g

j j
CD CD c CD c CD c
1 s111
Cl. Cll Cl. Cll Cl. CD
Cl.
:I i; ~ Cl.
:I i; ~ Cl.
:I
>
Cll
1 2 3
fl already reallzed 2 0 1,3 0,62 7 3 4,5 1,22 23 8 9,83 1,91

fE.s p1anned ror the 23 11 14 2,71 20 6 11 1,98 15 0 7,33 3,01

~2'
u 'next 2 vears
u
01 no realization 12 0 7,7 2,75 14 0 7,8 2,20 15 0 5,83 2,61

4 5 6
u already realized
CD 2 0 0,9 0,80 8 3 3,7 0,99 12 4 5,67 0,94
CD
Cll c
I! J!l planned for the 5 12 7,9 1,63 9 4 5,8 1,48 3 0 2,67 0,47
CD CL
> CD next 2 vears
01 801 no realization 18 10 14 1,54 16 6 13 1,61 19 8 14,7 0,47

7 8 9
CD already realized 1 0 0,3 0,43 1 0 0,3 0,45 6 2 3 1,30
u
c
~ J!l planned for the 2 0 1,1 0,75 6 3 4,4 1,02 4 0 1,54 1,28
rS! 2'
u
next 2 vears
u
01 no realization 23 20 22 0,85 20 16 18 0,99 21 13 18,5 1,15

realization Ievel
While group 4 enterprises have only realized one of the nine potentially
accepted possibilities and have 8 in planning stages, the "advanced" com-
panies have already realized nearly 4 applications (with an additional 5 in
planning stages).
Group 6 is in this category the most advanced group with 5-6 solutions.
However, since this group is composed of only three companies, it is not
possible to draw a reliable conclusion.

5 Adoption barriers and drivers for e-business

applications

There seem tobe four major barriers to e-business applications:

• Barriers caused by the specificity of the new technology ,
• Barriers caused by the specificity ofthe particular company,
 167

• Barriers caused by the general environment of the particular com-

pany,
• Barriers caused by the environment of the supply chain.
These barriers evoke controversial attitudes of different companies to e-
business depending on their actual situation (e.g. situation in competition,
attractiveness of the market, organizationallevel etc.). If the influencing
factors of these barriers are controlled by the particular company they may
turn into drivers to e-business realization.
In order to work out the particular aspects it seems to be useful to refer
to the well known results of innovation management (e.g. Rogers 1995).
Many points can be adopted because e-business solutions arenothing else
than a special kind of ordinary innovations.
In addition we found some further explanations in the results of our
investigation and in the knowledge of several empirical investigations
which were carried out all over the world.

5.1 Barriers caused by the specificity of the new technology

• Compatibility Barriers
The compatibility of an innovation shows to which extent it harmonizes
with already existing technologies and systems as well as experiences and
company resources (Rogers 1995, p. 224). In this sense a higher compati-
bility leads to a reduction in insecurity regarding the consequences of the
adoption. A compatibility barrier for e-business solutions exists, if the
principle applications in question do not complement and support the
existing IT systems.
Through a multitude of divergent standards, providers and solutions, a
barrier remains which should not be ignored. If a company has for instance
successfully introduced an ERP system, then e-business solutions which
are compatible to this system have considerably lower compatibility bar-
riers. A study of e-business adoption in the Australian mining industry
concluded that nearly all business which had positive attitudes about e-
commerce solutions were working with SAP. (Flynn!Purchase 2001, p. 6)
If the compatibility is not taken into consideration, it can lead to mis-
adoption, which results in the expensive exchange of existing solutions or
to the cancellation of new solutions (switching costs) (Rogers 1995, p.
226).
168

• Relative Advantage Barriers

The relative advantage barrier shows to what extent a new solutions is
distinguished as the better solution from that which is currently being used.
Monetary criteria as well as soft facts such as employee motivation or
image play in roll in such judgments. The advantages of a single e-busi-
ness solution often do not appear great enough to risk adopting it. Poor
performance is responsible for the resistance from companies especially
with B2B market places (Held 2001, pp. 10).
• Development Rate Barriers
Directly those hard- and software technologies upon which the most
various e-business solutions are established, are being developed at an
outrageous rate. A higher capability and constantly changing standards are
the result ("Moore's Law") (Moore 1965). The high development rate
represents an adoption barrier because it seems rational to companies to
skip buying current technology and delay making purchasing decisions.
This behavior is also described as Leapfrogging Behavior (Pohl 1996, p.
12).
• Complexity Barriers
Complexity describes how easy it is to understand innovation and it's
underlying principles. Innovations that are perceived as complex spread
much more slowly because there is greater uncertainty about the success
and consequences of the adoption.
The complexity of a solution depicts a large establishment barrier for
many e-business solutions. Neither the media nor the providers of e-busi-
ness solutions managed to successfully inform about the opportunities and
principles of the new technology (Rogers 1995, pp. 242 and
Tornatzky/Klein 1982, pp. 28).
• Security Barriers
A further barrier for the introduction of e-business solutions is in data
security. An online data interface furnishes the possibility of unauthorized
users to access, delete or manipulate data. Further dangers to data include
viruses or even one's own employees. Errors in the computer system as
well as intentional or unintentional misconduct of employees can also lead
to data leakage.
Especially smaller companies or other companies which use a classical
Defender Strategy (Miles/Snow, 1978) are afraid of losing their competi-
tive position if competitors can call up useful information on their intranet
site (Scupola 2002, p. 6). The internet is therefore not seen as a medium
 169

for winning new customers and maintaining a close commitment to exist-

ing customers, but is seen rather as an information source for competitors.

5.2 Barriers caused by the specificity of the particular

company

• Monetary Barriers
Monetary barriers appear more visibly in smaller and mid-size compa-
nies (Braßler/Schneider 2001). On the one hand they have less financial
power. On the other, SME find often themselves in private ownership
which, in the case of credit financing, could Iead to the personalliability of
ownership. Such companies therefore often invest more cautiously and
reservedly (Lanwes/Lehner 1998, p. 18).
• Structure Barriers
Often the organizational structure creates a barrier against the imple-
mentation of innovations, including e-business solutions. Kiesner/Kubicek
have developed five instrumental variables to describe organization struc-
tures. Ofthese five (formalizing, specializing, delegation, coordination and
configuration), specializing, formalization and centralization could be
proven as factors which had an influence on the innovation power of a
company (Kieser/Kubicek 1992). However, often only one relatively weak
connection to propensity to innovation can be observed (Rogers1995, p.
378).
Companies which have a strong specialization, formalization and cen-
tralization are often in a better situation to implement innovations. In the
case of decentralized unformu1ated companies on the other hand, new
innovations emerge with greater ease and are easier to achieve (Oelsnitz
2000, p. 105).
• Cultural Barriers
The employment of e-business solutions can equally be influenced by
company culture. Companies with lower error tolerance, poor communi-
cation, only average creativity and a slight inclination toward unconven-
tional methods are especially slow in applying innovative solutions.
Beyond that, cultural barriers are also expressed in that innovations
which are not developed directly in-house were viewed with reluctance
(Not-Invented-Here Syndrome) (Mehrwald 1999, p. 70).
170

• Competence Barriers
Lack of competence and available resources present a noticeable barrier
especially for small and mid-sized companies in the implementation of e-
business applications. Many small and mid-sized companies listed the lack
of know-how as the reasons for the cautious e-business adoption. The lack
of promoters also called champions, can also be a further part of the com-
petence barrier. Several studies have emphasized this as a key success
factor for the introduction of innovations. They possess adequate
knowledge as well as the power necessary to enforce these ideas in a
company (Rogers 1995 p. 398 and Nambisan/Wang 2000, pp. 129).
• Resistance Barriers
Active or passive resistance barriers can develop in members of an
organization, if the introduction of e-business solutions leads to the
restructuring of a business resulting in changes in status, power or income.
Those feeling that they may lose out or try to hinder restructuring and with
it the innovation.
Further reasons for resistance may be in negative experiences with past
innovations. Rogers depicts this behavior as "Innovation Negativity"
(Rogers 1995, p. 227) Various surveys about the satisfaction with IT
services in German business show that a general dissatisfaction exists
(Lawson/Alcock/Cooper 2002, p. 130).
An additional source of negative experiences lies in over investment of
the past years (Peterson/Welch/Liesch 2002, p. 214). Too much money
was invested in IT -innovations without the proper restructuring and
retraining.

5.3 Barriers caused by the general environment of the

particular company

• Political Barriers
A barrier to the introduction of e-business solutions can also be seen in
certain conditions in a non-existent political engagement. In order to push
innovation it is possible to apply tax advantages or support programs
which promote distribution. State institutions can in the same way produce
information regarding use, advantages, and costs as well as risks of e-busi-
ness possibilities through research assignments (Kshetri/Dholakia 2002, p.
123 and Lanwes/Lehner 1998, p. 38).
 171

• Legal Barriers
A further barrier to the introduction of some e-business solutions lies in
inflexible or unclear legal behavior. An example is the data protection law
also called the regulation of crossover transactions (OelsnitzJMüller 1996,
pp. 270).
• Infrastructure Barriers
Most e-business applications demand a certain infrastructure. Most
solutions are therefore not able to be realized without broadband intemet
access and favorable flat rates. Also betonging to the infrastructure are
worldwide backbones, which enable communication with the USA or Asia
at acceptable speeds.
A barrier was visible in the past years, however prices and broadband
have developed to such an extent (not the least through deregulation of the
telecommunications market since 0110 111988) that the infrastructure today
should no Ionger present a barrier. For intemationally active companies,
infrastructure in the target market could certainly be a barrier
(Kshetri/Dholakia 2002, p. 123 and Ling 2000, p. 7).

5.4 Barriers caused by the environment of the particular

supply chain

• Acceptance Barrier
The acceptance of e-business solutions is always necessary for two
sides. E-sales or e-procurements solutions naturally require the appropriate
level of acceptance and introduction from the relevant trade partners (Clark
et al. 2001, p. 18).
• Critical mass/network Barriers
Another barrier consists in technologies, which require use by a crucial
mass because network extemalities exist. Email was useless for the first
user because he could not communicate with anyone. Online market places
behave similarly. As long as certain technology has not reached or will
perhaps never reach a critical mass, there are good reasons not to adopt
this technology (Scupola 2002, p. 6 and Rogers p. 390).
172

• Competition Barriers
Even competitors can be seen as a barrier for the introduction of inno-
vations. Rogers describes this as "Observability", the opportunity to
observe and imitate innovations by others to remain competitive (Rogers
1995, p. 244). "Leaming increases over time with collective experience
and then it affects the likelihood ofadoption" (O'Callaghan 1998, p. 394).

6 Conclusions

In conclusion, the realization of e-business application in the field of

mechanical engineering is still in the beginning stages. Extensive planning
activities for most application possibilities are nevertheless visible. The
various attitudes of mechanical engineering companies towards the
challenges of e-business are most clearly illustrated in the cluster analysis.
The spectrum ranges from complete refusal to high acceptance.
In total, the yet low degree of e-business applications may be surprising,
but mechanical engineering has been a very conservative branch for a long
time. On the other hand the relatively high Ievel of planning activities
shows the perspectives of e-business and a comparison with other branches
like automotive industries show its potential.
It seems to be necessary to investigate the barriers mentioned in chapter
5 by another survey in still more detail to receive reliable information of
companies in the mechanical engineering branch on their attitude and
reservations to e-business investments.

7 Summary

An exploratory study was carried out in the summer of 2001 in con-

junction with VDMA, the German Association for the mechanical engi-
neering industry, on "E-business systems: current position and trends". The
main focus was on 23 e-business methods that are of intrinsic interest to
the mechanical engineering industry. The 195 retumed questionnaires
usable revealed two different pictures: that of the current situation and that
of the changes, which were felt to be imminent in the individual compa-
mes.
In addition, by using cluster analysis, significant variations between
companies in attitude and in the use of e-business methodology to which
the attitudes led were derived from the data obtained. The authors try to
 173

give an economic explanation for such a heterogeneaus situation. Issues

from production economics and organizational theory are taken into con-
sideration as well as approaches from innovation science and strategic
management.

References

Braßler A, Schneider H (200 1) E-Business-Lösungen im Maschinen- und

Anlagenbau. Journal für Betriebswirtschaft, 51 (2001) vol5-6: 200-207
Clark D (2001) Adoption and Implementation of E-Business in New Zealand:
Empirical Results. Research Report Series of the University of Waikato
Management School. Waikato
Flynn A, Purehase S (200 1) Perceptions of Barriers to E-Commerce, in:
Australian and New Zealand Marketing Academy Conference (ANZMAC),
Ist- 5th December, Massey University, New Zealand 2001, pp. 1-6
Held M (2001) Unternehmen kritisieren die Qualität der B2B-Marktplätze. Client
Server Computing select IT. vol2: 10-11
Kieser A, Kubicek H (1992) Organisation. Berlin/New York 1992
Kshetri N, Dholakia N (2002) Determinents of the Global Diffusion of B2B E-
Commerce. Electronic Markets, voll2, no 2: 120-129
Lanwes C, Lehner F (1998) Einsatzbarrieren für die Nutzung des Internet,
Ergebnisse einer Befragung von mittelständischen Unternehmen, Forschungs-
bericht no 18, Schriftenreihe des Lehrstuhls fiir Wirtschaftsinformatik III.
Universität Regensburg
Lawson R, Alcock C, Cooper, J (2002) Diffusion of Electronic Commerce in
Small and Medium Enterprises: Australian Journal of Information Systems,
vol9 no 2: 123-134
Ling CY (2000) Model of Factors Influences on Electronic Commerce Adoption
and Diffusion in Small and Medium Sized Enterprises, Pacific Asia
Conference in Information Systems, pp 1-11
Mehrwald H (1999) Das 'Not lnvented Here'-Syndrom in Forschung und
Entwicklung. Wiesbaden
Miles RE, Snow CC (1978) Organizational Strategy, Structure, and Process. New
York
Moore GE (1965) Cramming more components onto integrated circuits.
Electronics, vol38: 114- 117
Nambisan S, Wang YM (2000) Web Technology Adoption and Knowledge
Barriers. Journal of Organizational Computing and Electronic Commerce, vol
10, no 2: 129-147
O'Callaghan R (1998) Technology Diffusion and Organizational Transformation:
A Integrative Framework. In: Larsen TJ, McGuire E (eds), Information
Systems Innovation and Diffusion, IDEA Group
Peterson B, Welch LS; Liesch PW (2002) The Internet and Foreign Market
Expansion by Firms. Management International Review, vol42, 2: 207-221
174

Pohl A (1996) Leapfrogging bei technologischen Innovationen. Wiesbaden

Rogers EM (1995) Diffusion oflnnovation. New York
Scupola A (2002) Adoption Issues ofBusiness-to-Business Internet Commerce, in
European SMEs, Proceedings ofthe 35th Hawaii International Conference on
System Sciences, Published by the Computer Society, January 2002
Tornatzky LG, Klein KJ (1982) Innovation Characteristics and Adoption-
Implementation: A Meta-Analysis of Findings. IEEE Transactions on
Engineering Management, vol 29, no 1: 28-45
Oelsnitz D von der, Müller M (1996) Einsatzpotentiale und Diffusionsbarrieren
des Internet. Die Unternehmung, 50 J.,4: 261-278
Oelsnitz D von der (2000) Marktorientierte Organisationsgestaltung. Stuttgart
Elements of the Production of Services

Günter Fandei

Steffen Blaga

1 lntroduction

The article contributes to the current discussion on the development of a

theory of the production of services. The extent to which activity analysis
can be applied to the production of services will be discussed, and how it
can integrate the latter into a uniform approach together with the produc-
tion of physical goods.
In this context it will be shown that defined theoretical approaches to the
production of services correspond to certain aspects and terminological
embodiments of activity analysis. Other focuses will be on specifying the
output and the input system of service production as well as on the deter-
mination of terms for the area of service production familiar from goods
production: activity, technology, efficiency and production function.

2 The activity analysis as a concept of modelling

In developed economies service production is an important sector (Albach

1989). Just like any form of operational production it has tobe examined
for its economic efficiency (Albach 2000). With the activity analysis
(Koopmans 1951 ), an efficient instrument has been developed for exam-
ining and shaping efficient productions. It has proved to be of extremely
good value in the production of physical goods in assessing activities and
technologies, and in describing efficient productions through production
functions (Wittmann 1968, 1979; Dyckhoff1993, Steven 1998; Dyckhoff
et al. 2002). On the other hand, there are very few approaches that apply
this analysis to the production of services as well, even though these few
have by no nieans been unsuccessful (Fandel/Prasiswa 1988; Fandei
2001a, 2001b).
The difficulties that economic research into service production has with
the application of the input-output analysis can be seen by the fact that a
potential-, process- or stage-oriented perspective is proposed in addition to
176

or instead of output orientation (Corsten 1988, 1990; Hilke 1989). The in-
dividual perspectives are characterised as follows:
- Output-oriented perspective (Maleri 1973; Gerhardt 1987): services
are generated as intangible assets through the combination of re-
sources (e.g. haircut).
- Potential-oriented perspective (Meyer 1987): services are operative
capabilities (e.g. readiness to cut hair).
- Process-oriented perspective (Berekoven 1974; Rosada 1990): ser-
vices are created through processes in which the operative capabili-
ties of the producer are combined with the consumer or with objects
brought in by the latter and the external factor experiences a change
(e.g. cutting hair). The combination processes require a factual, spa-
tial and temporal coordination between the producer of the service
and the consumer.
- Stage-oriented perspective (Hilke 1989; Corsten 1990; Meyer 1991):
to characterise services this considers the above perspectives in iso-
lation (one-dimensional) or integrated (multi-dimensional).
Output, input combinations, activities or production processes and mod-
els of single or multistage production are, however, common elements or
procedures of the activity analysis. It therefore seems obvious that the
activity analysis should be examined to see whether it can be considered as
an efficient instrument for looking into the production of both services and
physical goods. Table 1 illustrates clearly the relations between the con-
trary perspectives in the Iiterature on the production of services and the
construction elements of the activity analysis. This enables the contrary
positions in the perspectives to be approximated to one another, even, in
fact to nullify the dispute between a output-oriented and a process-oriented
analysis of service production. These coherences are the object of the dis-
cussions in the following sections. How far terms used in activity analysis
can be filled out with meaningful contents for a production theory of ser-
vices will have to be clarified here. Production models for special services
(Farny 1975; Albach et al. 1978; Stieger 1980; Haak 1982) certainly en-
courage us to take this step.
 177

Table 1. Relations between perspectives of service production and elements of the

activity analysis
Perspectives of service production Elements of activity analysis
Output-ariented Output
Potential-oriented Input combinations
Process-oriented Activity I Technology
Stage-ariented Stage basis of production
- One-dimensional - Single-level
-Multi-dimensional - Multi-level
There are very practical considerations that speak for a link between the
production of physical goods and the production of services. Productions
of physical goods usually display elements of services creation as well, if
we just call to mind in general the operational functions of administration
and materials managements, and in particular the logistical processes of
transporting and storing parts, sub-assemblies, semi-finished and finished
goods. In the explicit modeHing of the production of physical goods with
the help of traditional approaches from the theory of production, these
elements are often unjustifiably overlooked, because they are presumed to
be contained implicitly in the input-output combinations of the activities
that are observed. They are only found as independent activities in appro-
priate network plans if they become relevant to the organisation of the
process (Neumann/Morlock 2002). On the other hand, activities from the
production of physical goods can be found in the production of services,
for example, the provision of services by a craft-based company, or by an
orthopaedic clinic. Many authors even refer to the close analogy that one
and the same production process in industrial companies represents the
production of physical goods on own account, but can be regarded as pro-
viding a service in the case of contract production (Engelhardt 1989;
Engelhardt et al. 1993, 1995). The fact that the transitions in the charac-
terisation of physical goods as against services are in fact in a state of flux
was shown very convincingly by Shostack (1982) in a graph in which
goods were assigned in accordance with how far the proportians of mate-
rial or immaterial elements dominate in them (see as well: Rushton/
Carson 1985). Because of the various types of connections, the economic
incentive is to attempt to integrate the production of services and physical
goods and to place them on a joint methodological base. This idea is not
new (Preel/De La Rochefordiere 1988).
178

3 Discussion of elements of a theory of service

production

3.1 Specification of the output

The objective of all production is to manufacture the desired goods

(Koopmans 1951 ), which are referred to as the output. It is in fact the
characteristics of these goods that define them (Krelle 1969) and that sat-
isfy consumers' needs (Marshall1961; Lancaster 1971, 1972). This is why
goods are in demand. Goods can be physical objects, and are then known
as physical goods. If they are intangible objects they are called services
(Say 1852; Lancaster 1972). In this way the output-oriented definition of
services is marked through their character as intangible assets, which will
be relevant below.
At the same time, an abundance of difficulties arises from the output-
oriented characterisation of services as intangible goods. This has triggered
numerous attempts to define the term "services" which in some cases have
even turned away from the output-oriented definition. These attempts may
be classified as follows.
- Enumerative definitions (Schär 1923; Rössle 1954; Wöhe 1973;
Schierenheck 1974): services are characterised by listing examples.
It is not possible to develop generally valid starting points for a the-
ory of services production from this, because the examination of the
fundamental common features of services is neglected.
- Negativedefinitions (Altenburger 1980): these combine all company
outputs into services as the tertiary sector which cannot be allocated
unambiguously to the primary or secondary sectors of the traditional
German economic output systematic. This makeshift solution
(Corsten 1988) is subject to the same critical evaluation as the enu-
merative definitions.
- Definitions based on constructive features of services (Gerhardt
1987; Corsten 1988; Rück 2000): these describe services through
their immateriality and the resulting fact that they cannot be stored,
and the necessity of including an extemal factor over which the con-
sumer and not the producer has the power of disposal (Maleri 1973).
Descriptive elements are also the readiness to provide the service,
the parallelism of production and sale and the synchronisation of the
supply of a service and the demand for it. This shows clearly that
 179

these definitions are increasingly moving away from the output defi-
nition and are taking up aspects of the input definition, the input
combination and the production process, which can also apply in this
way for the production of physical goods. The gradual turning away
from the output definition in the narrower sense manifests the em-
phasis on the individual perspectives shown in the previous section.
Under services here, "output-oriented" is tobe taken to mean intangible
goods that are created as the result of production processes. The result of
these production processes is therefore the service that is produced, and
not the production of the service, which, analogously to the production of
physical goods, would follow a process-oriented understanding.
A problern field which is closely connected to the property of immateri-
ality is seen in the lack of Operationability of the output service (Corsten
1986; Maleri 1998). This is differentiated into the aspects of insufficient
tangibility, quality measuring, determinableness and measurability
(Gerhardt 1987; Corsten 1998).
The lack of information on the product in contrast to physical goods is
often put forward as a reason for the insufficient tangibility of a service.
This makes it more difficult for the consumer to judge the quality. These
objections can be countered by considering that products are in fact de-
fined by their properties as the expression of objectively measurable quali-
ties, and that, to specify services, exact performance descriptions are re-
quired which stipulate as far as possible beyond reasonable doubt the
nature, scope and quality of the service. With construction services, and
with the services provided by doctors, lawyers and accountants, this is
done with the help of forms and specifications under the different official
fees regulations.
The quality aspect in the creation of a service is not different to that of
physical goods production. In the latter, quality controls are used to avoid
waste, and these would have to be set up for a service production. The cer-
tification of business processes in accordance with ISO standards which is
demanded by industry is taking this direction (Scheer 1995; Fandel2001c).
The analogy to services production is obvious.
The causes of the indeterminableness of a service that is occasionally
found are seen as being that the consumer hirnself (e.g. cutting hair), or
objects that hebrings in (e.g. car in need ofrepair), have tobe integrated in
the combination process of producing services as extemal factors
(Corsten/Stuhlmann 1998; Stuhlmann 1998) whose quality or condition is
insufficiently known at the beginning of the production process. If the in-
determinableness is the result of objectively measurable quality deficits of
the extemal factor (e.g. students with different educational levels at uni-
180

versities), this is to be evaluated in exactly the same way as with the pro-
duction of physical goods when a mistake is made with regard to the qual-
ity of the material. In contrast, if the indeterminableness of the output is
found in uncertainty with regard to the nature and quality of the input
"extemal factor" or of the production process, modeHing is a classical
problern of stochastic production theory. However, if the indeterminable-
ness of the service results from the situation that, because of the condition
of the extemal factor, clarity with regard to its final nature and scope is
only obtained in the course of its production (e.g. clinical diagnostics
[Fandel/Hegemann 1986; Hegemann 1986]), with an appropriate Ievel of
specification, less powerful output variables can be defined even for part
processes of services production (part examinations from which the final
diagnosis results, step by step [Hegemann 1986]).
If a service has been described as a product with a sufficient degree of
precision, it can then be measured as output in integral units. Here there
are direct common features with recording output quantities in the produc-
tion of physical goods such as motor vehicles or machines. Examples can
be found in the Iiterature for doctors' services on which figures were put by
the Bewertungsmaßstab-Ärzte (BMÄ) (Prasiswa 1979) and for services in
university teaching (Fandel/Paff2000; Fandel2001a).
The individuality of the order is occasionally indicated as a special fea-
ture of services (Corsten 1985, 1986; Garhammer 1988; Meyer 1991)
which, because it is incapable of standardisation, Ieads to there being only
heterogeneous products or no homogenous ones, which can be measured
as one in !arger units. This phenomenon is also found in the manufacture
of special machines, which are to be integrated into the particular produc-
tion technology ofthe investing company.
An argument that is occasionally used against the output-oriented defi-
nition of services is that certain services, such as concerts, pet shows or
similar events, cannot be defined by means of output, but in fact only
through their production process itself (Berekoven 1974; Rosada 1990). It
seems obvious here as well to draw a comparison with the process of pro-
ducing physical goods. The buyer does not pay for the production process
of the vehicle but for the manufactured vehicle itself. Shephard et al.
( 1977) have described the continuous production process through input and
output intensities for shipbuilding, and applied their efficiency analysis to
this with the help of the concept of correspondences, a process analogous
to the activity analysis. The actual input and output quantities are then cal-
culated through the time integrals ofthe intensities (Fandel1996a, 1996b).
This procedure can be transferred directly to the production of services as a
process and would eliminate the contrast between process- and output-ori-
ented views of service production. Thus a view is taken here that for the
 181

purposes of an activity analysis services can be handled as outputs simi-

larly to physical goods.

3.2 Extensions to the input system

Opinions on the taxonomy of a factor system with the production of ser-

vices differ greatly. Some authors believe that the factor classification of
the production of physical goods is in no way suitable for use as orienta-
tion in the production ofservices (Klein-Blenkers 1964, Carp 1974). Some
even share the extreme opinion that above all Gutenberg's (1994) factor
classification was conceived for the industrial production of physical
goods and that continuing to think in these categories tends to stand in the
way of an impartial new start to carrying out a factor classification for ser-
vice production (Carp 1974). These considerations culminate in the
demand that factor systems peculiar to services should be developed for
service production (Carp 1974). In contrast, Diederich (1966) takes the
position that, with some modifications, Gutenberg's factor system could
very well be used as a foundation for service production.
On closer observation there is more to be said in favour of the trend to
harmonisation than for polarisation. It is correct that the extemal factor, as
an input peculiar to services, is not found in Gutenberg's factor system and
that services are not listed explicitly as inputs; however, the dispositive
factor in his scheme and the associated handling of management functions
imply that services can certainly be components of his factor classifica-
tion! In addition, they have tobe integrated, because (as we have shown)
they also occur as inputs in the production of physical goods. On the other
hand, the view of Gutenberg's input system is improperly restricted if the
opinion is held that in the production of physical goods, which he focuses
on in his books (Gutenberg 1975, 1994), only physical goods occur as
inputs. Potential factors are not as such the inputs of production but the
services that they perform. This applies to both the operative (and for the
dispositive) output as much as to the output of the machinery as elemen-
tary factors. When Gutenberg (1994) talks of the subjective skills of the
workforce and, in extension of his concept of consumption functions, of
the z-situations of the technical properties of units through which the pro-
ductivity of the factor combination is essentially increased, we can see
how the product attributes of the inputs correspond to those of the outputs
of modern consumer economies. A vertical line can be drawn through
Gutenberg from the Engineering Production Functions (Chenery 1949,
1953; Smith 1961; Zschocke 1974) with their focus on the effects of the
engineering variables of the input and output on production
182

(Fandel 1996a), through to the product attributes that are equally relevant
for physical goods and services.
In contrast with the production of physical goods, the inclusion of exter-
nal factors into the system of resources on the production of services is
without doubt new and very plausible. Under extemal factors we under-
stand, in contrast to the definition of intemal factors, those types of input
whose appearance and involvement in the combination process of the pro-
duction of services with regard to time, nature, quantity and location are
not at the disposal of the services producer but are in fact determined by
the consumer of the service (Maleri 1973). Extemal factors are persons
(e.g. the customer as a person who wants further education, children who
are to be examined by a paediatrician) or objects (e.g. an animal to be
looked after or workpieces to be galvanised) which are brought by the con-
sumer of the service into the process of creating it (Maleri 1998).
Whether every service requires the inclusion of an extemal factor is
certainly questionable, because there appear to be enough cases in which
the integration of an extemal factor is not mandatory. These include, for
example, the offer ofbinding services by a public transport company with-
out anyone actually being carried. The transport service is provided but not
consumed. And finally, it is difficult to classify the provision of a
mechanical directory inquiries service by a telecommunications company
as anything other than the production of physical goods in store, without in
the one act including the consumer as an extemal factor.
Of the special features of extemal factors shown in the Iiterature
(Corsten 1998) one aspect should be illuminated which concems its quality
and the associated service profiles. With regard to the quality and the mode
of operation this causes in the combination process we can in principle
refer to appropriate considerations which take place on the efficiency of
employees, the productivity of resources and the profitability of materials
in the production of physical goods. However, with regard to the possibil-
ity of explaining the qualities of the extemal factor "humans" or substitut-
ing for the inadequate qualities in service production through the addi-
tional use of intemal factors, a new perspective of Substitution opens up
which is targeted at the examination of isoservice profiles or isoquality
profiles respectively (Corsten 1986). One example of achieving the same
results with a given quality profile by substituting the output of the intemal
factors by those ofthe extemal factor, would be the mixture ofservice and
self-service in restaurants of a certain category. Other examples are obvi-
ous and do not need to be mentioned here.
At the end of this section we want to discuss the idea of characterising
services in comparison with the production of physical goods through per-
formance potentials. We can see just how these performance potentials can
 183

also mark or depress physical goods production when the capacities held in
reserve in companies, their so-called "readiness to operate" (Kern
1975, 1992), arenot used to the full. The assembly lines ofan automobile
manufacturer then become a greater hindrance than the performance
potential of a snack bar. In the context of the input system performance
potentials can be described as a combination of potential factor types or of
their possible outputs respectively.

3.3 Activity, technology, efficiency and production function

The aim of the previous arguments was to show that the outputs and inputs
of service production can be described as to their quantities and that, as
with the production of physical goods, real numbers are possible, even
through, in the case of output, mainly integer numbers as subsets. In this
case, according to Koopmans' activity analysis (1951) the input-output
combination of a service production can be shown by an activity vector;
and the set of all activities that a services company can carry out on the ba-
sis of its knowledge and ability can then be described as technology, sub-
ject to the fact that this set of input-output combinations fulfils certain as-
sumptions (Fandel1994, 1996a, 2001a). In the theory of production these
activities are also referred to as production processes (Wittmann 1968;
Krelle 1969). It becomes clear here that the supporters of a process-ori-
ented view of service production find its location directly in production
technology and that from the perspective of the theory of production con-
tradictions with other perspectives do not have to exist.
The discussion whether service production can be described as a sched-
uled process (activity), or must rather be modelled as a period-related
process (Corsten 1988) does not lead us out of the familiar area of produc-
tion-theory analysis instruments either. Either we define the process dura-
tion as the production period, and can then calculate the continuous input
and output intensities through their time integrals into amounts, which are
assumed in the framework of static production theory to be realised at the
end ofthe period as a production point (Fandel2001a), or we map the ser-
vice production in the form of an approach of dynamic production theory
through chronicles or dynamic correspondences as a process taking place
in a time sequence (Shephard!Färe 1980).
The claim that service production is marked by certain inherent laws in
comparison with the production of physical goods has to be admitted in so
far the extemal factor, as a characteristic input (Maleri 1991, 1998), repre-
sents a resource category that up to now has not been found in the historic
field of traditional production theory. In contrast, the existence of inherent
184

laws going beyond this cannot be admitted. Service production is probably

more labour-intensive than physical goods production; but even this is not
a hindrance to the generation of activities. The aspect of performance
quality (Corsten 1986) occurs just as much in physical goods. It is possible
to get around the problern of indeterminableness (Gerhardt 1987), if it is
not founded on the as yet unclear conceptions of the consumer about the
service that he has demanded and not generally in stochastic elements of
production, by breaking a service down into part units, which are then pro-
vided consecutively (for example, a phase concept for the provision of
consultancy services on the introduction of a PPS system [Fandel!Franc;ois
2001]). Otherwise a stochastic production analysis has to be carried out
with the familiar complexity problems. Learning processes that come into
being through the extemal factor "humans" can be treated through
approaches of the dynamic production theory (Shephard/Färe 1980, May
1992) in the same way as learning processes of the intemal factor "human
resource" in physical goods production. The Synchronisation of production
and sales that may be required in certain circumstances, or the coordina-
tion ofservice provider and service consumer (Corsten 1986) is not a diffi-
cu1ty for a theory of production of services as far as the methodology is
concemed; it can be managed organisationally. And the fact that sales may
on occasion lie before production can also be found in production to order
of physical goods. Multistage production processes, such as those that can
occur because of the extemal factor and because of performance potentials
(Corsten 1984, 1998; Corsten/ Stuhlmann 1998), arereal empirical pheno-
mena of the activity analysis.
On the other hand, claiming that the foundation of the inherent laws is
that no production functions of the traditional typology (Kern 1992;
Gutenberg 1994) can be used for services production is a methodological
misunderstanding. The strength of Koopmans' activity analysis, and its
ability to act at the same time as a formal foundation for all types of pro-
duction function, is that it defines a mapping as a production function only
if it maps the efficient activities into zero (Koopmans 1951 ). This is just as
possible for discontinuous (and only finitely many) efficient productions;
for this purpose it does not require any explicit production function.
One aspect has surprisingly hardly been discussed at all in the Iiterature
up to now: whether production functions of services productions are sub-
stitutional or limitational (Gerhardt 1987)? The possible compensation of
an unsatisfactory activity Ievel of the extemal factor in services produc-
tions, in which the important matter is its interaction, by larger amounts of
intemal factors, suggests the assumption of substitutional tendencies.
There is also the consideration that, in comparison with physical goods
productions, services productions are much more labour-intensive. In con-
 185

trast to the limitational tendencies of industrial physical goods production,

which are characterised by a fixed allocation of people, machines and
materials flow, this tends to speak: for the substitutionality ofthe inputs. Of
course, these can be limited from the lower Ievels to the upper because of
the required qualifications, such as with treatment in hospitals. Doctors
could take over the work of nurses, but not vice versa. This means that in
certain circumstances there would only be a very limited peripheral sub-
stitutionality. Of course, in these descriptive cases the classical and neo-
classical production functions would experience a renaissance of economic
attractiveness which they lost as explanatory models to limitational pro-
duction models with greater emphasis on industrial physical goods pro-
duction.
The term efficiency can be applied directly to activities of service pro-
duction if input and output quantities are real numbers, as shown above.
However, the possibility of the explicit modelling of dispositive activities
as intemally produced services and their inclusion as intermediate products
in the extended input scheme opens up new perspectives of differentiating
in the efficiency definition between the usual allocational efficiency and a
planning efficiency or (generalised) a managerial efficiency (Fandet
2001a).
Many approaches have been formulated to describe special services
productions through production functions (Famy 1975; Albach et al. 1978;
Stieger 1980; Haak: 1982; Bode 1993). They all prove that the incentive to
go further along the road of a successful application of microeconomic
production theory is greater than the hesitance that results from the doubts
about not being able to operationalise input and output combinations in
these cases as satisfactorily as with physical goods production.

4 Summary

We have pointed out that concepts analysing the production of material

goods may (in principal) also be used to model the production of services
and to examine its efficiency. The activity analysis approach presents
many opportunities to combine and to integrate theoretical considerations
on the production of services with those devoted to the production of mate-
rial goods. It seems that more precision is needed in order to specify ser-
vices as outputs than in the case of producing material goods. Moreover it
is possible to design an input scheme which is appropriate for both service
production and material goods production. The notions of activity, tech-
nology, efficiency, and production function preserve high significance.
186

References

Albach H (1989) Dienstleistungen in der modernen Industriegesellschaft Mün-

chen
Albach H (2000) Allgemeine Betriebswirtschaftslehre. Wiesbaden
Albach H, Fandei G, Schüler W (1978) Hochschulplanung. Baden-Baden
Altenburger OA (1980) Ansätze zu einer Produktions- und Kostentheorie der
Dienstleistung. Berlin
Berekoven L (1974) Der Dienstleistungsbetrieb: Wesen- Struktur- Bedeutung.
Wiesbaden
Bode J (1993) Betriebliche Produktion von Informationen. Wiesbaden
Carp HJ (1974) Der Transformationsprozess in Dienstleistungsunternehmungen-
Eine Untersuchung der Leistungserstellung im außerindustriellen Bereich,
dargestellt am Beispiel der Verkehrsunternehmung. Berlin
Chenery HB (1949) Engineering Production Functions. The Quaterly Journal of
Economics: 507-531
Chenery HB (1953) Processand Production Functions from Engineering Data. In:
Leontief W, Chenery HB, Clark PG (eds) Studies in the Structure of the
American Economy - Theoretical and Empirical Explorations in Input-Output
Analysis. New York-Oxford, pp 297-325
Corsten H (1984) Zum Problem der Mehrstufigkeit in der Dienstleistungsproduk-
tion. Jahrbuch der Absatz- und Verbrauchsforschung 30: 253-272
Corsten H (1985) Die Produktion von Dienstleistungen. Grundzüge einer Produk-
tionswirtschaftslehre des tertiären Sektors. Berlin
Corsten H (1986) Zur Diskussion der Dienstleistungsbesonderheiten und ihre öko-
nomischen Auswirkungen. Jahrbuch der Absatz- und Verbrauchsforschung
32: 17-41
Corsten H (1988) Dienstleistungen in produktionstheoretischer Interpretation.
WISU 17:81-87
Corsten H (1990) Betriebswirtschaftslehre der Dienstleistungsunternehmen: Ein-
fiihrung. München Wien
Corsten H (1998) Ansatzpunkte fiir ein integratives Dienstleistungsmanagement
In: Bruhn M, Meffert H (eds) Handbuch Dienstleistungsmanagement Wies-
baden, pp 73-92
Corsten H, StuhlmannS (1998) Zur Mehrstufigkeit in der Dienstleistungsproduk-
tion. In: Bruhn M, Meffert H (eds) Handbuch Dienstleistungsmanagement
Wiesbaden, pp 141-162 ·
Diederich H ( 1966) Zur Theorie des Verkehrsbetriebes. Zeitschrift fiir Betriebs-
wirtschaft 36, 1. Ergänzungsheft: 37-52
DyckhoffH (1993) Aktivitätsanalyse. In: Grochla E, Wittmann W (eds) Hand-
wörterbuch der Betriebswirtschaft. Stuttgart, col 57-68
DyckhoffH, Ahn H, Souren R (2002) Übungsbuch Produktionswirtschaft. Berlin
et al.
Engelhardt WH (1989) Dienstleistungsorientiertes Marketing - Antwort auf
Herausforderungen durch neue Technologien. In: Adam D, Backhaus K,
 187

Meffert H, Wagner H (eds) Integration und Flexibilität: Eine Herausforderung

fiir die Allgemeine Betriebswirtschaftslehre. Wiesbaden, pp 269-288
Engelhardt WH, Kleinaltenkamp M, Reckenfelderbäumer M (1993) Leistungs-
bündel als Absatzobjekte: Ein Ansatz zur Überwindung der Dichotomie von
Sach- und Dienstleistungen. Zeitschrift fiir betriebswirtschaftliche Forschung
45:395-426
Engelhardt WH, Kleinaltenkamp M, Reckenfelderbäumer M (1995) Leistungsty-
pologien als Basis des Marketing - ein erneutes Plädoyer fiir die Aufhebung
der Dichotomie von Sachleistungen und Dienstleistungen. Die Betriebswirt-
schaft 55: 673-682
Fandei G (1994) Analysis ofproduction planning and control (PPC) systems as an
efficient combination of information activities. Operations Research-
Spektrum 15 (4): 217-224
Fandei G (1996a) Produktion I. Berlin et al.
Fandei G (1996b) Produktionstheorie, dynamische. In: Kern W, Sehröder HH,
Weber J (eds) Handwörterbuch der Produktionswirtschaft, 2nd edn. Stuttgart,
col 1557-1569
Fandei G (2001a) Interdependencies between network and activity-analytical
descriptions of production relationships in the implementation of large-scale
projects- illustrated by textbook production. International Journal ofProduc-
tion Economics 70 (3): 227-235
Fandei G (2001b) Einige produktionstheoretische Überlegungen zur Veröffent-
lichungstätigkeit eines Universitätsprofessors. Horst Albach zum 70. Geburts-
tag. Zeitschrift fiir Betriebswirtschaft 71: 731-7 42
Fandei G (2001c) Stärkung der Disposition als Wettbewerbsfaktor in mittel-
ständischen Unternehmen durch Enterprise Resource Planning (ERP)-Sys-
teme. In: Sadowski D (ed) Entrepreneurlai Spirits. Wiesbaden, pp 203-221
Fandei G (2003) Zur Beurteilung der Leistungsfahigkeit von Hochschulen aus
produktionstheoretischer Sicht: Eine Anwendung der Data Envelopment
Analysis auf nordrhein-westfalische Universitäten. In: Backes-Gellner U,
Schmidtke C (eds) Hochschulökonomie. Beiträge der Jahrestagung 2002 des
Bildungsökonomischen Ausschusses des Vereins fiir Socialpolitik. Berlin
(forthcoming)
Fandei G, Fran9ois P (2001) IT-gestützte Entscheidungen bei der Ein:fiihrung von
PPS-Systemen. In: Jahnke B, Wall F (eds) IT-gestützte betriebswirtschaftliche
Entscheidungsprozesse. Wiesbaden, pp 271-293
Fandei G, Hegemann H (1986) Kapazitätssteuerung im Krankenhaus- Ein Ansatz
zur Verbesserung der Kapazitätsauslastung in der klinischen Diagnostik.
Zeitschrift fiir Betriebswirtschaft 56: 1129-1147
Fandei G, Paff A (2000) Eine produktionstheoretisch fundierte Kostenrechnung
fiir Hochschulen- dargestellt am Beispiel der FernUniversität Hagen. Zeit-
schrift fiir Betriebswirtschaft 70, 3. Ergänzungsheft: 191-204
Fandei G, Prasiswa A (1988) Planning and Organization ofEconomic Units in the
Field of Out-Patient Medical Care. In: Fandei G (ed) Management Problems
in Health Care. Berlin et al., pp 113-138
Farny D (1975) Produktions- und Kostentheorie der Versicherung. Karlsruhe
188

Garhammer M (1988) Die unbezahlte häusliche Dienstleistungsproduktion - ein

Beitrag zur Diskussion über Dienstleistungsbesonderheiten. Jahrbuch der
Absatz- und Verbrauchsforschung 34: 61-94
Gerhardt J (1987) Dienstleistungsproduktion: Eine produktionstheoretische Ana-
lyse der Dienstleistungsprozesse. Bergisch Gladbach Köln
GutenbergE (1975) Einfiihrung in die Betriebswirtschaftslehre. Wiesbaden
GutenbergE (1994) Grundlagen der Betriebswirtschaftslehre (1): Die Produktion.
24th edn. Berlin et al.
Haak W (1982) Produktion in Banken. Möglichkeiten eines Transfers industrie-
betrieblich-produktionswirtschaftlicher Erkenntnisse auf den Produktions-
bereich von Bankbetrieben. Frankfurt am Main
Hegemann H (1986) Kapazitäts- und Prozessplanung in der klinischen Diagnostik.
Berlin et al.
Hilke W (1989) Dienstleistungs-Marketing. Wiesbaden
Kern W (1975) Kapazität und Beschäftigung. In: Grochla E, Wirtmann W (eds)
Handwörterbuch der Betriebswirtschaft. Stuttgart, col 2083-2089
Kern W (1992) Industrielle Produktionswirtschaft. Stuttgart
Klein-Blenkers F (1964) Die Ökonomisierung der Distribution. Köln-Opladen
Koopmans TC (1951) Analysis of production as an efficient combination of
activities. In: Koopmans TC (ed) Activity Analysis of Production and Allo-
cation. New York, pp 33-97
Krelle W (1969) Produktionstheorie. Tübingen
Laueaster K (1971) Consumer Demand - A new approach. New Y ork
Laueaster K (1972) Introduction to modern microeconomics. Chicago
Maleri R (1973) Grundzüge der Dienstleistungsproduktion. Berlin et al.
Maleri R (1991) Grundlagen der Dienst1eistungsproduktion. Berlin et al.
Maleri R (1998) Grundlagen der Dienstleistungsproduktion In: Bruhn M,
Meffert H (eds) Handbuch Dienstleistungsmanagement Wiesbaden, pp 117-
139
Marshall A (1961) Principles ofEconomics. London New York
May E (1992) Dynamische Produktionstheorie auf der Basis der Aktivitätsanalyse.
Heidelberg
Meyer A (1987) Die Automatisierung und Veredelung von Dienstleistungen -
Auswege aus der dienstleistungsinhärenten Produktivitätsschwäche. Jahrbuch
der Absatz- und Verbrauchsforschung 33: 25--46
Meyer A (1991) Dienstleistungs-Marketing. Die Betriebswirtschaft 51: 195-209
Neumann K, Morlock M (2002) Operations Research. München
Prasiswa A (1979) Die optimale Betriebsgröße einer Arztpraxis. Hagen
Preel B, de la Rochefordiere C (1988) Indikatoren einer Symbiose zwischen
Industrie und Dienstleistungen in Frankreich. In: Clement W (ed) Die
Tertiärisierung der Industrie. Wien, pp 207-236
Rosada M (1990) Kundendienststrategien im Automobilsektor: Theoretische
Fundierung und Umsetzung eines Konzeptes zur differenzierten Vermarktung
von Sekundärdienstleistungen. Berlin
Rössle KF (1954) Allgemeine Betriebswirtschaftslehre. Stuttgart
Rück HRG (2000) Dienstleistungen in der ökonomischen Theorie. Wiesbaden
 189

Rushton AM, Carson DJ (1985) The Marketing of Services: Managing the

Intangibles. European Journal ofMarketing 23 (8): 23-44
Say JB (1852) Cours Complet d'economie politique pratique. 3.eme edition, tome
I. Paris 1852, zitiert nach: Maleri R (1973) Grundzüge der Dienstleistungs-
produktion. Berlin et al.
Schär JF ( 1923) Allgemeine Handelsbetriebslehre. Leipzig
Scheer AW (1995) Wirtschaftsinformatik (Studienausgabe)- Referenzmodelle fiir
industrielle Geschäftsprozesse. Berlin et al.
Schierenheck H (1974) Betriebswirtschaftliche Grundlagen. Gießen
Shephard RW, Al-Ayat RA, Leachman RC (1977) Shipbuilding production fimc-
tion. In: Albach H, Helmstädter E, Henn R (eds) Quantitative Wirtschaftsfor-
schung. Tübingen, pp 627-654
Shephard RW, Färe R (1980) Dynamic theory of production correspondences.
Königstein
Shostack GL (1982) How to Design a Service. European Journal ofMarketing 16
(1): 49-65
Smith VL (1961) Investment and Production. Cambridge (Mass.)
Steven M (1998) Produktionstheorie. Wiesbaden
Stieger H (1980) Zur Ökonomie der Hochschule. Gießen
StuhlmannS (1998) Zur Genese des externen Faktors in der Dienstleistungs-
produktion. Schriften zum Produktionsmanagement des Lehrstuhls :fiir Pro-
duktionswirtschaftder Universität Kaiserslautem (22). Kaiserslautem
Wittmann W (1968) Produktionstheorie. Berlin et al.
Wittmann W (1979) Aktivitätsanalytische Ansätze dynamischer Produktionstheo-
rie und ihre Beziehungen zur Planung. In: Mellwig W (ed) Unternehmens-
theorie und Untemehmensplanung. Wiesbaden, pp 273-304
Wöbe G (1973) Einführung in die allgemeine Betriebswirtschaftslehre. München
Zschocke D (1974) Betriebsökonomie. Würzburg Wien
Strategie Supply Chain Management: A New
Approach to Analyze Product Life Cycles

Günter Fandei

Markus Stammen

1 lntroduction

For many years companies managed their logistical processes in procure-

ment, production and distribution functionally and organizationally inde-
pendently of one another. This focuses of investigation in designing
logistical processes have been displaced, because of increasing competi-
tion, advances in information technology and the onset of globalization.
Meanwhile many companies and especially global players have gone over
to adjusting these processes to one another and to their suppliers and
customers to opening up new business fields and value-added stages. We
refer to this adjustment as supply chain management.
The objective of this paper is to extend the perspective of traditional
supply chain management of procurement, production, distribution and
sales by the business processes of development and recycling to a product
life cycle. This business processes and their facilities are analyzed and de-
scribed in the formulation of a general model to design the strategic supply
chain of one company.

2 Strategie supply chain management

2.1 Definitions

In manufacturing companies the objective of operational activity is to

manufacture products and to arrange the associated business processes
optimally in a company. A process is defined here as a sequence of tasks
and functions which have to be dealt with in a defined order and which
Iead to a result or to a change in condition (cf. Thaler 1999, p. 17 ff.). Key
 191

or business processes which contribute directly to value are created from a

combination ofseveral processes (cf. Schönsieben 1998, p. 24).
Supply chain management covers the short- and long-term collaboration
of a company with other companies to develop and manufacture products
with the required intemal and inter-company organization, planning and
control of the flows of materials, financial value and information along the
business processes of the production of goods or services (cf. Stadtier
2000, p. 9, and Schönsieben 1998, p. 7). Here the production of goods or
services is not limited to the business processes of procurement,
production, distribution and sales but is extended to the operational
functions of development and recycling (cf. Fleischmann et al. 2000, p. 62
f.). The business processes of several companies are linked to one another
via their flows of materials, financial value and information. Through the
combination of business processes networks are designed in which goods,
financial values and information are dispatched and exchanged in a manner
which overlaps the companies involved. These networks are to be referred
to as the supply chain network. Along with the products, essential elements
of the supply chain network are the development centers, suppliers, p1ants,
distribution centers, retail outlets, customers or sales areas and recycling
centers, as well as the transpoft paths between the elements.
Strategie supply chain management is the long-term part of supply chain
management, where the supply chain network is designed and the elements
and processes are fixed for a period of time. It is focussed to the goals and
tasks of company policy and a component of strategic planning. The
alignment of company activities towards customers is of great importance
in strategic supply chain management, so the first step is to develop the
strategies and defined goals and targets for the best way to reach the
customers (cf. Ballou 1999, p. 35). Starting from customer-related strate-
gies, the products and services program and the strategic targets for the in-
dividual business processes must be determined as essential tasks for the
development of the supply chain network. The results of this planning con-
sists of the definition of the planned products, the customers and sales
areas supplied and the number, locations, capacity and type of possible de-
velopment centers, plants, distribution centers, retail outlets and recycling
centers. In addition, the strategically available suppliers have to be selected
from the set of possible sourcing options and the material requirements to
be obtained from them have to be stipulated. In a detailed analysis it may
also be necessary to fix the transpoft paths, the means of transpoft and the
minimum and maximum transpoft volumes between the elements of the
supply chain network (cf. Fleischmann et al. 2000, p. 62f.;
Vidal/Goetschalckx 1997, p. 2). The extended supply chain network with
their basic business processes and connecting flows is shown in Fig 1.
192

Development Procumment Prcductlon Distribution Mal1<eting Sales Recycling

Fig. 1. Extended supply chain network

---· -
lrlomwllcn jby

2.2 Business processes and elements of the extended supply

chain

Elements of the extended supply chain are found in more than one busi-
ness process for the creation of goods and services. These elements may
come from the logistics area as weil as from the financial area. Goods are
the most important process-overlapping elements in strategic supply chain
management. From the logistical aspect they have physical properties such
as weight, size and volume.
Together with goods, the process-overlapping elements of the material
flow include transport paths and warehouses. This means that different
transport times for the means of transport and the paths have to be taken
into account (cf. Kasilingam 1998, p. 51). Consumption and manufacture
of the goods are synchronized through stockpiling in one or more ware-
houses (cf. Schönsieben 1998, p. 8).
To determine the process-overlapping financial flow, goods have to be
evaluated with prices and costs. Customs duties and rates of exchange
must also be included in the individual business processes (cf. Arntzen et
al. 1995, p. 76 f. and p. 86 ff.). Allbusiness processes can completely take
place within a currency union or a tax union. In contrast, if the business
processes occur in countries in which different forms of corporation tax or
different tax rates are found, taxes have to be taken into account. lf the
business processes are not to be limited to a currency union, the financial
variables must be standardized with regard to a reference currency.
The starting point for the creation of goods and services is formed by
the development process. Using customer requirements as a foundation,
planning and developing new products, testing the technical and economic
feasibility and the costing of potential new products all have to be carried
out in collaboration with suppliers and customers (cf. Thaler 1999, p. 81
 193

ff.). Essential elements of this business process are the development

projects from which the new marketable products are created and the
development centers at which the development projects are carried out.
Personnel or technical resources and development costs can be allocated to
both elements.
The materials which are required to carry out the subsequent production
process are obtained in the procurement process. Suppliers who can supply
the materials in the required time and quality are regarded as the central
elements. The basis of the supplier relationships is formed by long-term
contracts with the suppliers in which the material prices are fixed with
minimum and maximum order quantities (cf. Schönsieben 1998, p. 51).
There are varying procurement strategies available when selecting
suppliers, whereby, from a strategic perspective, strategies in particular for
determining the number of sourcing options and the origin of the suppliers
are very important (multiple sourcing, single sourcing, local sourcing,
global sourcing; cf. Kuh/1998, p. 183 ff.).
In the production process input materials are converted or combined into
products. Materials that were acquired in the procurement process are used
for this purpose and they form the material flow into the production
process. The products represent the flow of outgoing materials and goods.
The central elements are the plants which, depending on the aggregation
Ievels, may be complete factories as well as smaller production segments
such as production lines (cf. Kasilingam 1998, p. 51).
The most important tasks in the distribution process are receiving and
distributing products to customers in accordance with their demand. Re-
strictions may be set up with regard to the quantity, weight and volume of
individual products as well as for the complete flow of material at a distri-
bution center. The strategies which are familiar from the procurement
process play an important part in the assignment of distribution centers to
the downstream retail outlets, customers and sales areas (cf. Erengü~ et al.
1999, p. 221 f.).
From the strategic aspect, the preconditions for optimum sales of the
products must be created in the marketing process. The tasks of the mar-
keting process cover organizing sales areas, planning the distribution cen-
ters in the sales areas and drawing up sales forecasts for individual
products. In this context, the sales prices must be fixed for the products in
each of the sales areas. Central elements of the marketing process are the
retail outlets, either in the form of independent sales and service branches
or in the form of cooperation with other companies. The marketing and
service targets must be defined in the sales process in close collaboration
with the distribution process.
194

The sales process, through which products are sold to the final
customers, is associated with the marketing process. The central elements
of this business process are the customers and the sales areas. This paper
does not share the widespread opinion found in the Iiterature on this sub-
ject that the sales process should be seen as a component of the marketing
process, because selling products to customers should be seen as a separate
component of the creation of goods and services.
If the customer is unable to make any further use of the products, the
products are passed into the recycling process. In contrast to the business
processes which have been dealt with up to now, it is not essential to link
the recycling process with the upstream material flow, because the recy-
cling volume depend basically on the expected service life of the products.
The essential elements in the recycling process are the recycling centers
where products can be collected, evaluated and disposed of in environ-
mentally-friendly systems (cf. Thaler 1999, p. 187).
An overview on the business processes and elements gives Fig 2.

Development Supplier Plant Distribution Retail Customer Recycling

center center outlet center

Fig. 2. Business processes and elements in a supply chain network

3 Modeling the problern of optimizing a strategic supply

chain

In the following sections a linear single-period optimization model will be

formulated for the strategic supply chain on the basis of the business
processes shown here. The considerations refer to the business processes in
a company which dominates the supply chain and is in possession of all
decision-making powers with regard to planning for subsidiaries. The
arguments can also be transferred to cooperating organization forms, if all
cooperating companies are joined in a (virtual) company to evaluate the
supply chain.
 195

3.1 Assumptions

In the models of the strategic supply chain management various

assumptions are necessary to reduce the complexity. In upstream sales
planning the companies determine the sales quantities and the sales prices
for the products, differentiated in accordance with sales areas. The sales
price is included in the optimization model as data. There is no assessment
of stocks in most single-period models and it is assumed that the initial
stocks of goods correspond to the closing stocks in the business processes
and that there are no volume lasses between the business processes. In
strategic supply chain management variable and fixed costs are given for
the elements of the business processes. The variable unit costs are allo-
cated to an item and to the development centers, suppliers, plants, distri-
bution centers, retail outlets and recycling centers. Fixed costs occur with
the operation of an element. The net prices planned in the marketing
process and the net prices expected as revenues from recycling are used as
the prices. Todetermine the net income it is necessary to take into account
corporation tax, which usually relates to a country, and which has to be
paid in the respective country. In this optimization model, all relevant
taxes are simplified into a single variable which relates either to an ele-
ment or to a country. Because all processes can take place in different
currency areas, in the models for an international supply chain the prices
and costs shown for all elements must be standardized to a uniform
evaluation base or currency through an exchange rate or an exchange rate
factor. Customs duties can be included through the return transport costs
for the transport of goods between the elements or, as in the following
model, through a product-related duty (cf. Cohen et al. 1989, p.77).

3.2 Model description

Tagether with the rather clear models of strategic supply chain manage-
ment cited shown in the previous chapter, other very extensive models
emphasizing different business processes have been developed, but it
would go beyond the bounds ofthis paper to describe them individually. In
addition, modeling of the whole product life cycle with its business
processes has not been carried out in the current literature. To reproduce
the whole output process of the production of goods a general basic
framework of a linear optimization model for a strategic supply chain will
be developed. The basic framework is based on components of the models
by Amtzen et al. (1995, p. 69 ff.), Cohen et al. (1989, p. 67 ff.),
Huchzermeier/Cohen (1996, p. 100 ff.) and on the representation on strate-
196

gic distribution by Geoffrion/Graves (1974, p. 822 ff.). Their optimization

models will be supplemented by newly developed components.
The description of the supply chain is carried out in the form of a static,
mixed-integer multi-product model. The goal of this strategic supply chain
model is to calculate maximum company operating result together with the
associated determination of the active and inactive goods and elements of
the business processes. While the financial flow is calculated in the objec-
tive function, the material flows can be shown for each business process in
the form of restrictions such as volume and capacity restrictions. From a
strategic perspective, the business processes are planned centrally by a
company. The indices, volumes, parametersandvariables oftbis model are
shown below.

Indices:
c Sales area
d Distribution centers
e Development centers
i,j Goods
p, q Plants
r Recycling centers
s Suppliers
t Countries
V Retail outlets

Sets:
c Set of sales areas
ct Set of sales areas in the country t, Ct c C
D Set of distribution centers
Dt Set of distribution centers in the country t, Dt c D
E Set of development centers
Et Set of development centers in the country t, Et c E
J Set ofgoods
Je Set of marketable products and services, Je c J
Jes Set ofmarketable services, Jes c J
Je Set ofproducts that are created from development projects,
JecJ
Set of goods produced at plant p, J P c J
 197

Set of different materials available from suppliers, Js c J

Union of sets of all supply chain elements
O=(EuSuPuDuVuC uR)
p Set of plants
P; Set of plants at which item i can be produced
Pr Set of plants in country t, Pr c P
Pre(j) Set of goods required to manufacture product j
R Set of recycling centers
R, Set of recycling centers in country t, R1 c R
s Set of suppliers
s, Set of suppliers in country t, S1 c S
T Set of countries
V Set of retail outlets
Vt Set of retail outlets in country t, f7t c V

Logistical parameters:
Production coefficient which indicates the units of the item i
which are required to manufacture a unit ofthe itemj
B Large number
Distribution coefficient which indicates how many time units
are required at the distribution center d to distribute a unit of
quantity of the product i
Development coefficient which indicates how many resource
units are required at development center e to develop product i
Production coefficient which indicates how many time units
are required at the production center p to manufacture a unit
of quantity ofthe item i
Recycling coefficient which indicates how many time units
are required at the recycling center r to manufacture a unit of
quantity of the item i
Cl Minimal capacity of the element I E 0 (in units of time)
Maximal capacity of the element I E 0 (in units of time)
Maximum demand for product i in the sales area c (in units of
quantity)
Minimum throughput of the item i at the element I E 0 (in
units of quantity)
Di! Maximum throughput of the item i at the element I E 0 (in
units of quantity)
198

sc Degree of customer service which indicates the minimum

share ofthe product quantities tobe fulfilled in the sales area c
s.
-I
Minimum number of suppliers for product i

Costs, prices and other financial parameters:

dut1m Mark-up for impoft and expoft duties which have tobe paid
for transpofting the goods from element l E 0 to element
mEO,m-:;:.1
Exchange rate factor for convefting costs and prices in country
t to the standard currency
Exchange rate factor for convefting costs and prices of the
element I E 0 to element m E O,m ::;:.1
Fixed operating costs for the element I E 0 (in GE)
Fixed development costs for the product i at the development
center e (in GE)
Variable unit costs for the product i at element I E 0 (in GE
per unit)
Transfer price of the product i distributed at distribution center
d (in GE per unit)
Transfer price ofthe product i manufactured at plant d (in GE
per unit)
Share of the fixed procurement costs allocated to plant p
Pie Sales price for product i in sales area c (in GE per unit)
Pir Price for the recycling of product i at recycling center r (in GE
per unit)
Fixed operating costs for using the transpoft path from ele-
ment I E 0 to the element m E O,m ::;:.1 (in GE)
Variable transpoft unit costs for the transpoft of item i from
element l E 0 to element m E O,m ::;:.[ (in GE per unit)
Percentage corporation tax rate in country t

Variables:
!Nett Pre-tax Operating result in country t (in GE)
INetD,t Pre-tax operating result ofthe distribution processes in coun-
try t (in GE)
INetE,t Pre-tax operating result ofthe recycling process in country t
(in GE)
 199

JNetP,t Pre-tax operating result ofthe production process in country t

(in GE)
JNetv,t Pre-tax operating result ofthe marketing process in country t
(in GE)
JNetw,t Pre-tax operating result ofthe development process in country
t (in GE)
Losst Pre-tax losses in country t (in GE)
Profttt Pre-tax profits in country t (in GE)

Taxest Taxes on pre-tax profit in country t

xilm Quantity of the item i which is transported from element l E 0
to element m E O,m l *
Ylm Binary variable which indicates whether a material flow takes
place ( Yzm = 1 ) or not ( Yzm = 0 )over the transport path be-
tween the elements l E 0 and m E O,m l *
y1 Binary variable for representing the active ( y1 =1) and inac-
tive ( y 1 =0 ) element l
zisp Binary variable for the material-related allocation of suppliers
to plants with zisp =1 , if supplier s supplies plant p with
material i, and zisp =0 otherwise
ZrJv Binary variable for the allocation ofthe distribution centers to
the retail outlets with ZrJv =1 , if the distribution center d
supplies the retail outlet v, and ZrJv = 0 otherwise
Zvc Binary variable for the allocation of the retail outlets to the
sales areas with Zvc =1 , if the retail outlet v supplies the sales
area c, and Zvc =0 otherwise

3.3 Model formulation

(3.1) z =max L e0t(INett- Taxest)

tET

with the equations

(3.2) Profitt- Losst =!Nett, teT,

(3.3) Taxest =taxt Profttt, teT,

200

(3.4) !Net1 = !Netw, 1 + !Netp,1 + !NetD,t + !Netv,t + !NetE,t, t E T,

(3.5) !Netw,t =L L (-KieYi- KeYe), t E T,

eEEt iEJe

(3.6) !Netp,1 = L (-KpYp + L L (Mip -kip)xipq

pEI't iE(JnJ8 )qEP;,qotp
+L L (Mip -kip)Xipd
dEDiEJc

- L L ((eqp(l + dutqp)M1q +t1qp)x1qp +TqpYqp)

iE( J nJ8 ) qEP; ,qot p
- L L ((esp(l+dutsp)kis +mpKsYs +tisp)Xisp+I'spYsp)), fET,
sESiEJs

(3.7) INetD,t = L (-KdYd + L (L (Mid -kid)xidv

dEDt iEJc vEV
-L ((epd(l+dutpd)Mip +t1pd)x1pd +TpdYpd)), tET,
pEP;

(3.8) !Netv,t =L (-KvYv + L L ((ecv(l-dutvc)Pic -kiv)Xivc)

VEVf CEC iEJc

- L ((edv(l+dutdv)M;d +tidv)xidv +TdvYdv)), tET,

dED
(3.9) INetE,t = L (-KrYr + L L (ecr(l+dutcr)P;rXicr -k;rXicr
rERt CEC iEJc

-ficrXicr - TcrYcr )), f E T,

under the restrictions
(3.10) L Xttm::;; B Ytm• l,mEO,l-:;:.m,
iEJ

(3.11) CeYe::;; L beiYi S Ce Yn eEE,

iEJe

(3.12) DisYs S L Xisp ::;; Dis Ys' iEJs,sES,

pEP

(3.13) Xisp = XispZisp ' iEJs,SES,pEP,

(3.14) L L Ztsp ~Si' i E Js,

pEPsES
 201

(3.15) DipYp:<S; L Xipq+LXipd:<: ::;DipYp· iEJ,pEPi,

qEP,,q*p dED

(3.16) [;_pyP:<:::; L Lbpixipq+ L

LhipXipd:<:::;C PyP, pEP,
qEP,q"*piEJ dEDiEJ

(3.17) LXisp= L ay· L Xjpd• iEJsopEP,

sES jEJc dED

(3.18)

(3.19) DidYd :<:::; L Xidv :<:::; DidYd,

VEV

(3.20) [;.dyd::::; L L bdixidv::::; CdYd, dED,

iEJc VEV

(3.22) L zdv =1, VEV,

dED

(3.23) L Xipd = L xidv'

pEP vEV

(3 .24) Sc Die Yi :<: :; L Xivc :<:::;Die Yi'

- vEV

(3.25) Sc Die:<:::; L Xivc :<:::;Die,

- VEV

(3 .26) L Xjdv = L Xivc ,

dED cEC

(3.27) DivYv :<:::; L Xivc :<: :; Div Yv,

cEC

(3.28) Xivc =Die Zvc,

(3.29) L Zvc =1, cEC,

VEV

(3.30) L Xivc = L Xicr ,

vEV rER
202

(3.31) crYr :$; L briXicr :$;er Yn rER,

CEC

(3.32) Xnm ~ 0, iEJ,/,mEO,

(3.33) Loss1 ,Taxes1 ,Profit1 ~ 0, tET,

(3.34) Yi>Ye>Ys>Yp,yd,Yv•Yr E {0,1}' iEJ,eEE,sES,

pEP,dED,vEV,rER,

(3.35) YtmE{O,l}, l,mEO,l-::t:m,

(3.36) Zisp E {0,1},

(3.37) zdv E {0,1}, dED,vEV,

(3.38) Zvc E {0,1}, vEV,cEC.

3.4 Supply chain network

The supply chain network can be illustrated by a graphical network repre-

sentation where the nodes symbolize all elements of the different business
processes of the strategic supply chain in which a transformation of the
material, financial and information flows takes place. With the notation
used here the branches make the direction of material flows clear on the
transport paths and the direction of the financial flows between the ele-
ments of the business processes. In contrast to the material flows, the
financial flows cannot be allocated definitely to a clear direction because
they do not always run in the opposite direction to the material flows, for
example as a result of the recycling process. The essential structure of the
supply chain network is shown in Fig. 3; a description of the information
flows will not be provided here. The upper area shows clearly the demar-
cation of the business processes in accordance with product-related pro-
duction areas. The representation of the materials flow through the
different elements of the business processes is linked to this and starts
from the procurement process through to the recycling process. The ele-
ments of the business processes are named in the middle. The bottom area
shows the financial flow in the supply chain network.
 8
8
Developmen t Supplier Plant Distribution Marketing Customer/ Recycling
center center center Sales areas center

8 e ~ e e 8 e
8 @ e. 8· 6· E9 e

Fig. 3. Graphical network representation ofthe supply chain network N

0
w
204

3.5 Objective function

The definitions of the objective function (3.1) to (3.9) contain an integra-

tion of the process-related results for an international application supple-
mented by exchange rates, customs duties and taxes and the determination
of the maximized overall operating result a:fter taxes. The maximization of
the operating result is achieved through the aggregated objective function
in equation (3.1), in which the country-specific operating result less the re-
spective taxes are aggregated and standardized to a reference currency.
This model follows the taxation practice that companies pay taxes if there
is a profit in a country, but no taxes are to be paid if there is a loss (cf.
Huchzermeier/Cohen 1996, p. 103). This calculation is carried out with the
relationships (3.2) and (3.33) in that either the variable Profit1 stands for a
positive operating result or the variable Loss1 for a negative operating re-
sult not equal to zero and correspond to the pre-tax operating result /Net1
in country t, t E T. The tax yield Taxes1 can be determined via a simple
multiplication of the positive operating result Profit1 by the tax rate tax1
in country t in equation (3.3). The business process related operating re-
sults of a country are aggregated in equation (3.4) to a country-specific
operating result.
The country-specific costs for the development process in equation (3.5)
result from the fixed costs K;e ofthe active development projects i, i E Je,
and the costs of the active development centers Ke, e E E1 • For the
production process the result is determined in equation (3.6) for all plants
in country t, t E T, from the quantities of manufactured goods evaluated
with transfer prices M;p , which have to be included less production and
procurement costs including the respective transpoft costs (cf. Cohen et a/.
1989, p. 78). The production costs consist of the fixed costs KP for
operating the active plants p, p E E: , and the variable production unit costs
k;p for each unit produced of product i, i E J, which is supplied to further
plants q, q E P;, q -:t:. p, or to the distribution centers d, d E D. In the case
of multi-stage production the plants p, which receive goods from other
plants q, bear the costs ofthe deliveries. These result from the quantities of
goods evaluated with the transfer prices M;q and the variable transpoft
unit costs f;pq and from the fixed transpoft costs Tpq • The procurement
costs of the materials i, i E 1 8 , delivered by suppliers s, s ES , to the
plants p are made up of the supplied material quantities evaluated with the
 205

material unit price kis and the transport unit price tisp . If a supplier s is
activated (Ys =1) and supplies to several plants, allocation of the fixed
procurement costs to the receiving plants cannot be done conclusively. In
this case the fixed procurement costs Ks are distributed through the
parameter mP , whereby the allocation of all fixed procurement costs can
be ensured through the relationship L mP =1 . In contrast, the fixed trans-
peP
port costs Tsp for operating the transport path in material procurement can
be allocated directly to a plant. The outgoing financial flow of the pro-
curement process and other plants are to be multiplied by the exchange
rate factors esp and eqp and tobe extended by customs duties dutsp and
dutqp.
The products i, i E Je, distributed by distribution centers d, d Dt, to
E

sales centers v, v E V, and evaluated with their transfer prices Mid provide
the positive contribution to the determination of the operating result for the
distribution process in equation (3.7). In contrast, the distribution fixed
costs Kd for operating the active distribution centers and the variable
distribution unit costs kid for each unit of product i delivered to the retail
outlets reduce the operating result. The product i delivered to the distribu-
tions centers is acquired at the transfer price Mip from the plant p, p E P ,
in which the product is manufactured. The fixed costs for operating the
transport path Tpd and the variable transport unit costs tipd for each trans-
ported unit are included for the transport of the products from the produc-
tion to the distribution centers. In addition, the financial flow of the pro-
duction process must be multiplied by the exchange rate factors epd and
extended by the customs duties dut pd .
The country-specific pre-tax operating result of the sales process of
country t, t E T, is calculated from equation (3.8). The positive contribu-
tion to the operating result is given through the sales quantities evaluated
in the marketing process with the sales prices Pie . The fixed sales costs for
operating the active retail outlets Kv and the variable marketing unit
prices kiv for selling the product i, i E Je, must be included on the costs
side. In addition, the incoming products evaluated with the transfer price of
the distribution centers Mid , the variable transport unit prices tidv for the
supply of the retail outlets and the fixed costs for operating the transport
path Tdv reduce the operating result. The incoming financial flow of the
206

distribution process and the financial flow caused by the sale of the
products must be evaluated with the exchange rate factors edv and ecv and
reduced by the customs duties dutdv and dutvc to determine the operating
result in the marketing process. It should be noted that the outgoing finan-
cial flow is converted to the currency of the retail outlet with the exchange
rate factor ecv .
The operating result for the recycling process arises for each country t,
t E T, from relation (3.9). On the one hand, a positive contribution per
product to the operating result can be achieved in the recycling process
through the sale of the recycling service by the recycling centers r, r ER,
to the customers c, c E C . On the other hand, the recycling costs and the
transport costs reduce the operating result. With the recycling costs, the
variable recycling unit costs kir and the fixed costs for operating the active
centers Kr should be noted. The transport costs consist of the variable
transport costs ticr and the fixed operating costs for the transport paths
Tcr . The operating result of the recycling process must also be evaluated
with the exchange rate factor ecr and reduced by the customs duties dutcr .

3.6 Constraints

Various constraints must be set up in a process-overlapping system to cal-

culate the variables of the material and financial flows. The non-negativity
restriction (3.32) applies for the volume variable Xifm which marks the
material flow between two elements. In addition, the various binary
variable are taken up through restrictions (3.34) and (3.35). The modeling
of the material flow via the transport paths and the condition of the trans-
port paths is done with the help of restriction (3.10) (cf.
Huchzermeier/Cohen 1996, p. 103).
The product creation and development process has not been taken in
account in previous models of strategic supply chain management in the
literature. Restrietion (3 .11) can be used to record the consumption of re-
sources of the development projects i, i E Je, at the development center e,
e E E , through the development coefficients bei and limited by the
planned development capacities.
To model the procurement process, restrictions on delivery quantity
Iimits ((3.12)), on the procurement strategies of "single sourcing" and
"multiple sourcing" ((3.13), (3.14), (3.36)) and on the procurement strate-
gies of "local sourcing" and "global sourcing" must be set up. The mini-
 207

mum and maximum delivery quantities for a material are fixed in long-
term framework agreements with suppliers which are mapped through
relation (3.12) (cf. Cohen et al. 1989, p. 81). The procurement strategies on
local sourcing and global sourcing are to be realized through delimiting the
available suppliers.
The strategic production process is shown through the volume and
capacity restrictions of the plants ((3.15), (3.16)) and the material re-
quirement assessment. In single-stage production the volume of the re-
quired materials can be determined from equation (3 .17) (cf. Cohen et al.
1989, p. 81). For each material i, i E Js, the equation links the material
flow retention at a plant p, p E P , with the program-based material
requirement assessment. Equation (3 .18) is to be used for the case of
multi-stage production through several, including alternative, plants and
for the case of the transpoft of the products between different plants (cf.
Amtzen et al. 1995, p. 85 ff.). The left-hand side ofthe equation shows the
incoming materials and intermediate products for each plant, the right-
hand side gives the outgoing products to the distribution centers and to
other plants.
In the model of the strategic distribution process the volume and
capacity restrictions of the distribution centers can be mapped analogously
to the production process through relations (3.19) and (3.20). Restrietions
(3.21), (3.22) and (3.37) mark the distribution strategy of single sourcing.
For retail outlets the materials flow maintenance in the distribution centers
is modeled with the equation (3.23). The restrictions on demand fulfill-
ment, materials flow maintenance, single sourcing and quantity Iimits
serve the description of the marketing and sales processed. The restriction
on demand fulfillment is of great importance in the models on strategic
supply chain management and is used in different forms (cf. Cohen et al.
1989, p. 82 and Arntzen et al. 1995, p. 88). Two possibilities should be
used to model demand fulfillment. Restrietion (3.24) can be used for the
products which can be created in the development process. In the case of
products are already developed and are marketable, demand fulfillment can
be modeled using restriction (3.25). The necessary product volumes are to
be determined using restriction (3.26) on materials flow maintenance. lt
should be noted that service products i, i E Jcs, which are not subject to
materials flow maintenance can also be offered in the marketing.
Restrietion (3.27) Iimits the minimum and maximal materials flow in the
active retail outlets. The marketing strategy of "single sourcing" can be
shown in the marketing process, comparable to the distribution process, if
customers aretobe supplied from a single retail outlet only. A retail outlet
208

is allocated to a sales area through restrictions (3.28), (3.29) and (3.38) on

the basis of the same system as in the distribution process.
The recycling process can be described through the restrictions on mate-
rials flow maintenance and on capacity Iimits. Because of the alignment to
the product-related output process of products, it materials flow mainte-
nance is assumed in restriction (3.30), so the complete produced quantity
enters the recycling process after the products are used. Alternatively,
comparable to restriction (3.24) demand for the recycling of a product
could be determined for the recycling process through which the materials
flow into the recycling process is measured. The capacity limit at the
recycling centers is considered in restriction (3 .31 ).

4 Applications and outlook

The applications of this model are varied because it was deliberately con-
figured to the supply chain and can be restricted to the respective required
business processes for each concrete application. One main industry with
interest in an extended supply chain management perspective is the auto-
motive industry. Car manufactures became global company with a global
development, sourcing, manufacturing and selling. These companies face
high efforts in developing new cars, but have very slow innovation strate-
gies between 8-10 years. In the European Union the manufactures are
forced to recycle the new sold cars by EU law from 2007 onwards. There-
fore these firms have a rising interest in optimizing their supply chain net-
work from the developing up to the recycling process.
The advantage of this model is that investment decision between con-
current products and projects can be made. The model determines the op-
timal product combination with considering the restriction of a consisting
supply chain network with its limiting resources and elements of the
product life cycle. Minimum sales volumes for the chosen product combi-
nations can also be identified.
There are also wide-ranging options for extensions, whereby the most
important are found in the inclusion of stochastic variables for the forecast
of customer demand and exchange rates and in an extension to include
multi-period observations. Some of the model's components can also be
developed further, such as the detailed modeling of taxes and customs du-
ties and of the transport system. In addition, an extension of the basic
framework to include sales-policy decisions is also conceivable. An addi-
tional emphasis might also be found in the optimization and causation-re-
lated allocation of the operating result of overlapping supply chains.
 209

On the whole, strategie supply ehain management represents a new, ex-

tended perspeetive in the framework of strategie planning whieh will be-
eome inereasingly important for globally aetive eompanies. Beeause of the
eurrent advanees in information teehnology it is possible to ealeulate even
extensive optimization models. Forthis reason, the eapability toset up and
solve models of Strategie supply ehain management will develop into an
important eomponent of strategie planning.

References

Amtzen BC, Brown GG, Harrison TP, Trafton LL (1995) Global Supply Chain
Management at Digital Equipment Corporation. Interfaces 25: 69-93
Ballou RH (1999) Business Logistic Management. 4.th edn. New Jersey
Cohen MA, Fisher M, Jaikumar R (1989) International Manufacturing and
Distribution Networks: A Normative Model Framework. In: Ferdows K (ed)
Managing International Manufacturing. Amsterdam, pp 67-93
Erengüc ~P, Simpson NC, Vakharia AJ (1999) Integrated Production/Distribution
Planning in Supply Chains: An Invited Review. European Journal of Opera-
tional Research 115: 219-236
Fleischmann B, Meyr H, Wagner M (2000) Advanced Planning. In: Stadtier H,
Kilger C (eds) Supply Chain Management and Advanced Planning -
Concepts, Models, Software and Case Studies. Berlin et al., pp 57-71
Geoffrion AM, Graves GW (1974) Multicommodity Distribution System Design
by Benders Decomposition. Management Science 20: 822-844
Ruchzermeier A, Cohen MA (1996) Valuing Operational Flexibility under Ex-
change Rate Risk. Operations Research 44: 100-113
Kasilingam RG (1998) Logistics and transportation - Design and Planning.
Dordrecht et al.
Kuhl M (1998) Wettbewerbsvorteile durch kundenorientiertes Supply Manage-
ment. Wiesbaden et al.
Schönsieben P (1998) Integrales Logistikmanagement Planung and Steuerung
von umfassenden Geschäftsprozessen. Berlin et al.
Stadtier H (2000) Supply Chain Management - An Overview. In: Stadtier H,
Kilger C (eds) Supply Chain Management and Advanced Planning - Con-
cepts, Models, Software and Case Studies. Berlin et al., pp 7-29
Thaler K (1999) Supply Chain Management, Prozessoptimierung in der Logistik
Kette. Köln
Vidal CJ, Goetschalckx M (1997) Strategie Production-Distribution Models: A
Critical Review with Emphasis on Global Supply Chain Models. European
Journal ofüperational Research 98: 1-18
An Analysis of Service Output Based on
Production Theory

Ralf Gössinger

1 Basics

The basic two-stage model of service production (e.g. Corsten 1984, pp.
263 ff.; Pranz 1969, p. 87; Haak 1982, pp. 173ff.; Herzig 1975, p. 292; for
a multi-stage model see Altenburger 1979, pp. 868ff.) serves as a starting
point for considerations. At the prior combination stage, the production
factors of a services enterprise are combined to create a readiness to
provide a service (for the readiness to operate see Gutenberg 1983, pp.
2ff., pp. 96ff. and pp. 171ff.). In thefinal combination an intended benefi-
cial alteration (e.g. Carp 1974, p. 37 and pp. 41f.; Mengen 1993, pp. 25f.)
is then made to the features of the extemal production factor (for a com-
prehensive analysis see Stuhlmann 1999, pp. 25ff.) contributed by one or
more customers. This alteration constitutes the output of the final combi-
nation and is referred to below as service output. Due to the specific
features of services special importance is attached to the term service
quality (see Meyer/Mattmüller 1987, pp. 189f.): on the one hand the
service quality is determined by production factors on the part of the
service provider and customer and, on the other hand, the ascertainment of
service quality is rendered more difficult by the high proportion of features
which elude physical measurement. Both aspects result in majorpotential
being created that the service provider and customer will perceive dif-
ferences in the quality. The task of the following contribution is thus to
propose a model of output and quality of the final combination based on
production theory which takes account of the perspectives of both the con-
sumer and the service provider.
 211

2 Products as solutions to problems

2.1 General specification

In the literature, the view has established itself that products are a bundle
of values ("Leistungsbündef'; for a discussion of the translation problern
caused by this term see Kleinaltenkamp/Jacob 2002, p. 152) (e.g.
Arbeitskreis 1975, pp. 759ff.; Bressand 1986, p. 78; ChisnaH 1985, pp.
48ff. and pp. 326 ff.; Engelhardt 1976, pp. 79ff.; Engelhardt/Kleinalten-
kamp/Reckenfelderbäumer 1993, pp. 407ff.; Haak 1982, p. 77; Harnmann
1974, pp. 136ff.; Shostack 1977, p. 74ff.), the components of which are
offered as a unity to satisfy the needs of the customer. This view is in
accordance with the output definition of production theory. On the one
hand the bundle concept represents a similar notion to that of products in
production theory which are defined as goods resulting from the combina-
tion process ofproductive factors (e.g. Chmielewicz 1967, p. 14). On the
other hand, the process of product formation in both positions is charac-
terised by the stages of creating capability and of utilising this capability
(see Kleinaltenkamp 1993, pp. 108f.). However, it must be emphasised
that the bundle concept explicitly takes both the perspective of the
customer and that of the service provider into account. Viewing an analy-
sis of services from the angle of production theory, this aspect gains
importance because a part of the services is characterised by an at least
partial simultaneity ofproduction and sales (see Corsten 1985, pp. 110ff.).
One possibility of including the customer and supplier perspectives in a
production theoretical manner is to modify the product notion so that
products have to have the potential to satisfy the wants ofthird parties (see
Haak 1982, p. 81). If a want is thereby characterised as a customer's indi-
vidual problern and the satisfaction of wants is characterised as a problern
solution accepted by consumer, it is then possible to generaHy interpret
products as being problern solutions which are found by the producer for
the consumer (see Kern 1979, col. 1434ff.).
In the possibility shown below of aHowing for this product notion in the
modeHing based on production theory, a feature-oriented modelling of the
production of goods is taken as a basis which refers to the microeconomic
consumer theory formulated by Lancaster (see Lancaster 1966, pp. 132ff.;
in the context of services e.g. Ehret 1998, pp. 214ff.; Schade 1996, pp.
75ff.; Woratschek 1992, pp. 38ff.) and to the process-oriented production
theory formulated by Bebrens (see Bebrens 1999, pp. 303ff.; Bebrens
212

2000, pp. 167ff.; for process-oriented production theory seealso Bea 1995,
pp. 38ff.).
A problern P ensues for an economic unit from a discrepancy perceived
as negative and intolerable between the actual state E1 and the target state
E of an object under consideration in its sphere of disposal (e.g. Bretzke
1980, pp. 33f.; Fisk 1981, p. 191; for a similar modeHing seeSchneeweiß
1995, p. 100):
(1)

The states can be described by means of feature vectors which at least

must take account of the relevant dimensions on the customer's side and
on the service provider's side (for differences in the perception of features
see Chamberlin 1953, p. 4 and pp. llf.; Chmielewicz 1967, pp. 37f.).
Within this modeHing two classes of feature cornponents are to be
considered. Static cornponents es describe the state of an object in a point
of time. Whereas dynarnic cornponents ed are used to describe the changing
of the corresponding static components within a period. These changes are
caused by exogenous processes affecting the object (e.g. aging processes):

E ...
= [(e~, ,e~)l
(2)

(el , ... ,ez)

A problern solution P' then consists in causing a reduction of the dis-

crepancy to a tolerable extent (see Johnston 1995, pp. 46ff.):
(3)

A reduction of the discrepancy between actual state and target state

occurs as a result of problem-solving processes which are defined by an
actions systern ~ which can be modeHed by the actions AI = (aj.I, ... , afN)
and their relations Rj(AI) to one another:
(4)

For the output analysis, an action afn is described by the vector E(afn) of
the alterations accompanying it to the features of the object considered. In
respect of these alterations, substitutional and limitational relations can
exist between the actions. The alterations to features E(Sj) accompanying
the realisation of an actions system are thus also dependent on the links
between the actions defined in the actions system:
(5)
 213

Because of the two types of feature components, it has to be differenti-

ated between two types of feature alteration:
- The state a/teration is used if the problern arises from a discrepancy be-
tween actual static components and target static components in a point
of time. A suitable actions system is targeted on an alteration of relevant
static components.
- In the case of process a/teration, also dynamic feature components are
relevant. Because they will Iead to a'n alteration of static components
within a period, in this case the problern arises from the discrepancy
between a fictitious future state and the target state. A suitable actions
system is targeted on a compensation of the dynamic component's
effect.

2.2 Service-oriented specification

To pointout one way in which the service output and service quality can
be modelled, the macrostructure shown below in Figure 1 is used which
shows the stages involved in the transformation of a customer's service
problern into a solution. Simultaneously this macrostructure serves as a
preliminary structuring for a detailed modelling of the individual elements
at a micro Ievel to be carried out in the course of further research.

Q .v
•

0 A
..

.. 0 ..

~O N ---

Fig. 1. Transformation of a problern in the course of the final combination

As a consumer of problem-solving services, an economic unit

approaches a service provider by making the problem-beset object under
consideration available to the service provider and by describing the
perceived problern P. The articulation represents a mapping of the
214

perceived problern in a model P (articulated problem) which the

customer generates through the transformation of information fN{ 0 ) (for
modeHing see e.g. Bretzke 1980, pp. 37ff.; Dyckhoff 1994, pp. 23ff.;
Kosiol 1961, pp. 319f. ). In the context of services production one has to
consider that the customer not only gains an advantage :from the result
remaining after the completion of the production process but also :from the
process itself and the resources employed in this connection (see
Donabedian 1980, pp. 79ff.; Hilke 1984, pp. 7ff.). The featurestaken into
account when articulating the problern can thus refer to the final state (E)
to be achieved by means of the problem-solving process, to the course of
the problem-solving process (V) and to the resources (M) used to imple-
ment the problem-solving process:

E= (
s.E s.E s.V s.V s.M
el , ... ,ex ,eX+I>···•eY ,eY+I•····ez
s.M J (6)
d.E d.E d.V d.V d.M d.M
el , ... ,ex ,eX+I•· .. ,ey ,eY+l , ... ,ez

The articu1ation of final state-related, process-related and resource-

related features requires that the customer has already formed an idea
about how the service has to be produced to solve his problem. This
bundle of expectations is called consumer's script (see Hubbert/Sehorn/
Brown 1995, pp. 71ff.). For the articulated problern the following applies:
(7)

By means of the information transformation function fN{ 0 ) the situation

is also taken into account that the economic unit only refers to those fea-
tures to describe the problern which it can perceive and which it regards as
relevant. Moreover, it has been noted that the description of the problern
can change in the course of the production of services due to the commu-
nication process which simultaneously takes place and due to the partici-
pation in the production process of services (see Scharitzer 1993, p. 102;
Weiber/Jacob 2000, pp. 564ff.) ifthe perception of features or the assess-
ment of their relevance is influenced by information provided by the
service provider h
It is the service provider's task to interpret the problern on the basis of
the incomplete problern articulated by the customer, i.e. a model P is
created ofthe customer's problem, which is named the perceived problem:
(8)

by combining the information about the problern P articulated by the

consumer with further information JA that is derived :from service
provider's knowledge (see Weiber/Jacob 2000, pp. 528ff.).
 215

In the course of the transformation .fA( 0 ) of the service provider' s infor-

rnation features describing the problern are repeatedly selected. On the one
band, it is possible in this connection for features to gain in irnportance of
which the custorner was not previously aware or which were classified as
irrelevant; on the other band, it is possible for the features articulated by
the consumer to lose irnportance due to their poor perceptibility or due to
their classification as irrelevant by the service provider. As it is possible
for the problern articulated by the consumer to change in the course of the
creation of services, it is also possible for the perceived problern to change.
The perceived problern P is the starting point for the creation of
services by the service provider. Basedonhis knowledge ofhow to solve a
problern, he develops a target notion P s of the solution to the problern.
The features of the problern throw light on which of the actions systerns Sj
known to hirn can be applied and how such a systern is to be organised in
detail. This actions systern represents a description of the service and is
called a provider's script (see Srnith/Houston 1983, p. 60) or a blueprint
(see Shostack 1981, p. 226). As an actions systern can be applied to
problerns in a specific class of problerns, one can assume, particularly for
services, that appropriate degrees of freedorn exist within an actions
systern to organise it which can be used depending on the specific problern
(for an application ofnon-linear production plans in service production see
Corsten 2002, pp 64f.). It is therefore necessary for an actions systern to
state ranges (e.g. rninirnum and rnaxirnum values) ofpossible alterations to
features:

["'E es.V ...

-•MJ
(9)
~s.E ~s.V ~s.M
j.l ej.X ej.X+l j.Y ej.Y+t ej.Z
es.E
j.l
es.E
j.X
-s.V
ej.X+l
es.V
j.Y
-s.M
ej.Y+l
... e~·M
J.Z
E(Sj) = -d.M ...
["".E
ed.E ~d.V ed.v
j.l
ed.E
j.X
ed.E
ej.X+l
-d.V
j.Y
-d.V
ej.Y+l
-d.M ...
e".MJ
J.Z
-d.M
j.l j.X ej.X+t ej.Y ej.Y+t ej.Z

The set of relevant actions systerns SCO) which can be altematively

applied to solve the problern then results frorn:
.= =,s
S(P) = {Sj I p = f(P= ,E(Sj )) /\]. =1, ... ,J} (10)

After the application of an actions systern, a solution P' then exists to

the problern perceived by the service provider. The next step of the trans-
formation process is that the service provider articulates the solution to the
custorner (articulated problern solution P'). For this, he specifies the
features of the solution which are relevant for hirn. F or the problern solu-
tion perceived by the consurner the following applies:
216

(11)

The actual service output 0 consists in the difference between the

problern articulated by the customer P and the solution P' articulated by
the service provider:
(12)
However, it is only possible for the service provider to ascertain the
output 0 A on the basis of the alterations to features perceived by him and
regarded as relevant:
(13)

The perceived problem-solving contribution ON then results for the

customer by comparing the problern originally experienced with the prob-
lern solution perceived on the basis of the features relevant for him:
(14)
The output is thus mapped as an alteration vector which represents the
contribution to solving a customer's problern and is perceived differently
by the customer and the service provider (see Scharitzer 1993, p. 96;
Grönroos/Ojasalo 2000, pp. 8ff.).
If quality is understood tobe a set ofthe features assigned to a good (see
Chmielewicz 1967, p. 37; Kawlath 1969, pp. 67ff.; Klatt 1961, pp. 22ff.;
Lücke 1973, p. 266), then it is expedient to take the feature-related
modelling of service output as a basis for quality modelling. The assess-
ment of quality by an economic unit ensues from a comparison between
the actual and the target features of the solution to the problern (see
Grönroos 1982, pp. 60ff.; Hentschel 1992, pp. 35f.; Parasuraman/
Zeithaml/Berry 1985, p. 48) whereby the target features express the
expectations that an economic subject has regarding the alterations to
features to be realised by the service. These expectations can be illustrated
as the vector of the target features of the problern solution for the customer
P's and the service provider P's (see Scharitzer 1993, p. 101) which can
vary due to the increase in information during the creation of the services
(for the alteration oftarget features over time see Lücke 1973, p. 271). The
quality assessments then constitute appraisals of the differences between
the expected and achieved problern solution:
(15)
 217

(16)

The different perceptions of output and qua1ity are due to agent-induced

and prob1em-induced factors. One important agent-induced factor can be
seen in the problern evidence of customer and provider. Problem evidence
describes the ability of an agent to recognise the problern he is faced with
and to recognise the suitability ofproblem-solving actions (see Engelhardt/
Schwab 1982, pp. 506ff.; Emenputsch 1986, pp. 50ff.; Fließ 2001, pp.
68f.; Kleinaltenkamp 1993, pp. llOff.). Cornrnunication capability is
another agent-induced factor. The more efficiently the agents are able to
articulate/perceive the features relevant to the problem, the smaller the
differences in output and quality perception will be. One essential problern
induced factor is the rneasurability of the features. For services one can
assume that the differences perceived in output and quality are all the
greater the smaller the share is of physically measurable features (see
Scharitzer 1993, pp. 96f.). It is possible to confirm for services that there is
a tendency towards a decline in the ability to physically quantify the fea-
tures of the result, course and resources of the problem-solving process
considered here: while it is oftentimes possible to physically quantify the
features of the resources used in the problem-solving process, the degree
of quantification decreases for the features of the course and even more so
for the features of the result of the problem-solving process. As a service
also always has features which can be physically quantified, one can
assume that the output and quality of a service can be quantified in princi-
ple even if the accuracy and the effort required to quantify these differ
depending on the service (see Fließ 2001, p. 184f.). Ifonly a small share of
relevant features is used for the output evaluation, the evaluation results
have tobe interpreted with utmost caution (see Corsten 1994, pp 56ff.).

3 Final remarks

Due to the influence exerted by information and communication processes

on service output and its evaluation, it is necessary to consider these
processes in addition to the basic problem-solving processes. Taking into
account the circumstance that both the service provider and the customer
are involved in the process of service production, it appears necessary to
explicitly include the perspectives of both. Simultaneously, to pay respect
to the importance of service quality in this contribution, a feature-oriented
modelling of the production of services is taken as a basis. The analysis of
service output in the course of final combination is orientated on a macro-
218

structure of stages involved in the transformation of a customer's problern

into a solution. This approach will serve as a preliminary structuring for a
detailed modeHing of the individual elements at a micro Ievel at a later
stage of research.

References

Altenburger OA (1979) Potentialfaktoren als derivative Produktionsfaktoren der

Dienstleistungsproduktion. Einige Gedanken zur Produktionstheorie der
Dienstleistungen. Zeitschrift für Betriebswirtschaft 49:863-872
Arbeitskreis "Marketing in der Investitionsgüter-Industrie" der Schmalenbach-
Gesellschaft (1975) Systems Selling. Zeitschrift für betriebswirtschaftliche
Forschung 27:757-773
Bea FX ( 1995) Prozeßorientierte Produktionstheorie und Lernen. In: Albach H,
Wildemann H (eds) Lernende Unternehmen, ZfB-Ergänzungsheft 3.
Wiesbaden, pp 35-47
Behrens S ( 1999) Grundlagen der prozeßorientierten Produktionstheorie. In:
Albach H et al. (eds) Die Theorie der Unternehmung in Forschung und Praxis.
Berlin Heidelberg New York, pp 297-312
Behrens S (2000) Produktionstheoretische Perspektiven der Virtuellen Unterneh-
mung. In: Albach H, Specht D, Wildemann H (eds) Virtuelle Unternehmen,
ZfB-Ergänzungsheft 2. Wiesbaden, pp 157-176
Bressand A (1986) Dienstleistungen in der neuen "Weltwirtschaft": Auf der Suche
nach einem konzeptionellen Bezugsrahmen. In: Pestel E (ed) Perspektiven der
Dienstleistungswirtschaft. Göttingen, pp 73-82
Bretzke W-R (1980) Der Problembezug von Entscheidungsmodellen. Tübingen
Carp H-J (1974) Der Transformationsprozess in Dienstleistungsunternehmungen.
Eine Untersuchung der Leistungserstellung im außerindustriellen Bereich,
dargestellt am Beispiel der Verkehrsunternehmung. Diss. Berlin (FU)
Chamberlin EH (1953) The Product As An Economic Variable. The Quarterly
Journal ofEconomics 67:1-29
ChisnaH PM (1985) Strategie Iudustrial Marketing, Englewood Cliffs et al.
Chmielewicz K (1967) Grundlagen der industriellen Produktgestaltung, Berlin
Corsten H (1984) Zum Problem der Mehrstufigkeit in der Dienstleistungs-
produktion. Jahrbuch der Absatz- und Verbrauchsforschung 30:253-272
Corsten H (1985) Die Produktion von Dienstleistungen. Grundzüge einer
Produktionswirtschaftslehre des tertiären Sektors. Berlin
Corsten H (1994) Produktivitätsmanagement bilateraler personenbezogener
Dienstleistungen. In: Corsten H, Hilke W (eds) Dienstleistungsproduktion.
Wiesbaden, pp 43-77
Corsten H (2002) Controlling der Dienstleistungsproduktion. In: Lingnau V,
Schmitz H (eds) Aktuelle Aspekte des Controllings. Festschrift für Hans-Jörg
Hoitsch. Heidelberg, pp 49-72
 219

Donabedian A (1980) The Definition of Quality and Approaches to its

Assessment. Ann Arbor
Dyckhoff H (1994) Betriebliche Produktion. Theoretische Grundlagen einer
umweltorientierten Produktionswirtschaft, 2nd edn. Berlin et al.
Ehret M (1998) Nutzungsprozesse als Ausgangspunkt des Innovations-
managements. In: Engelhardt WH (ed) Perspektiven des Dienstleistungs-
marketing. Ansatzpunkte fiir Forschung und Praxis. Wiesbaden, pp 189-241
Engelhardt WH (1976) Erscheinungsformen und absatzpolitische Probleme von
Angebots- und Nachfrageverbunden. Zeitschrift fiir betriebswirtschaftliche
Forschung 28:77-90
Engelhardt WH, Kleinaltenkamp M, Reckenfelderbäumer M (1993) Leistungs-
bündel als Absatzobjekte. Ein Ansatz zur Überwindung der Dichotomie von
Sach- und Dienstleistungen. Zeitschrift fiir betriebswirtschaftliche Forschung
45:395-426
Engelhardt WH, Schwab W (1982) Die Beschaffung von investiven Dienst-
leistungen. Die Betriebswirtschaft 42:503-513
Emenputsch MA ( 1986) Theoretische und Empirische Untersuchungen zum
Beschaffimgsprozeß von konsumtiven Dienstleistungen. Bochum
Fisk RP (1981) Toward a Consumption!Evaluation Process Model for Services.
In: Donnelly JH, George WR (eds) Marketingof Services. Chicago, pp 191-
195
Fließ S (2001) Die Steuerung von Kundenintegrationsprozessen. Effizienz in
Dienstleistungsuntemehmen. Wiesbaden
Franz J (1969) Die Labilität der Nachfrage bei investitionsintensiven, kunden-
präsenzbedingten Dienstleistungsbetrieben und die Möglichkeiten einer
betriebspolitischen Anpassung, Diss. München (LMU)
Grönroos C (1982) Strategie Management and Marketing in the Service Sector.
Research Report no 8, Swedish School of Economics and Business
Administration. Helsingfors
Grönroos C, Ojasalo K (2000) Service Productivity: Toward a Conceptualisation
of the Transformation of Inputs into Customer Value in Services. Research
Report no 419, Swedish School of Economics and Business Administration,
Helsinki!Helsingfors 2000
Gutenberg E (1983) Grundlagen der Betriebswirtschaftslehre, vol 1: Die
Produktion, 24th edn. Berlin Heidelberg New Y ork
Haak W (1982) Produktion in Banken. Möglichkeiten eines Transfers industrie-
betrieblich-produktionswirtschaftlicher Erkenntnisse auf den Produktions-
bereich von Bankbetrieben. Frankfurt a.M. Bem
Harnmann P (1974) Sekundärleistungspolitik als absatzpolitisches Instrument. In:
Harnmann P, Kroeber-Riel W, Meyer CW (eds) Neuere Ansätze in der
Marketingtheorie. Festschrift zum 80. Geburtstag von Otto R. Schnutenhaus.
Berlin, pp 135-154
Hentschel B (1992) Dienstleistungsqualität aus Kundensicht Vom merkmals-zum
ereignisorientierten Ansatz. Wiesbaden
Herzig N (1975) Die theoretischen Grundlagen betrieblicher Instandhaltung.
Meisenheim a.G.
220

Hilke W (1984) Dienstleistungs-Marketing aus der Sicht der Wissenschaft.

Diskussionsbeiträge des betriebswirtschaftliehen Seminars der Albert-
Ludwigs-Universität Freiburg im Breisgau, Freiburg i.Br.
Hubbert AR, Sehorn AG, Brown SW (1995) Service Expectations: The Consumer
Versus the Provider. International Journal of Service Industry Management
6(1):6-21
Johnston R (1995) The Zone of Tolerance. Exploring the Relationship Between
Service Transactions and Satisfaction with the Overall Service. International
Journal ofService Industry Management 6(2):46-61
Kawlath A (1969) Theoretische Grundlagen der Qualitätspolitik Wiesbaden
Klatt S ( 1961) Die Qualität als Objekt der Wirtschaftswissenschaft. Jahrbuch der
Sozialwissenschaft 12:19-57
Kern W (1979) Produkte, Problemlösungen als. In: Kern W (ed) Handwörterbuch
der Produktionswirtschaft, 1st edn. Stuttgart, col. 1433-1441
Kleinaltenkamp M (1993) Investitionsgüter-Marketing als Beschaffung externer
Faktoren. In: Thelen EM, Mairarnhof GB (eds) Dienstleistungsmarketing.
Eine Bestandsaufnahme, Tagungsband zum 2. Workshop fiir Dienstleistungs-
marketing, Innsbruck, Februar 1993. Frankfurt a.M. et al., pp 101-126
Kleinaltenkamp M, Jacob F (2002) German Approaches to Business-to-Business
Marketing Theory. Origins and Structure. Journal of Business Research
55:149-155
Kosiol E (1961) Modellanalyse als Grundlage unternehmerischer Entscheidungen.
Zeitschrift fiir handelswissenschaftliche Forschung, N .F ., 13:318-3 34
Laueaster KJ (1966) A New Approach to Consumer Theory. Journal ofPolitical
Economy 74:132-157
Lücke W (1973) Qualitätsprobleme im Rahmen der Produktions- und
Absatztheorie. In: Koch H (ed) Zur Theorie des Absatzes, Erich Gutenberg
zum 75. Geburtstag. Wiesbaden, pp 263-299
Mengen A (1993) Konzeptgestaltung von Dienstleistungsprodukten. Eine
Co:l1ioint-Analyse im Luftfrachtmarkt unter Berücksichtigung der Qualitäts-
unsicherheit beim Dienstleistungskauf. Stuttgart
Meyer A, Mattmüller R (1987) Qualität von Dienstleistungen. Entwurf eines
praxisorientierten Qualitätsmodells. Marketing- ZFP 9:187-195
Parasuraman A, Zeithaml VA, Berry LL (1985) A Conceptual Model of Service
Quality and lts Implications for Future Research. Journal of Marketing
49(4):41-50
Schade C (1996) Standardisierung von Beratungsleistungen. Eine ökonomische
Analyse integrativer Produktionsprozesse. In: Meyer A (ed) Grundsatzfragen
und Herausforderungen des Dienstleistungsmarketing. Wiesbaden, pp 69-96
Scharitzer D (1993) Das Dienstleistungs-'Produkt'. Der Markt 32:94-107
Schneeweiß C (1995) Stufen hierarchischer Abhängigkeit innerhalb und zwischen
Organisationen. In: Bühner R, Haase KD, Wilhelm J (eds) Die
Dimensionierung des Unternehmens. Stuttgart, pp 97-118
Shostack GL (1977) Breaking Free from Product Marketing. Journal ofMarketing
41(2):73-80
 221

Shostack GL (1981) How to Design a Service. In: Donnelly JH, George WR (eds)
Marketing of Services. Chicago, pp 221-229
Smith RA, Houston MJ (1983) Script-based Evaluations of Satisfaction with
Services. In: Berry LL, Shostack GL, Upah GD (eds) Ernerging Perspectives
on Services Marketing. Chicago, pp 59-62
Stuhlmann S (1999) Die Bedeutung des externen Faktors in der Dienst-
leistungsproduktion. In: Corsten H, Schneider H (eds) Wettbewerbsfaktor
Dienstleistung. Produktion von Dienstleistungen - Produktion als Dienst-
leistung. München, pp 23-58
Weiber R, Jacob F (2000) Kundenbezogene Informationsgewinnung. In:
Kleinaltenkamp M, Plinke W (eds) Technischer Vertrieb. Grundlagen des
Business-to-Business-Marketing. 2nd edn. Berlin et al., pp 523-612
Woratschek H (1992) Betriebsform, Markt und Strategie. Wiesbaden
E-Business and New Forms of Collaboration
along the Supply Chain

Stefan Kayser

1 lntroduction

During the last two years several new types of collaboration pattems
within the supply chain have emerged. This development was highly influ-
enced by the impact of e-business on the processes within companies, but
also on transactions between the various partners of the supply chain. 1
Digitalization of services in the logistics industry enables the partici-
pants of the supply chain to integrate their information systems by data
exchange via joint platforms and EDI/ web linkages. By this means the
companies pursue a seamless flow of information in parallel to the flow of
goods. The logistics and transportation industry drives this development
further by collaborating across companies' borders for the exchange of
data. New types of collaboration such as the sea and air carrier portals
INTTRA, GT Nexus, and GF-X, as well as freight exchanges in land
transportation imply both, the partnership between competitors (carriers,
forwarders) and the partnership between several participants of the supply
chain: the carriers, the freight forwarders, and the consignee/ shipper. 2
This paper aims at analyzing the impact of new electronic fulfillment
services within the logistics industry on the development of new types of
collaboration along the supply chain. In particular, it focuses on three
questions: 1.) In which way have changes in the supply chain on the one
hand, and the digitalization of services on the other hand affected the
collaboration in the logistics industry? 2.) Has the Internet changed the
determinants of strategic alliances in the logistics industry? 3.) What are
the opportunities and risks of new forms of collaboration along the supply
chain? Chapter 2.1 describes the shift towards more collaborative pattems
along the supply chain that is based on digitalization. Chapter 2.2 focuses
on three types of collaboration that have emerged during the last years.
Chapter 3 provides a framework for the analysis of the strategic decisions
for entering a collaborative relationship.

1 See e.g. Weber et al. (2002), Roberts (2000), Kerr (2001), Browne (2000).
2 See UBS Warburg (2000).
 223

2 Cellaboration along the supply chain

The traditional supply chain is highly fragmented. There are various

activities along the supply chain that can be listed for the three groups such
as customers, suppliers and logistics service providers. Activities carried
out by a customer encompass the creation of demand as well as the intemal
and extemal order. On the supplier side, activities include the receipt of an
order and its control check, inventory management, the transport/dispatch,
and the shipment. After having received the shipment order from the sup-
plier, the logistics service provider (LSP) arranges the pick-up, the storage,
the transport, and the delivery ofthe goods to the final customer. 3 The final
customer hirnself performs the delivery check, the registration in his ERP
system, and he initiates the payment to the supplier.
In the traditional supply chain these activities are carried out separately
and not interlinked with each other. There are informational interfaces
between the customer's extemal order and its receipt by the manufacturer,
between the shipment order and the pick-up by the LSP as well as between
the delivery by the LSP and the delivery check by the final customer.
These interfaces imply the transfer of information regarding the shipment
such as the characteristics of the goods, the package type and its measures,
the details of the seller, the buyer, the shipper, and the consignee. The
more interactions between the various parties take place the more infor-
mation is transferred through these interfaces and the higher are the trans-
action costs of handling these information transfers. Since in the traditional
supply chain the integration of processes and activities has been limited the
information flow needed more interfaces. Outside of these interfaces the
transfer of information was limited to the intra-company transfer but not
extended to the inter-company exchange of data.

2.1 Digitalization in the logistics industry

Collaboration along the supply chain has to be analyzed within two

developments: increasing integration of activities of the participants in the
supply chain and the digitalization of the information linked with these
activities.

3 The term logistics service provider encompasses the carriers (airlines, truck
hau-liers, shipping companies), forwarders and other 3rd party logistics
providers, see UBS Warburg (2000), p. 7-8. For a further distinction among
LSPs see Delfinann/ Albers/ Gehring (2002), p. 4-8.
224

Modem supply chains are characterized by a high degree of process

integration. More and more activities of the first tier as weil as the second
tier companies are integrated into manufacturing companies. In the same
way the information flow from customers to suppliers and further to LSPs
has become more integrated. Effective supply chain management is highly
connected with cooperative management, process integration over com-
pany borders, and integration of all supply chain activities. Furthermore,
the planning, measurement and controlling of flow of goods, information
and payment contributes to the overall effectiveness of supply chain
management and reduction ofprocess and transaction costs. 4
This development requires sophisticated IT-systems enabling the inte-
gration of all data along the supply chain. In the traditional supply chain
the information exchange between the customer and the supplier is mostly
based on the data exchange via Electronic Data interchange (EDI). This
exchange of data implies either the bilateral connections for long-term use
ortheusage ofValue Added Networks (VAN) in which specialized com-
panies provided networks to trading partners for their data transfer. 5
The Internet has facilitated the extension of the integrated supply chain
and enables the electronic exchange of data on the 2nd to the n-tier Ievel: on
the one hand, it electronically links suppliers and manufacturers on the
procurement side, on the other hand, the Internet connects the retailer and
the final customer on the distribution side. This development has allowed
for the rise of marketplaces during the last years: 6
• many-to-one-marketplaces: EDI networks, that are closed, expensive
and non-scalable, e.g. several suppliers sell products to one buyer,
• one-to-many-marketplaces: brochure ware based marketplaces that
publicize online, but sell offline; web shops for selling from web site,
• many-to-many-marketplaces: b2b e-commerce that enables com-
merce through the aggregation of many suppliers and many buyers
via an Internet platform.
Besides the extension of the supply chain and its increasing integration
via the intemet there has been a fast expansion of e-fulfillment services of-
fered by freight forwarders in order to integrate manufacturers, suppliers,
and carriers. By looking at the transport, warehouse and inventory
management activities handled by the acting companies in the logistics
market we find a variety of services enabling the electronic fulfillment of

4 See Chopra/Dougan/Taylor (2001), pp. 51-54; see Weber et al. (2002) for a
detailed description of the different phases of the logistics development.
5 For an overview on the EDI technology see Delfinann (2000), p. 63-65.
6 See Phillips/Meeker (2000), p. 25, Keller (2001).
 225

the traditional services offered by forwarders: 7 online price request, online

pricing, online booking, online tracking, online purchase order, electronic
warehouse services, electronic documents, and electronic payment ser-
vices.
The e-fulfillment services are an essential component that enables the
seamless data flow between the participants in the supply chain. 8 Some of
the non-transport related services are offered by a partnership with other
service providers, but the transport related services are offered by the for-
warders themselves. 9
All of these services aim at the integration of forwarders with their cus-
tomer group, the shippers, that are e.g. manufacturers and their suppliers.
Besides the forwarders also the carriers in the transportation sector have
developed e-fulfillment services in order to integrate with their customers.
This opens up the possibility of a seamless electronic transfer of transport
related data along the supply chain between all participants in the chain.
Since forwarders and carriers have strongly driven the development of
integrated e-fulfillment services these two parties mainly follow the ap-
proach of logistics-related collaboration. 10 Both, manufacturers and sup-
pliers act as shippers in this environment. By utilizing the means of
electronic and online data exchange various forms of collaborative pattems
have emerged that will be described in the next chapter.

2.2 Types of e-business collaboration along the supply chain

By looking at the three groups ofparticipants in the supply chain (shipper,

forwarder and carrier) the forwarder generally acts as an intermediate
between the shipper and the carrier. The traditional task of a forwarder is
to bundle the shipment orders of shippers, to reserve the freight capacities
for the bundled orders to carriers and to handle the paper transactions
linked with the orders. In particular the bundling of shipments is efficient
when several smaller shipments can be combined to a bigger shipment
order to carriers.
The development of e-commerce as described in chapter 2.1 has led to
several innovations in the field of logistics: web-enabling platforms, pri-

7 See Kayser (2001a) and Keller (2001). For further logistics-related online
services see Roberts (2000), p. 20-22.
8 See for a detailed overview for an example of the importance and the technical
realization of e-fulfilment servicesLock (2003) and Kayser/Seifert (2001).
9 See Kayser (2001b).
10 See also Chopra/Dougan/Taylor (2001, pp. 54-56) who focus on the success
factors of e-business implementation.
226

vate and public electronic marketplaces and hubs, freight exchanges and
carrier portals. With respect to these innovations several types of e-busi-
ness collaboration can be distinguished (see figure 1.).
The first type of collaboration are shipper centric platforms, which serve
as private platforms for web-based procurement and sales activities of
manufacturers and retail companies. These platforms are open only to a
limited group of companies and therefore designed for cooperation within
their network. In many cases these companies switch their type of data
exchange with their suppliers or customers from EDI-based to a web-based
linkage. Examples for shipper-centric platforms can be given by SupplyOn,
(Bosch), click4suppliers, click2 proeure (Siemens), Electronic Supplier
Link and VW Group Supply (Volkswagen).

Type 1: shipper-
centric platforms

Freight Forwarder Type 2: Virtual

forwarder/ neutral
platforms

Carriers Carriers Carriers Type 3: carrier-

Land Air Sea centric platforms

Fig. 1. Types of e-business collaboration along the supply chain; see Keller
(2001).

The second type of e-business collaboration are virtual forwarders or

neutral platforms. They function as freight exchanges that are not domi-
nated by a single user group of the logistics industry, but intend to form a
strategic alliance between several user groups. The functionalities of these
platforms imply a fast and focused search to filter shipments based on
various criteria such as geography, nature of goods, goods that can be
shipped together, and preferred business partners. The public section of
freight exchanges is mostly dealing with spot market transactions, whereas
transactions under long-term contracts are processed in the private section.
In particular, in its public sections freight exchanges aim at the optimiza-
 227

tion of freight capacities; generally, the freight exchanges intend to reduce

process costs of all participants who are trading freight capacities or
processing freight orders via these platforms.
There are examples of different platforms for the three transport modes
serving as virtual forwarders: Portivas, Teleroute, Benelog (Land Trans-
port), GF-X (air freight), and LevelSeas (sea freight). 11 In the Land Trans-
port business 45% of all platforms can be considered as virtual for-
warders.12 For the traditional forwarding companies the virtual forwarders
represent a possible threat towards their position in the basic freight com-
modity market, since they are targeting the same groups of carriers and
shippers, and thus, they are able to disintermediate the role of the tradi-
tional forwarder. 13
The third collaborative type of e-business along the supply chain
encompasses the carrier-centric platforms. These are carrier driven portals
forming complex strategic alliances between carriers, shippers, and for-
warders. This type of platforms aims at standardizing and supporting
transport services for customers who place freight orders, in general full-
container-load (FCL), with multiple carriers. Furthermore, the platforms
connect the user's intemal systems directly to the transportation network.
In this way, it contributes to the enhanced visibility of the process, a
seamless data flow and the reduction of process costs. The carrier portals
are dominant in the sea freight business, since there has been a structure in
the sea transport industry suitable for web-based business. 14 The services
of the carrier portals include e.g. sailing schedules, booking, tracking,
event management, documentation, instructions of bill of ladings as well
as their exchange of data files for print out. There are three carrier portals
in sea freight, each encompassing a group of carriers who have joined the
portal as founders or members: INTTRA (15 member carriers), GT Nexus
(13 member carriers), and CargoSmart (3 member carriers). 15 After having
been founded in 200112002 the carrier portals approached shippers as well
as freight forwarders in order to offer them the online services and the

11 See KPMG Consulting AG (2001).

12 See Keller (2001), UBS Warburg (2000).
13 This mostly holds for basic freight services. There are strong arguments against
the disintermediation of forwarders due to the complex services they nowadays
offer. Fora discussion ofthese arguments see Roberts (2000), p. 62-66.
14 There is a high degree of fragmentation of buyers and suppliers in both
chartering and procurement. See UBS Warburg (2000). Furthermore, the cost
for each carrier of builidng and maintaing its own website can be saved by a
joint portal. See also Kayser (2002).
15 Websites ofiNTTRA, GT Nexus, and CargoSmart on March 14, 2003.
228

integration of their IT systems. These efforts resulted in a collaborative

situation that is described by figure 2.:
• the biggest sea carrier have joined carrier portals and collaborate in
the same market with their competitors with respect to a joint IT
technology,
• the five biggest global freight forwarders have joined in different
ways the carrier portals. All forwarders have joined the INTTRA plat-
form as members, whereas only two forwarders have become mem-
bers oftheGT Nexus platform, 16
• since shippers use to place orders for freight capacities with all car-
riers, they are also active on all three platforms by directly ordering
and processing theirshipments via the portals.

INTTRA CargoSmart GT Nexus

15 sea carriers 3 sea 13 sea carriers

carriers

Fig. 2. Sea carrier portals and collaboration

In general, shippers will give their orders for full-container-loads (FCL)

to carriers via the portals whereas they place order for less-container-loads
(LCL) directly with the forwarders. Therefore, the threat of disintermedia-
tion of forwarders refers only to FCL orders. Forwarders are exp.osed on
the platform to the supply of services and capacities offered by their com-
petitors as weil as by the carriers as members of the portal. In this way the
market situation implies competition and collaboration on several Ievels:
Firstly, the carriers themselves simultaneously compete and cooperate;
secondly, the freight forwarders compete against each other and they face
the competition of the carriers on the platform. But they also cooperate
with respect to the acceptance of the same technology, the same data
format, and the usage of the same online services.

16 See websites ofiNTTRA, GT Nexus, and CargoSmart on March 14,2003.

 229

This situation has developed over the last years and results from strate-
gie decisions of all partieipants. Obviously, they expeet substantial savings
in proeess and transaetion eosts, whieh is the inherent goal of all eollabo-
rative aetivities deseribed above. The next ehapter eontains an analysis of
the strategie decisions that form the basis of eollaboration.

3 Analysis of strategic decisions on collaboration

All different types of e-business eollaboration require deeisions that have

to be taken in advanee before entering these kinds of partnerships. The
deeisions have to take into aeeount the different features of the freight
industry. Table 1 lists the main eharaeteristies of the freight industry and
its implieations for portals and freight exehanges in this field.

3.1 Characteristics of the freight industry and its implications

for e-business collaboration

There are four main eharaeteristies: The freight industry is large, it is

fragmented in various parts, it faees ineffieieneies, and there are near-
eommodity produets as well as eustomized produets. Eaeh eharaeteristie in
table 1. is deseribed by some examples. 17

17 See Pompeo (2000).

230

Table 1. Freight industry characteristics and its implications for e-business

collaboration, based on Pompeo (2000).
Characteristics of the freight industry lmplicatioo for
portals aod
freigbt excbaoges
I. Large market • Trucking Europe: • Spot market size • Sport-market vs.
EUR 200 billion varies with sector. contracts
• Ocean chartering: • Liquidity by net
EUR 120 billion work extemalities
• Air cargo:
EUR 70 billion
2. Partly frag- • 450.000 truckers • In some segments • Lock-in by IT in-
mented market in Europe a low concentra- tegration with
• Top five forwar- tion exists. users
ders hold 20-25% • Liquidity by net
of market, in work extemalities
air freight 70-80%
• 5.000 ocean
charterers
3. Inefficiencies • High interaction • Data flow by IT
• Importance of re-
costs lationships and integration
• Poor information contracts is en-• Price-orientations
tlow hanced. vs. relationship
• Price distortion orientation
• Operational ineffi- • Intermediation vs.
ciencies streamlining of
existing processes
4. Near-commodity • Full Truck Load • Customization and • Ability to handle
and customized (FTL) coordination is complex products/
products • Container (FCL) needed in order to networks
• Tailor-made Jogis- serve complex • Product-oriented
tics solutions supply chains. vs. customer-
oriented

1. The large size of the market is indicated by the sales in the three
transport modes. In the air cargo industry the market is significantly
smaller than in the trucking segment. Since each market consists of a spot
market and a contract market the relevance of freight exchanges varies
from transport mode to transport mode. Due to its larger size the land
transpoft market contains more spot business than the ocean or air freight
market. This implies a higher traded volume of freight and therefore, a
higher liquidity on these markets. On the contrary, the share of contract
business is higher in the air and sea freight markets, thus facilitating the
development of private marketplaces on which markets participants with
long-term relationships interact with each other.
 231

2. The level of concentration in electronic transport markets depends on

the degree of fragmentation in these markets. The higher the market is
fragmented the higher is the prabability that high investments in an IT
standard linked to a platform prevents the parties from switching to
another marketplace. The reason is that due to high market fragmentation,
the liquidity in all marketplace is low, and a lock-in by IT integration pro-
hibits a low-cost switch to another marketplace that might not be able to
generate a high freight volume. 18 In particular, this is true for the land
transport market whereas in the sea freight segment the existence of three
carrier portals implies that the need for a switch is diminished due to the
high commitment of the investing carriers respectively by high network
extemalities. 19
3. Due to the complexity in the transportation and logistics business
there is a high level of inefficiency along the supply chain implying e.g.
high transaction costs, lack of information along the supply chain, and
price distortions. In order to avoid these inefficiencies business partners
are interested in contracts and durable relationships and engage in private
marketplaces where they limit their activities to the electronic exchange of
order-related information with long-term business partners. Besides this,
they can also look for price advantages on public spot markets. This com-
bination has consequences for the extent to which a portal or freight
exchange can disintermediate other supply chain partners like forwarders.
4. The last characteristic refers to the product range of the logistics
industry. It encompasses commodity-related products such as FTL, FCL,
or tailor-made logistics solutions such as complex supply chain services.
The demand of supply chain partners for these products will influence the
structure of portals and freight exchanges. Since they are not able to handle
complex logistics products the portals may develop to points of entry for
alllogistics partners and offer beside online services also offline activities.

3.2 Relevant determinants of strategic alliances and their

effects

The analysis of the strategic decision for entering a collaborative relation-

ship such as a carrier portal or a freight exchange will be based on the
framework used by Porter in his research on strategic alliances and the

18 In particular in land transpoft there were more than 200 exchanges in 2001. See
for types oflock-in and affected parties ShapiroNarian (1999), pp. 116-131.
19 Ifthe carrier portal is perceived as valuable, the positivefeedbackwill make the
portal strong and the risk offailure is lowered. See also ShapiroNarian (1999),
pp. 173-179.
232

Internet. 20 According to Porter the Internet has only changed the front end
of processes. lt had its greatest impact on the reconfiguration of industries
that faced high costs for communication, the collection of information, and
transactions. Porter states that the factors determining the competitive
Iandscape have still remained the same. 21
Since the logistics and transportation companies are exactly facing the
same high cost structure mentioned above it can be analyzed in which way
the Internet has changed the competitive forces in this industry. Does the
emergence of new forms of e-business collaboration along the supply
chain reflect new reasons for entering strategic alliances or are these alli-
ances still determined by the same forces Porter has named? Can we link
with each of Porter's determinants those aspects that are beyond the rea-
soning for joining collaborative new platforms?
Porter identifies five underlying forces of competition: "the intensity of
rivalry among existing competitors, the barriers to entry for new competi-
tors, the threat of substitute products or services, the bargaining power of
suppliers, and the bargaining power of buyers."22 By combining these fac-
tors they determine the economic value that results from a product, service,
technology, or way of competition and that is allocated to any of the par-
ties involved in the industrial structure. 23
In order to carry out the analysis for the logistics industry I will apply
the determinants identified by Porter to the logistics industry. For each e-
business-related implication of the four characteristics described in chapter
3.1 the factors are singled out that can be found beyond the decision in
favor or against joining a portal or a freight exchange as a strategic alli-
ance. The analysis is done in a matrix for each of the three groups, the
shippers, the carriers, and the freight forwarders. Each group faces dif-
ferent opportunities and risk by entering the alliances, and within the indi-
vidual groups each member faces a different decision situation as weil.
The decision situation itself can be ambivalent, and the final result depends
on the weight the decision makers attribute to each individual factor. In the
matrices shown in tables 2. to 4. the general positive or negative impact of
the factors on each of Porter's competitive determinants is indicated by a
(+) or a (-).

20 See Porter (200 I). I do not use the concepts of a "strategic group" or "strategic
family" since these concepts do not apply in this context. See Albach (1992).
21 See Porter (2001), p. 66.
22 Porter (2001), p. 66.
23 See Porter (2001), p. 66. For the variety of collaborative relationships and the
factors for successful continuation of alliances see Moss Kanter ( 1994).
 Table 2. Opportunities and risks of e-business related strategic alliances tor shjppers
Relevant determinants of Strategiealliancesa nd tbeir effec ts

lmplications for portals Bargaining power of s bippers Bargaining powe r of carriers Rivalry a mong Bar ri ers to e ntry Tb real of
and freigbt excbanges and fonvarders existing com- Substitute of
netitors se rvices
Size: (+) Higher bargaining power (-) contract conditions become (+) reduced (+) low barrier to
• Spot-marke t vs. Con- by direct access to various more transparent transportation participate in platform
tracts LSPs and carriers on spot costs
• Liquidity by network exter- market
nalities (+) increased transparency
by price comparisons

Fragmentation: (-) risk of limited geographical (-)lock-in effect by adapting to

• Lock-in by IT integration scope of plotform the IT standard of the platform
with uscrs (-) veto power on decisions too (-) switching costs for different
• Liquidity by network low, risks of overrulement traffic modes
extemal ilies (-) low Ievel ofneutrality: de- (-) enhanced transparency on
pendency on asingle party shippe rs' data
(-)not many alternatives (esp. sea
traffic)

lnefficiencies: (-)high Ievei ofneutrality: Iack of (+) reduced ( +) due to low entry bar-
• Data flow by !T inte- liquidity on the platform (low net- Iransaction and rier higher network ex-
gration/ standards work extemalilies) intemal process temalities and higher
• Price-orientations vs. (+) low Ievel of liquidity: higher costs chance of freigh t
relationship orientation bargaining power due to compara- (+) saved in- matehing
• intermediation vs. tively high contribution vestrnent costs (-) due to low entry barrier
streamlining of existing (+) disintermediation of forwarder in EDI techno- cost ad vantage open to
processes possible logy competitors (e.g. if sales
price on Ci for DDU
basis)

Products: (+) increased Ievel ofValue-added (+) low entry barriers

• Ability 10 handle com- services on platforms might Iead to higher
plex prodts/networks (+) increased possibility 10 com- product diversity
• Product oriented vs. pare produc1s N
w
customer oriented w
 N
Table 3. Opportunites and risks of e-business related strategic alliances for carriers w
.j::>.

Relevant determinants of stratcgic alliances and their effeds

lmplicanons for portals and Bargaining power of Bargaining power of car- Rivalry an10ng existing Burriers 10 enlry Threat of substitute
freight exchanges shippers riers and forwarders compelitors of scrvices

Si7.e: (-) risk oflower bargaining (-) Iack of shippers • and for- (-) (+) Rivalry is shifted to (+) low barriers do (+) carriers and for-
• Spot-market vs. ContraciS power on spot market warders' panicipation Ieads electronic platform not reduce contract warders can substirute
• Liquidity by network exter- (-) increased price transpa- to reduced liquidity business, but forwarders in FTL and
nalities rency on spot markets enhance liquidity FCL business contract
(+) impact ofsingle shipper business
is low due to !arge nerwork

Fragmentation: (+) lock-in of shippers (+) power of carriers is en- (+) member profits from (+) SealAir (-) (+) spot rates allow
• Lock-in by IT iotegralion with (-)Land Transpon: due to hanced by lock-in of for- liquidity generated by Traffic:only big for easy Substitution of
users high fragmentation exit can warders competitors carriers join other suppliers and by
• Liquidity by oetwork extema- endanger survival (-) forwarders can threat with (-)lock-in by high IT in- alliance due to other suppliers
lities shift to other alliance vestrnents and membership high invesunents (+)Land Trame: easy
(-) Sea Traffic: participation (+) Land Traffic: access to new markets
of forwarders in several plat- Small carriers can and shippers
forms have easy access

lnefficiencies: {+) Sea Traffic: relationship (+) disintermediation of for- (+) high cost avings (+) Sea Traffic:
• Data ftow by !T inte- orientation reduces price warders possible (-) differences in processl cost advantage
gration/ standards t.ransparency (-)Land transpon: price ori- transaction costs are dimi- towards other
• Price-orientations vs. entation on spot markets re- nished within the alliance extemal com-
relationship orientation duces acceptance by other (+)Seal Air: cost savings petitors due to
• interrnediation vs. suppliers are high due to bundling of high entry barriers
streamlining of existing (+) SealAir Traffic: data now carriers
proccsses is steered by ponal (+) Land Transpon: com-
(+) forwarders cannot offer plementary services imply
alternatives to shipper interdependent benefits

Products: (-) higher pressure for (·)Land Transpon: com- (+) competitive ad-
• Ability to handle complex competition on Value- modization might Iead to vantage by customized
prodtslnetworks Added Services lower rates on spot margins VAS outside alliance
• Product oriented vs. (-) serviceswill be com-
-· ------~
customer oriented moditized
 Table 4. Opportunites and risks of e-business related strategic alliances for freight forwarders
Relevant determinants of strategic alliances and tbeir effects

lmplications for portals and Bargaining power of Bargaining power of carriers Rivalry among existing Barriers to entry Th reat of Substitute
freight e:rchanges shippers and forwarders competitors of services

Size: (·) increased price transpa- (· ) Iack of shippers' and (-)(+) Rivalry is shifted to ( +) low barriers do (· ) carriers and for-
• Spot-market vs. Contracts rency on spot markets carriers' participation Ieads to electronic platform not reduce contract warders can substitute
• Liquidity by network exter- (+) the impact ofsingle reduced liquidity business, buten- FTL and FCL business
nalities shipper is low due to a hance liquidity contract business
!arge network

Fragmentation: (+)lock-in ofshippers (·)high fragmcntation (+) member profits from (+) SealAir Traf- (·) (+) spot rates allow
• Lock-in by IT integration with (·) Land Transpon: due to requires high liquidity liquiditygenerated by fic:only big for- for easy Substitution of
users high fragmentation exit can (+) forwarders can threat with competitors warders join other suppliers and by
• Liquidity by network extema- endanger survival shift to other alliance (·) lock-in by high IT alliance due to other suppliers
lities (-) high fragmentation (+) Sea Trafl'ic: panicipation investments and member- high investments (+) Land Traflic: easy
requires high volume of forwardcrs in scveral ship (+)Land Traflic: access to new markets
(liquidity) platforms Small LSPs can and shippers
have easy access

Inefliciencies: (-) forwardcrs cannot offer (-) disintermediation of for- (+)high cost savings (+) Sea Traffic: (· ) threat of disinter-
• Data flow by IT inte· alternatives to shipper, they warders possible (-) differences in process/ cost advantage mediation can reduce
gration/ Standards have to follow them (·)Land transpon: price transaction costs are towards other the cost saving effects
• Price-orientations vs. (+) Sea Traftic: relationship orientation on spot markets diminished within thc extemal com- (+) ol ng Iasting rela-
relationship orientation orientation reducesprice reduces acceptance by other alliance petitors due to tionships result in
• intermediation vs. transparency forwarders (+) Sea traflic: cost savings high entry bar- higher efliciency and
strearnlining of existing (·) Sea Traflic: data flow is are high due to bundling of riers enhance customer
processes steered by ponal carriers retention
(+)Land Transpon: com-
plementary services imply
higher benefits
Pro<lucts: ( ·) higher pressure for (+) forwarders offer complex (·)Land Transp.: commo- (+) competitive
• Ability to handle complex competition on Value- logislies solution offiine that diz.ation will Iead to lower advantage by cus-
prodtslnetworks Added Services can be combined with online margins on spot margins tomized VAS provided N
• Product oriented vs. (-) serviceswill be com- orferings (+) good way to optimizc outside ofthe alliance w
Vl
customer oriented moditized network -
236

The matrices do not claim to be complete, but they show that the resul-
ting impact of opportunities and risks on the determinants can be derived
from the e-business relevant characteristics of the logistics industry. This
can be illustrated by an example in table 3.: the low fragmentation of the
market on the sea carrier side with three portals implies that there is a
strong lock-in effect by the IT standard set by each portal and the IT inte-
gration with the forwarders. This results in a more powerful bargaining
position of the carriers as members. However, since there is a high li-
quidity also on the other carrier portals, the forwarders can threat to switch
to another carrier portal or to use this more frequently. This threat weakens
the bargaining power of the carriers.

4 Conclusions

The analysis Ieads to the following conclusions: 1.) The digitalization of

services as weil as the changes in the supply chain have resulted in various
forms of collaborative e-business platforms in this field. 2.) The Internet
has not changed the pattems of strategic alliances in the logistics industry.
The determinants as described by Porter can still be seen as. the driving
forces beyond the decisions of supply chain participants for entering stra-
tegic alliances. 3.) It can be shown that the characteristics ofthe transpor-
tation market and its e-business implications do have various impacts on
the relevant determinants of Strategie alliances. The resulting opportunities
motivate some parties to start collaboration activities in this field; the
related risks prevent other parties from joining a collaborative e-business
platform.

References

Albach H (1992) Strategische Allianzen, strategische Gruppen und strategische

Familien, Zeitschrift fiir Betriebswirtschaft 6: 663-679
Browne M (2000) e-commere, freight distribution and the truck industry, Discus-
sion paper, 4th ACEA SAG Meeting, Brussels
Chopra S, Dougan D, Taylor G (2001) B2B e-Commerce Opportunities, Supply
Chain Management Review May/June: 50-58
Delfrnann W, Albers S (2000) Supply Chain Management in the Global Context,
Working Paper no 102, Köln
Delfrnann W, Albers S, Gehring M (2002) The Impact ofElectronic Commerce on
Logistics Service Providers, Köln
 237

Kayser S (2001a) Wettbewerbsvorteile durch eine integrierte Logistik für Markt-

plätze, presentation on the International E-commerce Congress, Bad Hornburg
Kayser S (200 1b) Integration of online financial services into the logistics chain,
presentation on the IBC Structured Commodities and Trade Finance Forum,
Amsterdam
Kayser S (2002) Strategie alliances in the transport and logistics industry, presen-
tation at the WHU, Otto-Beisheim Graduate School of Management,
Vallendar
Kayser S, Seifert A (2001) Case Study: From old economy "de1ivery" to new
economy "fulfilment", presentation on the eyefortransport conference, Miami
Keller K (2001) E-commerce in the transport sector. Markets and Technology,
presentation on the Conference of Swedisch International Freight Asociation
"Transportdagen", Stockholm
Kerr F (2001) Technology Utilization and Collaboration within the Transportation
Industry, London
KPMG Consulting AG (2001) Die "dot.com"-Welle in der Transportindustrie,
Frankfurt et al.
LockT (2003) Schenker Logistics IBM Case Study, London
Moss Kanter R (1994) Collaborative Advantage, Harvard Business Review
July/August: 39-48
Phillips C, Meeker M (2000) The B2B Internet Report, New York
Pompeo L (2000) Presentation on the eyfortransport conference, Amsterdam
Porter ME. (200 1) Strategy and the Internet, Harvard Business Review March: 63-
78
Roberts R (2000) E-Global Logistics. The Engine Powering Globalization, Little
Rock
Shapiro C, Varian HR (1999) Information Rules. A Strategie Guide to the Net-
work Economy, Boston
UBS Warburg (2000) A Guide to B2B Transport Marketplaces, London
WeberJet al. (2002) E-Commerce in der Logistik: Quantensprung oder business
as usua1, Pau1 Haupt, Bern et.al.
Capabilities of the Firm and their Effect on
Performance - Production of Information and
Communication Technology Services as an
Example

Olli Kuivalainen

Sanna Taalikka

1 Abstract

In this study the relationships between the capabilities and strategies and
the performance of the small ICT firms are explored. The empirical data
from the survey of 124 Finnish firms show that there is partial support for
positive relationship between the capabilities and performance, as mar-
keting capabilities seem to have a positive effect on tumover.
Keywords: capabilities, resource-based view, ICT, performance

2 lntroduction

This study explores the relationships between the capabilities and strate-
gies and the performance of small firms operating in the information and
communication technology (ICT) domain. In our view, sustainable
competitive advantage is primarily based on the capabilities or knowledge
the knowledge-intensive firms (e.g. ICT firms) possess. Thus, firm-
specific resources and capabilities and the barriers that prevent their
erosion by competitors are central to understanding performance in inter-
national environments (Fahy 2002).
In this study, the resource-based view of the firm (RBV) is used as the
theoretical background. Although there have recently been a number of
sturlies dealing with empirical verification of links between resources and
capabilities and firm performance (Helfat 2000), there is a paucity of
sturlies employing survey methodology data. Consequently, we do not
know enough about the relationships between different capabilities and
sustainable competitive advantage.
 239

The present study extends the relevant empirical research by proposing

a testable framework where the effect of capabilities and strategic orienta-
tions to performance can be tested. Our framework tries to capture the
different types of capabilities and strategies employed. In the theoretical
part of the work we notice the importance of knowledge based on capa-
bilities, and the effect of strategy on performance.
The empirical part ofthe paper includes data from a survey of 124 small
Finnish ICT-firms. The CEOs of the firms were asked to indicate the
extent to which different capabilities constituted strengths in comparison
with competitors. These results are re:flected to the different types of
strategies and the performance of the firms. Although marketing capabili-
ties are generally seen as the weakest assets in comparison to competitors,
they seem to have significant effect on performance (turnover). In the end
of the paper conclusions, as well as managerial and theoretical guidelines
are presented.

3 Theoretical background

3.1 The resource-based view of the firm

The resource-based view of the firm (RBV) has become one of the domi-
nant approaches in sturlies of strategy, business policy, and strategic
management in the past few years (see e.g. Peteraf 1993; Helfat 2000;
Barney 1996; Barney et al. 2001). The RBV is based on the premise that
sustained competitive advantage is derived from the resources and capa-
bilities of the firm that are valuable, rare, inimitable and non-substitutable
(Barney 1991 ). These so-called VRIN attributes can be viewed as a unique
bundle of resources the firm has in its possession. Such resources or capa-
bilities include both intangible and tangible assets. Management, mar-
ketingor technological skills, firm's organisational routines and processes,
as well as the information and knowledge it controls are some of the capa-
bilities whose importance has been stressed in many sturlies (see Kogut
and Zander 1992; Grant 1996; Miller and Shamsie 1996; Spanos and
Lioukas 2001 ). Recent emphasis on the role of knowledge as a source of
value in building organisational capabilities has also led to the emergence
of the 'knowledge-based view of the firm' (Grant 1996). Here, however,
we use the term resource-based view as a general term for the large variety
ofviews which emphasise a firm's intemal resources and capabilities as a
240

source of competitive advantage 1• These capabilities and the links between

different capabilities reconfigure a firm's competitive advantage con-
tinuously. Eisenhardt and Martin (2000) see dynamic capabilities (i.e.
capabilities which are dynamic by nature) as "best practices". These could
include different types of routines, based on the knowledge the firm
possesses or constantly creates.

3.2 Knowledge, internal capabilities and performance

Millerand Shamsie (1996) claim that in dynamic, i.e. changing and unpre-
dictable environments, the knowledge-based resources are seen to con-
tribute most to the performance. As mentioned above, many other
researchers share this view (Grant 1996; Teece 1998). The problem,
however, is how to define and measure knowledge-based resources or
capabilities and their value, as even the concept of 'knowledge' is often
undefined. Grant (1996:110) in building his suggestion for the knowledge-
based theory of the firm notes that the question of "what is knowledge" has
intrigued some ofthe world's greatest thinkers from Plato to Popper with-
out the emergence of a real consensus. His "simple tautology" [his own
wording] 'that which is known' isthat there are many types ofknowledge
that are relevant to a firm. Wilkins et al. (1997) see knowledge assets
consisting of facts, assumptions and heuristics which provide economic
value to their possessor. This definition is not always helpful, as
know1edge is extremely intangible by nature, and the evaluation of the
value ofknowledge assets is often done by using different types ofproxies,
the usability of which often depends on the situation. As an additional
dimension, the nature of the knowledge and its effect on the ability to
create competitive advantage is for many researchers the most important
issue (Kogut and Zander 1993). The critical question is the transferability
of knowledge, and distinction is often made between 'explicit or codified
knowledge' which is capable of articulation, and 'tacit knowledge' which

1 The recent definitions of the resource-based view (e.g. Barney 1991, 1996)
overlap broadly with the so-called knowledge-based view (Grant 1996, Miller
2002) and dynamic capabilities view (Teece, Pisano and Shuen 1997). Allthese
views focus on knowledge inventories, capabilities or resources as a source of
competitive advantage (see e.g. Miller 2002). Furthermore, Barney et al. (2001)
notice that a number of papers published suggest that resources, dynamic
capabilities, and knowledge are closely interlinked. Thus, dynamic capabilities
and the knowledge-based view can be seen as extensions to the "classic" RBV
as they focus more on the evo1ution/deve1opment of capabilities over time.
 241

is embedded within people and/or its application and not easily trans-
ferable (e.g Grant 1997; Blomqvist and Kyläheiko 2000). In general,
though, knowledge can be seen as organisational capability, and the
challenge for a firm is how to develop, integrate and share "new"
knowledge within the firm, to be able to gain a sustainable competitive
advantage and superior performance.
There are some classifications of different types of capabilities
presented in the literature. For example, Miller and Shamsie (1996) have
categorised resources into two classes, i.e. to property-based and
knowledge-based resources. Lee et al. (2001), in their study of Korean
technology-based start-ups, use a classification consisting ofthree types of
intemal capabilities, i.e. entrepreneurial orientation, technological capa-
bilities and financial resources. Spanos and Lioukas (200 1) with their
focus on RBV and emphasis on knowledge have divided capabilities into
organisational, technical and marketing capabilities. The intensive search
for resources and capabilities shows that empirical research has not yet
reached maturity (Miller and Shamsie 1996; Spanos and Lioukas 200 I;
McEvily and Chakravarthy 2002) and there are many different types of
measurements and classifications of capabilities used. In this paper a divi-
sion similar to that of Spanos and Lioukas (2001) is used. However, we
have added the financial capabilities -dimension, as for example Lee et al.
(200 1) have found the financial resources invested and having linkages
within the venture capitalists important indicators of performance.

3.3 Performance and strategy as an outcome of capabilities

The performance of a firm can be defined as the extent to which the objec-
tives of the firm are achieved as a function of specific strategies (Spanos
and Lioukas 2001). Porter (1991) sees performance as a function of in-
dustry and firm effects (which include the firm's actions in the market, e.g.
positioning). Traditional performance measures include economic goals
(e.g. profitability and sales). Depending on the purpose of the measure-
ment, these economic goals can be combined with more non-economic
(i.e. more miscellaneous non-financial measures related to e.g. market and
product) measures.
The performance of the firm can be considered as dependent on the
orientations and strategies the firm follows. Strategie orientation itself
refers especially to the dominant strategy or strategies that management
adopts on a continuous basis (e.g. Knight 2002). The industrial organisa-
tion (i.e. the Porterian view) -approach to strategy is that it is primarily
industry driven, whereas the RBV posits that the essence of strategy
242

should be defined by resourees and eapabilities (Rumelt 1984; Spanos and

Lioukas 2001 ). Thus, as mentioned above, in the RBV the retums stem
from aequiring and deploying VRIN assets rather than from the industry
strueture. The different strategies the firms pursue should be ehosen
depending on the strength of the eompetitive advantage they are estimated
to bring. Classie strategie orientation options inelude Porter's generie
strategies of differentiation, eost leadership and foeus (Porter 1980).
However, it is important to notiee that for eertain types of firms, e.g. for
the small ICT-firms in our sample, the traditionallow eost strategy may
not be viable. There are no eeonomies of seale available yet, as firms are
often still developing their produets, and leaming eurve effeets have not
taken plaee.

4 Model development and hypotheses

On the basis of the arguments above we propose a model ineorporating the

following effeets: 1) strategy effeets that are a neeessary basis for above-
the-average performanee, and 2) capabilities-based (firm speeifie) effects
that aecording to the logie of the RBV provide the eonditions for the
sustainability of performanee. We have built several hypotheses related to
ourmodel.

Hypotheses - capabilities
lt is argued that the capabilities of a small firm are positively associated
with its performanee. Thus, it is hypothesised that:
H 1a Technological capahilifies are positively associated with the
performance of small ICT jirms.
H1 h Marketing capahilifies are positively associated with the
performance of small ICT jirms.
H1c Organisational capahilifies are positively associated with the
performance of small ICT jirms.
H 1d Financial (management) capahilifies are positively associated with
the performance of small ICT firms.

Hypotheses - strategy
It is argued that a well-defined strategie orientation is positively assoeiated
with the performanee of a small ICT firm. Following Spanos and Lioukas
(200 1) we make a distinction between marketing and teehnologieal
differentiation. Thus, we hypothesise that:
 243

H2a: The adoption of technology differentiated strategy is positively

related to a firm 's performance.
H2b: The adoption ofmarketing oriented strategy is positively related to
a firm 's performance.
The principal aiin of the model is to test the effect on different capabili-
ties and strategies, and identify the significant linkages. The model 1s
presented in Fig. 1 and it is hypothesised to fit for Finnish ICT SMEs.

Capabilities MarketiniJ Orpanisational Financial TechnoloQical

Strategie
I
orientation

Fig. 1. Hypothesised model covering a firm's capabilities and strategies, and their
effect on performance

5 Data collection

A single industry was chosen to be able to isolate possible industry effects

on differences in performance (Hirsch 197 5). The population of interest in
our study was defined as small and medium-sized Finnish firms providing
value added services in the ICT-sector. Cohen et al. (2002:35) define ICT
as "a family of electronic technologies and services used to process, store
and disseminate information, facilitating the performance of information-
related human activities, provided by, and serving the institutional and
business sectors as well and the public-at-large". Due to the rapid
development of the ICT-sector and the unsuitability of the standard indus-
try classification codes, there was no single up-to-date sampling frame
available. A final sampling frame was devised through intensive search of
multiple sources, e.g. the Kompass Finland Database, The Statistical
Bureau of Finland database of Finnish companies, and the Internet sites of
the firms themselves.
244

Altogether 493 firms were identified and contacted by telephone in

November-December 2001. During this stage of the research 33 compa-
nies were found ineligible, and 74 firms refused to participate. The 386
firms which agreed to participate during the telephone conversation,
received on the following day an e-mail message containing instructions
for answering the web-based questionnaire. The actual, rather extensive
questionnaire was structured. lt was carefully pretested in a number of
firms. A reminder message was sent to those who had not returned their
answer within two weeks ofthe initial telephone conversation. In total124
firms responded, resulting in an effective responserate of 32%, which is
considered acceptable for the intended analytical approach. Following
Armstrong and Overton (1977) the comparison of early and late respon-
dents (with late respondents being assumed similar to nonrespondents) was
conducted to assess nonresponse bias. Here no significant differences were
found, and therefore nonresponse bias is not expected to have an effect on
the results. Most of the respondents were chief executive officers/
managing directors or members of the executive team, as it is believed that
they have the best overall knowledge conceming the resources and
capabilities of their firms. In this we followed the earlier studies related to
the RBV (see e.g. Fahy 2002).

6 Measurement constructs

The measurement scales are summated scales formed from the statements
used in the survey (see Appendix). The statements were mainly adopted
from previous studies (mainly from Spanos and Lioukas 2001, apart from
the financial capabilities -scale), but their reliability was assessed, and
factor analysis was applied to confirm the scales. Principal component
analysis with no rotation was conducted, and based on the factor loadings,
some scales were refined. The reliabilities of the fmal measure scales are
presented in Table 1. The generally applied Cronbach's alpha has the value
.70, but the benchmark of .60 can be also applied (e.g. Hair et al. 1998).
Thus, the measure scales in Table 1 were considered to meet the reliability
criteria for the analysis.
 245

Table 1. The reliabilities ofthe measurement scales

Variable N ofitems Mean Cronbach's a

Organisational capabilities 9 3.946 .8565
Marketing capabilities 9 3.562 .7726
Financial and resources capabili- 3 3.472 .7330
ties
Technological capabilities 5 3.942 .6964
Marketing differentiation 3 2.661 .7076
Innovative differentiation 8 3.374 .6759
Likert scale 1-5 in all scales
The variables measuring different capabilities measure the extent to
which the firm perceives it has the capabilities, i.e. high values for
marketing capability indicate that the firm perceives that they have good
marketing capabilities. Correspondingly, the variables for strategic dif-
ferentiation (here marketing or innovative, i.e. more technology-based)
measure the extent to which the firm perceives its strategic orientation to
be better than its competitors' orientation; i.e. a high value for marketing
differentiation indicates that the firm is more marketing oriented. Per-
formance was measured simply financially, with the firm's turnover: in the
questionnaire the respondents were asked to give the exact turnover of
their firm.
The use of perceived and subjective responses in our study can be ques-
tioned. There are, naturally, practical considerations supporting their use.
For example, in Finland small firms arenot obliged by law to report all the
necessary information related to this study in public, and thus "objective"
information is not available. However, there are also theoretical argu-
ments, which support the use of subjective data. Lefebvre et al. (1997)
notice that CEOs' views of the environment may "override factual
characteristics ofthe environment", and Spanos and Lioukas (2001) argue
that " ... managerial perceptions shape to a very important extent the
strategic behavior of the firm". Following this rationale, our use of self-
reported measures may be justified.

7 Results

The model illustrated in Fig. 1 was tested through structural equation

modeHing using the AMOS software package. The input data was in the
form of a correlation matrix (see Table 2). Hypotheses 1 and 2 are dis-
cussed at first, and after this the overall model is discussed.
246

Table 2. Correlation matrix ofthe measures in the model

ToO 1 Markdiff Inndiff FincaE OrgcaE TechcaE MarkcaE

To01* 1.000
Markdiff .062 1.000
Inndiff -.090 .084 1.000
Fincap .136 .157 .148 1.000
Orgcap -.028 .053 .360 .240 1.000
Techcap -.035 -.041 .364 .366 .463 1.000
MarkcaE .233 .343 .160 .313 .481 .248 1.000
*toO 1= firm turnover in 2001
The first hypothesis H 1a coneems the effeet of teehnieal eapabilities on
firm performanee. This hypothesis is not supported, meaning that the teeh-
nologieal eapabilities do not have a positive effeet on the performanee in
our sample. Aeeording to the seeond hypothesis, H 1b, the marketing eapa-
bilities are assumed to have a positive effeet on the performanee, and in
this ease the hypothesis is supported, i.e. the marketing eapabilities have a
positive effeet on the performanee. In general, the eompanies that pereeive
themselves as having high marketing capabilities have also higher per-
formance. However, the next hypothesis, H1e, suggesting that organisa-
tional eapabilities have a positive effeet on the performanee, is not sup-
ported. As in H 1a, the estimate from the organisational eapabilities to
performanee indieates a negative effeet, but this is not statistieally eon-
firmed. The last hypothesis related to capabilities, H1d, assumes that
financial eapabilities have a positive relationship to performanee. As with
the previous hypothesis, Hld is not supported. However, the estimate from
finaneial eapabilities to performanee indieates to the right direetion,
though without statistieal signifieanee.
Strategie orientation is diseussed in two hypotheses, H2a and H2b.
Aeeording to H2a, innovative differentiation as a strategie orientation was
expeeted to affeet positively on the firm performanee. This hypothesis is
not supported. This is also true for the next hypothesis, whieh states that
the marketing differentiated strategy affeets positively to the performanee.
Based on the analysis also H2b was found to be not supportive to our
claims.
 247

Table 3. Estimation ofthe model

Hy- From To Estimate t-value
poth
esis
H1a Technical capabilities ~ Performance -6.305 -.560
H1b Marketing capabilities ~ Performance 27.547 2.900**
H1c Organisational capabili- ~ Performance -14.575 -1.335
ties
H1d Financial capabilities ~ Performance 7.319 1.192
H2a Innovative differentiation ~ Performance -8.335 -.816
H2b Marketing differentiation ~ Performance -3.750 -.552
Technical capabilities ~ Innovative differentia- .250 2.567**
tion
Marketing capabilities ~ Innovative differentia- -.021 -.263
tion
Organisational capabili- ~ Innovative differentia- .235 2.479*
ties tion
Financial capabilities ~ Innovative differentia- .002 .032
tion
Technical capabilities ~ Marketing differentia- -.192 .189
tion
Marketing capabilities ~ Marketing differentia- .473 3.947**
tion
Organisational capabili- ~ Marketing differentia- -.135 -.949
ties tion
Financial capabilities ~ Marketing differentia- .094 1.147
tion
** p < .01, * p<.05

Fitindices for the model (df = 1)

x 2 = 1.375 (p = .241)

RMSEA= .056
CFI = .997
NFI = .990
GFI = .997

The model itself exhibits an acceptable goodness of fit. The ·l-value

and the values of RMSEA, CFI, NFI and GFI were studied, and based on
these indices our model provides an adequate fit to our data. Besides the
findings related to the previous hypotheses, the results of the analysis pro-
vide three additional significant paths or linkages in the model. The first of
these is from technical capabilities to innovative differentiation. According
to the RBV it is quite naturalthat high perceptions related to technical ca-
pabilities lead to higher perception of technological orientation, and this
proved to be true in our sample. The second significant linkage goes from
organisational capabilities to innovative differentiation. It seems that com-
panies with high organisational capabilities consider themselves highly
technology oriented. The third significant linkage leads from marketing
248

capabilities to marketing differentiation; this result is as weil supportive to

the logic of the RBV. High marketing capabilities support the perception
of strong marketing differentiation in Strategie orientation. The significant
linkages found are presented in Fig. 2.

Marketing Organisational Technological

Capabilities

(+) (+)

Marketing

Strat.or-
ientation

Fig. 2. The significant linkages of the tested model

8 Discussion, conclusions and further research

This study investigated the effects of intemal capabilities and strategic

orientation on the performance of small ICT -firms. We considered
different types of knowledge-based capabilities as the main assets in the
production of ICT services. This main tenet, i.e. the positive linkage
between different types of capabilities and performance was partially
supported, as our results indicate that strong marketing capabilities have an
effect on performance. However, our hypotheses related to other types of
capabilities were not supported. Both these findings can be related to the
current market situation. In our view, one of the main reasons for the Iack
of power in the hypotheses related to technical and organisational capa-
bilities can be found from the current stagnating economic climate. At the
momentsmall high-technology firms in the ICT-industry are coping with a
situation in which venture capital is no Ionger easily available and many
products and services do not have markets which are ready to utilise the
product yet, or their demand has slumped. In a Ionger time interval the
 249

situation might be different and the effect of the capabilities should be

more significant.
It seems that perceived technological skills do not affect the actual per-
formance. Good technological skills, although often prerequisites for
starting a business, were not enough to bring success in our sample. As
most of the firms studied here can be seen as technology oriented, there is
a lesson for managers, i.e. to put more emphasis on marketing skills and
resources. Most of the managers also saw that they were lacking more
skills/capabilities in marketing than in technology.
Although most of our hypotheses were not supported, we still see capa-
bilities as important building stones for sustainable competitive advantage.
The measured strategic orientations did not have direct/significant effect
on performance, although the linkages between different types of capabili-
ties and strategic orientations were statistically significant. This supports
the RBV approach, as capabilities define the strategic paths by which the
firms aim for competitive advantage.
The second factor which may have contributed to the weak results is the
cross-sectional nature of the study. The average year of establishment in
our sample was 1994; most of the firms are young. The capabilities
develop over time through learning and longitudinal study should therefore
be interesting. Other future research possibilities and additions into our
study are related to measures, e.g. performance measures. Our per-
formance measure was limited, as it was done only through turnover. More
measurements for different types of strategic orientations/strategies such a
entrepreneurial orientation should also be included into the model in the
future.
Although there are limitations, we believe that our study can signifi-
cantly contribute to our knowledge of linkages between capabilities and
performance. In general, the results, although only partially, support the
notion that better performance and self-perceived skills are linked. The
study is also useful as an exploratory type of work, as there are still only a
few sturlies which have tried to link knowledge to performance empirically
(see e.g. Miller and Shamsie 1996), and often the results have been only
partially supportive. Our study aimed to be one of the sturlies which would
increase empirical research in this area as we looked also at the linkages
between strategic orientations, capabilities and performance. There are still
many possible future research directions in this area. Spanos and Lioukas
(200 1) even notice that it seems that we do not have theoretical perspective
mature enough in knowledge capabilities to allow large scale empirical
testing through quantitative surveys. Also, the international context and
performance differences between domestic and international firms and
their capabilities is an interesting research direction.
250

References

Armstrong S, Overton TS (1977) Estimating non-response in mailed surveys.

Journal ofMarketing Research 18:263-264
Barney JB (1991) Firm resources and sustained competitive advantage. Journal of
Management 17:99-120
Barney JB (1996) The resource-based theory of the firm. Organization Science
7/5:469
Barney J, Wright M, Ketchen DJ Jr. (2001) The resource-based view ofthe firm:
Ten years after 1991. Journal ofManagement 27:625-641
Blomqvist K, Kyläheiko K (2000) Main challenges of knowledge management:
telecommunications sector as an example. In the CD-ROM Proceedings ofthe 8th
International Conference on Management ofTechnology, Miami, USA, 21-25.2.
2000
Cohen G, Salomon I, Nijkamp P (2002) Information-communications technologies
(Icn and transport: does knowledge underpin policy? Telecommunications
Policy 26:31-52
Eisenhardt KM, Martin JA (2000) Dynamic capabilities: what are they? Strategie
Management Journal21: 1105-1121
Fahy J (2002) A resource-based analysis of sustainable competitive advantage in a
global environment. International Business Review 11 :57-78
Grant RM (1996) Toward a knowledge-based theory of the firm. Strategie
Management Journal, Winter Special Issue 17:109-122
Grant RM (1997) The knowledge-based view of the firm: implications for
management practice. Long Range Planning 30(3):450-454
Hair JF, Anderson RE, Tatharn RL, Black WC (1998) Multivariate Data Analysis.
Prentice Hall, New Jersey
Helfat CE (2000) Guest editor's introduction to the special issue: the evolution of
firm capabilities. Strategie Management Journal. 21:955-959
Hirsch PM (1975) Organisational effectiveness and the institutional environment.
Administrative Science Quarterly 20(3):327-344
Knight G (2002) A taxonomy of born-global firms. Research workshop
proceedings, volume 2, McGill Conference on International Entrepreneurship:
researching new frontiers, Montreal, Canada 13-16.9.2002
Kogut B, Zander U (1992) Knowledge ofthe firm, combinative capabilities and
the replication oftechnology. Organization Science 3:383-397
Kogut B, Zander U (1993) Knowledge ofthe firm and the evolutionary theory of
the multinational corporation. Journal oflnternational Business Studies 24(4):
625-646
Lee C, Lee K, Pennings JM (200 1) Internat capabilities, external networks, and
performance: a study on technology-based ventures. Strategie Management
Journal22:615-640
Lefebvre L, Mason R, Lefebvre E (1997) The influence prism in SMEs: the power
of CEOs' perceptions on technology policy and its organizational impacts.
Management Science 43:856-878
 251

McEvi1y SK, Chakravarthy B (2002) The persistence of know1edge-based

advantage: an empirica1 test for product performance and technologica1
know1edge. Strategie Management Joumal23:285-305
Miller KD (2002) Know1edge inventories and manageria1 myopia. Strategie
Management Journal (forthcoming, in press)
Miller D, Shamsie J (1996) The resource-based view of the firm in two
environments: the Hollywood film studios from 1936 to 1965. Academy of
ManagementJournal39:519-543
Peteraf MA ( 1993) The Cornerstones of competitive advantage: a resource-based
view. Strategie Management Journall5:179-191
Porter ME (1980) Competitive strategy: techniques for analyzing industries and
competitors. Free Press, New York
Porter ME (1991) Towards a dynamic theory of strategy. Strategie Management
Journal, Winter Special Issue 12:95-117
Rume1t R (1984) Toward a strategic theory of the firm. In: Lamb R (ed)
Competitive strategic management. Prentice-Hall: Englewood Cliffs, New
Jersey, pp 556-570
Spanos YE, Lioukas S (200 1) An examination into the causal 1ogic of rent
generation: contrasting Porter's competitive strategy framework and the
resource-based perspective. Strategie Management Journal22:907-934
Teece DJ (1998) Research directions for know1edge management. Califomia
Management Review 40(3): 289-292
Teece DJ, Pisano G, Shuen A (1997) Dynamic capabi1ities and strategic
management. Strategie Management Journa118(7):509-533
Wilkins J, Van Wegen B, De Hoog R (1997) Understanding and va1uing
knowledge assets: overview and method. Expert Systems With App1ications
13: 55-72

Appendix - construct items used in the study

Organisational capabilities
Our resource management has become more efficient through experience.
We enjoy an encouraging atrnosphere.
W e have a 1eaming organization.
Our organisationa1 structure functions weil.
Our different functions areweil co-ordinated with one another.
Strategie p1anning is our strong force.
W ehave succeeded in our recruitment.
The turnover of our staff is 1ow.
We can utilize the expertise of our staff in different tasks.
Marketing capabilities
Wehave a specia1 understanding ofthe industry.
We have good connections to players in the industry (for examp1e, retailers,
suppliers etc.).
252

Wehave good and ftrm client relationships.

Our clients value our good service.
We are known in our market segment.
Our marketing expertise can be utilized for many different types of
products/services.
Our target market expertise can be utilised by other companies.
We have a strong brand.
We acquire a Iot of information on the development trends in the market (related
to, for instance, legislation, the economy etc.).
Technical capabilities
Wemaster the technology we use weil.
Our technical expertise is better.
Wemaster the development ofnew technologies better.
Wehave a clear vision ofhow technology will develop in the future.
Our special expertise can be adapted to new technologies.
Financial (planning) resources
Wehave excellent investment expertise.
We have good connections to different investors.
We constantly follow the company's fmancial condition
Strategy - innovative differentiation
R&D expenditure for product development
R&D expenditure for process innovations
The push to maintain a technological Iead (or emphasis being ahead of the
competition)
Number of product/service innovations
The company concentrates on developing its own technology
We are able to quickly adopt new technological know-how from the market
The automation and development of production processes
By developing our processes we will achieve economies of scale in production
Marketing differentiation
The innovations in marketing techniques
Emphasis on marketing department organization
Advertising expenditures
The Transaction Costs of eProcurement

J oachim Reese

Bjöm Saggau

1 lntroduction

Electronic procurement (eProcurement) has tumed out to be a much no-

ticed alternative to traditional procurement strategies taking into account
that the implementation of specific information and communication tech-
nologies is of essential importance for this strategy type. Undoubtedly,
eProcurement can often help to establish a fast and economic procurement
process after the available technology has efficiently been implemented
and additional measures of re-organization have taken place. lt is the aim
of this contribution to extend the evaluation of eProcurement beyond the
narrow frontiers of the theory of the firm, though. There exist two simple
reasons for such a broadening of the decision fundaments. First of all, the
real behavior of decision agents is not only based on production costs and,
thus, there sometimes seems to be a contradiction between the eProcure-
ment decisions which are derived from a comparison of the production
costs and the firm's behavior in reality. Second, the production theoretic
view does not reflect enough on the strategic consequences of a procure-
ment decision.
In order to thoroughly characterize the corresponding procurement
problem, the results of institutional economics, especially the transaction
cost theory, can be adopted, interpreted and further developed. Conse-
quently, the transaction costs of eProcurement have to be determined.
Adding production and transaction costs of procurement activities will
allow for far more reliable conclusions regarding the procurement strategy
and a detailed construction of the eProcurement instruments which have to
be installed in a single firm due to its particular characteristics.

2 Characterization of eProcurement

The eProcurement concept comes from the so-called new economy.

Similar to the problern of characterizing the new economy, a definite
254

description and classification of eProcurement is obviously rather

complex. In the following, eProcurement is characterized and distin-
guished from other procurement concepts closely related to eProcurement
by six criteria. The results are visualized in Figure 1.

Agents 82C 828 82A

eProcurement
eCommerce
-------öirectföii-öfflle-------------------------------------------------------------------
backward torward
Transaction
Processs
eProcurement eSales
------------ ................................................................................................................................................................................. ..
Impact of
long-term
Decisions
eProcurement

modern information and communication

technology
Technology
internet intranet I
extranet I EDI I
eProcurement
eCommerce
Transaction
Phase
initialization I agreement monitaring I
eProcurement
eServices I eCatalogue I online tracking
................................................ -................................ .. ---------------------------------------------------------------
Direction of the Co-
ordination Process
I~---------------r--------------~

vertical I
horizontal
I eProcurement
I cooperative sourcing

Fig. 1. Characterization of eProcurement

The characterization of eProcurement is concentrated on those features

which are supposed to have considerable influence on a transaction cost
specific analysis. For instance, the agents involved in an eProcurement
environment are on both sides firms which act professionally over a long
while and according to strict economic goals only. The transactions them-
selves are oriented backward. They are always induced by the successor in
the supply chain. This is why eProcurement is basically considered from
the buyer's point ofview, whereas eSales reflect the seller's point ofview.
The analysis of eProcurement is furthermore reduced to long-term irre-
versible transactions. Short-term measures are a direct consequence from
the long-term decisions. One central and constitutive feature of eProcure-
ment is the choice of an adequate information and communication tech-
nology. In literature, eProcurement is occasionally coupled with an intemet
engagement (Wirtz and Eckert 2001 , p. 152; Brennerand Lux 2000, p. 47)
 255

which would ignore other promising technological alternatives yet so that

this restriction is not regarded in this context. The chosen technology has
to be applied in at least one transaction phase, but there is no focus of
eProcurement on one specific phase. As it will be demonstrated in the fur-
ther discussion here, the differentiation of vertical and horizontal coopera-
tion partners plays a central role regarding the efficiency of eProcurement
activities.

3 The central impact of information and communication

technology on the transaction costs and phases

As mentioned before, eProcurement has been developed in order to con-

tribute to cost reductions in the procurement area. The past research in
Iiterature has been mainly focused on the importance of modern informa-
tion and communication technologies and its use in the different phases of
a procurement transaction (Koppelmann, Brodersen and Volkmann 2001;
Wirtz and Eckert 2001). If one follows Williamson (1985, p. 20) a trans-
action can roughly be divided in pre-contractual phases and post-contrac-
tual phases. In the pre-contractual phases, the internet technology has
proved to be very successful for implementing eProcurement processes.
Especially, the world wide web (www) serves as a technique which per-
mits comfortable information supply as concerns cooperation partners,
product qualities, test reports, supply conditions, etc. Negotiating the sup-
ply conditions can be supported by e-mail or e-post. As compared with
traditional correspondence, lead times as well as logistics costs are con-
siderably reduced.
In the post-contractual phases, the choice of an appropriate technology
results in cost decreases as well. For example, invoices can be received via
electronic data interchange (EDI) and, subsequently, electronically
processed in the firm's own ERP system. Many supply systemsalso dis-
pose of online-tracking routines so that the supply status of goods which
have been ordered from a supplier can permanently be surveyed.
Nevertheless, there exist a multitude of such offers of support tech-
niques available in www which finally leads to considerable increases in
the costs of initializing a partnership and can be described as "second-or-
der-effect" (Malone and Rockart 1992, p. 636).
These examples from literature, on the one hand, demonstrate obvious
cost advantages of using a modern information and communication tech-
nology, but, on the other hand, they neglect the integration of those other
important features of eProcurement presented in Figure 1. The current dis-
256

cussion is extremely focused on single phases and technological equipment

(e.g. Bogaschwesky and Kracke 1999; Dolmetsch 2000). This allows for
measuring the direct cost effects, but does not reveal the indirect conse-
quences of eProcurement which result from organizational and behavioral
issues. In particular, the interdependence of the cooperation partners as
well as their striving for unambiguous contractual agreements (Bauer and
Stickel 1996, 1998) is not considered.

4 Organizational aspects and types of eProcurement

The following analysis is centered around some relevant organizational

issues of eProcurement. With respect to the different directions of the CO-
ordination process, a two-dimensional idealistic concept is proposed. Hori-
zontal and vertical coordination activities which should improve the
procurement process provide different alternatives that have to be identi-
fied, analyzed and combined to a firm's definite eProcurement strategy.
The vertical organization of eProcurement can be divided into market-
oriented, cooperative, and hierarchical activities. Nonrecurring order
processes using an electronic platform are usually based on classical con-
tracts and thus are market-oriented ("spot markets"). The organizational
design "cooperation" is, however, founded on the relational or neoclassical
contract law (Macneil 1978, 1980; Williamson 1979). The partners aim at
a long-term business relationship without defining all its details. Hierar-
chical eProcurement means that the supplied goods come from a firm (de-
partment) which is legally connected with the buyer. It is again based on a
relational contract though different specific technologies, as e.g. the intra-
net technology and the firm's individual ERP system, can be applied and
the hierarchical planning structure can be adopted.
The organizational alternatives of a horizontal eProcurement process
can be similarly structured. With individual sourcing, a firm's unit has its
own procurement function which works independent of the other units
which become competitors. The sole coordination mechanism is the pro-
curement market. All participants are in direct competition with each
other. Cooperative sourcing, however, characterizes the systems of non-
market coordination mechanisms between legally independent partners
(Eßig 1999) (external coordination). Hierarchical sourcing appears when-
ever two or more units of a firm cooperate and act as a collective (internal
coordination).
A two-dimensional typology of eProcurement organization can now be
developed from the different forms of vertical and horizontal organization
 257

as shown in Figure 2. eProcurement of type 1 is practiced in online auc-

tions. There is no cooperation with vertical or horizontal partners at all.
Type 2 describes the vertical integration of firms such as a just-in-time
partnership with the support of an EDI technology. Type 3 means a verti-
cal integration of multiple stages within the firm and allows for the appli-
cation of intemal planning systems. In general, an intetnal integration
takes place as soon as highly specific production factors are concemed
which cannot be delivered from extemal sources at lower costs. eProcure-
ment of type 4 often appears as an online syndicate. Several firms which
are usually competitors on the procurement markets aggregate their indi-
vidual demands in order to invite bits in the intemet or viaother electronic
platforms. On principle, an online syndicate aims at price advantages
("power shopping"). As far as it is restricted to lowering the production
costs mainly non-specific goods are concemed. Type 5 is practically
realized as a virtual procurement network. Above all, small and medium
sized firms will join such a network on their search for suppliers who are
otherwise not available (e.g. global sourcing). Such networks may be
organized as joint ventures or strategic alliances. As a rule, one member of
the network is chosen as the coordinator and "lead" firm, respectively,
regarding a special goods type. Contractual agreements are arranged as
defined by the neoclassical law. With eProcurement of types 7 and 8, the
demands of a firm's different units regarding one group of materials are
aggregated in order to realize production cost advantages by way of a
centralized procurement process - whether by classical market transaction
or by long-term cooperation with selected suppliers. With both types, the
production cost advantages must be balanced with the investment in an
efficient information and communication technology - e.g. the design of an
extranet or an intranet solution which meets specific requirements of the
firms involved in that cooperation. The considerable organizational efforts
inherent to types 7 and 8 make things rather complicated. Just-in-time
procurement, e.g., cannot be reasonably realized under those restrictions.
Therefore, both types are focused on materials of low specifity which are
used in high frequency. The types 6 and 9 of eProcurement do not play any
role when implementing the strategy in a real-life situation.
258

-c_
Qlc:
c:w
WE
.i:: Q)

-
type 1
o~

c -•::::J
u type 4 type 7
~e
0 (;je..

.
E w
ro
N w"E
c
>Q)
:;::;E
ro <UQ)
Qi ::; type 2 type 5
0> a.u type 8
II.... 0 0
0 ~

0 uc..
Q)

-
ro
(.) _c:
mw
II....
.2 E
.J;:Q)
Q) u~
::::J type 3 type 6 type 9
"''-'0
~

> ~
w~

:.C:rt
individual cooperative hierarchical
sourcing sourcing sourcing

horizontal organization
Fig. 2. eProcurement matrix

5 Transaction parameters and transaction costs in the

eProcurement typology

The eProcurement matrix in Figure 2 reflects different degrees of market

penetration. While any eProcurement of type 1, e.g. online auctions, are
exclusively carried out in a procurement market, other types of eProcure-
ment (e.g. type 9) are not at all influenced by market procedures. Ac-
cordingly, the different types of eProcurement as well as the corresponding
contract law can be assigned to the vertical and horizontal degree of mar-
ket penetration. Since the classification of eProcurement types is based on
two criteria, it happens that different kinds of contracts may occur in one
and the same type of eProcurement. F or instance, contracts with horizontal
partners are due to relational law, whereas vertical partnerships are
founded on neoclassical contracts. Obviously, the more market penetration
is realized, the less elements of classical law will be adapted and vice
versa.
 259

+ horizontal market penetration -

+ classical
c:: contract
0
law
~
Ci)
c::
Q)
0.
Q) combination
.....
.::1! of classical,
CO neoclassical and
E relational contract law
elements with
~ increasing part of
t:Q) relational contract law
> contract
law

Fig. 3. Matrix of market penetration for different types of eProcurement

A special feature considering the transaction costs according to the

eProcurement typology thus refers to the different types of cooperation
partners. There may arise mutual dependencies between the individual
agreements so that their isolated consideration may cause inefficiency.
The asset specifity, the number of procurement transactions as well as
market and behavioral uncertainties were identified in Iiterature as the
main cost factors with regard to the transaction process (Picot, Reichwald
and Wigand 2001; Williamson 1985). They are characterized as transac-
tion-specific parameters. The relevance of the number of transactions is
comparab1e to the case of traditiona1 procurement. Its genera1 importance
needs not be discussed here, but is acknow1edged by extensive transaction
cost theoretic research (Wemer 1997, p. 72; Williamson 1985). The other
parameters, however, have to be analyzed regarding the design of an ePro-
curement strategy. Basically, specifity and uncertainty are positively
correlated with transaction costs. The more specifity and uncertainty is
observed, the more transaction costs arise. Consequently, as it is also
shown in Figure 4, the relevance of relational contract law grows with
increasing specifity and uncertainty, while classical contract law loses its
importance.
260

asset specifity +

classical

increasing
transaction
costs

Fig. 4. Matrix ofthe transaction factors asset specifity and market uncertainty

Comparing Figure 3 and Figure 4 reveals that those types of eProcure-

ment which create market penetration only cause low transaction costs. On
the opposite side, transaction costs increase with diminishing market pene-
tration.
Generally, asset specifity is identified with the specifity of tangible
objects which are used for production and which can only be transferred to
other production processes with difficulty (Windsperger 1996, p. 38). lt is
evident that in the case of eProcurement, asset specifity is first of all
connected with the procurement technology. Prominent technological
solutions are the intemet, intranets, extranets, and EDI. The use of the
intemet for eProcurement is not specific at all because this solution is
highly standardized and widely spread all over the world. However, the
other solutions involve certain specifity risks. For example, coupling of
intranets causes considerable investment in configuration measures and
access rights. Special areas of the intranets have to be released for extemal
use by the cooperation partners. The technological specifity reaches its
maximum when adopting an EDI solution. Such a type of an eProcurement
exclusively serves the purpose of a single partnership (cf. Huemer and
Quirchmayr 1996). The solution cannot be used otherwise. By the way,
EDI partnerships also create behavioral uncertainties which may result in
the opportunistic behavior of one or more partners and, last not least, in a
fundamental transformation (Amold 1999, p. 291).
 261

In addition, factor specifity within an eProcurement strategy can also be

explained to a certain degree by the importance of the procurement object.
The more a machine or material is important for the product quality, the
more specific is the partnership (Metcalf, Frear and Krishnan 1992). This
is why objects of utmost importance are mainly procured by adopting the
neoclassical or relational contract law.
Market uncertainty is closely connected with all intemal and extemal
channels of procurement which are used by the buyer. It mainly depends
on the diversity, the heterogeneity, and the interdependence of extemal
factors. Diversity is created by technological developments, e.g., the rate
and security of communication and further communication network
parameters. Transaction costs arise in so far as each parameter has to be
evaluated as concems the correctness of the data processing and trans-
mitting procedures. The uncertainty caused by the system' s fundamental
complexity will further increase as soon as the parameters under con-
sideration are heterogeneous, i.e. have their origin in different disciplines
as e.g. informatics, engineering, and economics, and mutually influence
each other. For example, a forgery of an electronic signature will initiate,
besides other measures of improving the security, a check-up of the Co-
operation partners. As a consequence, additional transaction costs occur.

6 Conclusions

The transaction cost theoretic analysis of eProcurement reveals that an

adequate institutionalization by means of technological, organizational,
and legal measures allows for controlling the transaction costs to a con-
siderable degree. Additionally, it must always be taken into account that
eProcurement, e.g. within a global sourcing strategy, also contributes to a
decrease of the firm's production costs. The effects on the transaction
costs, as one regards them isolated, induce an improvement of the firm's
flexibility towards the market and cooperation partners. Combining these
effects with a simultaneous decrease of the production costs - e.g. caused
by horizontal procurement alliances - may finally lead to long-term part-
nerships within a complex supply network which is based on a specific
electronic platform. In total, the results of the analysis do not permit con-
crete recommendations how to generally establish eProcurement activities.
Nevertheless, it should have become evident that, with an efficient use of
the eProcurement instruments, the procurement costs can be considerably
reduced.
262

References

Amold U (1999) Nutzung elektronischer Märkte für die Beschaffung. In: Nagel
K, Erben RF (eds): Produktionswirtschaft 2000: Perspektiven für die Fabrik
der Zukunft, Gabler: Wiesbaden, pp 285 - 299
Bauer S, Stickel E (1996) Contributions of IT to the Genesis of Network
Organizations. In: Khosrowpour M (ed): Information Technology
Management and Organizational Innovations, Idea Group Publishing:
Harrisburg, pp 46-52
Bauer S, Stickel E (1998) Auswirkungen der Informationstechnologie auf die
Entstehung kooperativer Netzwerkorganisationen. Wirtschaftsinformatik
40:434-442
Brenner W, Lux A (2000) Virtual Purchasing - Die Revolution im Einkauf,
Konradin Verlag: Leinfelden-Echterdingen
Bogaschewsky R, Kracke U (1999) Internet- Intranets- Extranets: strategische
Waffen für die Beschaffung, Deutscher Betriebswirte-V erlag: Gemsbach
Dolmetsch R (2000) eProcurement - Sparpotential im Einkauf, Addison:
München et al.
Eßig M (1999) Cooperative Sourcing, Lang: Frankfurt
Huemer C, Quirchmayr G (1996) The Dilemma of EDI- Problem Analysis of
Current Standards. In: Khosrowpour M (ed.): Information Technology
Management and Organizational Innovations, Idea Group Publishing:
Harrisburg, pp 10 1 - 107
Koppelmann U, Brodersen K, Volkmann M (2001) Electronic Procurement im
Beschaffimgsmarketing. Wirtschaftswissenschaftliches Studium 30: 79 - 86
Macneil IR (1978) Contracts: Adjustment of Long-Term Economic Relations
under Classical, Neoclassical and Relational Contract Law. Northwestem
University Law Review 72: 854-905
Macneil IR (1980) The New Social Contract - An lnquiry into Modem
Contractual Relations, Yale University Press: New Haven
Malone TW, Rockart JF (1992) Information Technology and the New
Organization. In: Nunamaker JF, Sprague RHJ (eds) Proceedings of the
Hawaii International Conference on System Science, Hawaii, pp 636-643
Metcalf L, Frear C, Krishnan R (1992) Buyer-Seller Relationships: An
Implication of the IMP Interaction Model. European Journal of Marketing
26:27-46
Picot A, Reichwald R, Wigand RT (2001) Die grenzenlose Unternehmung -
Information, Organisation und Management, Gabler: Wiesbaden
Wemer H (1997) Relationales Beschaffimgsverhalten, Gabler: Wiesbaden
Williamson OE (1979) Transaction-Cost Economics - The Govemance of
Contractual Relations. The Journal ofLaw and Economics 22: 233-261
Williamson OE (1985) The Economic Institutions of Capitalism - Firms,
Markets, Relational Contracting, Free Press: New York
Windsperger J (1996) Transaktionskostenansatz der Entstehung der Unter-
nehmensorganisation, Physica-Verlag: Heidelberg
 263

Wirtz BW, Eckert U (2001) Electronic Procurement - Einflüsse und

Implikationen auf die Organisation der Beschaffung. Zeitschrift Führung +
Organisation 70: 151- 158
A Complexity-Based Approach to Production
Management in the New Economy

Michael Reiss

1 A complexity view of the new economy

Production in the New Economy is very often production of services and

informational products. This is especially true for that sector of the New
Economy that is focused on e-business. Several elements or domains of
production management (eg. utilization of Internet technology, virtual
network organization, ) are shaped according to the generic features of the
New Economy. Figure 1 illustrates that the New Economy differs from the
traditional Old Economy (only) by degree, not principally. Entrepreneurial
orientation for example exists in both economies. Still, there is a difference
in degree, both quantitative (incidence of entrepreneurs) and qualitative
(variants ofentrepreneurship such as "netpreneurs", "intrapreneurs", etc).
low high

I PEED slow ruc

I VOL.ATILITV cablt lurbultnl

bureaucratic O'llanic

managers entreprtneurs

low ltch l«hnology drivtn

brick Ould
local global
tustomer&
capilal markel
c.apital market
marktl resurcb prosumers

Fig. 1. New and Old Economy

Various features in Figure 1 signal a high complexity of this sector of

the economy in generat i.e. also of production management. Complexity is,
to an extent, part of the identity of the New Economy (cf. Wood 2000;
Lissack/ Roos 2000; Underwood 2001 ). This refers both to the specific
complexity Ioad (pressure, requirements, "stress" etc.) in the New
 265

Economy and the available complexity potential for dealing with this load.
Complexity load on the one band is defined by the high pace of "internet
time", by volatility of businesses, personalization of products, global scale
of Internet economy, chaos in business development, hypercompetition
and coopetition. Complexity potential on the other band is defined by en-
trepreneurial culture, self-organization, flexibility, a powerful Internet in-
frastructure, virtual organization, rapid learning and a dual growth option.
Potential for managing the complexity load in the New Economy com-
prises both hard factors (Internet infrastructure, mobile computing etc.)
and soft factors (culture, self-organization, learning, agility, ambiguity
tolerance etc). The dual growth option serves as a perfect illustration of
complexity potential. Since New Economy companies operate embedded
in a network, they have two options for growth (cf. Reiss/ Bernecker 2003)
to deal with the complexity load of markets (e.g. reaching critical mass on
a global market): traditional growth via expansion of node size by having
more employees on tbe one band or virtual growth by finding new partners
(i.e. increasing the number ofnodes) on the other band.
In a way, hypercompetition and coopetition represent the essence of
complexity requirements in the New Economy context. Hypercompetition
(cf. d'Aveni 1994; Bruhn 1997; Zahn/ Foschiani 2000) comprises
complexity factors such as speed, surprise or shifting the rules of the game.
Coopetition (cf. Brandenberger/ Nalebuff 1996; Bengtsson/Kock 2000;
Reiss/ Beck 2000) represents a chanenging mix of competition and Co-
operation due to loose network structures which do not exclude multiple
engagementsindifferent networks. The combination "Hyper-Coopetition"
represents an extremely chanenging complexity load. The complexity of
strategy is not mitigated by strategic alliances or similar structures that (try
to) eliminate competition amongst former competitors by turning to a
complexity-reducing cooperative structure (according to the motto "lf you
can't beat them, join them! "). Instead, coopetition underlines an iBherent
strain due to "competition amongst partners" whicb is typical of organiza-
tional networks (as opposed to traditional cooperative organizational
structures).
Against this background, the thing to do would be to examine the use of
models, which work with "complexity" as the input and/or output
parameter, for production systems in the New Economy. In modelsoftbis
type, attempts are made at putting down typical performance phenomena
in the New Economy (ie excellence, "hype", rise and fall, ups and downs,
breakdown etc.) to complexity phenomena. A particularly interesting
question is whether hyper-complexity is the major reason behind the
numerous tombstones in the New Economy: is complexity overload or
"overkill" a specific failure factor in the New Economy?
266

2 Outline of the complexity approach

Complexity-based modeHing already has a tradition in formal, natural,

social and economic sciences. Often, this form of modeHing is Iabelied
"complexity theory" (cf. Brown/Eisenhardt 1997; Allen 2001;
Allison/Kelly 1999) or less ambitiously "complexity management" (cf.
Adam 1998; Ahlemeyer/ Königswieser 1998; Gharajedaghi 1999; Lewin/
Regine 2001; Schwenk:-Willi 2002), or "complexity handling/ mastering"
(cf. Jost 2000). To avoid any controversy about whether complexity-based
reasoning is a "science" or Gust) an "art", this mainstream area will be
referred to from now on as the "complexity approach". The complexity
approach serves as a holistic framework or "umbrella" for various focused
types of modelling. It utilizes complexity as a key unit of analysis. The
spectrum comprises mass production, risk management, descriptive statis-
tics (i.e. methodology to handle mass data), management of conflicts
(between objectives or stakeholders), probability theory, fuzzy set theory
and management (cf. Grint 1997), as well as chaos theory and change
management. All of these approaches have (specific facets of) complexity
as a common denominator (cf. Stacey 2000). At the moment only a vague
preconception of complexity is used. There is no doubt that more scruti-
nizing of the complexity concept is needed to answer questions such as
"What is meant by the complexity of a strategy, a structure or a market?"
(cf. chapter 4).
As far as its contribution to the theory of the firm is concemed, well
thought-out positioning ofthe complexity approach reveals that it does not
replace economic modelling such as the productivity paradigm, informa-
tion economics, "economic-behavioral amalgam" (bounded rationality),
agency theory, property rights theory, corporate govemance or transaction
cost theory. Rather, it serves as an interpolated support stage in a three-
stage modelling process: (1) implicit modelling, (2) formal complexity
modeHing and (3) economic modelling. Within the modelling of transac-
tion processes for example, complexity-based modelling cannot furnish
optimal combinations between task features and forms of coordination
within the market-hierarchy continuum in the way the Coase theorem can.
However, the complexity approach can support the systematic search for
drivers of transaction costs. Most characteristics of transactions such as
frequency, specifity, (un)certainty or (in)stability of tasks referred to in
transaction cost theory turn out to be complexity features. So the com-
p1exity approach supports the identification of the "quantity structure" of
transaction costs but is incapable of generating any optimization based on
economic valuation, i.e. prices derived from performance measures such as
 267

costs, profit or cash flow. Interestingly enough, the complexity approach

depends on input from economic models, for example when it comes to the
"optimal" management of complexity: handling of complexity in the form
of decoupling or standardization has cost effects ("complexity manage-
ment costs") which frequently serve as criteria for choosing the "best" way
of managing complexity. The same economic reasoning is behind the idea
of "optimal complexity": take for instance the notion of an "optimal degree
of conflict" which is supposed to stimulate fair-play-based competition
without causing the cost-driving effects of interpersonal quarrels, mistrust
and hostility. Metcalfe's law of optimal network size also belongs to this
kind of complexity modeHing based on the economic effects of complexity
(cf. Weiber 2002, p. 280).
There are parallels between the complexity approach and the system
approach (cf. Grothe 1997; Stüttgen 1999)- and probably other "transdis-
ciplines" such as the evolution theory (cf. Malik 2000; Hermann-Pillath
2002): all of them represent formal methods of modelling, operating on
formal units such as elements, relationships, variety, decoupling, loose
coupling, selection, retention, feedback loops, etc. And all of them func-
tion as heuristics in the model-generating process, helping to improve eco-
nomic modelling, eg theories of the firm.
The complexity approach mainly comprises three areas which are
focused on different aspects of complexity respectively:
Level of complexity: Major interest here is in the level of overall
complexity and in optimal complexity. Complexity becomes a problern
when the level of complexity is either too high or too low.
Course of complexity: Major concem in this area is the response of
complexity to certain drivers of complexity - which can be complexities
themselves. Examples of these drivers include linear growth of comp1exity
or trade-offs between two categories of complexities. Complexity becomes
a problern for instance when there is over-reaction, such as a complexity
escalation comparable to the "butterfly effect" in chaos theory.
Composition of complexity: Here focus is on the optimal composition of
different types of complexity. These can be relationships between com-
plexity load and complexity potential, various domains of complexity
(complexity of markets, products and process) or dimensions of com-
plexity (eg diversity, discontinuity). Complexity is considered a problern
when there are disproportions in the composition of the overall com-
plexity, regardless ofthe level oftotal complexity.
A generic type of composition model in the economic and social sci-
ences is the complexity-based fit model (Emery/ Trist 1965, Child 1972;
Roters 1988). In systems theory, Ashby' s law of requisite variety (cf.
Ashby 1956) represents an archetype of complexity-based reasoning: it
268

postulates the unilateral "complexity" fit between environment (surround-

ing systems) and system (i.e. company). Picot/ Reichwald/ Wigand (1996,
p.l4) outline a type of fit model relevant to organizing production systems
in the New Economy.
There are numerous models of complexity fit. The core fit model of the
complexity approach is based on balancing the complexity Ioad (need, re-
quirement, burden) and complexity potential (capabilities, resources,
capacities and the like for complexity handling). In Figure 2 three areas of
fit and misfit reflect the static view of fitting: complexity fit signals an
adequate balance of Ioad and potential, whereas the two misfit areas are
either determined by a surpluspotential (e.g. idle CPU capacity, ERP
functionalities, entrepreneurial power etc.) or an overload (inability to go
the pace of the Internet time, not meeting demands for personalization of
services etc.).

Fig. 2. Generle Fit Model

The area of complexity fit is depicted by a corridor, not just a line. This
implies the existence of tolerable imbalances in terms of slight overload
("challenge" to mobilize potential), as weil as slight surplus ("slack" to
respond to unexpected increases in complexity Ioad).
From the dynamic perspective, four basic mechanisms or processes for
establishing fit (balancing Ioad and potential) exist. On the one band,
''pul/" and "cut" processes adjust potential to Ioad by increasing ("pull" via
growth, diversification etc.) or reducing ("cut") potential (e.g. via down-
 269

sizing or body Ieasing of surplus manpower). "Push" and "ease" mecha-

nisms on the other hand adjust load to potential by increasing load (addi-
tional projects, more output volume, more variants of a product, higher
rate of innovation etc.) or reducing load (e.g. by means of standardization
or extending life cycles).
Classical fit approaches in the theory of the firm can be remodelled in
terms of complexity fit. Merging the strategic fit approach and the com-
plexity approach helps refine the notion of strategic fit. These models of
strategic fit belong to the contingeny, congruency or consistency
approaches in management (cf. Staehle 1999, pp. 60-66). The fit between
strategy und structure within Chandler's model (cf. Schewe 1999) repre-
sents one of the most popular paradigms of "strategic fit". A simple illus-
tration of Chandler's paradigm is the complexity fit between diversifica-
tion strategy (e.g. Amazon entering online clothing business) and
divisionalized structures. From the static perspective, this is accomplished
by establishing (strategic) business units (divisions) corresponding to busi-
ness segments (e.g. products, customers and regions on the e-learning
market). But even this "simple" complexity fitting is challenging: in the
case of related diversification for example, the number of business units
may be smaller than the number of business segments because products
can be dustered into product families or categories.
From the dynamic complexity viewpoint, the legendary unilateral fitting
formula ("structure follows strategy") covers both pulland cut complexity
processes (eg clearing the intracompany jungle of committees). Likewise,
the "strategy follows structure" formula (cf. Miles/Snow 1978) comprises
push activities (eg. structure enabling entry into new markets) and ease
activities. In addition to the strategy-structure fit model, several other two
domain-bilateral fit modelsexist (cf. Venkatraman 1989; Scholz 2000, pp.
112), such as the afore-mentioned Ashby model, the market-structure fit
models (Staehle 1999, pp. 433) and the businesses-resources fit models (cf.
Albach 1978, p. 709).
The outlined state-of-the-art fit model of the complexity approach has a
number of limitations and weaknesses. Some of these have to do with the
overly simplistic conception of the model.
The dynamic fitting process is undetermined since we do not know
which of the four mechanisms will be applied, nor the appropriate sequen-
tial combinations (eg first pull, then ease, ... ), let alone the simultaneous
combinations of these mechanisms. To clarify this additional guidelines
are needed:
Principles of handling complexity: This could be accomplished by ge-
neric maxims for dealing with complexity. Take for example principles of
parsimony or simplicity, from Ockham's razor ("What can be done with
270

fewer [... ] is done in vain with more"), lean management and downsizing
to the KISS principle ("Keep it simple, stupid") and the "less is more"
principle. The plea for simplicity might stem from the need to avoid high
overall complexity (aggregated over all domains, such as complexity of
strategy plus complexity of structure) in order to prevent a "complexity
overkill". Another option is provided by simplex-complex compensation
guidelines (cf. Reiss 1993a). These are not devoted to the idea of com-
plexity reduction. Instead they recommend the compensation of complex
activities in one area by simplex activities in another area. Finally the
afore-mentioned idea of "optimal complexity" can serve as a guideline. For
example, this kind of reasoning is reflected in the optimal degree of
conflict or decoupling as well as in the distinction between (optimal)
"economies" and (non-optimal) "diseconomies" of scale, speed or scope. lt
has to be considered that most of these optimization models yield isolated
local optima that are valid in one area of complexity only: the optimal
extent of conflicts is normally estimated without taking into account
existing competences in handling conflicts.
Guidelines from sequential planning: Here orientation is derived from
procedures developed in integrated planning (cf. Gutenberg 1965, p.146).
In the short run this would advocate a complexity Ievel determined by the
"bottleneck domain", i.e. minimum sector-potential (capacity for
manufacturing, power on markets etc) to deal with complexity. An existing
organizational structure might be the restricting factor. In the long run
bottlenecks have to be abolished to facilitate a complexity Ievel which
matches the complexity Ioad demanded by markets or provided by com-
petence. To accomplish this, complexity reasoning must rely on economic
modelling in terms of costs and benefits of complexity management.
For companies in the New Economy, it can be assumed that a possible
imbalance of complexity Ioad and complexity potential has to do with the
fact that these companies frequently operate as network-embedded "inter-
preneurs" (Reiß 2000). As individual entrepreneurs these young and small
companies (sometimes known as micro-companies or "one-person shows")
certainly Iack market power, as well as lobbying power. This fact hampers
the execution of some ease-procedures, e.g. reducing pressure from power-
ful customers and competitors. By networking they want to attain virtual
size i.e. access to partner resources via loose contracting. However it
remains unclear whether this networking enables ease-procedures at the
customer and competitor level(s). Furthermore, any networking gives rise
to an intemal demand for coordination reflected in transaction costs
amongst network partners. The resources required for this intemal coordi-
nation are not available for establishing potential to handle market-driven
 271

complexity loads. In other words, pulling-procedures to reach a complexity

fit may be less successful (ie may not foster a fit).
Multi-domain fit models: The bilateral fitting of two respective domains
(eg strategy and structure) could be determined by fitting processes
between other domains (eg. market and strategy) with markets forcing a
high level of complexity. This "complex" fitting requires holistic models
of complexity fit.
Hence, without further refinement of complexity models some funda-
mental generic questions conceming complexity fit, such as whether a
balance is reached at a low level of complexity (contraction of complexity
via ease and cut) or at a high level of complexity (expansion of complexity
via pull and push) cannot be answered. Although there are hints of a
potential risk of complexity overload in the New Economy (despite the
available potential captured in slogans such as "complexity is our busi-
ness!"), the evidence of complexity reasoning remains vague. Considering
this, at least two extensions of the "state-ofthe-art" complexity approach
are required.
Multilateral fit model: Conceiving several isolated two-domain fit
models (eg strategy and structure, structure and competence, market and
strategy) can only produce bilateral complexity fits between two domains.
However, the idea of strategic fit is supposed to cover all domains of
management (strategy, structure, culture, systems etc.) simultaneously (cf.
Mintzberg 1989, Scholz 2000). So a more holistic and hence more com-
plex model of interdomain complexity fit in terms of a multilateral com-
plexity fit is needed, in accordance with "structure follows process follows
strategy" or similar principles.
Complexity profilefit model: There is no strict empirical evidence for
the validity of some standard fitting models, e.g. for the validity of
Chandler's formula conceming structure-strategy fit (cf. Schewe 1999).
This lack of corroboration leaves room for several random and even Con-
tradietory formulas, such as "structure follows structure" or "structure fol-
lows fashion". The holistic complexity approach offers a model of multi-
dimensional intradomain complexity to cope with this problem. Since both
structure and strategy have their own multidimensional structural com-
plexity profile and multidimensional strategic complexity profile respec-
tively, an appropriate profilefit model has tobe conceived.
It is obvious that in both cases a more holistic complexity approach is
needed, in general, and for the New Economy in particular.
272

3 Multilateral complexity fit

A multi-domain model of complexity fit has to provide a holistic view of

all relevant domains, as well as an idea of a multilateral fit between all
domains. Figure 3 combines four complexity domains (markets, strategy,
structure and competency). Competency comprises soft factors (such as
learning, culture, skills, tolerance for errors and conflicts, relationship as-
sets, systemic thinking etc.), as well as hard factors such as process tech-
nology (hardware and software), Internet infrastructure, business intelli-
gence and control systems. Both sets of factors foster the requisite
infrastructure to cope with extemal, strategic and structural complexity.
Bilateral fitting procedures yield several, normally incompatible subop-
tima, such as the Chandler-fit, Ashby-fit, and market-structure contin-
gency. The bilateral strategic fits are quite often reached on different Ievels
of complexity: the degree of divisionalisation (number of business units)
may be higher when determined via market pull (strategy-structure fit)
than via competence push (competence-structure fit). lt is very unlikely
that the principle of suboptimization guarantees an overall optimum. In-
stead, sophisticated models are needed to warrant such ambigious fitting.
Take for example a strategy-structure-technology fit model fostering a
trilateral fit (Miles/ Snow 1978).
The multilateral interdependency reasoning is based on a multi-stage
complexity load/complexity potential interdependency. The model
provides an overall fit and not just several isolated bilateral fits. Multi-
lateral interdomain complexity fit obeys the laws of complementarity: high
Ievels of market complexity Ioad are proliferated into high Ievels of strat-
egy, structure and competence complexity. The same principle holds for a
competence-pushed interdependency: competence (derived from the Inter-
net, an integration of ERP software and CRM software as well as from
entrepreneurial culture and self organization) enables powerful organiza-
tional structures (flexible response to changes in customer demands and
competitor activities, efficient and effective mastering of business
processes etc.) which support hypercompetitive strategies to deal with
challenging markets.
 273

Load
Complexity of
Structure
Potential

Potential
Complexity of ~/ Load
Complexity of

7 ~
Competence Strategy
Potential Potential

Load
Complexity of
Market

Fig. 3. Multilateral complexity fit

Multilateral fitting starts with diagnostic act1v1ty, i.e. specifying the

existing overall misfit. The diagnostic procedure has two starting points.
Market-pulled from market complexity load, resulting in the requisite
complexity of competence. Competence-pushed from competence poten-
tial, yielding a permissible level of market complexity. The overall gap
between these two fitting operations can be visualized in the southwest
quadrant of Figure 3. Within this framework, the fitting can be performed
by applying all four fitting procedures (ie. push, pull, ease and cut) in
every sector. This (admittedly) complex procedure increases the likelihood
of finding an overall multilateral fit.
It is possible that New Economy companies with a one-sided compe-
tence-push approach have difficulties finding an overall fit. Whenever e-
business is more about "e" than about "business" the pulland cut-options
of fitting are not fully exploited. In the production of e-learning this is
typical of companies that are unable to merge technology-based virtual
leaming products (WBT, CDs, etc.) with traditional face-to-face leaming
products (print materials, classroom teaching etc.) in a so-called blended
leaming offering.
The multilateral model also covers fit between non-neighboring
domains in Figure 3. Take for example classical market-structure contin-
gencies and businesses-resources fit. These "short circuits", by omitting
interpolated domains, must not however mislead managers into forgetting
that fitting jobs still have to be done in the omitted domains.
If we take the total (overall) complexity viewpoint complementarity
triggers an accumulation of complexities (both loads and potentials) within
274

one institutional entity, e.g. a company or a cross-company supply chain:

complexity (potential) is the answer to complexity (load). This accumula-
tion may cause fears of a complexity escalation resulting in a complexity
"overkill" (provoking simplistic complexity reduction attempts, cf.
Brandes 2002, Trout/Rivkin 1999). However, complementarity inhibits
attempts torelax overall complexity by applying complex-simplex pattems
(Reiss 1993a): for instance, there is no way of responding to a complex
strategy with simple structures. This would provoke an (interdomain) mis-
fit, or more precisely an overload of (strategic) complexity not covered by
the potential of adequate structures. Likewise, neither culture nor com-
puters function as substitutes for structure. Instead these competence fac-
tors provide emergent structures, ie powerful "spontaneous orders" such as
"invisible hand"-controlled coordination pattems and informal networks.
These self-organized structures are capable of supporting complex strate-
gies (cf. Rycroftl Kash 1999).
Interdomain complementarity does not exclude however substitutional
simplex-complex compensation between subdomains of strategy or struc-
ture or competency. Take for example the classical paradigm of substi-
tuting technical resources for human resources as potential for information
processing. Likewise, there are substitutional relationships between dif-
ferent kinds of Ioads: with outsourcing, the intemal complexity of
insourcing is replaced by "extemal" complexity, i.e. contracting with
extemal suppliers. An e-leaming-specific arena of passing on Ioad
concems customization which is charaterized by a substitutional relation-
ship between customization by the provider, customization by the cus-
tomers as "prosumers" (self-customization), and third party customization
by intermedianes (cf. Reiss/ Koser 2001 ).

4 Complexity profilefit

Every domain in the multilateral complexity fit model contains different

aspects of complexity. This intradomain complexity cannot be adequately
incapsulated in one single complexity ratio (eg "the" complexity of
organizational structure). Organizational networks for instance - a typical
form of organizing production in the New Economy (cf. Bellmann/
Mildenherger 1996) - are entities with several complexity features, such as
connectivity, openness, coopetition, heterarchy, heterogenous relationships
and dynamics (cf. Reiss 2001).
Behind the various complexity features such as coopetition, connec-
tivity, openness, and the like are Jour generic dimensions of complexity (cf.
 275

Figure 4): multiplicity, variety (not to be confused with Ashby's notion of

variety which covers more or less all dimensions of complexity), fuzziness
and dynamics (Reiss 1993b). Same ubiquitous complexity phenomena
address only one complexity dimension while others are linked to more
than one MESS dimension at the same time. This is the case, for instance
with "hybrids": hybrid corporate visions (eg "glocals", a combination of
"global" and "local"), hybrid strategies combining two competitive advan-
tages (outpacing, mass customization cf. Pine 1983, J enner 2000, Proff/
Proff 1997), or hybrid organizational structures (alliances, joint ventures,
networks etc.) between market and hierarchy. Hybrids contain multiplicity
(overlay of sales contracts, relational contracts, cultural bonds etc. in
organizational networks, cf. Reiß 1996), variety (Contradietory elements
such as "mass" and "customization"), fuzziness (lack of identity, blur of
borders cf. Picot/Reichwald/Wigand 2001), and dynamics (instability due
to internal tension).
The four-dimensional MESS model covers a plethora of ubiquitous
complexity aspects (cf. Figure 4). Any of these variables ("scale", "con-
flict", "speed" etc.) can be considered a load or a potential depending on
the respective context and their extent. This, for instance, is the message
behind the "economies" and the "diseconomies" of scale, speed or net-
works. Diversity in terms of "arbitrage" is a business (not a burden) for
many a company. So too is diversity in terms of conflict for lawyers, and
in terms of suboptimal integration of materials flow for logistics providers.
The MESS dimensions can be dustered into two groups requiring two
distinct, partly Contradietory species of competence: integration, where
mass phenomena have to be handled by requisite capacity and variety, and
flexibility which deals with change phenomena by tolerating ambiguity
and chaos (i.e. patterns of "disorder").
276

I COMPLEXITY I
IINTEGRATION
"And linkage"
I I FLEXIBILITY
"Orlinka11.e"
I
I I I I
MULTIPLICITY VARIETY FUZZINESS DYNAMICS

•Multitude •Diversity •Ambiguity •Growtb

•Numerosity •Heterogenity •Uncertainty •Venions
•Holism •Piuralism •Risk •Volatility
•Piurality •Scope •Randomness •Iostability
•Volume •Options •Paradox •Cbangefulness
•Mass •Variance •Vagueness •Discontinuity
•Frequency •Divergence •Options •Amplitudes

....
•Scale •Conßict
•lndividualization
•Personalization
•Biur
•Entropy
•Rise & Fall
•Lifecycles
•Question Marks •Chaos

....
•Specifity •Intransparency

....
•Redundancy I Stack
•Surprise

....
•Speed

Fig. 4. Dimensions of complexity

The MESS concept helps elaborate the hitherto vague idea of what is
meant by complexity of strategy, structure or markets. Complementary re-
lationships between the MESS dimensions are quite frequent. This results
in an escalation (or de-escalation) oftotal intradomain complexity. In pro-
duction processes for example, manufacturing variants (variety) also
requires flexibility: it has an impact on fuzziness ("Which variant is
scheduled next"?), and dynamics ("people and/or technical facilities
switching from one variant to another"). Likewise, a multi-objective goal
system (multiplicity) is bound to create conflict between severallogistics
targets (variety). Conflict can be the source ofinstability (dynamics) ofthe
production system. Conversely, low uncertainty can keep instability down.
However, intradomain complexity relationships - unlike interdomain
complexity relationships - are not necessarily dominated by the laws of
complementarity. Although multiplicity is, by definition, a precondition
for variety ("It takes at least two elements to create variety"), and likewise
fuzziness a precondition for dynamics ("No ambiguity, no change"), there
is no suchthing as an automatic escalation or accumulation ofmultiplicity,
variety, fuzziness and dynamics in one domain. In other words, a high
level of multiplictiy (e.g. many customers from all over the world) does
not automatically trigger a high Ievel of variety (different offers for each
region or country). By means of a globalization strategy this multitude of
customers is offered standardized world products ("commodities") with
very little geographical diversity ("customization"). Correspondingly,
uncertainty is not always reflected in dramatic changes (such as disconti-
 277

nuities) in the course of time. Continuous development can be accom-

plished by stabilizing (e.g. via step-by-step implementation rather than big
bang implementation). Finally, diversity in the form of conflict does not
necessarily go along with instability and permanent changes, as long as the
conflicting interests are uncoupled (self-sufficiency that minimizes mutual
resource dependence) or harmonized on the basis of compromises (making
deals). It would seem that with appropriate measures (such as uncoupling,
standardization, proactive avoidance of conflicts etc.) the intradomain
proliferation of complexity can be suppressed. As for the corresponding
potentials (integration versus flexibility), substitutional relationship is the
rule since integration is connected to big units, monocentric structures,
close coupling (via contracts), variants (broad scope ofproducts at a time),
centripetal movements, and the like, whereas flexibility is connected to
small units, polycentric structures, loose coupling, versioning (updates,
upgrades of products in the course of time), centrifugal movements (e.g.
spin-offs), etc. Still, some hybrid management concepts try to foster both
potentials, ie go beyond the either-or-altemative: this holds for hybrid
competitive strategies, holding models (integrative corporate center plus
flexible business centers) and similar concepts.
In a sophisticated intradomain complexity model, markets, strategies,
structures and competences are represented by their respective (four-
dimensional) complexity proflies (MESS profiles). The MESS profile of a
matrix organisation - another relevant pattem for organizing production in
the New Economy, especially in the form of a project matrix organization-
is depicted in Figure 5.
278

e actual profile
• target profile

Fig. 5. Complexity Profile of a Matrix Organisation

In a multidimensional profile approach, complexity fitting is about

matehing Ioad profiles and potential profiles (complexity profile fit). As a
rule, the measurement of correspondance (fit) between Ioad profile and
potential profile is accomplished separately in each dimension, ie no
compensation is permitted. Consequently, the overall degree of fit is
defined by multiplicity-fit + variety-fit + fuzziness-fit + dynamies-fit A
deficit, say, in personalization cannot be compensated by speed, nor can
(linear) growth pattems (versions of software products) make up for a Iack
in diversity. Similarly, an excess potential in mass production is no accept-
able substitute for required slack to handle risks. Consequently, testing
interdomain fit (eg structure-strategy fit) should be executed on the basis
of complexity profiles and not on the basis of one undifferentiated catch-
all ratio.
Before the multidimensional measurement procedure is executed, a
closer Iook at the actual complexity profiles is advisable. One discovers
that not all complexity measures of a matrix organization, network organi-
zation, and the like serve as genuine potential to meet the requirements of
strategies. Instead, some complexity measures represent excess com-
plexity. This type of complexity is referred to as "hidden" complexity since
it is not visible unless a profile view is applied. Hidden complexity of
matrix structures is reflected in conflicts and scattered responsibilities (cf.
Reiss 1994b): these are not required to support functional activities and
 279

project performance simultaneously. On the contrary! They absorb energy

which is then no Ionger available for marketing activities since it is used
up for intemal communication, coordination, as well as "warfare".
Respanding to complex strategies with a complex matrix structure would
appear to be only superficially convincing. The complexity profile model
reveals that this line of complexity thought implies a fallacy since existing
hidden complexities are ignored.
To discover hidden complexity, the actual MESS profile has to be
compared to the requisite MESS profile which is derived from other
domains, such as strategy or markets (cf. Figure 3). Discrepancies between
the two represent either a deficit or an excess in complexity. With excess
complexity, a differentiation has to be made between a) over-load and b)
surplus potential. How can hidden complexity as an intradomain misfit
between required potential and actual complexity be accounted for within
a multi-domain multilateral fit model (cf. Figure 3)? On the one band,
complexity overload of matrix structures can pull the development of
additional competence to deal with matrix-specific conflicts. On the other
band, it can be countered by ease-fitting. In the case of matrix structures
this can eventually mean abandoning this structure because of its negative
inherent complexity. Since the extra load of complexity is not justified by
strategic requirements, let alone by market complexity, ease-procedures
make more sense. They are also an antidote against the risk of an overall
complexity overkill due to hidden complexities in certain domains.
The matrix structure example outlined a comparatively simple situation.
Tranferring this line of thought to the New Economy has to take into
account that all domains are characterized by a high level of complexity.
Markets are turbulent and global. Strategies reflect hypercompetition. This
implies hybrid competitive strategies (such as "mass customization" or
"mass personalization") which are enabled by competences fastered by the
Internet, as well as by people. Production is embedded in organizational
networks or virtual companies. The complexity profile of organizational
networks needs thorough scrutinization, especially since there is something
akin to a network "ideology" within the scientific community: networks
are treated as a myth or magic formula, rather than as an organizational
tool.
Networks are highly complex structures with regard to all four MESS
dimensions. On the one band, their inherent complexity is welcome in
terms of connectivity and fluidity. These support both integration (com-
munication channels) and flexibility (low entry and exit costs for project
networks). On the other band, coopetition and heterogeneity are mixed
blessings: Moderate forms of competition and heterogenity amongst part-
ners foster creativity and efficiency while high levels absorb a great
280

amount of energy and generate high complexity costs. Since networks

contain a considerable amount of hidden complexity Ioad this widespread
structure could turn out to be another fallacy in complexity management
unless the inherent complexity Ioad is covered by competence.
Without doubt entrepreneurial competences together with holistic
thinking (cf. Ulrich/ Probst 1991) help to cope with complexity challenges
hidden in networks. However, application of the complexity profile model
to these competences reveals that there is also hidden complexity load to
be taken care of. The dark side of self-organization comprises risks of free
riding, loafing, moral hazard, fraud and corruption. This is true not only
for teams, but also for inter-company networks. So it would be both naive
and inconsistent with empirical evidence of New Economy failure to as-
sume that competences ofNew Economy firms are a panacea for all com-
plexity troubles. The lesson learned from analyzing the typical structure
and competence situation in the New Economy isthat these companies can
well be the victim ofhidden complexity in networks and self-organization.

5 Conclusions and outlook

With production management in the New Economy the holistic complexity

approach furnishes arguments to answer the crucial question "ls hyper-
complexity (hyper-coopetition) responsible for New Economy failure?"
The following arguments advocate a positive answer: the inability ofNew
Economy companies to achieve a complexity fit may be due to deficits in
establishing a multilateral fit between market, strategy, structure and
competence because of a bias towards competence-pushed fitting as
weil as restrictions for ease-fitting caused by a Iack of power. Conse-
quently crucial fitting parameters for attaining a complexity fit are not
applied.
dealing with hidden complexity in organizational networks and
competences of self-organization.
High overall complexity per se is not the "killer" of the New Economy.
High complexity is no problern as long as there is a balance between
complexity Ioad and complexity potential. In other words naive
approaches to complexity in terms of reducing the Ievel of complexity are
inadequate. As are naive models of strategic fit that ignore hidden
complexity. Only the merging of state-of-the-art complexity reasoning and
Strategie fit into a holistic multi-domain multi-dimensional complexity
approach can support conceiving models that help us understand New
 281

Economy failure. Despite these useful contributions to understanding how

the New Economy works "seen through complexity glasses", the holistic
complexity approach has quite a number of drawbacks and limitations.
These are primarily the result of existing drawbacks of a) strategic fit
modelling and b) complexity modelling. Take as an example the lack of
complexity metrics: There are undoubtedly some available measurements
of complexity. The spectrum comprises counting (numerosity), N:K-ratios
(N=number of agents; K = number of Connections between agents, cf.
Eisenhardt/ Bhatia 2002, pp. 433-447), statistics of variance, range,
standard deviation (variety), probabilities and entropy (fuzziness) and
volatility (dynamics). Still, the lion's share of complexity can only be
measured on ordinal scales, clarifying "more" and "less" complexity at
best. This also holds for the underlying concepts of "fit" and "misfit". They
represent measures of meta-complexity incapsulating the discrepancy
(variety) between (1st order) complexity parameters.
This lack of "operationalisation" is a major drawback of complexity rea-
soning, especially when it comes to "estimating" fit or misfit. It would
seem that the complexity approach is only capable of providing a number
of "soft heuristics". These may be unacceptable from the "Only what is
measured can be managed" viewpoint. They are at any rate useful in
fostering awareness for the impact of complexity, and they support eco-
nomic modelling.

References

Adam D (ed, 1998) Komplexitätsmanagement, SzU - Schriften zur Unter-

nehmensnwhrung,vol61, VViesbaden
Ahlemeyer HVV, Königswieser R (eds) (1998) Komplexität managen: Strategien,
Konzepte und Fallbeispiele, VViesbaden
Albach H (1978) Strategische Unternehmensplanung bei erhöhter Unsicherheit.
In: Zeitschrift fiir Betriebswirtschaft, 41: 702-715
Allen P (200 1) VVhat Is Complexity Science? Knowledge of the Limits to
Knowledge. Emergence - Journal of Complexity Issues in Organizations and
Management, 3: 24-43
Allison MA, Kelly S (1999) The Complexity Advantage: How the Science of
Complexity Can Help Your Business Achieve Peak Performance, New York
(NY)
Ashby VVR (1956) Introduction to Cybernetics, New York (NY)
Bellmann K, Mildenherger U (1996) Komplexität und Netzwerke. In: Bellmann
K, Hippe A (eds) Management von Unternehmensnetzwerken - Interorga-
nisationale Konzepte und praktische Umsetzung, VViesbaden, pp 121-156
282

Bengtsson M, Kock S (2000) "Coopetition" in Business Networks - to Cooperate

and Compete Simultaneously. Irrdustrial Marketing Management 29: 411-426
Brandenburger A, NalebuffB (1996) Co-opetition, New York (NY)
Brandes D (2002) Einfach managen. Klarheit und Verzicht - der Weg zum
Wesentlichen, Wien
Brown SL, Eisenhardt KM (1997) The Art of Continous Change: Linking
Complexity Theory and Time-paced Evolution in Relentlessly Shifting
Organization. Administrative Science Quartely, 42: 1-34
Bruhn M (1997) Hyperwettbewerb: Merkmale, treibende Kräfte und Management
einerneuen Wettbewerbsdimension. Die Unternehmung, 51: 339-357
Child J (1972) Organizational Structure, Environment and Performance: the Role
ofStrategic Choice. Sociology, 6: 1-22
D'Aveni R (1994) Hypercompetition. Managing the Dynamics of Strategie
Maneuvering, New York (NY)
Eisenhardt K, Bhatia MM (2002) Organizational Complexity and Computation.
In: Baum JAC (ed) The Blackwell Campanion to Organizations, Oxford
Malden (MA), pp 442-466
Emery FE, Trist EL (1965) The Causa! Texture of Organizational Environments.
In: Human Relations, 18: 21-32
Gharajedaghi J (1999) Systems Thinking: Managing Chaos and Complexity: A
Platform for Designing Business Architecture, Boston (MA) et al.
Grint K (1997) Fuzzy Management - Contemporary Ideas and Practices at W ork;
New York (NY)
Grothe M (1997) Ordnung als betriebswirtschaftliches Phänomen - Die Bedeu-
tung von Koordination und Komplexität; Wiesbaden
Gutenberg E (1965) Grundlagen der Betriebswirtschaftslehre, Band 1 Die
Produktion, 10th edn, Berlin Heidelberg New York (NY) et al.
Jenner T (2000) Hybride Wettbewerbsstrategien in der deutschen Industrie .:...
Bedeutung, Determinanten und Konsequenzen für die Marktbearbeitung. Die
Betriebswirtschaft, 60: 7-22
Kelly K (1998) N ew Rules for the N ew Economy: 10 Radical Strategies for a
Connected World, Reading (MA)
Kelly S, Allison MA (1998) The Complexity Advantage: How the Science of
Complexitiy Can Help Your Business Achieve Peak Performance, New York
(NY)
Lewin R (1999) Complexity: Life at the Edge ofChaos, 2nd edn, Chicago (IL)
Lissack MR, Roos J (2000) The Next CommonSense: The e-manager's guide to
mastering complexity, London
Malik F (2000) Systemisches Management, Evolution, Selbstorganisation: Grund-
probleme, Funktionsmechanismen und Lösungsansätze für komplexe
Systeme, 2nd edn, Bern Stuttgart Wien
Mintzberg H (1989) Mintzberg on Management: Inside Our Strange World of
Organizations, New York (NY)
Picot A, Reichwald R, Wigand RT (1996) Die grenzenlose Unternehmung.
Information, Organisation und Management, 2nd edn, Wiesbaden
 283

Picot A, Reichwald R, Wigand RT (200 1) Die grenzenlose Unternehmung.

Information, Organisation und Management, 4th edn, Wiesbaden
Pine BJ (1993) Mass Customization. The New Frontier in Business Competition,
Boston (MA)
Proff H, Proff HV (1997): Möglichkeiten und Grenzen hybrider Strategien. In:
Die Betriebswirtschaft, 57. Jg., 1997, 6: 796-809
Regine B, Lewin R (2000) Leading at the Edge: How Leaders Influence Complex
Systems. In: Emergence - Journal of Complexity Issues in Organizations and
Management, 2 (2): 5-24
Reiss M (1993a) Komplexität beherrschen durch "Orga-Tuning". In: Reiss Met al.
(eds) Komplexität meistem- Wettbewerbsfähigkeit sichern, Stuttgart, pp 1-41
Reiss M (1993b) Komplexitätsmanagement (1/II). Das Wirtschaftsstudium, 22: 54-
60, 132-137
Reiss M (1994a) Optimale Komplexität von Führungsorganisationen. Die
Unternehmung, 48 (2): 83-97
Reiss M (1994b) Schlanke Matrix. Zeitschrift Führung und Organisation, 63 (1):
6-10
Reiss M (1996) Grenzen der grenzenlosen Unternehmung. Perspektiven der
Implementierung von Netzwerkorganisationen. Die Unternehmung, 50 (3):
195-206
Reiss M (2000) Unternehmertum in Netzwerken. In: Reiss M (ed) Netzwerk-
Unternehmer. Fallstudien netzwerkintegrierter Spin-offs, Ventures, Start-ups
und KMU, München, pp 1-37
Reiss M (2001) Netzwerk-Kompetenz. In: Corsten H (ed): Unternehmungs-
netzwerke. Formen unternehmungsübergreifender Zusammenarbeit, München
Wien, pp 121-187
Reiss M, Beck TC (2000) Netzwerkorganisation im Zeichen der Koopkurrenz. In:
Foschiani S, Habenicht W, Schmid U, Wäscher G (eds) Strategisches
Management im Zeichen von Umbruch und Wandel, Festschrift für Prof. Dr.
Erich Zahn zum 60. Geburtstag, Stuttgart, pp 315-340
Reiss M, Koser M (2001) Third Party Customization. In: Blecker T, Gemünden
HG (eds) Innovatives Produktions- und Technologiemanagement, Berlin et
al., pp 55-85
Reiss M, Bemecker T (2003) Entwicklungsdynamik von Dienstleistungs-
netzwerken. In: Bruhn M, Stauss B (eds) Jahrbuch Dienstleistungs-
management 2003: Dienstleistungs-Netzwerke, Wiesbaden, pp 87-112
Roters M (1988) Komplexität und Dynamik als Einflußgrößen der Effizienz von
Organisationen, Frankfurt am Main et al.
Rycroft RW, Kash DE (1999) The Complexity Challenge, London
Schewe G (1999) Unternehmensstrategie und Organisationsstruktur. Die
Betriebswirtschaft, 59:61-75
Scholz C (2000) Strategische Organisation. Multiperspektivität und Virtualität, 2nd
edn, Landsberg am Lech
Schwenk-Willi U (2000) Integriertes Komplexitätsmanagement : Anleitungen und
Methodiken für die produzierende Industrie auf Basis einer typologischen
Untersuchung, St. Gallen
284

Stacey RD (2000) Strategie Management and Organisational Dynamics, London,

New Y ork (NY) et al.
Staehle W (1999) Management: eine verhaltenswissenschaftliche Perspektive, 8th
edn, München
Stüttgen M (1999) Strategien der Komplexitätsbewältigung in Unternehmen- Ein
transdisziplinärer Bezugsrahmen, Stuttgart Bem, Wien
Trout J, Rivkin S (1999) The Power of Simplicity- A Management Guide to
Cutting Through the Nonsense and Doing Things Right, New Y ork (NY)
Underwood J (2001) Thriving in E-Chaos, Roseville (CA)
Venkatraman N (1989) The Concept ofFit in Strategy Research: Toward Verbal
and Statistical Correspondence. Academy of Management Review, 14 (3):
423-444
Wood R (2000) Managing Complexity- How businesses can adapt and prosper in
the connected economy, London
ZahnE, Foschiani S (2000) Strategien und Strukturen für den Hyperwettbewerb.
In: Wojda F (ed) Innovative Organisationsformen, Stuttgart, pp 89-113
Advanced Planning Systems - Basics and
beyond

Hartmut Stadtier

1 Abstract

Advanced Planning Systems (APS) have proved tobe a valuable means for
effective intra-organizational Supply Chain Management. This paper
firstiy shows the modules oftoday's APS and their interrelations.
Secondly, we will focus on inter-organizational Supply Chains and ways
to align decentrally generated (master) plans. Starting from the collabora-
tive planning concept of today's APS we present a model-based negotia-
tion scheme and its relation to multi-agent systems.

2 lntroduction

Many companies have realized that Euterprise Resources Planning (ERP)

Systems are excellent (transactional) information systems (Vollman et al.
(1997)) but fall short when it comes to planning. As a result, employees
often created individual software solutions decentrally for decision support
based on spreadsheets. However, these individual software solutions often
lack integration and suffer from inconsistent data bases and uncoordinated
decision making. In light of the above deficiencies companies now are
implementing Advanced Planning Systems (APS) expecting state-of-the-
art decision support.
This paper firstly shows the architecture of APS by means of the Supply
Chain Planning (SCP) Matrix. Thereby we will show the modules of
today's APS, their planning functionality and their interrelations. Since
APS are based on the principles of hierarchical planning (see Anthony
(1965), Hax and Meal (1975), Stadtier (2002)), APS are best suited to
support planning within a single company.
However, due to the formation of Supply Chains (SC, see Christopher
(1989)) there is a growing need to coordinate plans even across legally
separated companies. The problern is that these companies usually cannot
be controlled in a hierarchical manner. Even worse, they often are reluctant
to share the data necessary for centralized planning (Kersten (2002)).
286

Hence, the second aim of this paper is to focus on inter-organizational

SCs and ways to align decentrally generated plans. Starting from the
collaborative planning concept of today's APS we present a negotiation
scheme and its relation to agent technology.
Alternative approaches for aligning decentrally generated plans have
been developed in areas like SC contracting (Tsay et al. (1999)), inventory
control combined with transfer payments (Lee and Whang (1999)), pur-
chasing with quantity discounts (Corbett and Groote (2000)), and hierar-
chical planning oriented coordination (Schneeweiss and Zimmer (2003)).
However, in order to keep this paper concise these approaches will not be
dealt with in this paper.
The structure of the paper is as follows. In Section 3 the architecture and
modules of today's APS are introduced. Section 4 is devoted to the
coordination of plans in inter-organizational SCs. It starts with recalling
the essence of collaborative planning in today's APS. Then we describe a
negotiation scheme intended for coordinating master plans. It is shown that
this scheme can be regarded as a specific building block to be used in
agent technology. Ideas for future research are provided in Section 5.

3 Architecture of APS - the supply chain planning

matrix

Although developed independently by different software vendors APS

exhibit a common architecture based on the principles of hierarchical
planning. The main focus is on supporting the material flow across a SC
and related business functions: procurement, production, transport and
distribution as well as sales (see figure 1, x-axis). The associated planning
tasks can be considered at different Ievels of aggregation and planning
intervals ranging from 'aggregated long-term' to 'detailed short-term'
planning (see figure 1, y-axis). These two axes span the SCP matrix. lts
contents are the planning tasks, which also correspond to software modules
constituting an APS. These planning tasks and associated functionality of
softwaremoduleswill now be described briefly.
 287

procurement produc11on

long-term Strategie Network. Planning

mid-term
I
Master Plannlng
I Demand
Plannlng
Purchaslng Productlon Distribution
& Plannlng Plannlng
Materlai
Demand
Requirements Transport
short-term Schedullng Fulfllment
Planning Plannlng &ATP

Fig. 1. Softwaremodules covering the SCP matrix (Meyr et al. (2002), p. 99)

Demand planning
Aceurate demand forecasts form the starting point of effective SC plan-
ning. The further upstream the customer decoupling point is located in the
SC (e.g. deliver-to-stock, assemble-to-stock, make-to-stock . . .(Kolisch
(2001))), the !arger is the portion of already known customer orders com-
pared with forecasted demand. Also, the shorter the planning horizon, the
!arger this portionturnsout (Meyr (2003)).
Apart from well-known methods for univariate time series - like
Winters exponential smoothing for seasonal and trend demand (Silver et
al. (1998))- there arealso multivariate methods and life cycle models. The
step from pure demand forecasting to demand planning is made by adding
to the formal demand forecasts those exceptional influences expected to
happen in the future and their impact on future sales.
Demand planning is closely related to a data warehouse where time
series can be retrieved in different granularity and dimensions (product,
geographic region and time). Expected future demands are the input to
several modules in various aggregations and forecast intervals.

Strategie network p/anning

A planning interval of several years can be assumed when designing the
structure of a SC. Here the location of production sites, warehouses, geo-
graphical customer areas to serve are laid out. Also, capacities of these
288

facilities as well as the transportation means (ships, trucks, railway etc.) to

be used are decided upon.

Master planning
Given the structure of the SC, Master Planning looks for the most efficient
way to fulfill demand forecasts over a medium-term planning interval,
which often covers a full seasonal cycle. Master Planning not only
balances demand forecasts with available capacities but also assigns
demands (production amounts) to sites in order to avoid bottlenecks. Due
to the medium-term planning horizon it is often possible to adjust available
capacities to a certain extent (e.g. by overtime). Usually, the amounts tobe
procured may result in special arrangements with suppliers (e.g. standing
orders).

Production planning and detailed scheduling

While Master Planning coordinates flows between sites, Production Plan-
ning and Detailed Scheduling is run within each site, or even each produc-
tion department based upon directives of Master Planning. In production
planning the level of detail are shifts, machine groups or flow lines which
may become a bottleneck and operations to be performed on these poten-
tial bottlenecks. In case the loading of machine groups - including lot-
sizing decisions - is strongly related with the sequencing of jobs, both
Production Planning and Detailed Scheduling have to be performed
simultaneously (which often applies to the process industry).

Purchasing and material requirements planning

Master planning as well as short-term Production Planning and Detailed
Scheduling provide directives for calculating procurement quantities to be
planned within the module Purchasing and Material Requirements Plan-
ning. After disaggregating product types or product families into items, a
bill-of-material (BOM) explosion is applied to derive required quantities of
procured items.
Furthermore, this module is needed in the short term for planning of
non-bottleneck operations because only potential bottleneck operations are
planned for in Production Planning and Detailed Scheduling. In order to
find out which operations have to be performed at which points in time
also a simple BOM explosion is executed.
 289

Distribution planning
So far we have mainly concentrated on production operations. Now the
flow of goods between sites as well as in the distribution network comes
into play. Seasonal stock levels at different stocking points in the SC have
already been planned for in Master Planning. Here, we have to take care of
transports of goods to customers (directly) as well as via warehouses and
cross docking. This now happens at a greater detail than in Master Plan-
ning. In case production amounts do not exactly match a current period's
demand, rules and procedures are applied to guide the flow of goods
within the SC (e.g. in case of shortage, transport of goods will be such that
target inventories of an item at different distribution centers are filled at an
equal percentage).

Transport planning
Basedon production orderstobe completed the next day (or shift) truck-
loads for different destinations have to be formed (so called vehicle
loading). This also requires detailed knowledge of outstanding orders from
warehouses and customers. Also, the specific needs of customers (like
time windows for delivery) and legal restriction for drivers have to be
obeyed. Sequencing customer locations on a vehicle's trip is accomplished
in (models of) vehicle routing

Demand fulfillment and available-to-promise

Last but not least there is the interface to the customers, via the Demand
Fulfillment and Available-to-Promise module. One task is to track cus-
tomer orders from order entry, via order execution to order delivery.
Furthermore, order promising, due date setting and shortage planning are
considered here.

Execution
Although not considered a module of APS we would like to point out the
issue of executing decisions made at short term planning levels. The only
link from APS to those in charge of executing plans (like employees on the
shop floor or process controls of machines) is via an ERP system. Since
transactional ERP systems cannot provide real time control, additional
systems, e.g. manufacturing execution systems (MES), may be required
(MESA International (1997)).
A further interface has to be considered, if plans are generated decen-
trally in an inter-organizational SC which is the topic of the next section.
290

4 Coordinating plans in inter-organizational supply

chains

4.1 Collaborative planning by today's APS

While managing an intra-organizational SC may already be a difficult task,

the real challenge arises in an inter-organizational SC. Here, one has to
assume that each (legally separated) organizational unit has its own plan-
ning domain and objectives. Each organizational unit keeps its own data
base (data warehouse) and creates its own plans. Still, plans along a SC
have to be aligned in order to remove waste and thus to improve competi-
tiveness.
Assuming that two adjacent parties, i.e. a supplier and a buyer, make use
of their own APS, coordination of plans can take place at different Ievels
ofthe planning hierarchy (see figure 2).

Strategie Network P lanning

Master Planring

ProdJction Distribution Purt:hasing

Planning &

Fig. 2. Collaboration between APS (Meyr et al., 2002, p. I 03)

We define collaborative planning as the alignment ofplans on the same

hierarchical (planning) Ievel between self-contained, usually legally sepa-
rated organizational units linked by a supplier-buyer relationship. Collabo-
rative Planning is supported by specific modules available from (APS)
software vendors. In essence the basic idea is as follows:
A long term negotiated Ievel of demand is agreed upon by the supplier
and buyer for each item (see figure 3). Starting from "today" there are
 291

three phases, the frozen phase, the commit phase and the forecast phase. In
the frozen phase no changes to the shipment plan are allowed. During the
commitment phase amounts to be shipped should remain unchanged too.
However, if needed one party may ask for alterations, if it compensates
cost increases faced by the other party. In the forecast phase there is a
widening exception corridor the further one Iooks into the future. Only if
the exception corridor is violated, e.g. by planned demand for an item, an
alert is created automatically, showing both parties that they should contact
each other to find out whether actions have to be taken to cope with this
exception.

nstory Iorecast
phase pha.se

Fig. 3. Time phases of collaboration (example) (Kilger and Reuter (2002), p. 236)

Thus exception handling rests with the users and is not supported by the
software. While it might seem easy to find a solution for one item, it might
turn out that interdependencies between items (e.g. via a bottleneck
resource) Iead to new "problems" while resolving a given alert. This obser-
vation calls for a more sophisticated approach simultaneously taking into
account all bottlenecks and associated items within an organizational unit.
A corresponding negotiation scheme which can make use of sophisticated
optimization models available for Master Planning is presented next.
But before tuming to model-based negotiations an advantage of collabo-
rative planning - even in its basic form as described above - compared
with pure market interactions should be made clear: Now the buyer not
only informs the supplier about the current order to be placed but also
about future demands (orders). Thus the supplier does not have to resort to
her own (vague) demand forecasts and may even get a commitment on
future sales. The buyer on the other band can be sure to receive the raw
materials and components needed to fulfill her customers' demands at
292

known cost. Hence, due to more reliable information on demands one can
expect a smaller bullwhip effect (see Lee et al. (1997)).

4.2 A model-based negotiation scheme

The negotiation scheme presented here is intended for bilateral negotia-

tions between one supplier and one buyer. Note, that one supplier and one
buyer form the smallest inter-organizational SC possible (for an extension
see Dudek and Stadtier (2003)). The negotiation scheme is based on the
following premises:
- Both the supplier and the buyer create their own master plans
individually. Theseplans are assumed to be generated with the help
of Mathematical Programming models; hence it is a model-based
negotiation scheme.
- Only purchasing orders and supply proposals in the planning interval
are exchanged between the supplier and buyer. Purehaseorders must
be fulfilled completely but can be shipped in several consignments at
different points in time.
- Latest deliveries of cumulated supplies over the planning interval are
made available to the supplier for each item.
- Monetary compensations are requested by the buyer if the supplier
asks for modifications of orders. Transfer prices have been fixed
beforehand.
- Both parties are assumed to be fair, i.e. there is true information
providing and no cheating.
- Once a procurement/supply plan is agreed upon it is binding
(committed) for both parties.
The second and third premise depict the least information exchange
necessary for aligning procurement orders and supply proposals. These
requirements respect the findings of an empirical study by Kersten (2002),
reporting that suppliers in the automotive industry are very sensitive with
respect to information sharing with buyers. E.g. they are reluctant to report
capacity utilizations to the buyer. On the other hand suppliers rate an
exchange of demand plans favorably.
The buyer is able to calculate latest deliveries of cumulated supplies
over the planning interval by a simple bill-of material (BOM) explosion of
expected end-item demands by utilizing a Material Requirements Planning
(MRP) module (see Vollman et al. (1997)).
Compensations are the result of negotiations which start with pure
upstream planning. l.e. the buyer generates her minimal cost master plan
 293

assuming unrestricted supplies (objective value i· 0). Associated with this

master plan is a set of purchasing orders from the buyer covering several
items and spanning the whole planning interval.
The supplier now creates her best master plan such that purchasing
orders are fulfilled exactly, resulting in cost z8' 0• Now the supplier tries to
reduce her cost by looking for "small modifications" to those purchasing
orders which yield large cost reductions. A "small modification" to an
order to be shipped in period t 1 means that it is postponed up to the next
planned delivery ofthat item, say t 2 at the latest. Obviously, postponement
of an order - or a portion of it - is further restricted to a point in time where
timely delivery of the buyer's end item demand is starting to be at risk.
Such a counterproposal may be generated with the help of a Mathematical
Programming model which is the basis of Master Planning in most APS
today. The counterproposal, with say an objective function value Z 8' 1 (with
~· 1 ~ ~· 0 ), is sent back to the buyer for evaluation.
The buyer then calculates a new minimum cost plan given the supply
proposals from the supplier (resulting in objective function value i· 1). Due
to restrictions on procurement, cost will increase (or at best remain the
same) compared with the previously generatedplan with cost i· 0 • Thus the
buyer will (at least) ask for a compensation of c 1 = i· 1 - i· 0 in ordernot to
be worse off than before. Now, the supplier has two different supply
proposals to choose from with different compensations (the first compen-
sation c0 being 0). This terminates the first iteration.
However, the supplier may ask for another iteration i (i = 1, ... , I) if
there is a chance for further cost reductions. Then the buyer will make a
counterproposal with additional "small modifications" to the supplier's
supply proposal and less cost on her side, say i· 2• Her second order
proposal, together with the requested compensation c2 = zb,z - zb,o, is
handed over to the supplier for evaluation.
Finally, the supplier will choose the supply plan which constitutes
minimum cost from her perspective over all iterations, i.e. min{~.i + ci II
= 0, ... , 21-1 }. It can be shown (by inserting ci = i·i- i· 0) that the supplier's
choice also yields the minimum cost among the solutions generatedfor the
SC as a whole.
A hypothetical example (based on a numerically solved test instance,
see figure 4) further illustrates the negotiation scheme:
294

Supplier Buyer Compensation

Iteration 1 ( :t'· 0~ 187,728 +------ i· 0= 77 464 0

'
zd= 177 998 ---+ !'· 1 ~ 80,524) 3,060
'
Iteration 2 ( :t·'~ 187,728 +------ i· 2= 78 099 635
'
z'· 3= 168 498 ---+ i· 3= 85 744 8,280
' '
Fig. 4. An example negotiation between a buyer and supplier (2 iterations)

The first counterproposal generated by the supplier (with cost Z8 ' 1=

177,998) results in a buyer's master plan with cost i·'= 80,524. Conse-
quently, the buyer will ask for a compensation of c 1= 80,524-77,464 =
3,060. Still, the supplier will be better off than accepting the initial (pure
upstream) plan. The best solution for the supplier, and thus for the whole
supply chain, is the last supply proposal depicted in figure 4.
As has been shown by Dudek and Stadtier (2002) near optimal solutions
can be achieved after only a few (e.g. 4-5) iterations.
By adopting the above negotiation scheme the buyer will perform at
least as good as in pure upstream planning (due to compensations) while
the supply chain as a whole will operate at almost minimal cost. In order to
achieve a win-win situation it seems wise for the supplier to offer a
premium to the buyer on top of compensations for her willingness to
cooperate in the negotiation process.
The following section will show how the above model-based negotia-
tion scheme fits into multi-agent systems.

4.3 Multi-agent systems

Exchanging information, aligning and updating plans in a stochastic,

dynamic environment between partners of an inter-organizational SC asks
for a flexible software architecture - one such solution are multi-agent
systems (MAS).
According to Fox et al. (2000, p. 166) "an agent is an autonomous, goal
oriented software process that operates asynchronously, communicating
and coordinating with other agents as needed." An agent is in charge of a
specific task, hence one can discriminate e.g. (Grolik et al. (2001)):
 295

- production agents representing a single capacitated resource (e.g.

machine) which is capable of producing one (or several) products,
- logistics agents in charge for the transport of goods required between
a production agent A and B; negotiations concem the time window of
the shipment and the price,
- inventory agents being responsible for inventory holding of items,
- etc.
The above mentioned agents perform a single task which cannot be
divided any further. Grolik et al. (2001) also discuss the combination of
several agents (also of different types) to become an agent-holon, i.e. a set
of agents which "act as one" with respect to the outside world. Hence, the
decomposition of tasks along a SC becomes an important modeling feature
of an agent system. Another decomposition scheme and definition of
agents for SC management is advocated by Fox et al. (2000):
- order acquisition agents are responsible for acquiring orders from cus-
tomers,
- transportation agents are in charge of assigning and scheduling trans-
portation resources,
- scheduling agents perform scheduling and rescheduling activities in a
factory or answering "what-if' questions with respect to order accep-
tance,
- etc.
Although a software agent follows its own goals, makes choices and
decisions, agents cannot act entirely on their own. They have to make
arrangements with other agents and to find compromises (e.g. an end item
can only be produced if the required components are procured by a
supplier agent).
Advantages of an MAS are its flexibility, Sealability and reusability. It is
flexible in the sense that it can be adapted easily to the changing (network)
structure of a SC just by exchanging some agents, instead of remodeling
the total system.
Research in MAS so far has concentrated on the architecture of MAS
and related issues in the area of computer science. Less attention has been
paid to the quality of decisions generated. Applicability is sometimes
demonstrated by means of an example (Grolik et al. (2001)) or by com-
paring the performance of different kinds of rules for decision making
(Kjenstad (1998)).
Often simple rules are advocated for decision making in MAS (like
auction mechanisms, Fischer et al. (1998)). In the following we propose a
hybrid system consisting of an MAS which makes use of sophisticated
296

models and solution techniques, as they are known in Operations

Research, together with the negotiation scheme presented above. This
should improve the performance of a SC significantly without jeopardizing
the advantages of an MAS.
Standardized conversation plans and negotiation protocols play an
important role in MAS. Information exchange is executed via a specific
syntax laid out in e.g. a Knowledge Query and Manipulation Language
(KQML, Finin et al. (1995)).
Figure 5 shows an adaptation of conversation plans presented in Fox et
al. (200 1, p.177) to our mode1-based negotiation scheme: A system' s
(agent's) state is represented by a circle, a task which transfers the system
from one state to another by an arrow.
The process starts with an input of demand data. Then the "buyer agent"
is in state "demand data received". Error checking routines can now be
applied detecting e.g. negative demands. If an error is detected, the system
is transferred into state "failed". Otherwise the data are input to the Master
Planning module which creates a feasible master plan. In case no feasible
master plan is generated the buyer agent is transferred into the "failed"
state. Otherwise, purchase orders are derived from the buyer's masterplan
and sent to the supplier (agent).
Here, purchase orders are received and may also be checked for errors
(e.g. finding orders for items not produced by the supplier). Once checked,
orders are fed into the suppliers Master Planning module. Three outcomes
are possible. Firstly, it may not be possible to fulfill purchase orders within
the planning interval ("failed"). Secondly, it may turn out that the sum of
the current master plan's cost plus the compensation requested by the
buyer is larger than those associated with a previously calculated best plan.
Then the conversation is stopped and the set of purchase orders related to
the bestplan can be accepted (committed). Thirdly, a counterproposal is
generated by the supplier by utilizing Master Planning resulting in a new
supply proposal.
 297

Start

Buyer agent
Demand
data
reeeived
/ . . Failed
Masterpl Initiate eounterorooosa Masterplan
ereated + ereated +
eompensatio
........... Failed
Purehase Supply
plan proposal
derived reeeived

Purehase Supply
orders proposal
reeeived derived

./
Masterplan Masterplan
ereated ereated

........... FaHed
Purehase
orders
aeeepted
Initiate eounterorooosal
Fig. 5. Conversation plan for the model-based negotiation scheme

It is handed over to the buyer (agent) for evaluation, whieh terminates

one iteration. Evaluation will take plaee by generating a new master plan
taking into aeeount the supply proposal. Finally, the supply proposal will
give rise to generating a eounterproposal on the part ofthe buyer.
The eonversation plan eould be extended by e.g. inserting a "supplier
seleetion" task in ease there are several suppliers for an item. Also, there
might be a ehoiee of solution algorithms for ereating master plans differing
in solution quality and eomputational efforts. However, sinee it has been
298

our intention to only show that our model-based negotiation scheme can be
described in terms of an MAS conversation plan, we will not further com-
plicate the example.
We would like to add that workflows created to describe the information
flow between planning tasks in the SCP matrix (see Fleischmann et al.
(2002) pp. 82) can also be regarded as conversation plans in terms of
agent-oriented SC management.

5 Summary and future research directions

We have shown that the architecture of APS can be described with the help
of the SCP matrix. Strategie, tactical and operational planning tasks are
solvable with the help of specific modules. These modules contain state-
of-the-art modeling elements and sophisticated solution algorithms. Con-
sequently, one should expect a much better quality of plans than before
(when only transactional ERP systemswerein place).
Now that APS have been adopted by many companies the next step is to
support collaborative planning in inter-organizational SCs. While collabo-
ration has to take place at all Ievels of the planning hierarchy, we have
shown a model-based negotiation scheme for aligning master plans
between a supplier and a buyer. Only insensitive purchase orders and sup-
ply proposals are exchanged between the two partners. A great advantage
of the scheme proposed is the calculation of compensations requested by
the buyer. Starting from pure upstream planning improved master plans
can be generated resulting in a win-win situation for both partners.
The negotiation scheme can be used as part of a hybrid system
consisting of an MAS together with sophisticated modeling and solution
techniques.
An extension of this negotiation scheme to two Ievel SCs with divergent
and convergent structures is under way. However, one should be cautious
to apply this bilateral negotiation scheme in a SC with more than two
Ievels: Although possible, there are situations where favorable negotiations
between two partners may fix master plans in such a way that subsequent
negotiations with a third partner result in extremely high cost. Thus, the
SC as whole might be worse off than in pure upstream planning. Further
research is needed to cope with these unpleasant situations.
 299

References

Anthony RN (1965) P1anning and control systems. A framework of analysis.

Boston
Christopher M (1998) Logistics and Supply chain management. Strategies for
reducing cost and improving service. London et al. 2nd edn
Corbett CJ, de Groote X (2000) A supplier's optimal quantity discount policy
under asymmetric information. Management Science 46: 444-450
Duclek G, Stadtier H (2002) Negotiation-based collaborative planning in supply
chains, Working paper, Darmstadt University of Technology. Darmstadt
(submitted to EJOR)
Duclek G, Stadtier H (2003) Supply Chain Coordination by Model-Based
Negotiations. Working paper, Darmstadt University of Technology.
Darmstadt
Finin TLY, Mayfield J (1995) KQML as an agent communication language. In:
Bradshaw JM (ed) Software agents. MIT Press, Cambridge, MA
Fischer K, Ruß C, Vierke G (1998) Decision theory and coordination in
multiagent systems. Research report RR-98-02, Deutsches Forschungs-
zentrum für Künstliche Intelligenz GmbH, Kaiserslautern, Germany
Fleischmann B, Meyr H, Wagner M (2002) Advanced planning. In: Stadtier H.,
Kilger C (eds. ). Supply chain management and advanced planning - concepts,
models software and case studies. Berlin et al., pp 71-96
Fox MS, Barbuceanu M, Teigen R (2000) Agent oriented Supply Chain
Management. International Journal of Flexible Manufacturing Systems 12:
165-188
Grolik S, Stockheim T, Wendt 0, Albayrak S, Fricke S (2001) Dispositive
Supply-Web-Koordination durch Multiagentensysteme. Wirtschaftsinformatik
43: 143-155
Hax AC, Meal HC (1975) Hierarchical integration of production planning and
scheduling. In: Geister MA (ed) Studies in Management Science, vol I,
Logistics. Amsterdam: 53-69
Kersten W (2002) Internetgestützte Beschaffung komplexer Produktions-
materialien, presentation at the meeting ofthe "wissenschaftliche Kommission
Produktionswirtschaft am 13. und 24. September 2002 in München".
Kilger C, ReuterB (2002) Collaborative Planning. In: Stadtier H, Kilger C (eds).
Supply chain management and advanced planning - concepts, models
software and case studies. Berlin et al., pp 223-237
Kolisch R (200 1) Make-to-order assembly management, Berlin et al.
Kjenstad D (1998) Coordinated supply chain scheduling. Ph.D thesis., Norwegian
University of Science and Technology - NTNU. Department of Production
and Quality Engineering, Trondheim, Norway
Lee HL, Padmanabhan V, Wang S (1997) The bullwhip effect in supply chains.
Sloan Management Review 38 (Spring): 93-102
Lee H, Whang S (1999) Decentralized multi-echelon supply chains: Incentives
and information. Management Science 45: 633-640
300

Lee H, Whang S (1999) Decentralized multi-echelon supply chains: Incentives

and information. Management Science 45: 633-640
MESA International (1997) MES explained: A high Ievel vision. White paper
number 6 (September 1997)
Meyr H, Wagner M, Rohde J (2002) Structure of Advanced Planning Systems. In:
Stadtier H, Kilger C (eds). Supply chain management and advanced planning
- concepts, models software and case studies. Berlin et al. 99-104
Meyr H (2003) Die Bedeutung von Entkopplungspunkten fiir die operative
Planung von Supply Chains. Working paper, University of Augsburg,
Augsburg (submitted to Zffi)
Sehneeweiss C, Zimmer K (2003) Hierarchical coordination mechanisms within
the Supply Chain, European J. ofüperations Research (forthcoming)
Silver EA, Pyke DF, Peterson R (1998) Iuventory Management and Production
Planning and Scheduling, New York et al. (Wiley) 3rd edn
Stadtier H (2002) Basics of supply chain management. In: Stadtier H, Kilger C
(eds). Supply chain management and advanced planning- concepts, models,
software and case studies. Berlin et al. 7-28
Tsay AA, Nahmias S, Agrarwal N (1999) Modeling supply chain contracts: A
review. In: Tayur S, Ganeshan R, Magazine M (eds). Quantitative Models for
supply chain management. Boston et al. 300-336
Vollman TE, Berry WL, Whybark DC (1997) Manufacturing Planning and
Control Systems. New York et al. 4th edn
Hierarchical Planning Structures in Supply Chain
Management

Marion Steven

1 lntroduction

In the recent years, supply chain management has become a basic keyword
for production planning including procurement and delivery activities.
This development has several reasons:
• Procurement and sales relations of the enterprises become more com-
plex as production of parts and components is delegated to specialized
suppliers.
• Customer's requirements are increasing conceming product quality as
well as delivery time and flexibility.
• Improvements in information technology and planning systems lead to
changes in manufacturing organization and planning processes.
• E-business opens new possibilities to get into contact with partners,
exchange prevailing information and integrate planning activities along
the supply chain.
Research in the field of supply chain management is focused on dif-
ferent aspects, such as the organization of the supply chain, the flow of
materials and related information, the effectiveness and cost efficiency of
its various processes, and the planning activities required. As the relations
along the supply chain become more complex, sophisticated planning
methods have to be provided that allow to coordinate processes not only
within one enterprise, but also with the partners on the various levels of the
supply chain. The focus of this paper is to investigate how far the struc-
tures of hierarchical production planning can improve planning activities
within the supply chain.
302

2 Supply chain management

2.1 Development

Supply chain management can be defined as the integration of planning,

realization and control of production and logistics processes for the enter-
prises in a value chain. The value chain of a certain product comprises all
enterprises that contribute to the production and distribution of this product
and reaches from the suppliers of raw materials over several production
Ievels to the retailers and, at last, to the end customer. Although the
underlying problems of coordination in a multi-level delivery structure
have been discussed in Iiterature before, the term supply chain manage-
ment came up in the 1990s.
One important root of supply chain management is the work ofForrester
(1972), who described the bullwhip effect of stock Ievels in a multi-level
production and delivery structure increasing from stage to stage. This
effect is caused by information time lags between wholesalers, retailers,
producers and suppliers. Further sources of supply chain management lie
in production, logistics, purchasing, marketing, organization, and opera-
tions research. So supply chain management is an interdisciplinary
approach combining different methods and procedures.
The typical supply chain comprises the subsequent phases of procure-
ment, production, distribution and sales. Although the term supply chain
literally refers to a linear structure of the value chain, it usually describes a
network structure of enterprises on different process Ievels linking their
respective procurement and distribution activities, see fig. 1.

ENTERPRISE

Fig. 1. Supply Chain

 303

A central aspect of supply chain management is the point of view that

for every process the preceding activity is considered as a supplier and the
following activity as a customer, no matter whether they are situated in a
market partner or in the same enterprise. According to the SCOR-model
formulated by the Supply-Chain Council (1999) as a standard model for
supply chain management, suppliers and customers are linked by processes
that can be analyzed on different hierarchicallevels. The basic processes of
supply chain management are supply activities (source), production activi-
ties (make) and retailing activities (deliver) that are controlled by planning
activities (see fig. 2).

plan :::=:::::::==-
source ) make) deliver)

Fig. 2. SCOR-model

A supply chain can thus be characterized by the following dimensions

and criteria:
• Sourcing type: kind of materials
sourcing strategy
flexibility of suppliers
• Production type: organization of production processes
lot sizes
• Delivery type: structure of production network
frequency of deliveries
means of transportation
customer relations
demand forecasts
Ideally, the improvements in the flow of information caused by supply
chain management evoke considerable economic advantages that can be
distributed among the participants of a supply chain, so that a win-win-
situation can be achieved for all partners.
304

2.2 Planning in supply chain management

Planning and coordination of activities in a supply chain is a demanding

task that has to consider a multitude of partly conflicting goals reflecting
market requirements and increasing competition (see fig 3):

Fig. 3. Goals of supply chain management

• Cost reductions are required in order to be able to offer better prices

than the competitors.
• Reduction of time to market and delivery times is an important attribute
of customer service.
• Continuous improvement of product quality is indispensable in order to
offer high-dass products.
• Increase in flexibility conceming product specification and delivery
service is essential in dynamic markets.
Planning activities in supply chain management can be assigned to three
subsequent planning Ievels:
1. Supply chain configuration comprises long-term strategic decisions on
the structure of the underlying network.
2. The task of supply chain planning is to take tactical decisions on flows
of material and information for some or all partners in the supply chain
over a medium-term horizon. Supply chain planning comprises espe-
cially demand planning, production planning, inventory planning, trans-
portation planning and distribution planning.
3. Supply chain execution deals with short-term decisions, such as the
execution and control of operations, on the operative planning Ievel.
Typical tasks are production control, transportation monitoring, inven-
tory control and control oftime targets.
For all these planning tasks, lots of heuristic methods and optimizing
algorithms are available that have originally been designed in order to plan
and coordinate Operations in a single enterprise. The challenge of planning
 305

in supply chain management is to transfer and adapt these methods to the

planning of the whole supply chain.
Consideration has to be paid not only to the higher complexity of the
overall planning problern but also to the fact that the partners in the supply
chain are autonomous enterprises. Typically each of them uses different
methods, needs special data and some may shrink from disclosing intemal
information to extemal partners. A promising approach for the planning of
supply chains are advanced planning systems.

2.3 Advanced planning systems

Software solutions for supply chain management that are based on classi-
cal ERP systems are called advanced planning systems (see e.g. Stadtier
and Kilger 2000, Krüger and Steven 2002). Supply chain planning
functions implemented in advanced planning systems can be arranged
systematically in the supply chain planning matrix (see e.g. Rohde et al.
2000). The dimensions of this matrix are the stage in the value chain
(procurement, production, distribution, sales) and the planning horizon
(long-term, medium-term, short-term). As these tasks are typically
assigned to different responsibility levels in the organizational hierarchy of
the enterprise, a corresponding hierarchy of planning activities is installed.
Fig. 4 shows the typical structure of the supply chain planning matrix and
the planning activities on the subsequent stages of the supply chain.

> procuremen? production > distribution > > sales

long-term -materials program -physical -production program

-plant location
-supplier selection distribution -Strategie sales

.. ..
planning -production system
-Co-operation structure planning

H H H H
medium-term -material -master production
requirements scheduling -distribution -mid-term
planning planning ~ planning sales planning

.. ..
- contracts -capacity planning

n H n n
-personnet -Iot sizing -warehouse
short-term replenishment
planning -scheduling -short-term
planning
-materials -shop floor ~ -transport sales planning
orders control planning

:z::::=> flow of goods <==:> vertical flow of information

...._. horizontal flow of information

Fig. 4. Supply chain planning matrix (see Fleischmannet al. 2000, p. 63)
306

As the supply chain planning matrix is a generat structure for the

arrangement of planning tasks according to their planning horizon and the
planning stage, it has to be adapted and put in concrete terms according to
the individual needs of a certain network of enterprises.
In order to support the planning processes and the corresponding flow of
information along the supply chain, standard software solutions have been
developed by several suppliers of information systems. Among the big
suppliers, SAP APO (Advanced Planner and Optimizer) from the SAP AG
and Baan SCS from the Baan Company have to be mentioned. Both of
them offer a great variety of functions covering all three planning levels of
supply chain management. Other software systems for supply chain
management that cover certain parts of the overall planning task are
offered by Technologies, ILOG, Numetrix, Manugistics, Caps Logistics,
Pro Alpha, or Syte APS from Frontstep (see Steven et al. 2000, pp. 17-20).
Although the final aim of planning activities in supply chain manage-
ment is to integrate all partners in a comprehensive planning system that
allows to coordinate their production and logistics processes on a real-time
basis, implementation has to take place step by step. It makes little sense to
improve planning processes between partners as long as production plan-
ning is not optimized in every single enterprise, because a (supply) chain
can only be as strong as its weakest link. An important ratio by which the
performance of an enterprise can be assessed is its ability to meet delivery
dates. Here the use of an available-to-promise module (ATP) can produce
substantial progress.

3 Hierarchical production planning

3.1 Basic model of hierarchical production planning

The basic model of hierarchical production planning has been formulated

by Hax and Meal (1975) as a case study for a manufacturer ofrubber tires,
later it has been refined and theoretically founded by other authors. The
traditional field of application of hierarchical production planning is mass
production and continuous batch production which are both characterized
by large production lots and rather few set-up procedures, but applications
from other manufacturing areas such as job production (e.g. Tsubone and
Sugawara 1987), batch production (e.g. Oliff and Burch 1985), flexible
manufacturing systems (Villa and Rosetto 1986), group technology
(Kistner et al. 1992) etc. arealso reported in literature.
 307

The substantial principle of hierarchical production planning is the

combination of several problern simplification strategies that allow to
reduce the complexity of the planning problern to a manageable size (see
Steven 1994, pp. 25--60):
• Decomposition of a complex overall planning problern leads to less
complex sub-problems that can be coordinated by well-defined inter-
faces.
• A hierarchy of sub-problems is established that coincides with the
organizational structure of the enterprise so that existing control mecha-
nisms can be applied.
• Aggregation of products, production facilities and time periods is used
in order to simplify higher level planning problems.
• The use of a rolling planning horizon on all planning levels allows to
defer decisions for later periods to a point of time when better informa-
tion is available.
• Making use of a tailor-made combination of optimizing algorithms from
operations research and adequate heuristics helps to find reasonably
good solutions even for NP-hard problems.
Another characteristic of hierarchical production planning is the coordi-
nation of planning levels by using feed forward and feedback flows of
information (see e.g. Sehneeweiss 1999). Fig. 5 shows the typical structure
of a hierarchical production planning problern covering aggregate produc-
tion planning for a mid-term horizon and lot sizing and scheduling deci-
sions for the first period.
On every planning level, additional information about capacities and
demand is given. The respective planning results are taken as input for the
next planning level, and the results of the last planning level are realized
on the work floor level. Besides the "normal" feed forward information
that is given top-down in the form of instructions from an upper planning
level to the next planning level, also a bottom-up feedback flow of infor-
mation is possible. This feedback information is either given after the
realization of the plan or in advance. In the first case, it initiates a leaming
process and thus enables the upper planning level to make better decisions
in the future. Given in advance, it even helps the upper planning level to
anticipate the consequences of its present decisions, so that it may change
them before implementation.
308

demand forecasts
estimated capacitie
estimated stocks } aggregate
production ~-- ...
planning
aggregate
' ....
''
'\I
production I
I
program I
I
I

}
/

--...
/

detailed, actual Iot sizing ~,../ ""

demand decisions ...
' ....
'' \
I
production Iots J
I
I

}
I
/

sequencing "
actual capacities ~-----
decisions -- ... ' ....
''
detailed 'I I
production I
I
---1~~ feed forward schedule I
I

I
information I

I work floor Ievel /

•----- feedback
information

Fig. 5. Structure ofhierarchical production planning

Coordination in hierarchical production planning has to take care of

different kinds of interdependencies:
• On the one hand, many decisions that are taken simultaneously are inter-
dependent because they mutually need information which is only
available ifthe solution ofthe corresponding sub-problem(s) has already
been found. A frequent interdependency of this type is competition for
the same scarce resources, for instance in lot sizing and sequencing.
• On the other hand, decisions taken in the past have an influence on the
feasible area from which present decisions may be taken, and present
decisions determine the constraints for future decisions. A typical
example is stock that has been built up in former periods and that can be
used to fulfill actual demand.
A substantial advantage of hierarchical production planning in compari-
son to other planning methods is its ability to respect existing organiza-
tional structures. Although top-down instructions are prevailing in hierar-
chical production planning, different expressions of the feedback flow of
information allow the upper planning Ievel to evaluate the quality of its
decisions. Especially coupling methods with limited exchange of informa-
 309

tion which can be realized by means of anticipation, of slack or of

stochastic models are able to help subsequent planning levels to react
reasonably to each other's decisions (see Steven 1994, pp. 180-199,
Sehneeweiss 1999, pp. 25-70).

3.2 Hierarchical production planning in supply chains

Typically, every enterprise taking part in a supply chain uses its own plan-
ning tools, software systems and hardware equipment. So the first
requirement is to instaU general interfaces that allow a quick exchange of
information about required quantities, delivery times and possible prob-
lems between the partners, ideally by standardized transfer protocols as
they are used in e-business applications.
If the idea of hierarchical production planning is to be transferred to
supply chain management, additional coordination problems arise. Besides
coordination of intemal planning problems over a certain period of time,
all the partners in the value chain and their respective logistical decisions
have to be coordinated. The following problems of decentralized decision
making have tobe considered (see Zimmer 2001, p. 25-26):
• Uncertainty of prices, demand, production quantities and deliveries can
lead to inefficient decisions of the partners.
• Agency problems (hidden action, hidden information and hidden
characteristics) can arise because information is distributed asymmetri-
cally among the partners.
• Individual pricing of every partner and the resulting effect of double
marginalization can reduce overall profit if the resulting optimal prices
are not identical.
Coordination mechanisms used in order to harmonize planning activities
of the partners depend substantially on the type of the supply network:
• A hierarchical network is dominated by vertical structures. It is typically
managed by one enterprise that is responsible for strategic decisions,
and the partners on the other production levels depend directly or indi-
rectly on this focal enterprise. Long-term contracts are prevailing in this
situation.
• A polycentric network consists of a set of enterprises with mutual
interdependencies and equally distributed market power. Relations
between partners are mainly horizontal and based on non-repetitive
market transactions.
310

Empirical investigation on the structure of supply chains shows that they

normally take a position between those two extremes with a tendency
towards a hierarchical structure. For instance in the automotive industry
the network is normally Iead by the vehicle company which acts as focal
enterprise. Fig. 6 shows which coordination mechanisms can be applied in
supply chain management depending on the degree of hierarchy prevailing
in the various supply relations.

market hierarchy
prices contracts ru1es
• planned prices • common goals • general rules
• transfer prices • negotiations • directives
• market prices • organizationa1 s1ack • instructions
• discounts • incentives/motivation
• penalties

Fig. 6. Coordination mechanisms for supply chain management

So different forms of contracts are the most important coordination

instrument in supply chain management. The basic structure of hierarchi-
cal planning is a two-level setting with a producer and a supplier acting
within the framework of a long-term basic contract as usual in just-in-time
delivery relations. Fig. 7 shows the relevant flows of information in this
hierarchical planning situation.

top Ievel (producer)

planned

basic .§
contract
.i
basic Ievel (supplier)

Fig. 7. Coordination oftwo planning Ievels (see Sehneeweiss 1999, p. 29)

 311

The negotiation of a basic contract structuring all relevant actions of the

partners in advance is a long-term decision and thus situated on the strate-
gic planning level. Here minimum and maximum delivery quantities, price
margins, penalties, premiums and other details of the subsequent business
relation between the partners have to be fixed. On the tactical planning
level, detailed order respectively delivery quantities and times have to be
fixed and coordinated with each partner's intemal production planning.
The task of operative planning is to control the realization of plans. In the
following, coordination on the tactical planning level will be further
examined.
Typically, the producer is the actor on the top level, calling a certain
amount of material from the supplier. Before transferring his order to the
supplier as an obligatory instruction, he passes through an intemal feed-
back loop evaluating the probable reactions of the supplier to different
order levels. In the extemal feedback loop, the real reaction of the supplier
which has to lie within the limits defined by the basic contract, can be
observed and possibly actuates further actions ofthe producer.
In some cases, the supplier can take the top level position, e.g. offering
an special discount on his material and thus forcing the producer to react
with an order for a certain amount. In the intemal feedback loop he antici-
pates the reaction of the producer in order to reserve enough production
capacity. When the order comes in, he can evaluate this final reaction and
thus improve future discount offers.
In order to use the approach of hierarchical production planning for
decision making in supply chain management, appropriate models for the
top and the basic planning level and the extemal and the intemal feedback
loop have to be formulated.

4 Conclusions

Hierarchical production planning is a promising approach for structuring

the coordination and planning problems that occur in supply chain
management. In addition to the traditional coordination mechanisms used
in hierarchical production planning inside an enterprise, contracts have to
be employed as market-oriented coordination tools. Future research will
have to develop a wide range of coordination methods for different
situations of distributed decision making and give advice which tool is
appropriate in a certain situation.
312

References

Fleischmann B, Meyr H, Wagner M (2000) Advanced Planning. In: Stadtier H,

Kilger C (ed), Supply Chain Management and Advanced Planning, Berlin et
al., pp 57-71
Forrester JW (1972) Iudustrial Dynamics, Cambridge /Mass., 7. edn
Hax AC, Meal HC (1975) Hierarchical Integration of Production Planning and
Scheduling. In: Geisler MA (ed), Logistics, TIMS Studies in Management
Science voll, Amsterdam, pp 53-29
Kistner KP, Schumacher S, Steven M (1992) Hierarchical Production Planning in
Group Technologies. In: Fandei G, Gulledge T, Jones A (ed) New Directions
for Operations Research in Manufacturing, Berlin et al., pp 60-74
Krüger R, Steven M (2002) Advanced Planning Systems- Eine neue Generation
von SCM-Informationssystemen. Supply Chain Management 2, vol II: 7-14
Oliff MD, Burch EE (1985) Multiproduct Scheduling at Owens-Coming
Fiberglas. Interfaces 15, no 5: 25-34
Rohde J, Meyr H, Wagner M (2000) Die Supply Chain Planning Matrix. PPS-
Management, vol5, no 1: 10-15
Sehneeweiss C (1999) Hierarchies in distributed decision making, Berlin et al.
Stadtier H, Kilger C (2000) (ed), Supply Chain Management and Advanced Plan-
ning, Berlin et al.
Steven M (1994) Hierarchische Produktionsplanung, Heidelberg, 2. edn
Steven M, Krüger R, Tengler S (2000) Informationssysteme für das Supply Chain
Management. PPS Management 5, no 2: 15-20
Supply-Chain Council (1999), Supply Chain Council & Supply Chain Operations
Reference (SCOR) Model Overview, URL: http://www.supply-chain.org,
State: June 9, 2000
Tsubone H, Suguwara M (1987) A Hierarchical Production Planning System in
the Motor Industry. Omega 15: 113-120
Villa A, Rossetto S (1986) Towards a Hierarchical Structure for Production
Planning and Control in Flexible Manufacturing Systems. In: Kusiak A (ed)
Modelling and Design of Flexible Manufacturing Systems, Amsterdam, pp
209-228
Zimmer K (2001) Koordination im Supply Chain Management. Ein hierarchischer
Ansatz zur Steuerung der unternehmensübergreifenden Planung, Wiesbaden
 Section 4

Human Resource Management

and

Economic Organization
Firm Foundations and Human Capital
Investments: The 0-Ring Approach to
Organizational Equilibrium in an Ernerging
lndustry

Oliver Fabel

1 lntroduction

Definitions of the "New Economy" frequently refer to particular industries

- such as the bio-technology, computer, and ICT industries - in which
technological innovations spark off the foundation of new firms. However,
there also exists an organizational economics perspective conceming the
structural similarities of such New Economy firms. This view emphasizes
two common features. First, within these industries production appears to
be characterized by positive extemalities between specialized tasks.
Second, employees are exposed to incentive schemes which induce owner-
ship-like income claims. Thus, in corporate firms employees are typically
motivated by affering stock or stock option plans.
Conceming the first characteristic feature noted above, (Rajan and
Zingales 2000, 2001a) remark that the innovations themselves originate
from human capital rather than inanimate firm assets. The individual
member of the production team possesses necessary skills and knowledge.
This argument not only implies a fundamental shift of power towards
human capital. The production process is further characterized by a super-
modular technology, respectively positive complementarities among team
members assigned to specialized tasks. It can be shown that flat hierarchies
dominate in the organization ofsuch processes 1• Moreover, following (Prat
2002) the optimal recruitment policy aims at composing homogeneaus
production teams.
Given flat hierarchies, "up-or-out" -options set incentives to specialize
for young professionals2 • New Economy start-ups therefore reflect that the
outside opportunity has become dominant over the specialist's career. The
business idea is then typically based on innovations encountered during the

1 Rajan and Zingales 200lb.

2 Rajan and Zingales 2001 b.
316

founder's previous employment3• It follows that positive complementari-

ties also apply to the production in these new firms. Hence, "unusual
judgement and perceptiveness" in employee selection constitutes a prime
success factor in New Economy firms 4 • However, in contrast to well-estab-
lished firms, the start-ups are not restrained by "corporate culture" rules of
equal treatment when selecting their personnel5 •
Tuming to the characteristic second feature, ownership-like manage-
ment incentive schemes constitute the single most important characteristic
ofNew Economy firms 6 • (Holdemess et al. 2001) report that management
shareholdings have increased significantly over the 1980-90s. Such
schemes serve as self-selection devices when attracting employees 7 • How-
ever, the professionals in such firms are then also exposed to project risk.
The poor performance of the stock or stock option plans vis-a-vis soaring
financial markets thus induces the necessity of further compensation8 and
motivation problems 9 • These arguments imply that risk-aversion of the
New Economy participants limits the scope ofthe newly founded firms.
(Fabel 2002) has shown that these two characteristic propet!ies - posi-
tive complementarities in production and risk-aversion on the side of firm
founders - can induce a separating industry equilibrium. The particular 0-
Ring production technology 10, which is introduced in detail below, allows
to link individual abilities directly to project risk. Given this separating
equilibrium, high-ability individuals found entrepreneurial firms as part-
nerships of ability-matched teams. In this respect, (Bhide 2000, p. 94)
reports that New Economy firm founders are in fact characterized by supe-
rior educational backgrounds.
However, the existence of such an equilibrium requires that the ability
spread within the group of industry professionals must be sufficiently
large. The benefits of self-selecting by founding a new firm must more
than compensate the risk-premium. Clearly, the realized ability spread
depends on ex-ante individual human capital investments. At the same
time, the incentives to invest in human capital just as obviously depend on

3 Bhicte 2000, p. 54.

4 Bhicte 2000, p. 54.
s Bhicte 2ooo, p. 324.
6 Audretsch and Thurik 2001.

7 Bhicte 2ooo, p. 87, 200.

8 Zingheim and Schuster 2000.

9 Weinberg 2001.
10 The term "0-Ring production" refers to the Challenger accident which is

blamed on the malfunctioning of a simple earring. As an analogy, the

productivity of the team is taken to depend entirely on the weakest team
member's productivity.
 317

the industry's ex-post organizational structure. More precisely, the strength

of these incentives reflects the degree of entrepreneurial activity in the
industry. Only existing options to found new firms can yield a benefit of
self-selection.
Hence, the current analysis introduces human capital investments which
- as all other investment projects - must be carried out given ex-ante
uncertain returns. The return risk reflects that particular ex-post ability
realizations may or may not induce the possibility to engage in profitable
entrepreneurial activity. Managed firms which recruit randomly while
offering a certain wage-income cannot be kept from entering the industry.
However, if it is individually beneficial to seek employment in such firms,
given that human capital investments are sunk, the resulting industry
structure provides less incentives to invest than a purely entrepreneurial
industry.
While it remains true that entrepreneurial firms - if they exist - are
founded by teams of realized superior abilities, the equilibrium analysis
reveals a rather different perspective of the incentive mechanism. Human
capital investments are not so much necessary in order to induce entrepre-
neurial activity. Rather, entrepreneurial activity provides the effective in-
centives to invest in human capital. This distinction is important to note.
Due to the risk-shifting associated with employment in managed firms,
policies directed at maximizing entrepreneurial activities will generally fail
to implement an efficient solution. As with every other incentive mecha-
nism, incentive compatibility constitutes a constraint on the possibility to
achieve a first-best solution.
The remainder of the paper is organized as follows. The next section
introduces the basic assumptions and notations. Section 3 then investigates
the incentives to invest in human capital, given that the industry consists of
managed, respectively entrepreneurial firms only. The following section 4
investigates the ex-ante investment incentives which endogenously arise
given the induced ex-post industry equilibrium. Since all propositions
derived in section 4 are interpreted in detail, the final section 5 then only
provides a brief concluding discussion.

2 Basic assumptions

Suppose that K, K ~ 2, separate tasks must be performed simultaneously

in order to produce output. The output good becomes more sophisticated
and, thus, more valuable for consumers as the number of tasks defined in
the production process increases. Per-capita revenue is therefore given by
318

r(K), with r(K) > 0, for K > 0, and r'(K) > 0. Yet, in contrast to (Fabel
2002), the number oftasks is exogenous throughout the current analysis. It
reflects the state of the art in production in a newly ernerging industry.
Thus, for notational simplicity r( k) = r in the following.
Further, following (Kremer 1993), positive output can only be realized
if all K production tasks are performed perfectly. If a single member of the
production team fails in performing the task assigned to her, the output of
the whole group is destroyed. This so-called 0-Ring theory of production
implies that team production is characterized by positive complementari-
ties11.
Assuming that each task is carried out by exactly one team member, let
i then denote the probability of perfect task performance of team mem-

ber i, l =1, ... , K. Given the assumption above, total revenue thus equals
K (1)
Kr, with probability Jl qi
i=l
R= K
0 , with probability 1- JI qi
t=l

Team members stem from a population of ex-ante identical individuals.

Before being recruited and assigned to production tasks, each individual
can invest in human capital. The respective investment level is denoted
a j, j = 1, ... , N, where N refers to population size. It will be assumed that
0 ~ aj ~ 1. Individual quality if assigned to a productive task then consti-
tutes a random variable which is distributed uniformly over the interval
[0, aj] for individual}. Since E{qlaj }= a; , invest,ments in human capi-
tal therefore enhance the productive quality of the individual.
The human capital investments must be made before production com-
mences. Individual preferences are characterized by an identical instanta-
neous utility function u =ln{y - ca j), where y ~ 0 denotes income and c >
0 the investment costs. Assuming logarithmic utility - thus, constant rela-
tive risk aversion - serves to simplify the analysis. All individuals then
also possess exogenous income y > 0 .
Suppose that all individuals are assigned to productive tasks in NIK
fmns. Further, there exists an investment equilibrium implying

11 Positive complementarities constitute a special case of the more general

concept of super-modular technologies. Compare (Prat 2002).
 319

a~ =a, Vj =1, ... , N . Then, if firms select workers randomly, the expected
firm revenue can be obtained as
(2)

On the other hand, if individuals of identical qualities can be matched

within firms, the average expected revenue over all such firms equals

ERmatched = J
a
0
qK ;{; dq
a
r K =
~K
a
(K + 1)
rK
(3)

Due to K ~ 2, E~atched > EKandom. Hence, taking the ex-ante perspec-

tive, all individuals who have not yet realized this productive quality
would strictly prefer to enter an economy in which production teams are
quality-matched. Moreover,
a,v0 matched
Dft -
K2
~(K-1) O (4)
aa = r (K + 1) a >

and
a2E'nmatched
ft _-
K2(K
-
1) ~(K-2) 0 (5)
----=[a-a-=-y=--- - r (K + 1) a >

for K ~ 2 . Thus, the productive advantages of matehing teams of

identical quality imply economy-wide increasing returns to human capital
investments.
Note that production will be weakly preferred over no production, if

in (.v + (;~KI) - c'ii) ~ in (y). lnserting 'ii ~ I then reveals that

r (6)
---,>c
(K +1)-
is sufficient to induce the first-best investment level a = 1' due to the
increasing returns implied by (4) and (5).
However, implementing the first-best solution requires a transfer
scheme which provides a certain income equal to the average expected
per-capita revenue for all individuals. The project risk would have to be ·
320

distributed over the whole population. Yet, in this case all individuals
would obviously prefer aj =0, Vj = 1, ... , N.
Hence, the income opportunities of a competitive economy must
provide sufficient incentives to induce human capital investments. In the
following, it is therefore assumed that individuals may found partnerships.
Since every partner is a member of the production team herself, she can
also verify each other team member's productive quality. Furthermore,
partnerships can enforce individual attachments or Separations respec-
tively. Thus, partnerships can form matched production teams.
At the same time, they can only distribute realized revenue among their
partners. Due to the quality risk associated with each team member, part-
nership income is thus always subject to project risk. Partnerships will
therefore be denoted entrepreneurial firms in the following. Clearly, if
there exist such fmns in industry equilibrium, the first-best will not beim-
plemented. Partnerships preclude risk-sharing.
Since the production technology is common knowledge, the already
existing Old Economy may also integrate part ofthe industry. Within these
firms managers are responsible for employee selection. Since the managers
do not participate in the production process themselves, they cannot verify
individual qualities. Altematively, it can be assumed that the corporate
culture of these firms precludes the necessary wage-differentiation. Such
managed firms therefore hire randomly from the pool of potential em-
ployees in the industry. However, they will also pay a certain salary wm.
Thus, employees of managed firms do not face income risk, but their firms
are subject to quality risk.
Then, assuming that there exists an equilibrium such that
a; = a, Vj = 1, ... , N, with the superscript "*" referring to individually
optimal investment Ievels, three possible types of organizational equilibria
can be distinguished:
• the pure equilibrium in which there exist only entrepreneurial fmns;
• the pure equilibrium in which there exist only managed fmns;
• mixed equilibria in which entrepreneurial and managed firms coexist.

3 Firm types and human capital investments

Clearly, all firms are subject to the project risk associated with their
employees', respectively their partners' failure in task performance. Re-
alizing zero profit therefore does not signal a Iack of productive quality.
Moreover, if managed firms conditioned their salary on project success,
 321

individuals would always prefer to join a partnership. This follows from

the fact that managed firms cannot observe individual qualities and, thus,
do not implement quality-matched teams.
Hence, the only way to successfully compete with partnerships is to
offer risk-free salaries when contracting with employees. The expected
profit of a managed firm can be derived as
(7)

with am = { a;} denoting the ability profile. The expectation opera-

tor E {...} serves as a reminder that managed firms cannot observe neither
am
quality realizations nor human capital investment levels.
The firm's expectation conceming the ability profile of its employees
must be entirely based on anticipating the equilibrium human capital in-
vestments. lt is assumed that the non-cooperative Nash-equilibrium
concept applies. Recalling that all individuals are ex-ante identical, the
equilibrium i~vestments in human capital satisfy a; a,
= Vj = 1, ... , N.
Hence, a; denotes the individually optimal choice, given that all other
members ofthe population, k * j' choose ak = a.
lt then follows:
Proposition 1: The new industry cannot emerge as an equilibrium
consisting of managed firms only.
a,
Proo( Suppose an equilibrium with a; = Vj = 1, ... ,N, exists and the
new industry consists of managed firms only. Given the firms' anticipation
of the equilibrium human capital investment level, (7) implies
Eh= E { qi Ia} r K- wm K (8)

= {ttrK-wmK
Again, managed firms cannot observe the quality realizations of their
employees.
Clearly, offering a fixed income wm when recruiting employees,
managed firms cannot avoid to incur losses of size -wmK, with probability
f.
(a I 2 Competition among these firms implies zero expected profits.
Thus, the salary level equals
(9)
 322

Note once more that this salary Ievel only depends on the equilibrium
investment Ievel a. However, each employee actually possesses a fixed
income claim. Thus, the individuals choose their investment Ievel
ai, j =I, ... , N, such as to maximize tn( y + wm- caj). Clearly, a; = 0.
Hence, with managed firms only, a
= 0 also constitutes the only possible
equilibrium human capital investment Ievel. In this case wm (ala
=0) =0
and production in the new industry does not tak:e place. Q.E.D.
Obviously, individual incomes must intemalize at least part of the pro-
ject risk in order to provide incentives for human capital investments.
Given the distinction of only two firm types above, this requires the exis-
tence of some entrepreneurial firms in the industry. Thus, Iet
qP = {qf, qf, ... , qk} denote the ability profile of partnership p, p = 1, ... ,
P. The expected utility from participating in this partnership can then be
derived as

(10)

for each partner }, }=1, ... ,K. Without loss of generality, (10) assumes
that total revenue rK is distributed equally across partners.
Then, note that

(11)
aEuij = ~.qf~n(y+r -caJ-tn(y-c aJ]
8qj kofc.j

K
IIqf ~(K, aJ > 0
= kofc.j

for each partner j -:f:. k . Moreover,

(12)

Thus, joining partnerships which already consist of relatively high

quality team members will always be more attractive than entering part-
nerships characterized by a relatively poor quality profile q P = {q At f}.
the same time, existing partnerships will compete for the highest quality
team member available in the industry's pool of potential partners. These
arguments imply that, in equilibrium, all partnerships consist of team
 323

members of identical quality. Hence, qf =qP, Vi=1, ... , K.

Entrepreneurlai firms p, p = 1, ... , P, differ with respect to team quality,
while exhibiting a homogeneous ability profile.
Given that entrepreneurial firms always exploit the benefits of ability-
a
matching, let ~ 0 again denote an equilibrium human capital investment
level. Consider an organizational equilibrium in the industry consisting of
entrepreneurial firms only. The size of the population is taken to be suffi-
ciently large. Thus, the individual probability to realize quality
q; E (q, q + &] ], i =1, ... , N, translates into the relative frequency of
respective quality-matched partnerships among the total of N/K entrepre-
neurial frrms. Furthermore, divisibility problems associated with allocating
individuals characterized by quality realizations q e [o,a] to matched
groups of K partners are ignored.
Then, if individual j chooses the human capital investment level ai ~ a
and subsequently realizes quality q E [ 0, aj J, she will be able to join a
partnership of team quality q with probability one. If she decides on
a j > a, the probability to find and join a partnership of team quality ,

qe [o, a] is again one. However, she may also realize quality q e ( a, a j J,

in which case she cannot team up with partners of identical quality. Then
recall (11 )-(12). The expected utility of all (K- 1) partners increases when
a superior-ability partner is attracted. Hence, all existing partnerships will
accept individual j. She can therefore freely choose which firm to join.
Obviously, she will join a partnership which offers the highest expected
utility of all.
Hence, given her believe ak = a, k = 1, ... , N and j ::t:- k, individualj's
ex-ante expected utility can be obtained as:
w(a1 ,a)= (13)
ct>"Hn{y+r -caJqK +in{y -caJ{l-qx)] (., dq
0

+(1-ct>){pln{y +i' -caJqK + ln{y- caJ{I-qK )]y., dq

+ f[ln{y+r -caJ(a)<x-J)q+ln{y-caJ{I-(a)<x-J)q)]y., dq}

=ln(y- caJ+ il(K,aJ{ct> ((aJ) + (1-ct>)[( (a)<x;J) +!._(a)<x-J)(a1 _ (aY )]}

K+l K+la 1 2 a1
324

with
0, for aj >a (14)
<I>= {
1, for aj ~ a
Since all entrepreneurial firms implement perfect quality-matching, a
single individual's choice of human capital investment affects the
probabilitytobe able to join a team of a particular quality. Hence, partici-
pating in an entrepreneurial firm provides incentives to invest in human
capital. If individual j should choose a j =0, the probability of earning r
by joining an entrepreneurial firm equals zero with certainty.
The following can then be shown:
Proposition 2: Suppose the newly ernerging industry consists only of
entrepreneurial firms. Then, there exist two locally stable equilibria. One
equilibrium is characterized by aj = a = 0, 'Vj = 1, ... , N. However, for
ji > c and sufficiently large returns to productive activity r, there also
exists an equilibrium in which all individuals choose the efficient invest-
a
ment Ievel aj = =1 .
Proot
For <I>= 1,

_aw-'-"-(ai~,a)i<~>-J =---c
aai - y-ca1
+[-c__
y-cai
c
y+r-cai (K+l)
]-(ai_f (15)

+ ä{K,aJ-(_K_) (aJK-I)
K+l

(16)

Also, for <I> = 0 ,

 325

(17)

(18)

Note that (15) is monotonically increasing in aj. Thus assume that all
individuals k, k = 1, ... , N with k =t. j choose aj =Ci= 1. Moreover,
individual j correctly anticipates these investment choices. Then, the case
characterized by <I>= 1 must apply. Inserting aj = 1 into (15) reveals:

(19)

r
~c~ - -
Il y +r- c
IK+ -
y-c

by virtue of Jensen's inequality. Hence, a; = 1 is optimal for individual

j, if the final condition in (19) is satisfied (sufficient condition). Since this
applies to every individual, given that she anticipates that all others choose
this investment level, Ci= 1 constitutes a Nash-equilibrium. Noting that,
for y > c, the RHS of (19) is monotonically increasing in r yields the
qualification reported in the proposition.
326

The second possible Nash-equilibrium is more easily identified. If

ak = a = 0 , for all k = 1, . . . N with k -::1= j, only the case characterized by
<I> = 0 can apply. Obviously, individual j' s optimal choice then satisfies
a; = 0 as weiL Finally, both second derivations (16) and (18) are clearly
strictly positive. Hence, an interior optimum a; 0,1) for some E (

ak = a E (0,1), with k -::1= j, cannot exist. Q.E.D.

If individual j believes that all others will not invest in human capital,
she anticipates that all projects will be unsuccessful with certainty. Hence,
she will not invest herself either. In contrast, if she believes that everyone
eise chooses the efficient investment Ievel, this choice will also constitute
her individual optimum.
However, two conditions must then further be satisfied. First, she must
be able to finance her own education. Second, the activity must be suffi-
ciently productive such that per-capita revenue compensates for the cost of
education and the risk-premium. Hence, while such an organizational
equilibrium provides sufficient incentives to induce the efficient human
capital investment Ievel, it clearly does not implement the first-best.

4 Endogenaus organizational equilibrium and human

capital investments

Effective incentives for human capital investments can only stem from
entrepreneurial activity. However, propositions 1 and 2 assume the exis-
tence of either one of the two possible pure organizational equilibria. As
for mixed industry equilibria in which both firm types coexist, the fol-
lowing holds:
Proposition 3: Suppose the equilibrium human capital investments
satisfy aj = a > 0, 'tfj = 1, ... , N. Then, an organizational equilibrium
consisting of both managed and entrepreneurial firms is characterized by a
separating quality Ievel q , with 0 < q < 1 . Individuals realizing qualities
q > q team up to found ability-matched partnerships. In contrast, all indi-
viduals realizing qualities q ~ q seek employment in managed firms
which randomly recruit from the Iabor pool characterized by q E [0, q).
Proof Suppose that within the interval of possible quality realizations
[0, a] there exist four non-trivial subintervals Ql = [ qo, ql)'
Ql = [ql, ql)' Q3 = [ql,q3)' and Q4 = [q3,q4], with 0 ~ qo < q4 ~ a'
 327

such that individuals realizing qualities q E Q2 and q E Q4 voluntarily

seek employment in managed firms, while individuals of qualities q E Q1
and q e Q3 prefer to found partnerships.
The expected utility EUP ( q) derived from participating in ability-
matched partnerships monotonically increases with team quality q:
(20)
BE~:(q) =Kq(K-lJ!!.(K,a)>O

Thus, individuals characterized by q E [0, q1 ) will also prefer employ-

ment in a managed firm over founding partnerships. Since quality cannot
be observed in managed firms, these individuals cannot be prevented from
entering such firms. Hence, q1 =0 and Q1 constitutes an empty set.
Let wm ( Q2 Q4 ) denote the certain wage-income offered by such
managed frrms. They can randomly recruit among individuals charac-
terized by q E Q2 and q E Q4 • By construction, the following would be
implied:
EUP(q)<ln(y+wm(Q2 Q4 )-c), VqEQ2 (21)

EUP(q)>ln(y+wm(Q2 Q4 )-c), VqEQ3

EUP(q)<ln(y+wm(Q2 Q4 )-c), Vq E Q4
Again, according to (20), the expected partnership utility monotonically
increases in team quality q . Hence, the inequalities (21) cannot hold
simultaneously. The proposition thus follows by contradiction. Q.E.D.
If some ability-matched groups ever found partnerships, all individuals
characterized by higher qualities will also prefer this option over joining a
managed firm. Every mixed organizational equilibrium which can prevail
in the new industry is therefore characterized by a single quality level q
which separates managed firm employment from entrepreneurial activity.
Given this charactcrization, the Nash-equilibrium consequences for
human capital investments must be addressed. As in Propositions 1 and 2,
the behavioral effects can be obtained by maximizing the respective
expected utility:
328

(22)

+(Hf)[HY+(%)' r -ca}q }Ia,

+J[tn(Y +i' -ca }q' +in()' -ca
1 1 )(1-q') Jdq (., ]}

+(J-<P+Hl'+(%)' r -ca 1 )]

+(1-Y'l[Hl'+(%)' r-ca} }Ia,

ii

+ Kln(.Y +r -caj )qK +ln(.Y -caJ(l-qK) ]dq }laj

q

+ ·~ln(Y + i'- ca )M<-n q +in( y -ca, ){I- .1<-nq) ]dq /a, ])

1

with
(23a)
<I>={J,if
0' if
(23b)

and (ij,a) defined above. Note that, if aj ~ q,- hence \}'=I - indi-
vidual} willjoin a managed firm with certainty. Only if aj > q, the option
of teaming up with partners of identical quality q > q to found an
entrepreneurial firm arises. Then, if the individual' s optimal choice entails
a ~ ~ q, symmetry implies that a = 0 must constitute the unique Nash-
equilibrium. Thus, whether or not this case applies hinges on the incentives
to found partnerships. Since these incentives depend on equilibrium
investment level a again, it is useful to note the following:
 329

Proposition 4: Suppose the industry structure constitutes a mixed equi-

librium. Hence, q , with 0 < q < 1 , separates intervals of realized qualities
such that individuals characterized by q :::; q (q > q ) prefer to be
employed by managed firms (to found entrepreneurial firms). Then, the ex-
ante equilibrium human capital investment Ievel must satisfy a = 1 .
Proo[: The first-order condition for interior optimal with respect to aj
can be derived as
(24)

respectively
(25)

+~(K,a+ (K =~~·;:1)' + (K !~~·;:j + Y,+ Y,(: J

+[~IY-ccai Y+F~f-caJ
+[(::~;aj (::~;aj + ~aj-~( ~]] [y-ccaj- y+: -cJ
with ~(K,aj) =[zn(.Y + r(~}- caj) -ln(y -caJ] and 11.( K,aj)

defined above. Obviously, (24) and (25) again distinguish the conse-
quences of choosing a;:::; a,
respectively > a; a,
for j = 1, ... ,N. If
330

aE (0,1) constitutes a Nash-equilibrium human capital investment level,

both conditions must hold for a1 = a. Insertion in (24) and (25) can then
be seen to yield

- a [ W(a1 , aia1 =a) I'Y=JJ - -

a [ W(a1 , aia1 =a) I'Y=JJ = (26)
&.J ~ &. ~
J

1
11( K,a )[1-a(K-n(-
K +1
-+__!5_)]
K +1
=
11(K,a)(1-a(K-l) J>O

Clearly, (26) not only contradicts that a, a

with 0 < q < < 1, may
constitute a Nash-equilibrium. Since (25) reflects the change in expected
utility as the individual increases her investment level a1 , the positive sign
in (26) also implies that a1 = a = 1 constitutes the only possible equi-
librium choice in this case. Q.E.D.
If both firm types coexist in equilibrium, the individuals will always
enter managed firms with positive probability. In this case they cannot
benefit from the productive advantage of ability-matched teams. Hence,
deriving the second-order conditions associated with (24) and (25) respec-
tively, the possibility of an interior individual optimum a1* E (0, 1), for
a E (0, 1) Cannot be excluded apriori.
However, (22) highlights a second, unambiguously positive effect of
increasing the individual investment level aj above the level a which she
believes to be chosen by all others. She can increase the probability of
joining a partnership characterized by the highest realized quality level in
the industry.
While this effect also occurs in (15) and ( 17), it is not necessary to
prove Proposition 2. In contrast, it turns out to be decisive in generating
the positive sign in (26). The possibility to join an entrepreneurial firm
with a top-ability team more than compensates for the reduction in invest-
ment incentives due to the coexistence of managed firms.
Proposition 4 then allows to compute the equilibrium expected utility,
given a particular industry structure. It follows:
Proposition 5: (a) If r(ljt ~ c, there always exists an organizational
equilibrium with efficient human capital investments a = 1 . Hence, the
new industry will emerge as long as individual j , j =I, ... N , ex-ante
believes ak = a = 1' k =/:. j.
 331

(b) Given a1 = a = 1, j = 1, ... ,N, it further follows:

(i) If ~tK -1 ~ ~' there can only exist a pure organizational
y-c
equilibrium such that the industry consists entirely of entrepre-
neurial firms.

(ii) If ~tK - 1 ::S: ~,

y-c
there exists at least one, but possibly more

than one mixed equilibrium.

Proof Consider the difference between the expected utility derived
from a marginal partnership combining K individuals of identical ability
q and the certain utility associated with joining a managed firm which
randomly recruits among individuals characterized by q E [o,q). Assuming
q~O,

EU P (q) -ln(y + wm (q)- c) = (27)

ln(y+r -cFfK +ln(y-c)(l-qK )

-zn(y+-r(%r -c)?_

-Kr
rq 1
y+r-c
(fjf
---
y-c 1
by virtue of Jensen's Inequality. Note that (27) has been derived setting
a
a1 = =I, j = l, ... ,N, as implied by Proposition 4. Moreover, for r > 0,
(27) holds with strict inequality. Part (b )-(i) of Proposition 5 then provides
the condition under which this expression is positive for all possible values
of q > 0. If it satisfied, only entrepreneurial firms can exist in the industry.
However, even given that all individuals have chosen a1 = = 1, thea
reverse merely constitutes a necessary condition for the existence of a
mixed equilibrium. Moreover, EUP (0) = ln(y- c) = ln(y + wm (0)- c).
Hence, in order to show whether there actually exists an incentive to join a
a
managed firm upon investing aj = =1' the difference in first deriva-
tives must be accounted for:
332

(28)

Kij(K-l)~)K- -(q) <

y+r l

l
-c

-(K-1)[
rq
1 C!-it
y+r-c-y+r(% )-c
Again, (28) uses Jensen's Inequality and strict equality follows from
r ~ 0 . The term in quantities obviously becomes negative for
lim(q) ~ 0, if the condition noted in part (b )-(ii) of Proposition 5 is satis-
fied. Hence, for infinitesimal, but positive values of q , the certain utility
associated with managed firm employment increases at higher rates than
the expected utility derived from joining the marginal entrepreneurial firm.
Hence, there mustexist q E [0,8), with 8 > 0, suchthat employment in a
managed firm constitutes the preferred choice for individuals realizing
qualities q < q upon having invested a =1 .
Part (a) of Proposition 5 is more easily proved then. If r(ff ~ c, in-
vesting a1 = Zi = 1 yields an equilibrium in which the ex-ante expected
utility is weakly higher compared to the situation with no production in the
new industry. In the latter case, everyone consumes y > 0. The ex-ante
expected utility is even strictly higher, if there exist some entrepreneurial
firms. While (6) above therefore remains necessary for the emergence of
the new industry, the inequality reported in part (a) of Proposition 5
constitutes a sufficient condition.
If it is satisfied there exists an equilibrium with at least some entrepre-
neurial firms. Above, this equilibrium has been shown to be locally stable
with respect to the ex-ante individually optimal choices of human capital
investments. Also, it globally dominates the solution a = 0 which
necessarily emerges, if there exist only managed firms in the industry.
Q.E.D.
Conditions (i) and (ii) in Proposition 5(b) reflect the strength of the
incentives to found entrepreneurial firms in industry equilibrium. Thus,
r l(y- c) measures the benefit-loss ratio associated with organizing the
 333

firm as a partnership. If this ratio is reduced, the adverse effect of project

risk on expected partnership utility ceteris paribus decreases.
On the other band, for large ratios r l(y- c), risk-averse individuals
will always prefer to possess the option of receiving a certain utility in
managed firms. Their risk-premium associated with being exposed to
project risk as members of entrepreneurial firms exceeds the expected
income loss which - due to random recruiting - has to be incurred when
joining a managed firm.
The strength of this incentive to found partnerships as well as the possi-
bility to realize an equilibrium outcome with efficient human capital
investments both depend on the size of the per-capita revenue r. Thus,
very large r -values induce two effects which on first sight appear contra-
dictory.
First, the emergence of the new industry becomes more likely and,
given that production then actually becomes profitable, an equilibrium
with managed firms only cannot exist. Second, ceteris paribus relatively
large per-capita revenue associated with successful production al~o induce
a tendency towards profitable market-entrance of managed firms.
Hence, while entrepreneurial firms are necessary to provide human
capital investment incentives, highly productive new industries will also be
particularly attractive for managed firms. Their market entrance reduces
the incentives to invest in human capital again. The mere presence of
managed firms thus decreases the likelihood that the new industry will
actually emerge.
Comparing condition (6) with part (a) of Proposition 5, the emergence
· dustry reqmres
of th e new m · · revenue r-
per-caplta E (K +
1 fJ/)-K
~c,1.7 2 c) .

Hence, focussing exclusively on the potential productive benefits of a new

industry - in particular, assuming a first-best solution could be imple-
mented - generally underestimates the actual chances to experience the
emergence of such an industry.
Furthermore, part (a) ofProposition 5 additionally requires that all indi-
viduals actually believe that everyone else will invest efficiently in human
capital. As shown in Proposition 2, there always exists a stable equilibrium
a
characterized by investments aj = = 0, Vj = l, ... ,N, as well. Ifregu-
lating and subsidizing formal education would implement a lower bound
for realized productive qualities, this outcome could generally be avoided.
Hence, an educational policy which ensures that every individual in the
economy is characterized by a strictly positive probability of perfect task
performance will enhance the likelihood of new industry emergence. Yet,
334

given the incentives for managed firms' market entrance, ensuring mini-
mum educational standards alone cannot guarantee the emergence of a new
industry.
Since this result holds even if commencing production in this new in-
dustry is efficient, additional govemmental aid such as tax-benefits, or
subsidized venture capital for entrepreneurial firms can be welfare-en-
hancing. However, as shown below, the success of such subsidies cannot
be measured in terms ofmaximizing entrepreneurial activity.
Case (ii) of Proposition 5(b) still offers the possibility of multiple mixed
equilibria. In this respect, the following can be shown:
Proposition 6: Consider Proposition 5.
(a) Given part (b)-(ii) ofthe proposition, multiple mixed equilibria
will generally exist. Let 'ifs denote the respective separating
quality level ofthe mixed equilibrium s, s = 1, ... , S. Then, there
always exists a unique efficient equilibrium characterized by
(j* = max{qs} .
(b) If the initial wealth net of human capital investment costs
approaches zero, the equilibrium industry structure tends to
consist of entrepreneurial firms only.
Proof Increasing q implies

oln (y+wm(q)-c) (29)

aq
and

82 /n (r+wm(qJ-c)_ ~ K(K -1) r: (zur- ;(y-c)

2 (30)
[oq] 2 - (y+wm(qJ-cl
~)(rY(Zfh)K[(K;l) 1] >

+
(y + Wm (q) - cJ
= 0
<

Similarly, the expected utility associated with participating in a marginal

partnership of team quality q changes according to

(31)

in equilibrium with efficient human capital investments. Also,

 335

(32)

since K 2:: 2 .
In mixed equilibrium ln(y + wm (q)- c)= EUP (q). However, given the
ambiguous sign of (30), a single-crossing property of ln(y + wm ( q)- c)
and EUP ( q) cannot be established. Hence, in general there will be multi-
ple intersection ofthe two functions.
Each of these intersections - characterized by a separating ability level
qs, s = 1, ... , S- constitutes a mixed equilibrium. Since both the expected
utility derived from participating in a partnership and the certain utility
associated with managed firm employment are monotonically increasing in
q, the equilibrium characterized by (j* = maxfqs} then dominates. These
arguments imply part (a) ofProposition 6.
Finally, note that

1;(-)2 (q)lK((K -1) ]) (33)

tim [a2tn(wm(q)+y-c}]=14r 2 ~K~- <0
(y-cJ->O [Bq p (wm (q) J
Hence, (32) and (33) imply the existence of a unique mixed equilibrium
as initial wealth net of human capital investment costs approaches zero. It
further follows:
oEUP (ij) (34)
oq LI( K,IJ(y +wm(q)-c)
= -------'---::---_ _:_
~----,--~____!_---:-

Oin(y+wm(ijJ-c) (}'itr
aq

given Jensen's Inequality again. Thus,

336

(34)

/im
(Ji-c )-+0 oln(y +wm(q)-c)
aq
with strict inequality since r > 0. Recall that EUP (0) =ln(y- c) and
ln~+wm(O)-c}=ln(y-c). Thus, if there exists an equilibrium with
positive production and investments, it must be a pure equilibrium with
entrepreneurial firms only. This argument is summarized in part (b) of
Proposition 6. Q.E.D.
Part (a) of Proposition 6 demonstrates the danger of misperception of
the equilibrium mechanism by policy-makers. On the one hand, without
entrepreneurial activity there exist no incentives to invest in human capital.
On the other hand, maximizing the entrepreneurial activity in an industry
does not yield an efficient solution.
In general, there will exist multiple mixed organizational equilibria. The
equilibrium characterized by the highest separating ability Ievel dominates.
However, compared to all other possible equilibria, entrepreneurial activity
is minimized by choosing this equilibrium.
By paying a certain income independent of project risk, managed firms
provide an insurance against poor ex-post realization of abilities. At the
same time, their mere existence reduces the incentives to invest in human
capital. This typical adverse incentive effect of income insurance cannot be
avoided. Once the individuals have invested in human capital, managed
firms cannot be kept from entering the industry.
This dilemma is also highlighted in part (b) of Proposition 6. Suppose
policy-makers would actually wish to maximize entrepreneurial activity.
The proposition suggests that this can be achieved by increasing the indi-
vidual cost ofhuman capital investments. Thus, a policy of cutting the per-
sonal subsidies for educational programs may appear suitable.
However, the desired entrepreneurial activities will clearly only result, if
it remains individually beneficial to invest in human capital at all.
Increasing the private investment costs clearly also yields an incentive to
withhold such investments though. Thus, the attempt to maximize entre-
preneurial activity is likely to achieve the non-emergence of the new in-
dustry.
 337

5 Concluding discussion

Given the rather complex theoretical model, it appears worthwhile to dis-

cuss the restrictive nature of some of the assumptions introduced. Thus, as-
suming that the number of tasks in production is exogenously fixed for the
industry and across firms is certainly subject to doubts. However, (Kremer
1993) already shows that team sizes can be determined endogenously.
Higher team ability then implies an increasing team size.
(Fabel 2002) then considers team size and capital investments as con-
tinuously variable inputs given a standard Cobb-Douglas per-capita reve-
nue function. Separating industry equilibria will still emerge under rather
general conditions. In particular, the paper is based on general instanta-
neous utility functions.
However, this previous paper assumes an exogenous distribution of
abilities in the population. Hence, the current model adds the analysis of
ex-ante risky individual human capital investments. In this respect, the
assumption that such investments constitute a neutral "bad" exhibiting
constant unit investment costs possesses great simplifying virtue. Com-
bining this assumption with the particular 0-Ring revenue function, re-
alized investments can only be either efficient or equal to zero.
Altematively, marginal investment costs could be taken to increase. In
this case, both first-bestand second-best investment Ievels may constitute
interior solutions of the individual optimization problern. Yet, only the
discussion of Proposition 6 appears to require qualification. Comparing
different possible equilibria, varying degrees of entrepreneurial activity
will then also induce different Ievels of human capital investments. Hence,
stimulating entrepreneurial activity may actually possess a welfare
enhancing quality per se.
However, the main result of the above analysis appears robust and
warrantstobe reemphasized. Obviously, it is true that the emergence of a
new industry is driven by the entrepreneurial activities of high-ability
professional teams. Still, this observation does not lend support for pro-
grams directed at improving the entrepreneurial abilities of potential firm
founders. Rather, investigating the equilibrium mechanism, it is more
important to realize that such entrepreneurial activity is necessary in order
to provide effective incentives for human capital investments.
As with every other incentive mechanism, the art of policy-making must
then be concemed with achieving a balance between simultaneously
setting adequate investment incentives and allowing for sufficient risk-
shifting through employment in managed firms. Generally, it is question-
able whether public policy can actually "walk this narrow bridge" - for
338

instance, by founding public venture capital funds or providing consulting

services at no costs. Corporate venturing by established managed firms
offering the possibility of voluntary spin-offs for high-profile employees
should be considered the dominant mechanism.

References

Audretsch DB, Thurik AR (2001) What's New About the New Economy? Sources
of Growth in the Managed and the Entrepreneurlai Economies. Industrial and
Corporate Change 10:267-315
Bhide AV (2000) The Origin and Evolution of New Businesses. Oxford
University Press, New York
Fabel 0 (2002) Spin-Offs of Entrepreneurlai Firms: The 0-Ring Approach.
Discussion Paper Series I - 314, rev. version (first version entitled: The
Emergence of a New Economy: An 0-Ring Approach), Department of
Economics, University ofKonstanz
Holderness CG, Krosner RS, Sheehan DP (1999) Were the Good Old Days That
Good? Changes in the Managerial Stock Ownership Since the Great
Depression. Journal ofFinance 54:435-470
Kremer M (1993) The 0-Ring Theory of Economic Deve1opment. Quarterly
Journal ofEconomics 108: 551-575
Prat A (2002) Should a Team Be Homogeneous? European Economic Review
46:1187-1207
Rajan RG, Zinga1es L (2000) The Governance ofthe New Enterprise. In: Vives X
(ed.) Corporate Governance: Theoretical and Empirical Perspective.
Cambridge University Press, Cambridge, pp 201-227
Rajan RG, Zingales L (2001a) The Influence of the Financial Revolution on the
Nature ofFirms. American Economic Review 91: 206-211
Rajan RG, Zingales L (2001b) The Firm as a Dedicated Hierarchy: A Theory of
the Origin and Growth of Firms. Quarterly Journal of Economics 116: 805-
851
Weinberg CR (2001) Reward Strategies for Dot.Corp Organizations: Lessons
from the Front. Compensation and Benefit Review 33: 6-14
Zingheim PK, Schuster JR (2000) Total Rewards for New and Old Economy
Companies. Compensation and Benefit Review 32: 20-33
Training Strategies and Remuneration Systems
of Enterprises of the New and Old Economy -
Similarities and Disparities

Rosemarie Kay

Caroline Demgenski

1 lntroduction

At the core of the new economy are enterprises which create or sell infor-
mation technology and applications or use them intensively to provide ser-
vices (Anstötz 2002). Typically, they are young and small, but nonetheless
fast-growing. Although they differ from traditional enterprises primarily
by a more intensive use of and a higher dependency on information tech-
nology (Klodt 2001 ), it is the skills and the motivation of the employees
that ultimately determine a company's success or failure, even more so in
new economy enterprises than in traditional companies. These employees
with their innovative ability, service orientation and flexibility are the
decisive competitive factor (Marschlich/Pendt 2001).
Securing the necessary workforce hasn't been easy for new economy
companies up until now. The reasons were, firstly, a very intense competi-
tion for scarce skills, this competition existing not only between new
economy companies but also between the new and old economy.
Secondly, young, mostly unknown companies without any substantial
experience in personnet work have disadvantages in recruitment which are
possibly intensified by the fact that new economy firms can't provide their
(potential) employees with incentives on the same scale as traditional
firms. Finally, new economy firms seem to resort to formal vocational
training far less frequently as a strategy for securing skilled labour than
traditional firms.
In this paper we study two of the various possibilities exerting an influ-
ence on a sufficient supply of appropriately skilled labour: training strate-
gies and remuneration systems. The extent to which these strategies are
applied in the new economy and how they differ from the old economy are
analysed empirically, based on data collected by or made available to the
IfM Bonn in various surveys in recent years.
340

In the first section, we will deal with the training strategies and then, in
the second section, with the remuneration systems of new economy firms.

2 Training strategies

Traditionally, German firms train within the highly formalized dual voca-
tional system which plays the main role in initial training and absorbs
roughly two thirds of all school graduates each year. Options for formal
qualification and training outside the dual system are limited and seldom
officially recognized. 1 More specifically, this means that two thirds of
Germany's human capital is trained and qualified within the dual system.
And because securing highly skilled labour is absolutely essential for the
growth of new economy firms - indeed their survival often depends on it -
the question arises, if and how extensively new economy firms use voca-
tional training and, ifnot, what alternative training strategies they follow.
Despite the difficulty in obtaining statistical data on the new economy,
there is a strong indication that new economy firms participate far less in
initial vocational training than their Counterparts in the old economy. The
training participation rate in the business service sector - which reflects an
important part of the new economy - is below the average of the overall
economy (19.1 % compared to 23.1 %). Its training intensity, i.e. the ratio
between trainees and employees, is also lower (4.5 % compared to 5.9 %
for the overall economy) (Bundesanstalt fiir Arbeit 2000).
One reason for this under-average training participation rate could be
that firms in the business service sector train less due to their young age
and Iack of training experience, a point made in BIBB (2000) and by
Zwick/Schröder (2001). This hypothesis is examined in the following
using empirical data from a survey of 298 young firms from the business
service sector, carried out by the IfM Bonn in 2001.2 Firstly though, we
attempt to establish to what extent vocational training is considered an
adequate strategy for securing skilled personnel by business service firms.

1 Non state-recognized, but regulated initial training measures barely represented

1.3% of all initial training venues in 2000 (1999: 0.9 %).
2 The firms in the sample are mainly software companies and ICT users. The
sample was drawn from the NACE (System of Economic Sectors of the
European Union) sectors K72 and K74. 78% ofthe firms are less than 10 years
old. Firms from the ICT manufacturing sector are not included in the sample.
 341

2.1 The importance of vocational training

To identify strategies considered primordial in new economy firms for

securing skilled personnel, the firms were asked to rank various strategies
by degree of importance. Fig. 1 shows that training in young business
service firms still largely focuses on training-on-the-job, i.e. on informal
training, while the formal vocational training option comes in second.
Non-training alternatives, such as extemal recruitrnent on the labour mar-
ket and even active extemal poaching remain crucial for securing highly
skilled personnel and are considered significantly more important than
continuous training, be it in-house or extemal.

External recruitment

Training on the job

Continuous training

Vocational training

External poaching

im very important D important I?ZJ quite important

D unimportant D totally unimportant o:-::IIM:-:-:Bon:--n-I
'-
02 50 008

Fig. 1. The importance of various strategies for securing skilled personnet as

viewed by the firms in the sample (Demgenski/lcks 2002)

The overall ranking in fig. 1 highlights the importance of formal initial

vocational training: 43 % of the firms questioned consider this a very im-
portant, a further 32 % an important strategy for securing skilled person-
nel. Somewhat surprisingly, even a large majority of non-training firms
share this view. Thus, it does not seem to be the perceived inadequacy of
vocational training as an instrument for securing skilled workers that lies at
the root ofthe relative training abstinence in the business service sector.

2.2 Vocational training and the age of the firm

Young firms often Iack the necessary size, experience as weil as financial,
personnel, and temporal resources associated with training. Therefore, it is
often assumed that there is a positive relationship between the age of a
firm and its training activities. Because the share of young and very young
342

firms is much higher in the business service sector than on average, 3 this
hypothesis could explain why training participation is lower in the new
than in the old economy.
At first glance, our findings tend to support the assumption that training
participation is closely related to the age of the firm: on average, training
firms are slightly older (8.1 years old) than non-training firms (6.3 years
old). Broken down by age groups, fig. 2 shows that the percentage of firms
engaged in initial vocational training is markedly lower for very young
firms under 4 years of age (31.4 %) than for firms in the next age-category
(40.2%).

50.0
45.1
40.2 41 .1 40.1

······' ·'
; I : I ; I : ~;

1-3 4-6 7-8 9-11 12 and more All companies

n = 290 Company age (in years) C lfM Bonn
02 50 005

Fig. 2. Training participation rate offirms in the sample by age (in%)

Yet there is no statistical correlation between the age of a firm and

training participation. In fig. 2 this is apparent by the fact that subsequent
to the initial sharp increase in training between the first two age groups,
training participation does not rise significantly once a certain threshold -
here approximately 5 years of market presence - has been passed. Older
firms are not as active conceming vocational training as might have been
expected. While the hypothesis according to which training participation is
likely to increase as firms mature must therefore be rejected, five other
factors were found to have an impact on training participation in young

3 Approximately one third of all German companies are less than I 0 years old and
one fifth less than 5 years old (MIND-Survey 1999). Due to the high start-up
rates in the business service sector the share of young and very young firms is
higher in the new economy (Demgenskillcks 2002).
 343

business service firms. 4 In the following we concentrate on the impact of

size on training because, while the influence of size on training is well-
known and documented, there are some interesting differences between
young business service firms compared to the overall economy average.

2.3 Vocational training and firm size

The positive relationship between company size and training participation

rate verified for the overall economy (fig. 3) holds true for the new
economy firms in our sample, too. Whilst merely 28.9% ofthe small firms
with less than 10 employees train, this ratio surpasses 50 % in bigger com-
panies with 10 and more employees. On the whole, the bigger the com-
pany, the more lik:ely it becomes that it partak:es in initial vocational
training.
However, the data also shows that the relationship between company
size and training is much less clear-cut for the sample firms than for the
overall economy. Despite the altogether positive correlation between size
and training participation, fig. 3 shows that the largest companies in the
sample (with 50 and more employees) are not more actively engaged in
training than the companies that have between 10 and 49 employees, as is
the case for the average firm.

4 A multivariate analysis showed that five factors have a significant impact on

training: the number of employees, the share of university graduates, the ex-
pected evolution of skill demands, the start-up form and the sub-branch.
344

70.2

1-9 10-19 20-49 >50

Companies with ... employees

n =256 1<·;·;·1Sampie ~ overall economy (statistics) @11M Bonn

02 50 006

Fig. 3. Training participation rate of firms in the sample and the overall economy
by size in % (Bundesanstalt fiir Arbeit 2000; Demgenski!Icks 2002)

The training gap between old and new economy firms thus seems to
stem primarily from the under-average training participation rate of larger
firms. This raises the question, as to why the largest andlor oldest new
economy companies are not significantly more involved in initial voca-
tional training. One of the answers to this question lies in the institutional
training framework in place in Germany.

2.4 The institutional framework

The configuration of training profiles and the degree to which they cor-
respond to the qualification demands of the new economy have a direct
impact on training activities. Until the mid-nineties there were barely any
new economy-specific job profiles available. The only existing IT-related
job-profile ("Datenverarbeitungskaufmann"), for example, went back to
1967 and was hopelessly outdated. For the firsttime in 1997, four new IT-
specific job profiles were introduced. But until then, new economy firms
that wanted or needed to train were often obliged to resort to informal
training strategies outside the dual System or, altematively, to circumvent
training altogether. Thus, many firms satisfied their need for highly skilled
labour by recruiting over-qualified workers, i.e. university graduates.
 345

There is, accordingly, a notable difference in the skills structure between

new and old economy frrms. 5
lt is possible that firms with a long-standing and confirmed experience
with alternative strategies for securing skilled personnet tended to abide by
these strategies, even once new dual training venues had been opened in
the mid-nineties. Thus, in our sample, older and larger firms give more
weight to training-on-the-job than younger firms. Furthermore, they view
recruiting skilled workers on the labour market and extemal poaching
noticeably more important for securing qualified personnet than smaller
and younger firms. On the other hand, younger firms consider vocational
training a much more important training instrument than older firms. This
is true even for young firms that do not train.
Our findings also show that the youngest firms are the quiekest to
implement new training profiles: 73 % of all training firms up to 3 years
old train in these newly established profiles; only 27 % train in jobs with
"traditional" profiles. On the other hand, only 50 % of the older firms train
in the new economy-specific profiles.
At the same time as these new job-profiles were introduced, a number of
measures were taken to facilitate training by companies who, until then,
had perhaps never trained before. The first of these measures pertained to a
certain relaxation of the stringent training requirements. 6 By 2001,
approximately one fourth of all training companies were dispensed from
passing the formal training aptitude test (AEVO) (Statistisches Bundesamt
2001). Next to this, and perhaps more importantly, training networks were
widely introduced and subsidized ("Verbundausbildung"), enabling firms
to fulfil the necessary professional and technical training requirements
more easily. Today, 2 % of all German firms make use of this infrastruc-
ture (BMBF 2001 ).
Our findings for business service firms not only show that over a fifth of
them resort to network-training, but also that this instrument is used much
more intensively by young firms than by older ones. 33 % of the youngest
firms in our sample make use ofthese networks, compared to barely 4.5%
of the firms that are 9 years and older. Without this state-induced network
training structure, many small, young companies would not be in a posi-
tion to train.

5 The average share of university graduates in the skills mix lies at 15 % for the
overall economy, at 30% for business services (Zwick 2001), and at 43% for
the firms in our sample.
6 57% ofnon-training firms in the sample still rate fulfilling the necessary formal
requirements as a major hindrance to training.
346

Due to this, the younger generation of new economy firms are far
quicker to decide on - and to implement - dual training activities than were
more established firms, when they first decided to train. While it took
firms that were founded before 1990 approximately 7 years before they
first trained, young companies founded 1990 and later take no more than 3
years. We believe that this reflects the impact of institutional measures
taken to enhance training activities in the new · economy. That these
measures meet high standards is shown by the fact that firms anticipating
rising qualification demands in the future are much more likely to engage
in vocational training than firms that don't expect rising qualification
demands. Training in the dual system has become a chief option for new
economy firms that are particularly sensitive to future skill needs.
To conclude, one could say that by serving as a catalyst for a new insti-
tutional training framework the new economy has, in a sense, modeled the
dual vocational training system to suit its specific needs. New high-profile
professions have been created and training for small inexperienced compa-
nies facilitated. Young companies are quick to implement this strategy for
securing skilled personnel. Although the age of a firm does not have a
statistical impact on training, our findings give some support to the
hypothesis that formal vocational training participation will increase as this
group of young firms grows and matures and that the gap between the old
and new economy will gradually close.

3 Remunerationsystems

New economy firms are faced with a dilemma when designing a remu-
neration system (Scherer 2000). On the one hand, they have to offer com-
petitive incentive structures in order to both recruit scarce skills on the
labour market and tie them to the company afterwards. This generally
results in a comparatively high compensation level. On the other hand,
their financial resources often don't permit such a compensation level
because of high start-up investments and development costs, as weil as
delayed or irregular proceeds.
One possibility to resolve this dilemma is a remuneration system that
consists both of fixed and performance-related remuneration elements. In
order to retain as much liquidity as possible, it is recommended to
minimise the portion of the fixed salary and to defer the payment of the
performance-related remuneration as far as possible into the future. The
best ways to implement performance-related remuneration elements are
privileged share ownerships or granting stock options. Another possibility
 347

is profit sharing. However, a precondition to this is that the profit sharing

scheme be connected with the weil founded prospect of a high increment
and/or a high profit distribution (Backes-Gellner/Kay/Schröer/Wolff
2002). Finally, the benefit attainable by the whole remuneration package
has to exceed the benefit attainable by alternative jobs (Gomez-
Mejia!Balkin 1992). In the following we shall examine to which degree
new economy enterprises have adopted the above-mentioned policy on an
empirical basis and compare the results with those of other economic
sectors.

3.1 Fixed remuneration

First of all, we consider fixed remuneration. If an enterprise is subject to a

collective agreement, the fixed remuneration usually has to be at least as
high as the standard wage. Hence, the question is, to what extent the sala-
ries and wages paid by the companies match the standard wage. Fig. 4
shows no significant differences between companies in the various eco-
nomic sectors. In six out of seven firms the majority of employees receive
salaries corresponding to or above standard wage. This also applies to
business service companies. Of the one third of enterprises that pay wages
above standard, a third grants them to all employees. The rest of the enter-
prises reserves this supplementary allowance for selected employees or
employee groups. This pattem is to be found in all considered economic
sectors likewise.
348

1--:-----.,...----;--"..,----,n 31 '3
Business Services ~~~~~~~~~~~~~i?ß'i2l23:8823'83~ 56.3
15! '"'"~ "~"" 12.5
41.0
lndustry ')v<X:. 43.6

... 35.6
Handicraft ;><;'' ''>6?S: 49.5

33.1
Trade ~ 51 .0
15.9
35.7
'XY'I 49.5

35.3
•>I 49.8
Overall Economy ~~~~~~f~iis2!~s&ES2~~~21
~ 14.9
Majority of employees receive salaries/wages ...
[2] above ~ at ~ below
n =691 tariff Ievei I ~02IIM50Bonn
001

Fig. 4. Salaries and wages by economic sector (in%) (MIND-survey 1999)

These findings do not suggest that new economy enterprises pay lower
fixed salaries than other companies. However, one could argue that stan-
dard wages differ vastly between the various industrial sectors. For this
reason, the company statements regarding the wage Ievel in relation to the
standard wage do not permit conclusions regarding the actual wage Ievel.
Thus, it is possible that the remuneration paid by companies in the busi-
ness service sector is weil below that of other economic sectors, particu-
larly as it is not clear at which collective agreement new economy firms
orientate themselves. They are, after all, often not bound to a collective
agreement. But the results of another survey (Backes-Gellner/Freundl
Kay/Kranzusch 2000) suggest that the wage and salary Ievel paid by the
companies of the new economy actually does not differ overly from the
average of the overall economy. In the survey, companies were asked for
the average starting salary of a graduate. On average, it lies at about
62,150 DM per annum in the business service sector and exceeds the
average of the overall economy by 1,000 DM. The mean differences
between the sectors are not significant.
 349

3.2 Performance-related elements of remuneration

In the following we investigate the performance-related elements of

remuneration in the new economy. Fig. 5 shows highly significant
differences between the economic sectors, albeit most of the companies
with performance-related elements of remuneration are not to be found in
the business service sector but in the industry sector. In another step, we
pool the traditional companies and compare them with the new economy
companies. The resulting differences are not significant anymore. These
findings, also, fail to corroborate expectations conceming the configura-
tion ofremuneration systems in the new economy.

Business Services

lndustry 29.5

Handicraft

Trade

Other Services

Overall Economy

n = 948 Performance-related elements of remuneration O li M Sonn

02 50 002

Fig. 5. Performance related elements of remuneration by economic sector (in %)

For new economy company purposes, profit sharing seems to be the

ideal form of performance-related remuneration. Further information about
this can be gained from the family-owned companies survey
(Wallau!Kayser/Backes-Gellner 2001 ). As shown in fig. 6, profit sharing is
the dominating profit sharing scheme in all considered economic sectors,
whereas employee share ownership and stock options play an inferior role.
Furthermore, there are significant differences regarding profit sharing
plans between the economic sectors: they are granted more often by
business service companies and less often by mining and construction
350

companies. The two other pro fit sharing schemes show a similar picture. 7
Pooling the traditional companies again and comparing them with those of
the new economy, noticeable differences regarding the three profit sharing
schemes become apparent. The differences regarding profit sharing and
employee share ownership are statistically significant. These results
support the assumption that companies of the new economy use per-
formance-related remuneration elements more often, which, like stock
options or employee ownership plans, partly will not take effect until the
future. But: Just 25.9% ofthe new economy companies have implemented
one or more profit sharing schemes (Backes-Gellner/K.ay/Schröer/Wolff
2002), compared to 16.9% ofthe traditional companies. 8

Business Services

Mining

Manufacturing lndustry

Construction

Overall Economy

I:WI Profit sharing ~ Employee Share Ownership

~ Stock Options

n = 957 C lfM Bonn

02 50 003

Fig. 6. Profit sharing schemes, by economic sector (in %) (Family-owned enter-

prises-survey 200 1)

7 Due to the small number of companies which have implemented corresponding

profit sharing schemes statistical testing is impossible.
8 A logistic regression model that estimates the impact of several characteristics
on the existence of profit sharing did not result in a significant influence of the
economic sector (Backes-Gellner/Kay/Schröer/Wolff 2002). So the differences
between old and new economy firms identified by bivariate analyses aren 't due
to sectoral differences. Rather, behind these differences other variables such as
firm size, innovation Ievel or development trend of the recent years are
concealed.
 351

3.3 Remuneration packages

The analysis of the remuneration systems of new economy firms concludes

with a survey of their remuneration packages. The questionnaire of the
MIND-survey comprises ten remuneration items: fixed salary and nine
additional remuneration elements. The outcome of this is a multitude of
combinations. For lack of space we only present the morefrequent combi-
nations in the following.
But at first, we have a glance at the size of these packages. On average
such a package comprises 3.2 remuneration elements in traditional firms
compared to 2.9 for new economy firms. This mean difference is statisti-
cally not significant, whereas a further analysis shows significant mean
differences between business service and industry firms. The mean dif-
ferences between business service companies and the companies of the
remaining economic sectors are statistically not significant. The industry
sector offers on average 4.1 and the business service sector 2.9 remunera-
tion elements. As shown in fig. 7 the share of enterprises that so1ely pay a
conventional fixed salary amounts to 21.6 % in the new economy sector
and exceeds therewith the share in the traditional economy (15.8 %).
352

Solely fixed salary ~~~~-___,.___,

Fixed salary plus ...
Christmas allowance ~====~~~64.3
~====~ 54 .4

Vacation allowance b""""',._.,~~,.....-__:_j 61 .5

Company car

Commission

Other performance related remuneration

Christmas allowance plus vacation
allowance
Christmas allowance plus company car
Christmas allowance plus performance
related remuneration
Christmas allowance plus commission

Christmas allowance plus profit sharing

D Old Economy EE New Economy

n = 1.042 ClfM Sonn
02 50 004

Fig. 7. Remuneration packages of enterprises of the old and the new Economy
(Multiple responses, in%)

Traditional firms grant additional remuneration elements which are

named in collective agreements more frequently, namely Christmas and
vacation allowances. On the other band, new economy companies grant
commissions, company cars and other performance related remuneration
more frequently. Therefore it is no surprise that more than half of the
traditional firms pay Christmas and vacation allowances together, while
barely 40 % of new economy firms do so. All further combinations are
granted noticeably less often, bothin the new and the traditional economy.
These findings show, firstly, that new economy companies grant per-
formance related remuneration elements more frequently than traditional
companies. And, secondly, that they are, on average, apparently not
entirely able to compete with the remuneration packages of the old
economy. Anyhow it isn't tobe overlooked that, even though we have no
information about the Ievel of the overall remuneration, a not irrelevant
part of new economy firms can by all means tie competitive remuneration
packages. This impression is intensified by further results of the MIND-
survey 2001: The share of companies in the new and old economy which
 353

solely pay a fixed salary resp. a fixed salary plus Christmas allowance has
largely converged and the gap between the shares regarding the
combination fixed salary plus vacation allowance has decreased. At the
same time the gaps between the shares regarding remuneration combina-
tions which are offered more frequently by new economy frrms have
noticeably deepened.
All results taken together, the allegedly ideal remuneration strategy can
be empirically proven for only a relatively small part of new economy
firms; the majority ofthese firms don't ernhark on this strategy. In this re-
spect they don't really differ from traditional companies. All things con-
sidered, relative to remuneration systems, there seem tobe more similari-
ties than disparities between the old and new economy.

4 Discussion and conclusions

Although the general conditions of new economy firms drafted above

suggest that they would have to implement remuneration systems different
from those of traditional firms, these different systems cannot be proven
empirically. What is the conclusion? Some new economy firms have paid
wage Ievels exceeding their financial capacity in order to be in a position
to recruit much-needed skilled workers or in order to tie them to the com-
pany. This strategy cannot work out in the long run and finally ends in the
failure of the companies. This failure has been observed. lt cannot be
excluded, however, that the base assumption, according to which the
financial margin of new economy firms is more narrow than that of tradi-
tional :firms, does not hold true for all new economy firms.
Based on our findings, the question and implicit meta theme of this con-
ference, i.e. whether the new economy calls for a new theory of the firm,
has to be answered with no, it does not. The empirical data do not sustain
the assumption that the new economy requires a new theory of the firm to
explain or model personnel economics. The analysis of training strategies
reveals that once an adequate framework is put in to place, new economy
firms tend to resort to traditional vocational training instruments. The
analysis of remuneration systems indicates that there is no empirical evi-
dence to uphold the base assumption that there are, in fact, any significant
differences between the old and new economy in this respect. Rather the
combined evidence supports the hypothesis that the problems new
economy firms are confronted with are problems typical for small and/or
young companies. Due to the high concentration of both small and young
firms in the new economy there is an accumulation of disadvantages
354

relative to securing skilled personnel which should, in our view, diminish

as the new economy matures.

References

Anstötz K (2002) Die Finanzierung junger Unternehmen des Neuen Marktes.

WISU, 31: 1086-1092
Backes-Gellner U, Freund W, Kay R, Kranrusch P (2000) Wettbewerbsfaktor
Fachkräfte. Rekrutierungschancen und -prob1eme in kleinen und mittleren
Unternehmen. DUV, Wiesbaden
Backes-Gellner U, Kay R, Schröer S, Wolff K (2002) Mitarbeiterbeteiligung in
kleinen und mittleren Unternehmen. Verbreitung, Effekte, Voraussetzungen.
DUV, Wiesbaden
BIBB Bundesinstitut fiir Berufliche Bildung (2000) Auswirkungen wachsender
Dienstleistungen auf die Berufsausbildung. Referenz-Betriebs-System
Information no 17, 6/2000, Bonn
BMBF (2001) Berufsbildungsbericht 2000. Bonn
Demgenski C, Icks A (2002) Berufliche Bildung in jungen Unternehmen. DUV,
Wiesbaden
Gomez-Mejia LR, Baikin DB (1992) Compensation, Organizational Strategy, and
Firm Performance. College Devision, Soutb-Western Publ. Co., Cincinnati
Impulse (1999) MIND- Mittelstand in Deutschland, Köln
Impulse (200 1) MIND - Mittelstand in Deutschland, Köln
Klodt H (2001) The Essence ofthe New Economy, Kieler Diskussionsbeiträge Nr.
375, Kiel
Marschlieh A, Pendt G (2001) HR-Management in der New Economy. Human
Resource-Management in der New Economy - Derzeitige Praxis und
zukünftige Herausforderungen. Frankfurt a.M.
Scherer M (2000) Aktienoptionen in Wachstumsunternehmen. In: Achleitner A-K,
Wollmert P (eds) Stock Options. Finanzwirtschaft, Gesellschaftsrecht,
Bilanzierung, Steuerrecht, Unternehmensbewertung. Schäffer-Poeschel,
Stuttgart, pp 61-68
Statistisches Bundesamt (200 1) Bildung und Kultur. Fachserie 11, Reihe 3,
Berufliche Bildung. Metzler-Poeschel, Wiesbaden
Wallau F, Kayser G, Backes-Gellner U (2001) Das industrielle
Familienunternehmen. Kontinuität im Wandel, hrsg. vom Bundesverband der
Deutschen Industrie (BDI) und Ernst & Young Deutsche Allgemeine
Treuhand AG Wirtschaftsprüfungsgesellschaft, Berlin
Zwick T (200 1) Beschäftigungsmöglichkeiten von Fachkräften mit dualer
Ausbildung in informationsintensiven Dienstleistungsunternehmen. MittAB,
34: 74-81
Zwick T, Sehröder H (2001) Wie aktuell ist die Berufsbildung im
Dienstleistungssektor? ZEW Wirtschaftsanalysen Band 55, Baden-Baden
Training: A Strategie Enterprise Decision?

Thomas Zwick

1 Abstract

This paper analyses the training behaviour of German establishments on

the basis of representative and topical data. Especially large and co-deter-
mined establishments, investors and establishments bound by collective
wage agreements train. In addition, establishments with apprentices and a
large share of vacancies are more inclined to train. Also worker charac-
teristics play a role: establishments with a high share of qualified em-
ployees train more often while establishments with a high share of em-
ployees leaving train less. A participative work place organisation and a
state-of-the-art technical equipment have a positive impact on the willing-
ness of the establishment to invest in training. In addition to comparable
empirical studies, this paper also includes variables that indicate at which
time establishments invest in training. It argues that establishments invest
strategically when they expect additional skill demand and when profits
allow investments in human capital.

2 lntroduction

Human capital, knowledge, and skills are increasingly important competi-

tive assets within establishments. Continuous employee training sponsored
by the establishment is therefore perceived as one of the most important
measures to gain and keep productivity. Especially the German economy
that is based on a relatively high share of well qualified employees who
frequently work in flexible, complex and diversified quality production
derives its main competitive advantages from human capital and therefore
has a large demand for continuous training (Roth 1997). In 1998, German
firms accordingly invested on average the substantial amount of 1128 EUR
per year per employee in continuous training (Institut der deutschen
Wirtschaft 2002: 99).
It is not clear, however, in which economic situations firms offer
training. Although the importance of training is a widely accepted fact,
firms may be tempted to reduce costs in a temporary crisis by cutting
356

training expenditures first because the negative effects will be feit only in
later years whereas the positive cost effects are immediate. This would im-
ply that firms train when they can afford this investment in human capital,
i.e. in good economic times. On the other hand, training may be a measure
to regain competitiveness. This is a possible interpretation of the fact that
especially firms with productivity disadvantages train (Bartel 1994; Zwick
2002). This means that firms invest in training in bad economic circum-
stances. Training is therefore seen as a strategic attempt to regain competi-
tiveness, because training increases productivity at least during the two
following years (Zwick 2002). A third theoretical possibility for the timing
of training is that investments in training are a necessity when the work-
force is not adequately qualified and firms are forced to retrain workers
intemally instead of facing high labour turnover costs and a shortage of
skilled workers on the labour market (Zwick and Sehröder 2001).
Although there is an extensive Iiterature on the training behaviour of
firms and several empirical correlations between establishment, Organisa-
tion and employee characteristics and training incidence are known, the
timing of training seems to be under-researched. The contribution of this
paper therefore mainly is to analyse the timing of training investments.
while the insights from the Iiterature are also taken into account. This
paper therefore simultaneously takes into account employee characteris-
tics, employer characteristics, organisational structure and workplace
practices as weil as indicators for the timing of training. After a short
Iiterature review, hypotheses on the impact ofthe explanatory variables on
the inclination of the establishment to offer training are derived. The re-
sults of the multivariate regression model are interpreted with a special
focus on the timing of training and the last section draws some conclu-
sions.

3 The motivation of firrns to provide training - a short

Iiterature survey

Many papers on training incidence are based on household or individual

surveys (Lynch 1992; Bartel 1995). They fmd that training is more likely
to be received by better qualified employees, employees with higher job
tenure and employees in larger firms. In addition to firm size and sector,
frequently further data on the employer are absent, however.
Employer based data sets are better suited to analyse the training deci-
sions ofthe employers. Lynch and Black (1998) and Hughes et al. (2002)
include an extensive list of variables in order to explain training incidence,
 357

training intensity and differences in the types of training programmes of-

fered. Their exogeneous variables contain a wide range of employee and
employer characteristics as well as differences in workplace practices and
other characteristics between firms. They both find that large firms with a
highly qualified workforce, new machinery, participative workplace prac-
tices, investments in machines and communication and information tech-
nology, and recently hired employees train more.
For Germany, there are a couple ofrelevant sturlies on the motivation of
establishments to offer training using establishment or firm data (Gerlach
and Jirjahn 1998, 2001; Düll and Bellmann 1998; Bellmann and Büchel
2001). These sturlies all use roughly the same estimation approach and a
comparable list of explanatory variables. Gerlach and Jirjahn (1998, 2001)
include variables on the structure ofthe workforce, production technology,
organisation of work, incentives, industrial relations and the structure of
the firm on the basis ofthe Hanover Firm Panel. Düll and Bellmann (1998)
and Bellmann and Büchel (200 1) use the lAB establishment panel. While
Bellmann and Büchel estimate a very parsimonious model as a first step in
a two step productivity estimation, Düll and Bellmann explain training
intensity and training incidence by including a broad range of establish-
ment and employee characteristics. The correlations found in the sturlies
sketched above are replicated also in the German studies. Co-determina-
tion and coverage by collective wage bargaining have a stronger positive
role in training incidence in Germany, however.

4 Data basis

The empirical analysis is performed on the basis of the lAB establishment

panel (for detailed information seeBellmannet al. 2000 and Bellmann and
Büchel 2001 ). Establishments in this panel are drawn from all establish-
ments in Germany with at least one employee who has a social security
number. Therefore, only establishments consisting of employees not
covered by social insurance (mainly farmers, mine workers, artists, and
journalists) along with public enterprises with only federal employees are
excluded. There is a large set of questions that is included every year about
production, investment, industry sector, employee structure, personnet
problems, business strategy, and vocational training. The survey is held in
the middle of the year. Some questions like average employment during
one year, output and profit situation are therefore asked retrospectively in
the following wave. Every year, additional questions are added on an
irregular basis. In the wave 2000, additional information on the training
358

behaviour of the establishments was collected. We use this information

here.
We include only profit oriented establishments and those that have not
been merged with other establishments or have merged other establish-
ments. After deleting all establishments with item non-response we are left
with 9627 establishments from an original sample of 10418.

5 Factors likely to influence training provision

There are two agents in the training decision - the employer and the
employee and we need to know both sides in order to understand why a
firm trains or not (Lynch and Black 1998). Employers have to decide if
they train their employees or rather hire skilled employees from the labour
market. Some employees gain more from training than others and although
continuous training is mainly paid by the establishment also employee
characteristics may explain why some establishments train and others do
not.

5.1 Establishment size

There is are a number of reasons why larger establishments are expected to

provide more training than smaller establishments (Lynch and Black 1998;
Düll and Bellmann 1998; Gerlach and Jirjahn 1998, 2001; Bellmann and
Büchel2001; Hughes et al. 2002). First, establishments providing training
incur fixed costs. If these can be spread over a large number of employees
it will lower the average training cost per employee. Second, larger estab-
lishments tend to pay higher wages and have more extensive intemal
labour markets that facilitate retention and promotion of employees from
within. Consequently the returns to a training investment may be higher
for large than for small establishments because the pay-off period is Ionger
and trained employees can be better matched to appropriate work places.
Third, the loss of production because employees have to be trained may be
higher for small establishments than for large establishments (Hughes et al.
2002). As the correlation between the willingness of establishments to of-
fer training and the number of employees is not necessarily (log)linear, 5
establishment size categories have been introduced. If a small establish-
ment is part of a larger group or if it has several subsidiaries, it can also be
assumed that fixed training costs are easier to shoulder for the establish-
ment (Düll and Bellmann 1998).
 359

5.2 Qualificational structure of employees

The qualificational structure of the employees is expected to influence the

incidence of training. Higher qualified employees are more willing and
capable to accumulate additional human capital. Therefore establishments
employing larger proportions of highly qualified employees may invest
more frequently in their training since they expect a higher rate of return
from such investments (Lynch and Black 1998; Düll and Bellmann 1998;
Gerlach and Jirjahn 1998, 2001).

5.3 Economic sector

Economic sector is also expected to influence training. Traditionally the

manufacturing and construction sectors have engaged in higher-than-
average rates of training. Training rates are also higher in financial and
business services (Hughes et al. 2002; Zwick 2002).

5.4 Vacancy rate

In recent years, many German establishments were plagued by high

vacancy rates primarily for skilled or highly skilled employees (Zwick and
Sehröder 2001). One response which can provide a solution to a general
skilled labour shortage is to upgrade the skills of existing employees by
providing training in skills which are scarce in the labour market. If estab-
lishments do respond in this fashion, one would expect to find that estab-
lishments with high vacancy rates are more likely to provide training than
establishments with lower rates (Hughes et al. 2002).

5.5 Work place practices

Many organisational forms that increase the participation of non-manage-

rial employees also induce training needs (Lynch and Black 1998; Gerlach
and Jirjahn 1998, 2001 ). Typical examples of a more participatory work
organisation are flat hierarchies, autonomous workgroups and teamwork.
These measures widen the set of tasks for non-managers, increase their
social interaction and require at least a basic understanding of the produc-
tion process. In addition, all re-organisations render skills obsolete and
require new skills from the employees (Acemoglu and Pischke 1999).
Therefore it can be assumed that establishments introducing more partici-
patory workplace practices in recent years are more prone to offer training.
360

The lAB establishment panel comprises the introduction of three partici-

patory workplace re-organisations: the introduction of teamwork, delega-
tion of responsibility to lower hierarchical Ievels and autonomous work
groups.

5.6 Usage of modern technology

Advanced technology not only requires higher qualified employees but

frequently induces training (Düll and Bellmann 1998; Gerlach and Jirjahn
2001 ). Also investments in new machinery are expected to increase
training needs (Bellmann and Büchel 2001). A catalyst for increased
training needs seems to be investments in information and communication
technology, because establishments and employees indicate the highest
skill gaps in this area (Düll and Bellmann 1998; Jacobebbinghaus and
Zwick 2002).

5.7 Collective wage agreements and Co-determination

Collective wage agreements as well as Co-determination might increase the

willingness of establishments to offer training because they reduce
poaching by fixing a wage structure binding for most establishments in the
sector, increase trust between employees and employers by giving em-
ployees a voice, frequently include rules for (increased) training efforts of
the establishment and reduce job tum-over (Backes-Gellner, Frick and
Sadowski, 1997; Düll and Bellmann 1998; Gerlach and Jirjahn 1998,
2001).

5.8 Job turnover rate

Establishments hesitate to invest in training when they run a high risk to

lose the employees after the training, because other establishments poach
or employees change their employer (Lynch and Black 1998; Hughes et al.
2002). It therefore can be assumed that a high rate of employees leaving
the establishment decreases the inclination of establishments to offer
training. On the other hand, new employees frequently need training on the
job or other continuous training and therefore it can be assumed that
establishments that recently hired new employees also train more fre-
quently (Lynch and Black 1998). Finally, a high rate ofvacancies is a sign
of a demand for new skills in the establishment that can not instanta-
neously be fulfilled on the extemallabour market.
 361

5.9 Apprenticeship training

Vocational training may on the one hand be a substitute for further

training, on the other hand, further training and apprenticeship training
may be complements on the individual level when the apprenticeship
training entails skill gaps (Gerlach and Jirjahn 2001; Jacobebbinghaus and
Zwick 2002). The sign of this coefficient therefore does not seem tobe
clear a-priori.

5.10 Timing of training

Establishments that expect an increasing need for skills to keep their busi-
ness running effectively are more likely to engage in training than estab-
lishments which think that the need for skills is declining or stable (Düll
and Bellmann 1998; Hughes et al. 2002). Therefore the variable "in-
creasing skills need expected" is added as an indication for strategic
training decisions. When profits and training are negatively correlated,
establishments increase training when the economic situation turns bad and
this behaviour could be interpreted as using training investments as a last
resort. Finally, when establishments train because they cannot find ade-
quately skilled employees, this is an indication that establishments have to
fill their skill gaps and increase their competitiveness by investing in
training because the extemal labour market does not provide adequately
skilled job applicants.
The dependent variable is a dummy indicating if the establishment of-
fered training between January and June 2000 and therefore a bivariate
Probit to explain training incidence seems appropriate.
362

Table 1. Probit estimation to explain if a establishment trains or not, 2000

Exogenous variables Coefficients z-Value

Timing ofTraining
Problems to find appropriate skilled employees 0.165*** 4.67
expected
High qualification and training need expected 0.444*** 8.32
Profits in 1999 0.101 *** 3.46
Establishment Characteristics
Establishment has at least one subsidiary 0.266*** 6.28
Investor in information technology 0.367*** 10.78
Investor in machines 0.092*** 2.68
Collective wage agreement 0.239*** 6.62
Co-determination 0.355*** 7.96
Establishment size 20-199 0.244*** 6.14
Establishment size 200-499 0.728*** 8.57
Establishment size 500-999 0.949*** 6.29
Establishment size 1000+ 0.930*** 4.18
Workplace Practices and Employee Characteristics
Share of employees leaving the establishment - 0.225** - 2.63
in 2000
New employees in 2000 0.034 0.98
Share of vacancies 0.274** 2.46
Share of qualified employees 0.705*** 12.65
Reorganisation of departments 0.234*** 4.94
Employee participation 0.141 *** 3.11
Introduction of team work/independent work 0.134** 2.26
groups
Introduction of units with own cost accounting 0.167*** 2.62
State of the art technical equipment 0.277*** 8.37
Apprenticeship training 0.413*** 12.51
Control Variables
East German establishment 0.126*** 3.83
Agriculture - 0.307*** - 2.96
Mining and basic goods 0.210 1.22
Food - 0.351 *** - 4.14
Machines and ships - 0.437*** - 5.50
Production goods - 0.277*** - 4.78
Investment and consumption goods - 0.177*** - 3.17
Construction - 0.363*** - 6.69
Traffic and logistics - 0.258*** - 3.21
Credit, banking and insurance 0.623*** 4.79
Hotels, catering and restaurants - 0.511 *** - 5.97
Education 0.377** 2.49
Health and social services 0.555*** 7.83
 363

Table 1. (cont.)
Data processing, research and development, 0.395*** 4.94
consulting
Other business services - 0.01 - 0.01
Other personal services 0.076 0.91
Constant 1.599*** - 26.75
Wald-test X2(35) = 2751 Prob<0.01
Pseudo R2 0.29
Number of observations 9627
Comments: The significance level is marked by stars: *** significant at 1 % and
** significant at 5%, heteroscedasticity-robust standard errors.
Source: lAB Establishment Panel, Wave 2000, own calculations.

6 Results

According to the results in comparable empirical sturlies and our hypothe-

ses, larger establishments and establishments with subsidiaries train more.
In contrast to the Anglo-Saxon and Irish literature, manufacturing estab-
lishments train less than service establishments. Especially in sophisticated
business services, banking and insurance, education, the health services
and hotel and catering training is wide-spread. Also establishments that
invest in information and communication technology and machines have
more frequently a demand for training. The regression shows that estab-
lishments with apprenticeship training and collective wage agreements
have a higher inclination to train. Therefore apprenticeship training and
continuous training are complements, see also Düll and Bellmann (1998)
and Gerlach and Jirjahn (2001). It is also a well-established fact that estab-
lishments with a higher qualified workforce train more. We find support
for our hypothesis that establishments with a high share of employees
leaving the establishment do invest less often in human capital of their
work force. The fact that a establishment has new hires does not have a
positive impact on the training incidence while the share of vacancies
which is an indication for skill gaps on the labour market increases the
inclination of establishments to train. We can therefore state that the
results presented are mainly in accordance to the literature.
The paper additionally shows that even when we control for a broad
range of relevant factors, variables that indicate the timing of training are
important additional explanatory variables for training incidence. Most
Germanestablishments react on expected skill shortages by training efforts
because the extemal skilled labour market isthin (Roth 1997; Zwick and
364

Sehröder 2001). Therefore, it is not surprising that establishments ex-

peeting a rise in qualifieation needs have a higher inclination to train. The
same applies for establishments that expeet problems to fill skilled vaean-
cies. Both variables support the hypothesis that German establishments
invest strategieally in training already in advanee of skill problems. In ad-
dition, training efforts vary with profits. Establishments that realised
satisfaetory profits in the previous year train more often. This also indi-
eates that establishments do not invest in training as a last resort, but
whenever they ean afford it.

7 Conclusions

This paper shows on the basis of representative data that German estab-
lishments invest strategieally in human eapital when they ean afford it or
when they expeet skill gaps in the near future. Establishments that invest in
training on average have a relatively low produetivity at the time training
is offered. This should not be interpreted as an indieation that training is a
measure of last resort and not strategie, however. In addition, the multi-
variate regression model used also eontrols for a broad speetrum of other
well-known determinants of training. Hereby, the results of eomparable
studies have been replieated.
Table 2. Descriptive analysis ofvariables used

Exogenous variables Share

Establishment offers training (1 = yes, 0 = no) 0.35
Problems to find appropriate skilled employees expected (I = yes, 0.23
O=no)
High qualification and training need expected (1 = yes, 0 = no) 0.06
Profits in 1999 (1 = very good, 0 = satisfactory, sufficient, 0.35
unsatisfactory)
Share of employees leaving the establishment (divided by total 0.08
number of employees)
Reorganisation of departments during last 2 years(l = yes, 0 = no) 0.06
Introduction of employee participation during last 2 years 0.09
(delegation of responsibilities to lower hierarchicallevels, 1 = yes,
O=no)
lntroduction ofteam work/independent work groups during last 2 0.03
years (I= yes, 0 = no)
Introduction ofunits with own cost accounting during last 2 years 0.03
(1 = yes, 0 = no)
Establishment has at least one subsidiary 0.11
Establishment size 1-19 (reference category) 0.87
 365

Table 2. (cont.)
Establislnnent size 20-199 0.10
Establislnnent size 200-499 0.01
Establislnnent size 500-999 <0.01
Establislnnent size 1000+ <0.01
Share of qualified employees 0.48
State ofthe art technical equipment (1 = technical equipment state 0.69
ofthe art, 0 = technical equipment obsolete)
Investor in IT (1 = yes, 0 = no) 0.49
Investor in machines (1 = yes, 0 = no) 0.47
Co-determination (1 = yes, 0 = no) 0.07
Collective wage agreement (1 = yes, 0 = no) 0.65
Apprenticeship training offered (1 = yes, 0 = no) 0.30
New employees hired (1 = yes, 0 = no) 0.27
Share ofvacancies (divided by total number of employees) 0.03
East German establislnnent (1 = yes, 0 = no) 0.21
Agricu1ture 0.03
Mining and basic goods <0.0 1
Food 0.02
Machines and ships 0.03
Production goods 0.03
Investment and consumption goods 0.05
Construction 0.13
Trade (reference category) 0.24
Traflic and logistics 0.05
Credit, banking and insurance 0.03
Hotels, catering and restaurants 0.08
Education 0.01
Health and social services 0.09
Data processing, research and deve1opment, consulting 0.07
Other business services 0.10
Other personal services 0.04
Comments: All figures are for the first half of a year in 2000 if not noted
otherwise, figures are weighted and based on the sample used in the regression
presented in Table 1.
366

References

Acemoglu D, Pischke J (1999) Beyond Becker: Training in Imperfect Labour

Markets. The Economic Journal109: 112-143
Backes-Gellner S, Frick B, Sadowski D (1997) Codetermination and Personnet
Policies ofGerman Firms: The lnfluence ofWorks Councils on Turnover and
Further Training. The International Journal of Human Resource Management
8:328-347
Bellmann L, Kölling A, Kistler E, Hilpert M, Heinecker P, Conrads R (2000)
Codebook zum lAB-BetriebspaneL Nuremberg, Institut fiir Arbeitsmarkt- und
Berufsforschung
Bellmann L, Büchel F (2001) Betrieblich fmanzierte Weiterbildung und Unter-
nehmenserfolg. In: Backes-Gellner U, Moog P (eds) Bildungssystem und
betriebliche Beschäftigungsstrategien. Berlin, Duncker-Humblot, pp 75-92.
Bartel A (1994) Productivity Gains from the Implementation of Employee
Training Programs. lndustrial Relations 33: 411-425
BartelA (1995) Training, Wage Growth, and Job Performance: Evidence from a
Company Database. Journal ofLabor Economics 13:401-425
Düll H, Bellmann L (1998) Betriebliche Weiterbildungsaktivitäten in West- und
Ostdeutschland. MittAB 31: 205-225
Gerlach K, Jirjahn U (1998) Determinanten betrieblicher Weiterbildungsaktivitä-
ten: Eine empirische Untersuchung mit Daten des Hannoveraner Firmenpa-
nels. In: Pfeiffer F, Pohlmeier W (eds) Qualifikation, Weiterbildung und
Arbeitsmarkterfolg. Nomos, Baden-Baden, pp 311-338
Gerlach K, Jirjahn U (2001) Employer Provided Further Taining: Evidence from
German Establishment Data. Schmollers Jahrbuch 121: 139-164
Hughes G, O'Connell P, William J (2003) Company Training and Low-Skill
Consumer-Service Jobs in Ireland. International Journal of Manpower 24:
forthcoming
Institut der Deutschen Wirtschaft (2002) Deutschland in Zahlen. Deutscher.
Institutsverlag, Köln
Jacobebbinghaus P, Zwick T (2002) New Technologies and the Demand for
Medium Qualifled Labour in Germany. Schmollers Jahrbuch 122: 179-206
Lynch L (1992) Private Sector Training and the Eaming of Young Workers.
American Economic Review 82: 299-312
Lynch L, Black S (1998) Beyond the Incidence of Employer-Provided Training.
Industrial and Labor Relations Review 52(1): 64-81
Roth S (1997) Germany: Labor's Perspective on Lean Production. In: Kochan T,
Lansbury R and MacDuffie J (eds ), After Lean Production. ILR Press, Ithaca,
pp117-136
ZwickT (2002) Continuous Training and Firm Productivity in Germany. ZEW
Discussion Paper 02-50, Mannheim
Zwick T, Sehröder H (2001) Wie aktuell ist die Berufsbildung im Dienstleis-
tungssektor? ZEW Wirtschaftsanalysen 55. Nomos, Baden-Baden
Section 5

Finance
Some New Properties of Risk Measures

Luca Bonaccorso

Salvatore Greco

Benedetto Matarazzo

Pietro Platania

1 Abstract

Risk measures are more and more important in fmancial management. The
risk measure most widely adopted in nowadays practice is VaR (Value at
Risk). However, within specialized Iiterature some doubts about some
properties of V aR originated a large interest in axiomatic definition of risk
measures. Coherent risk measures were the first class of risk measures to
be proposed in this context and they are also the most well known within
the specialized literature. Coherent risk measures are characterized by the
following four axioms: monotonicity, positive homogeneity, translation
invariance and subadditivity. Recently, some new axioms have been pro-
posed to emphasize particular features of the financial risks. In this paper
we are mainly interested in conservatism, i.e. in the property that requires
that a risk measure should depend only on the losses, and not also on the
gains of an investment. We consider two classes of conservative risk
measures: the robust risk measures and the conservative coherent risk
measures. While the robust risk measures are based on some characteristic
axioms quite different from those of coherent risk measures, conservative
coherent risk measures take into account conservatism, trying to maintain
as much as possible the properties of coherent risk measures. We give also
the financial interpretations of all the axioms considered and we discuss
the specific relations between the robust risk measures and the Sugeno in-
tegral.
370

2 lntroduction

Recently the problern of correctly defining and measuring performances

and risks of an investment has been taken into consideration with in-
creasing interest, both from the theoretical and operational points of view.
Investors and intermedianes need to have synthetic and meaningful indi-
cators in order to efficiently compare, control and manage different in-
vestment opportunities and the correspondent risks. This ensures that the
financial decisions are actually based on an appropriate Ievel of return-risk
coherent with investors' preferences.
Presently, the most used methodology for risk measurement is Value-at-
Risk (VaR) [Value at Risk. (1996), RiskMetrics. (1995)]. Roughly
speaking, VaR equal to Y means that for the considered investment there is
a probability a% (generally a=5%) to have a loss greater than or equal to
the given amount Y. This type of information is quite simple and very
meaningful for the aims of risk management, especially in a financial
context. This explains the success ofVaR. However, VaR does not satisfy
one very important desirable property of risk measures: subadditivity.
Subadditivity means that the sum of the risks of two investments cannot be
larger than the risk of the investment obtained by the sum of the
considered investments. From financial viewpoint, this is very important
because it is related to the idea that diversification reduces (or, at least,
does not increases) the risk. An important answer to this problern has been
given by Artzner et al. [Artzner et al. (1997), Artzner et al. (1999)] that
recently proposed a new class of risk measures, defined coherent risk
measures, which are characterized by four properties: monotonicity,
positive homogeneity, translation invariance and subadditivity.
Coherent risk measures have been enthusiastically accepted in the
specialized Iiterature for their appealing approach to define risk measures
on the basis of simple axiomatic premises. In this paper we put ourselves
in the same perspective. More in detail, we are interested in defining risk
measures which adequately take into account a property which - in our
opinion - is very important for risk measure: conservatism. Conservatism
means that the risk measures should depend only on the possible Iosses of
an investment and not also on its gains, according to a principle which can
be synthesized as: "no loss, no risk". This way to define the risk empha-
sizes the difference between a risk measure and a performance index. A
risk measure is related only to the possible lasses. Within this downside
perspective, only the negative part of the cash flow is relevant to properly
measure the risk, which could be then understood as the insurance
premium to cover the expected lasses. The perspective of a performance
 371

index is quite different: in this case gains and losses participate together to
define an index, which measures the comprehensive attractiveness of the
investment. To give an extreme example permitting to distinguish the per-
spective of the risk measures and the perspective of the performance indi-
ces, consider two investments for which, in any possible state of the world,
there is no loss. In this case, a risk measure considering only the negative
part of the cash flow gives no indication about the preference in comparing
the investments. The reason is simple and even trivial: there is no possible
loss in both the investments and therefore they cannot be discriminated by
the risk measures we are considering. But this is not surprising, and on the
contrary it is concordant with common sense: indeed would be quite
singular to give a different risk measure to the two investments presenting
no risk. What instead is surprising is that actually all risk measures con-
sidered in the practice and in the Iiterature do not satisfy conservatism and
therefore, in general, discriminate the two investments. The reason is that
these risk measures consider both gains and losses, and therefore they dis-
criminate with respect to the "risk", even when there are no possible loss.
Another important aspect of conservatism is that, with respect to a risk
measure, losses cannot be compensated by gains. We think that this is also
a simple and quite intuitive axiomatic premise of a risk measure. However,
coherent risk measures do not satisfy this principle and to our knowledge
until now there has been no attempt to properly define risk measures
satisfying conservatism. We envisaged two possible approaches to define
risk measures satisfying conservatism:
1) to define new classes of risk measures which satisfy conservatism on
the basis of some axiomatic premises completely different with
respect to properties characterizing coherent risk measures;
2) to define risk measures which satisfy conservatism, trying to main-
tain as much as possible the axioms of coherent risk measures.
According to these two approaches, in this paper we consider two new
classes of risk measures:
a) robust risk measures [Bonaccorso et al. (2002)], characterized by
some axioms quite different with respect to coherent risk measures:
conservatism, continuity, sure loss principle and robustness;
b) conservative coherent risk measures [Greco et al. (2002)], which
adapt the axioms of coherent risk measures with the aim of
permitting also consideration of conservatism.
Let us remark that, apart from their financial interpretation, the robust
risk measures have a particular interest also within measure theory and
fuzzy set theory. This interest depends on the special role in representation
372

of robust risk measures of a very weil known fuzzy integral: the Sugeno
integral [Sugeno (1974)]. Actually, robust risk measures are the class of
the risk measures that can be represented as the Sugeno integral of the
possible Iosses of a project. In this sense, with respect to measure theory
and fuzzy set theory, the axiomatic characterization of robust risk
measures gives also a specific characterization of Sugeno integral.
The article has the following organization. Section 3 presents the nota-
tion and recalls the axioms characterizing coherent risk measures. Section
4 introduces the robust risk measures and their representation as Sugeno
integral of the Iosses of the considered investment project. Section 5 pre-
sents a characterization of the robust risk measures satisfying also subad-
ditivity. In Section 6 some modifications of properties characterizing
coherent risk measures are presented and the relative axiomatic bases are
stated, taking into particular consideration the property of conservatism.
Final Section groups conclusions.

3 Coherent risk measures

In the following we are using the same notation of [Artzner et al. (1999)].
More precisely:
a) We shall call n the set of states of nature, and assume it is finite.
Considering n as the set of outcomes of an experiment, we compute
the net worth of a position for each element of 0; it is a random
variable on 0, denoted by X. Its negative part is denoted by x- and,
for each roEO, x-(ro)=min(X(ro),O). Analogously, its positive part is
denoted by x+ and, for each roEO, X\ro)=max(X(ro), 0)
b) Let G be the set of risks, i.e. investments, that is the set of all real
valued functions on n. Since n is assumed to be fmite, G can be
identified by Rn, where n=card(Q).
Definition 1 ([Artzner et al. (1997), Artzner et al. (1999)]). A mapping
p from G into R is called a coherent risk measure if the following
properties hold for all X,Y EG:
Monotonicity: if X~Y, then p(X);:::p(Y). If an investment has returns
worse than (more precisely not better than) another investment for all
states of nature, then its risk measure is greater (more precisely not
smaller).
Subadditivity: p(X+Y)~p(X)+p(Y). Diversification does not increase
risk; no manipulation is allowed: it is impossible to reduce risk splitting
 373

financial assets (the risk of two investments is not greater than the sum of
their own risks).
Positive homogeneity: for A.~O, we have p(A.X)=A.p(X). Increasing (or
reducing) the size of a given investment will increase (or reduce) the risk
in the same size.
Translation invariance: for every constant function aeG such that
a=[a, ... ,a] with aeR, we have that p(X+a)=p(X)-a. Adding a positive
(negative) risk-free asset to an existing investment will reduce (increase)
the original risk by the same amount invested in the asset.

Remark 1. Subadditivity and positive homogeneity imply

Convexity: for each X,Y eG and for any A.e]0,1[, p(A.X+(l-A.)Y)
~A.p(X)+(1-A.)p(Y). A convex combination of different investments cannot
increase the risk (the risk of the combined investment is not greater than
the weighted sum of their own risks ).
Let us also observe that convexity does not imply subadditivity and
positive homogeneity. For example the function f(x)=x 2 is convex but it is
neither subadditive (as it easy to verify considering that, for example,
f(1)+f(3)<f(1+3), i.e. 12+3 2<(1+3) 2), nor positively homogeneous. Thus
subadditive and positively homogeneaus functions constitute a proper
subset ofthe convex functions and they are defined sublinear.

Theorem 1 ([Artzner et al. (1997), Artzner et al. (1999)]). A risk

measure p is coherent if and only if there exists a family P of probability
measures on the set of states of nature n, such that
I
p(X)=sup{Ep[-X] PeP]}=-inf{Ep[X] PeP]. I
Remark 2. This measure can be easily interpreted as the expected value
of the result of the investment minimized with respect to the considered
family of probability distributions, i.e. the insurance premium to cover the
expected loss in the warst case scenario. In this case, scenario means one
probability measure PeP. Therefore, given two seenarios represented by
probability measures P 1,P2 eP, P 1 represents a scenario worse than that
represented by P2 if
Ep 1[-X]> EP2[-X].
Let us observe that p(X) can assume also negative values. In this case
risk measure has the following meaning: the investment is not considered
risky and therefore no insurance premium should be paid and, on the
contrary, an amount of money equal to -p(X) can be withdrawn main-
taining the investmentnot risky.
374

Remark 3. It seems quite questionable that a (coherent) risk measure is

determined also by positive outcomes (see Introduction). They are rele-
vant, together with negative ones, when a performance index is required.
But risk measure should depend only on the possible Iosses of an invest-
ment and not also on its gains. The basic idea is that compensation among
gains and Iosses should be not allowed when we are dealing with risk
measures. Of course, good financial decisions require to take into account
both performance indices and risk measures, but it is always prudent to
distinguish the two perspectives.

4 Robust risk measures

This section and the following are based on [Bonaccorso et al. (2002)]. In
this section we introduce and characterize a new dass of risk measures: the
robust risk measures.
Definition 2 [Bonaccorso et al. (2002)]. A mapping p from GintoR is
called a robust risk measure ifthe following properties hold:
• Conservatism: for all XEG, p(X)= p(X );
• Sure lass princip/e: for every constant ftmction aEG, a=[a, .... ,a] with
a~O, we have that p(a)=-a;
• Continuity: for each X,Y EG, if p(X)<p(Y), then, for each aER such
that p(X)<a<p(Y), there exists AE]O,l[ suchthat p(A.X+(l-A.)Y)= a;
• Robustness: for each X,Y EG,
[ {mEO: Y(m)~-p(X)};;;2{mEO: X(m)~-p(X)} ]=>[ p(Y)~p(X)].
Conservatism has an interesting financial interpretation: risk measure
depends only on losses. Coherent risk measures do not satisfY conserva-
tism. However, conservatism was considered as one axiom of coherent risk
measures in a first draft of [Artzner et al. (1999)], as quoted in [Studer
(1997), pp.17-18]. Sure loss principle has also a simple financial
interpretation: if an investment gives a sure loss in each state of the world,
then the measure of its risk should be exactly equal to this loss. Continuity
has an immediate interpretation: if there are two investments having
different risk measures, then it is always possible to build up, by a linear
combination of these two investments, another investment having any
intermediate risk measure. This means also that, if the outcomes of an
investment change with continuity, then the correspondent risk measure
reacts without "jumps". Robustness has the following interpretation. Risk
measure can be interpreted as a margin to protect an investment from the
risk of losses. Let us suppose that the margin p(X) of an investment X
 375

covers the Iosses of X relative to states of nature from A<;;;;O (i.e. p(X)~­
X( ro) for each ro EA ), while the same margin p(X) covers the Iosses of
another investment Y only with respect to states of nature from BcA (i.e.
p(X)~-Y(ro) for each mEB). In this case, the margin p(X) would not be
able to cover the Iosses of investment Y relative to states from A-B (i.e.
p(X)<-Y(ro) for each roEA-B), while the correspondent Iosses of X would
be covered by p(X) (i.e. p(X)~-X(ro) also for each roEA-B). In this case
robustness requires that the margin of investment Y, calculated as p(Y),
should not be smaller than p(X), i.e. p(Y)~p(X). The following example
illustrates this point.
Example 1. Let us suppose that,
- 0={1, 2, 3, 4},
-X=[-200, -50, -30, 300] (i.e. X(l)=-200, X(2)=-50, X(3)=-30,
X(4)=300),
- Y=[-75, -100, -20, 150] and
- p(X)=70
Let us also suppose that the margin of X is calculated in terms of its risk
measure, p(X). In this case the margin p(X) does not cover the loss of X
relative to state of nature 1 (in fact p(X)=70<-X( 1)=200), but it covers the
Iosses relative to states 2 (in fact p(X)=70>-X(2)=50 ) and 3 (in fact
p(X)=70>-X(3)=30 ). Relatively to investment Y, the same margin p(X)
would not permit to cover neither the loss relative to state of nature 1 (in
fact p(X)=70<-Y(1)=75 ), nor the loss relative to state ofnature 2 (in fact
p(X)=70<-Y(2)=10 0), but it covers only the loss relative to state 3 (in fact
p(X)=70>-Y(3)=20 ). For robustness property, this implies that the margin
of Y, calculated in terms of its risk measure p(Y), should not be smaller
than the margin of p(X), i.e. p(Y)~p(X).
Lemma 1. Robustness implies monotonicity.
Remark 4. If we interpret the risk measure as a margin to protect from
the Iosses of the investment, then another interesting property is the
following:
• Non-negativity: p(X)~O, for each XEG.
Non-negativity means that the margin cannot be negative.
lt is interesting to remark that, under the hypothesis of sure loss princi-
ple and robustness, conservatism and non-negativity are equivalent. Indeed
sure loss principle, robustness and conservatism imply non-negativity.
Since, under the hypothesis of sure loss principle and robustness, con-
servatism is equivalent to non-negativity, Definition 2 is equivalent to the
376

following Definition 3, where conservatism is substituted by non-nega-

tivity introduced above.
Definition 3. A mapping p from G into R is called a robust risk
measure if the following properties hold: non-negativity, sure loss princi-
ple, continuity, robustness.
Theorem 2 [Bonaccorso et a/. (2002)]. A risk measure p is robust if and
only ifthere exists a capacity J.l on n, i.e. a function J.l:2°~R+u{+oo} such
that, for each A~B~n, J.l(A)~J.l(B), J.l(0)=0 and J.l(O)=+oo, and for each
XEG
p(X)=max { min{ min {-X-(ro)}, J.l(A)} }.
A<;;;Q IDEA

Remark 5. Given N={l, ... ,n}, a capacity on N is a function

J.l:2N~R+u{+oo}, suchthat

1) J.l(0)=0, J.l(N)=+oo (boundary conditions) and

2) J.l(A)~J.l(B) if A~B, A,B~N (monotonicity).

Given a=[a 1, ... ,a0 ]E R~, its Sugeno integral [Sugeno (1974)] with
respect to capacity J.l on N is defined as
S(a; J.l) = max {min{a1tQ), J.l(Altm)} },
j=l, ... ,n

where ax 0 ), ... ,ax(n) are values at, ... ,an reordered according to
permutation 1t:N~N suchthat ax(l)~ ... ~ax<nl• and A1tm={7tG), ... ,7t(n)} for
eachj=l, ... ,n.
The Sugeno integral of a=[at, ... ,a0 ]E R~ with respect to capacity J.l on
N can be calculated also as

S(a; J.l)= max { min{ min {ai}, J.l(A)} }.

A~N iEA

Therefore formulation of robust risk measure p(X) given by Theorem 2

coincides with the Sugeno integral of -X , the absolute value of the nega-
tive part of risk X, using the capacity J.l on 0. Thus Theorem 2 can be
reformulated as follows.
Theorem 2 '. A risk measure p is robust if and only if there exists a
capacity J.l on n such that, for each XEG,
p(X)= S(-X-; J.l).
 377

Remark 6. Theorem 2 has an intrinsic interest in terms of characteriza-

tion of Sugeno integral as aggregation function.
Any function F: R~ ~R+ is an aggregation function. The following
properties of an aggregation function F are useful to characterize the
Sugeno integral:
- F is idempotent iff, for each ae R~ suchthat a=(a, .... ,a), F(a)=a;
- F is continuous if for each a,be R~. if F(a)<F(b), then for each
aeR+ such that F(a)<a<F(b), there exists A.e]O,l[ for which
F{t..a+(l-A.)b)=a;
- F is monotonic with respect to the upper Ievel sets, iff for each
a,bER~
[{ieN: bi ~ F(a)};;;2{ieN: ai ~ F(a)}] ~ F(b)~(a).
Theorem 3. An aggregation function
F: R~~R+
is idempotent, continuous and monotonic with respect to the upper Ievel
sets if and only if there exists a capacity J.1 on N={l, ... ,n} such that, for
eachae R~.
F(a)=S(a; J.l).

Remark 7. From the financial point of view, the result of Theorem 2

has the following interpretation. Let us suppose that the risk measure p(X)
represents a margin to cover the eventual Iosses relative to X. For each
A~;;!l, one threshold J.l(A) is given. This threshold J.l(A) means that the
minimalloss (the "sure" one) relative to states of nature from A must be
covered within the amount J.l(A). Thus
p(X)~min {- X""(ro)}
meA
if min {-X""(ro)}:::;J.l(A); on the contrary
meA
p(X)~J.l(A).

This financial interpretation of capacity J.1 justifies its monotonicity

property. The following example illustrates this point.
Example 2. Let us suppose that 0={1,2,3}. Table 1 presents capacity
J.l.
378

Table 1. Capacity f.l

Subset of states of nature AcO Capacity f.l(A)
0 0
{1} 10
{2} 5
{3} 7
{1,2} 15
{1,3} 13
{2,3} 10
{1,2,3} +oo

N ow let us suppose that we should evaluate the risk measure of the

investment X=[-13, -14, 40]. According to the capacity f.l presented in
Table 1, the risk measure should cover the loss in state 1, i.e. X(l)=-13,
within the relative threshold f.l( {1} )= 10. Thus
i) p(X)~10.
Analogously, the risk measure should cover the loss in state 2, i.e.
X(2)=-14, within the relative threshold f.l( {2} )=5. Thus
ii) p(X)~5.
With respect to state of nature 3 there is no loss and therefore there is no
covering problem. With respect to states 1 and 2 together, Iet us remark
that, if one of the two states is verified, there is a minimal loss of 13, i.e.
the 1oss is of at least 13: in fact
min{-X(l),-X(2)}=13.
The risk measure p(X) should cover this minimal loss within the
threshold f.l( {1,2} )= 15. Thus we have that
iii) p(X)~ 13.
With respect to the other subsets of states of nature, there is no minimal
(i.e. "sure") loss. In fact, state ofnature 3 belongs to each set {1,3}, {2,3},
{ 1,2,3} and X(3 )=40 is a gain. Thus, for Theorem 2 and i), ii) and iii), we
have that p(X)=max{10,5,13}=13.

5 Robust risk measures and subadditivity

In the following we discuss relations between robust risk measures and a

fundamental property of coherent risk measures: subadditivity
 379

Remark 8. Subadditivity can be stated as "a merger does not create

extra risk". Subadditivity is often considered the most characterizing
feature of a risk measure when a portfolio, i.e. a set of investments, is
considered. lt is related to portfolio diversification, because only the
subadditivity of the risk measure ensures that the global risk of a portfolio
will not be larger than the sum of its partial risks, i.e. the risk measures of
all its component; in other words, risk cannot be reduced splitting financial
assets. Moreover subadditivity is very important in capital adequacy,
because, if subadditvity holds, then the risk of the whole bank cannot be
larger than the sum ofthe branches' risks. We do not introduced subaddi-
tivity among the axiomatic premises of robust risk measures because we
think that there can be specific situations in which subaddivity may not be
considered an important property ofrisk measures. For example, ifwe deal
with an "all-or-nothing" investment, with respect to which it is impossible
a diversification, subadditivity is not a very interesting property.
Example 3. Let us show that the capacity presented in Table 1 gives a
risk measure violating subadditivity. Let us suppose that W=[-30, -60, 30]
and Z=[O, 30, -60]. According to capacity J.l defined in Table 1, we have
that p(W)=15 and p(Z)=7. Considering W+Z=[-30,-30,-30], we have that
p(W+Z)=30. Thus, p(W+Z)>p(W)+p(Z), which means that subadditivity
is not satisfied.
Following Theorem 4 characterizes the subadditivity of robust risk
measures.
Theorem 4. A risk measure p is robust and subadditive if and only if
capacity J.l is subadditive, i.e. for each A,B~n, J.t(AuB)~J.t(A)+J.t(B).
Remark 9. Theorem 4 has an intrinsic interest in terms of characteriza-
tion of subadditivity of Sugeno integral. From this viewpoint, Theorem 4
can be reformulated as follows.
Theorem 5. Given a capacity J.l on N, the Sugeno integral is subadditive
(i.e., for each a,bE R~, S(a+b, J.t)~S(a; J.t)+S(b; J.t) if and only if
J.t(AuB)~J.t(A)+J.t(B).

Example 4. The capacity J.l presented in Table 2 satisfies condition of

Theorem 4 and therefore the correspondent robust risk measure is
subadditive.
If we consider risks W and Z presented in Example 3 and recalculate
risk measures on the basis of capacity from Table 2, we obtain: p(W)=30,
p(Z)=7 and p(W+Z)=30. Therefore, p(W)+p(Z)~p(W+Z), which means
that, in this case, subadditivity is satisfied.
380

Table 2. Capacity J..l

Subset of states of nature AcO Capacity J.l(A)

0 0
{1} +oo
{2} 5
{3} 7
{1,2} +oo
{1,3} +oo
12,3} 10
{1,2,3} +oo
Remark 10. Robust risk measures satisf)ring subadditivity present an
interesting characteristic pointed out by the following Corollary 1.
Corollary 1. If a risk measure p is robust and subadditive, then there
exists at least one ro E n such that Jl( {ro })=+oo.
In financial terms Corollary 1 has the following interpretation. If p(X)
represents a margin for investment X, Jl( {ro} )=+oo means that any loss of
X in the state ofnature wen must be completely covered. A quite natural
interpretation of Jl( {ro} )=+oo is that ro is considered a state of nature corre-
sponding to a "normal" evolution of the affairs. To cover any loss com-
pletely in this state of nature is absolutely reasonable.
Example 5. With respect to capacity presented in above Table 2, state
of nature 1 can be interpreted as a "normal" evolution of the affairs and
therefore any loss in this state of nature must be completely covered by a
margin calculated in terms ofthe risk measure p.

6 Conservative coherent measures of risk

We are now considering some other possible formulations of risk

measures, starting by the theorem characterizing coherent risk measures
and slightly modif)ring each time its formulation, in order to take into par-
ticular consideration the conservatism. The correspondent properties, satis-
fied (or not satisfied) by each proposed formulation, will be presented.
This section is based on [Greco et al. (2002)].
We start remembering that there is no risk measure satisf)ring
conservatism and translation invariance [Greco et al. (2002)]. Therefore,
besides the properties introduced in the previous Sections, we are now
 381

considering also the following smooth version of the translation invariance

principle:
- Conservative translation invariance: for all x:::;;o and constant
function a, suchthat X+a::s;O, we have that p(X+a)=p(X)-a.
Conservative translation invariance can be interpreted as follows.
Adding a risk-free asset to an existing "investment", which gives always
losses, will reduce the original risk by the same amount invested in the
asset, only if this further sum does not permit to change any loss into a
gain. In other words, all outcomes of the new investment (X+a) will
remain Iosses (as in the original investment X). For example, let us
consider the investment X=[-13, -14, -30] and let us add to its outcomes
(alllosses) a constant a::s;13. Since the outcomes ofthe investment X+a are
again losses, the original risk measure of X decreases of the same amount
a, i.e. p(X+a)=p(X)-a.
Theorem 6. A risk measure p satisfies monotonicity, subadditivity,
positive homogeneity, conservatism, conservative translation invariance if
and only if there exists a family of probability measures P on the set of
states of nature n, such that
I
p(X)=sup{Ep(-X-] PEP}.
In this case sure loss principle is also satisfied.
Remark 11. The risk measure considered in Theorem 6 can be inter-
preted as the insurance premium to cover the expected loss in the worst
case scenario, but taking into consideration only negative outcomes. In this
case, scenario means one of the probability measure PE P. Therefore, if
we are interested only in the negative outcomes (i.e. Iosses), given two
seenarios represented by probability measures P~,P 2 eP, P 1 represents a
scenario worse than that represented by P2 if
-
Ep 1(-X ]> EPZ[-X ].
-
Theorem 7. [Frittelli (2002a)] A risk measure p satisfies monotonicity,
subadditivity and positive homogeneity if and only if there exists Z~R+ xP,
where P is a family of probability measures on the set of states of nature
n, suchthat
I
p(X)=sup{aEp(-X] (a,P)EZ}.
Remark 12. The risk measure here considered is a more general case
than that considered in Theorem 1 and it can be interpreted as the in-
surance premium to cover a pos}tive multiple aEp(-X] ofthe expected loss
in the worst case scenario. In this case, scenario means one pair (a,P)EZ.
382

Therefore, given two seenarios represented by pairs (a~,P 1 ),(a2 ,P 2 )EZ,

(ahP 1) represents a scenario worse than that represented by (a2,P2) if
a1EP1 [-X]> azEPZ[-X].
Theorem 8. A risk measure p satisfies monotonicity, subadditivity,
positive homogeneity and conservatism if and only if there exists Z~R+ xP,
where P is a family of probability measures on the set of states of nature
n, suchthat
p(X)=sup{aEp[-X-] I(a,P)EZ}.
Remark 13. The risk measure considered in Theorem 8 can be inter-
preted as the insurance premium to cover a positive multiple of the
expected loss in the warst case scenario, but taking into consideration only
negative outcomes, i.e. aEp[-X]. In this case, scenario means one pair
(a,P)EZ. Therefore, given two seenarios represented by pairs
(ahP 1),(a2,P2)EZ, (a~,P 1 ) represents a scenario worse than that represented
by (az,Pz) if
- -
a1EP1[-X ]> azEPZ[-X].
Theorem 9. [Frittelli (2002)] A risk measure p satisfies monotonicity
and convexity if and only if there exists Z~R+ xRxP, where P is a family
of probability measures on the set of states of nature n, such that
I
p(X)=sup{aEp[-X]+b (a,b,P)EZ}.
Remark 14. The risk measure considered in Theorem 9 can be inter-
preted as the insurance premium to cover a positive linear transformation
of the expected loss aEp[-X]+b in the warst case scenario. In this case,
scenario means one triple (a,b,P)EZ. Therefore, given two seenarios repre-
sented by triples (a~,bJ,P 1 ),(az,b 2 ,Pz)EZ, (ahbJ,PJ) represents a scenario
worse than that represented by (a 2,b2,P 2) if
a!EPJ[-X]+b1> azEPZ[-X]+ bz.
Theorem 10. A risk measure p satisfies monotonicity, convexity and
conservatism if and only if there exists Z~R+ xRxP, where P is a family of
probability measures on the set of states of nature n, such that
I
p(X)=sup{aEp[-X-]+b (a,b,P)EZ}.
Remark 15. The risk measure considered in Theorem 10 can be inter-
preted as the insurance premium to cover a positive linear transformation
of the expected loss in the warst case scenario, but taking into considera-
tion only negative outcomes, i.e. aEp[-X ]+b. In this case, scenario means
 383

one triple (a,b,P)EZ. Therefore, given two seenarios represented by triples

(a~,b~,P 1 ),(a2 ,b 2 ,P 2)EZ, (a~,b~,PI) represents a scenario worse than that
represented by (az,bz,Pz) if
-
a1EPI[-X ]+b1> azEpz[-X ]+ bz.
-
Theorem 11. [Carr et al. (2002), Föllmer et al. (2002a), Föllmer et al.
(2002b), Frittelli et al. (2002)] A risk measure p satisfies convexity and
translation invariance if and only if there exists Z~RxP, where P is a
family of probability measures on the set of states of nature n, such that
I
p(X)=sup{Ep[-X]+b (b,P)EZ}.
Remark 16. The risk measure considered in Theorem 11 can be inter-
preted as the insurance premium to cover the expected loss in the worst
case scenario up to an additive constant. In this case, scenario means one
pair (b,P)EZ. Therefore, given two seenarios represented by pairs
(b~,P 1 ),(b 2 ,P2 )EZ, (b~,P 1 ) represents a scenario worse than that represented
by (bz,Pz) if

Theorem 12. A risk measure p satisfies monotonicity, convexity, con-

servative translation invariance and conservatism if and only if there exists
Z~RxP, where P is a family of probability measures on the set of states of
nature n, such that
I
p(X)=sup{Ep[-X-]+b (b,P)EZ}.
Remark 17. The risk measure considered in Theorem 12 can be inter-
preted as the insurance premium to cover the expected loss in the worst
case scenario up to an additive constant, but taking into consideration only
negative outcomes, i.e. Ep[-X]+b. In this case, scenario means one pair
(b,P)EZ. Therefore, given two seenarios represented by pairs
(b~,P 1 ),(b 2 ,P2 )EZ, (b~,P 1 ) represents a scenario worse than that represented
by (bz,Pz) if
- -
Ep1[-X ]+b1> EPZ[-X ]+ bz.

7 Conclusions

In this paper we investigated risk measures and their properties. Firstly, we

recalled the most well-known risk measures: VaR and the coherent risk
measures. Our starting point were the following remarks with respect to
risk measures:
384

1) in our opinion, is quite questionable that risk measure is determined

also by positive outcomes of investments: to evaluate the per-
formance of an investment, both negative and positive outcomes
must be considered; but, to evaluate the risk of an investment, is
quite reasonable and concordant with "common sense" to consider
only negative outcomes;
2) in our opinion, is quite questionable to consider as an admissiblerisk
measure only a function satisfying the properties characterizing
coherent risk measures (monotonicity, positive homogeneity, trans-
lation invariance and subadditivity); we think that further desirable
properties could be very usefully introduced and justified by the
financial point of view.
On the basis of these premises, we proposed a new class of risk
measures, called robust risk measures. With respect to coherent risk
measures, robust risk measures are characterized by quite different axioms
(conservatism, sure loss principle, continuity and robustness ), reflecting
each one some particular claim, easy to be understood and accepted.
Apart from their financial interpretation, we showed that the robust risk
measures have a specific interest also within fuzzy sets theory and measure
theory; actually they are the class of the risk measures that can be
computed as the Sugeno integral of the possible Iosses of a project. In
addition, there is also another interest in these risk measures within fuzzy
set approach: indeed the axioms of robust risk measures give a specific
characterization of Sugeno integral. Due to important financial meaning of
the axiom of subadditivity, we particularly investigated also the relations
between Sugeno integral and this property, obtaining a characterization of
subadditive Sugeno integral.
Finally, by slight modifications of the definition of coherent risk
measures, some other possible formulations of risk measures and their
axiomatic bases have been stated, giving particular emphasis to the
property of conservatism.
As a final message ofthe main results oftbis study, we like to underline
that a Iot of new and interesting risk measures can be defined, consistently
with different and desirable financial properties. In our opinion, there is
not a privileged class of measures of risk, but any of them reflects
particular and reasonable financial meanings. Thus it is very important to
carefully analyse the axiomatic premises of considered risk measures in
order to choose and use in each occasion the most appropriate for the
problern at hand.
 385

References

Artmer P, Delbaen F, Eber J-M, Heath D (1997) Thinking Coherently. RISK 10:
68-71
Artzner P, Delbaen F, Eber J-M, Heath D (1999) Coherent Risk Measures.
Mathematical Finance 9 (3): 203-228
Bonaccorso L, Greco S, Matarazzo B (2002) Robust measures of risk. University
ofCatania
Carr P, Geman H, Madan DB (2002) Pricing and hedging in incomplete markets.
Journal ofFinancial Economics 62: 131-167.
Föllmer H, Schieid A (2002a) Convex measures of risk and trading constraints.
Finance and Stochastics 6: 429-447
Föllmer H, Schieid A (2002b) Robustpreferencesand convex measures of risk.
In: Sandmann K, Schönbucher PJ (eds) Advance in Finance and Stochastics.
Springer-Verlag, Berlin, pp 39-56
Frittelli M (2002) Representing sublinear risk measures and pricing rules, working
paper 10, University ofMilano Bicocca
Frittelli M, Rosazza Gianin E (2002) Putting order in risk measures. Journal of
Banking and Finance 26: 1473-1486
Greco S, Matarazzo B, Platania P (2002) Conservative Coherent Measures of
Risk. University of Catania
Risk Special Supplement. Value at Risk. (1996) Risk Magazine. June
RiskMetrics. (1995) Technical Document Morgan Guarantee Trust Company,
Global Research, New York
Studer G (1997) Maximum Loss for measurement of Market risk. PhD thesis,
ETH Zurich. http://www .ifor.math.ethz.ch/pub/diss/1997-1 00-studer-diss.pdf
Sugeno M (1974) Theory of fuzzy integrals and its applications. PhD thesis,
Tokyo Institute ofTechnology
Discovery in the New Economy - Why
Entrepreneurs may not Contract with Investors

Helmut M. Dietl

EgonFranck

Stefan Winter

1 lntroduction

As highlighted by Figure 1, New Economy fmns usually run through three

typical development stages: development of a business plan, venture capi-
tal investment and Initial Public Offering (IPO). Bach single development
stage typically coincides with a specific phase of the innovation process.
During the discovery phase, an entrepreneur finds a new business opportu-
nity and develops a business plan. In the New Economy, this phase is usu-
ally not financed by investors. The entrepreneur typically stays self-em-
ployed. Investors enter the scene only after a new business plan has been
discovered. The first investors are venture capitalists, who verify the eco-
nomic potential of the business model. Usually, only a small number of
business plans passes the verification test and is actually financed by ven-
ture capitalists. Of the business plans that have passed the verification test,
again only a very limited number reaches the public capital market as an
IPO. It typically takes several years until a business plan reaches the
capital market and public investors finance the large scale exploitation of
this business plan.

Discovery Verification Exploitation

Business Venture
~
IPO
Model Capital
~

Fig. 1. Typical Development Stages ofNew Economy Firms

 387

There is an exploding amount of literature dealing with venture capital

investment and IPOs. 1 Our paper will not add to this literature. Instead, we
will exclusively focus our analysis on the discovery stage and try to
answer a preliminary question: Why do entrepreneurs in the New Econ-
omy stay self-employed during the discovery stage? Or, to put it dif-
ferently: Which factors produce incentives for entrepreneurs not to con-
tract with investors of any kind in the discovery phase of the innovation
process?
This question is relevant because there are alternatives to self-employ-
ment during the discovery stage. In the Old Economy, for example, many
firms spend substantial amounts of their cash flows on R&D activities. 2
These R&D activities usually also encompass the discovery of new busi-
ness opportunities, such as new products, new technologies, new distribu-
tion channels, new forms of organization etc. Accordingly, discovery in
the Old Economy is funded by investors and conducted by employees such
as engineers, scientists, designers etc .. In addition, many owner-dominated
firms in the Old Economy perform whole innovation cycles from dis-
covery to exploitation "in house." Again, in these firms discovery is done
by employees who contracted with the owner of the firm as the relevant
investor. Why is discovery in the Old Economy financed by investors
whereas entrepreneurs in the New Economy remain self-employed because
they are not willing or not able to contract with investors of any kind? Our
paper tries to answer this question based on a semi-formal analysis.
The remainder of our paper is organized as follows. Section 2 develops
our semi-formal analysis starting from rather restrictive assumptions and
working our way through towards a more realistic scenario. At the end of
this section, we identify the conditions under which entrepreneurs will not
contract with investors in the discovery stage of the innovation process.
Section 3 explains why these conditions are more likely to be met by busi-
nesses in the New than in the Old Economy. Hopefully, this discussion
also sheds some light on the still blurred phenomenon New Economy.

1 See e.g. Bartlett (1988), Sahiman (1990), Megginson!Weiss (1991), Hellmann

(1998), Weimerskirch (2000), Schmidt (2002).
2 e.g. Siemens 4.6 billion €, Bayer 2.3 billion €, General Electric 2.2 billion US$,
Daimler Chrysler 7.4 billion € (all figures for 2000) See the respective annual
reports at www.siemens.com, www.bayer.com, www.ge.com, and
www.daimlerchrysler.com.
388

2 The problern of discovery

We start by assuming that discovery is the sole problem. Once discovery

happened, all payoffs are distributed and the game ends. The value of
business opportunities does not depend on further exploitation activities.
Assurne that there is a set B ={B;}, i =I, ... ,! ofbusiness opportunities not
discovered so far. Bach opportunity is unique. On the other hand, there is a
population of potential entrepreneurs. Bach of those entrepreneurs is char-
acterized by a technical skill. The set of skills available in the population
of entrepreneurs is S = { Sj} , j = n, ... , J. There may be only one or there
may be more entrepreneurs of skillj available in the population. In order to
be able to discover business opportunity B; , the entrepreneur needs skill
S; . Given an entrepreneur of skill i, she has to invest an amount of C; to
find the opportunity. As a point of reference, think of this amount as the
cost of living of the entrepreneur while searching for the opportunity. Ad-
ditionally, the cost ofliving is at the sametime the net present value ofthe
reservation wage of each entrepreneur. Any one of them is able to leave
the discovery market and earn a wage of W; = C; , leaving her with a total
utility of U; = W; - C; =0 .
Once the discovery has been made, the gross present value of the op-
portunity B; is V; . The net present value of the opportunity therefore is
V; = v; - C; . The net present value may or may not be positive. Anyone in-

volved can invest or borrow at the risk free interest rate r. The timing is as
follows. The cost of living has to be spent at t =0 . The gross value of a re-
alized opportunity is v; (I + r) and is received at t = I .
The question arises who will finance C; and who will get the returns of
the discovery. We will start to explore this question by making some very
strong assumptions. Later on we will Iook at the consequences of relaxing
those assumptions.

2.1 Common knowledge without incentive problems

Under this regime, we assume that everything is common knowledge. All

entrepreneurs and all potential investors know the business opportunity set,
the gross value of each opportunity, the skills of each entrepreneur, and the
costs of living. There are no transaction costs. Risk preferences play no
role due to a Iack of randomness in any of the variables. Once an entrepre-
 389

neur of skill S; starts to search for the opportunity, she will find it with
certainty. The process of search itself does not cause any disutility of ef-
fort. Therefore, there is no incentive problern in the search process. The
consequences of these assumptions are straightforward. Any possible dis-
covery with a positive net value will be rnade. Possible are those dis-
coveries for which a corresponding skill exists in the population of entre-
preneurs. The costs of living as well as the returns frorn the discoveries are
bome by the entrepreneurs. There is no need for investors in this setting.
If n > 1 , then there are opportunities which can not be discovered due to
a lack of corresponding skills in the entrepreneurial population. If J > I
then there are skills in the entrepreneurial population that can not be used
in discovering new opportunities. Since it is cornrnon knowledge that those
skills are useless in the discovery rnarket, no one would want to pay the
cost of living for such an entrepreneur to stay in that rnarket. Entrepreneurs
with useless skills will leave the rnarket and eam their reservation wages
sornewhere eise. lt follows that only entrepreneurs with skills needed to
rnake a discovery will stay in the rnarket. Whether all of thern would want
to stay depends on whether their skill is unique or not and whether the net
present value of their possible discoveries is nonnegative. Since no one
would want to finance a discovery with a negative net present value, we
can lirnit further argurnents to opportunities B; of nonnegative net present
values.

2. 1. 1 Unique ski/ls
First of all, consider an entrepreneur of useful skill S; which is unique. If
the entrepreneur invests in his cost of living and keeps all returns, her total
utility would be V;. Since V; ~ 0, her utility of staying in the discovery rnar-
ket is at least as high as her utility received when leaving that rnarket. She
will stay and rnake the discovery. V; is her retum to skill, which is
nonnegative and can only be earned in the discovery rnarket.
An investor would only want to pay the cost of living when his return
frorn doing so exceeds the risk free interest rate r. Assurne that the investor
hires the entrepreneur, pays her cost of living and keeps a share of
0 < a ~ 1 of the gross present value v;, leaving a share of ( 1- a Jv; to the
entrepreneur. The entrepreneur would be better off cornpared to leaving
the rnarket. However, the investor would require that a v; - C; > 0 . If the
entrepreneur would stay in the rnarket and pursue the discovery on her
own, she would end up with a total utility of V; • If she accepts the contract
390

with the investor, she will end up with ( 1- a )V; . Therefore, she will only
accept the contract as long as ( 1- a )V; > v;. Since by definition
V;= V; -C;, the condition becomes ( 1- a)V; >V;- C, or C; -aV; > 0.
Since the investor requires aV;- C; > 0 and the entrepreneur requires
C; - a V; > 0 , there is no feasible contract. The entrepreneur will always
pursue discovery on her own.

2.1.2 Competing skills

Things will not change significantly when competition is introduced. Re-
member that there may be more than one entrepreneur in the population
with skill S; . However, since opportunities are defined to be unique, the
market for discoveries needs only one manger of skill S; . Assuming
symmetrical equilibrium, the entrepreneurs of skill S; will simply split up
v; between themselves. One of them stays in the market and buys out the
others. The arguments developed above still hold. There is no room for in-
vestors.

2.2 Asymmetrie information

The assumption of common knowledge of all variables is more than ques-

tionable. Introducing private information might Iead to much more realistic
descriptions of the problern. It would be more realistic to assume better in-
formed entrepreneurs. Better information should be expected with respect
to business opportunities and/or skills. Furthermore, the process of search
suggested above is hardly convincing. The probability of discovery should
depend on the intensity of search by the entrepreneur. The intensity then
should require costly efforts by the entrepreneur, effort not being observ-
able by investors. Thus, financing by investors creates an incentive prob-
lern. Additionally, when uncertainty is introduced, risk preferences will
start to play a role. Investors are assumed to be risk neutral, while entre-
preneurs may be risk averse. We start by assuming that entrepreneurs can
observe allvariables with certainty.

2.2.1 Uncertain business opportunities without incentive

problems
Assurne that investors know the set of potential business opportunities.
However, while an entrepreneur of skill S; knows with certainty whether
 391

there exists an opportunity Bi which she will be able to discover at fixed

costs Ci , investors believe that this opportunity is available only with
probability q; < 1 . They do not know for sure whether a given opportunity
really exists. Any formal argument is obviously futile. Since investors do
not know whether an opportunity actually exists, they will only finance the
trial of discovery when they receive an additional return covering the ex-
pected loss from nonexistent opportunities. As shown above, even without
that additional return, entrepreneurs will prefer to pursue discovery on
their own.

2.2.2 Uncertain skills without incentive problems

Again, assume that investors know the set of potential business opportuni-
ties. Any of those is known to exist with certainty and can be discovered at
a fixed cost Ci by an entrepreneur of skill Si. However, whether a given
entrepreneur does or does not have the specific skill can not be verified
with certainty by investors. The result will be the same as above. Entrepre-
neurs will never need investors to finance discoveries and share their re-
turns to skill. It follows that any entrepreneur trying to contract with an in-
vestor can only have a useless skill. But entrepreneurs with useless skills
will have no incentive to lie. Their costs of living is paid for by the inves-
tor. Additionally, they get a share of (1- a)V; of the gross present value.
However, lacking the needed skill, the discovery will not be made, i.e.
V; =0 . The entrepreneur ends up with a total utility of zero which she can
also achieve in her alternative employment. Therefore, she has no incen-
tive to lie.

2.2.3 Uncertainties with incentive problems

The introduction of incentive problems is easily modeled by introducing
the necessity of search effort. In order to find the opportunity, the entre-
preneur has to spend costly effort on search activities. But then, the re-
sulting incentive problern is best solved by letting all entrepreneurs pursue
their discovery projects on their own. Again, no investors are needed.

2.2.4 Risk averse entrepreneurs

So far, entrepreneurs were assumed to know all relevant variables with
certainty. In such a setting, risk preferences of the entrepreneurs play no
role. In what follows, the entrepreneur has only incomplete information.
We will focus on a situation in which the entrepreneur does not know for
392

sure whether there exists an opportunity she can discover with her skill.
She then has to compare her expected utility of staying in the discovery
market with her alternative employment utility of zero. We focus on op-
portunity B; and skip subscripts. To keep things simple, assume again that
the cost of discovery is C and does not depend on any kind of search ef-
fort. The entrepreneur believes that the opportunity B exists with
probability p. In what follows, we assume that her utility of wealth w is
U ( w) =1- EXP(- w) , the utility function being common knowledge. If
she pursues on her own, her expected utility of staying in the discovery
market is
E(U) = p(1-EXP(-v J]+(J- p)(1-EXP(CJ]

She will stay as long as E(U) ~ 0. lf she contracts with an investor, she
will receive an expected utility of
E(U1_ 0 ) =p(1-EXP(-(J -a)V J]+(J- p J[1-EXP(OJ]
=p(1-EXP(-(J -a)V J]
She will strictly prefer the contract as long as E(U1_ 0 ) > E(U). She
will prefer the contract to leaving the market when E (U1_a) ~ 0 . Given
p ~ 0, a ~ 1 and V~ 0, the second condition holds anyway.
The investor believes that the opportunity exists with probability q. He
would like to invest whenever qa V ~ C . If investors compete for invest-
ment opportunities, market equilibrium would drive a down to a level
where qaV =C or

a=-.
c
qV
We ignore cases in which a > 1 would result. We can do so because
a ~ 1 would imply a non-positive expected net present value of the busi-
ness opportunity from the investors point ofview. He would never want to
finance such a project.
It is Straightforward to see that under this regime contracting will always
happen as long as the investor and the entrepreneur share the same beliefs.
 393

c
Assuming p =q implies a =- . The expected wealth of the entrepre-
pV
neur without contracting is E(w)= p(V -C)+(l- p)(-C)= pV -C,
while the expected wealth under the contract is
E(w1_a)=p(l-a)V+(l-p)(O)=pV-C. So both Situations provide
the same expected wealth, while the situation with the contract does pro-
vide that expected wealth with less risk. Therefore, with shared beliefs the
risk averse entrepreneur will always contract up to a level of p =1 , where
she is indifferent.
Additionally, even with differing beliefs, contracting will always happen
as long as the investors belief in the chances of success is higher than the
entrepreneurs belief, i.e. q > p . In that case, the contract is not only less
risky, it even carries a higher expected wealth for the entrepreneur.
Only when the entrepreneur believes more strongly in the chances of
success, contracts may not be feasible anymore. If p > q , then
E ( w1_a) < E ( w) and there is a trade-off between expected wealth and
insurance.
The emergence of conflicts of belief is not modeled here. If beliefs
satisfy p > q , then comparative static analyses yield the following results.
The entrepreneurs inclination to contract declines:
• as p increases. lf p =1 the entrepreneur will always pursue the
process of discovery on her own.
• as q decreases. Decreasing q implies that the investor reduces his
estimated probability of success. He therefore would require an in-
creased share of V. From the entrepreneurs perspective that means
that the cost of capital increases.
• as V increases (C decreases). The value of pursuing on her own in-
creases. Consider the extreme example where C is zero. Then,
there is no initial investment needed. The entrepreneur will never
share V with an investor in exchange for an initial investment of
zero.
• as the degree of entrepreneurial risk aversion (not modeled above)
declines.
Basedon our arguments developed above, we can't rule out the possi-
bility of contracting, unless entrepreneurs and investors have different be-
liefs with p > q . Whenever this is true, limited risk aversion and limited
initial outlays C willlead to the non-contracting solution.
394

3 Why discovery in the new economy may be different

Are there reasons to assume that conflicts of belief with p > q will sys-
tematically emerge in the New Economy? Moreover, are there reasons to
believe that, on average, V will be higher and/or C lower for business pro-
jects in the New Economy?
In order to illustrate our arguments we introduce the following meta-
phor.3 Let us compare entrepreneurs to fishers. The business field they
explore for projects corresponds to a pond with a restricted population of
fish. Fishers may choose to fishin ponds already known (Old Economy) or
move on to try out new ponds (New Economy). Whatever the fish popula-
tion in a pond may be at the beginning, continuous fishing activity is likely
to have the effect that fish caught later will be smaller. If fishers have a
choice, they take big fish first. Moreover, big fish is easier to catch
because less sophisticated technology is required to detect it. The Ionger
fishing activity goes on at a pond the more sophisticated technology will
be needed to catch ever smaller fish. Improving fishing technology is
costly. Fishers (entrepreneurs) will face increasing costs Cto catch smaller
fish (discover projects of lower value V). Consequently, fishers (entrepre-
neurs) will prefer to finance their "fishing gear" by investors in exchange
for a share in the "catch." Similarly, entrepreneurial search becomes less
productive at "old ponds" because many profitable ideas have already been
"taken out" and are being exploited. To the extent that old ponds are being
"over-fished" there comes a point when new pondswill provide entrepre-
neurs with higher V and lower C, no matter what their initial "fish popula-
tion" is. The chance to catch "bigger fish" (higher V) with "less effort"
(lower C) in the New Economy not only has to do with favorable "fish
growing conditions" in the "new ponds," but follows from the exhaustion
of discovery opportunities in the Old Economy.
Firms in the Old Economy have run through innovation cycles re-
peatedly. At least apart of the assets built up in the stages subsequent to
the discovery of a business opportunity are reusable in later innovation cy-
cles. These assets are specific to the business field of the firm but not to
the single innovation. Or to stay within the fishing metaphor: They are
specific to the pond, not to the single fish. A common example is the case
of a pharmaceuticals firm who built up resources and capabilities to man-
age the approval process of new drugs through the FDA regulation. Such
assets that are reusable in subsequent innovation cycles are called com-

3 See Franck/Jungwirth (2001) who use this metaphor in their analysis of

research activities.
 395

plementary assets. 4 Obviously, the value of a certain discovery is sys-

tematically higher for a firm already holding the complementary assets
suited to verify and exploit the discovery as compared to an entrepreneur
who would have to build up these assets from scratch.
Now a crucial difference between businesses in the Old and New Econ-
omy should be obvious. Because the "ponds" are new, there are no estab-
lished firms with complementary assets in the New Economy. Without
"pond specific" assets the value of a discovery is not systematically higher
for an established firm than for an entrepreneur. In contrast to this, the en-
trepreneurs of the Old Economy have a systematic disadvantage compared
to the established firms in control of the complementary assets. They are
only able to draw a lower V (respectively would have to invest higher C)
from the same discovery. In our pharmaceuticals example the discovery of
a formula for a new drug may actually be worthless for an entrepreneur
without the resources and capabilities to turn it into an approved drug. En-
trepreneurs in the New Economy pursue discovery on their own more
often because there are no established firms which can simply boost the
value of discoveries without additional cost simply through pre-established
ownership of complementary assets.
N ow what about differences in p and q? When fishing gets started at a
new pond, fishers and potential investors will build their expectations on
subjective probabilities of success. At the beginning, these expectations
may diverge substantially. The Ionger fishing is going on at a certain pond,
the more these initially subjective probabilities of fishers and investors will
converge, based on common experiences. Therefore, we should expect a
convergence of the success expectations of entrepreneurs and investors in
a certain business field over time. Following from this logic, conflicting
expectations between entrepreneurs and investors are more likely to be
found with regard to success probability of fishing in the "unexplored
ponds" of the New Economy than in the "over-fished ponds" of the Old
Economy. As a result, conflicts of beliefs should be more frequent for
projects in the New Economy than in the Old Economy.
Conflicts of belief allow for differences between p and q in both direc-
tions. However, there are arguments why p > q should occur more often
in the New Economy than p < q . In the Old Economy investors can infer
q from the observation of past fishing activities at certain ponds. This way
they get to know the business opportunity set as well as the skills of certain
fishers. In the New Economy such observation of past fishing activities is
not possible. Therefore, it is plausible to assume that entrepreneurs are

4 See e.g. Besanko/Dranove/Shanley (2000), p. 498.

396

better informed with respect to their skills and also with respect to the
business opportunity set. Metaphorically speaking, entrepreneurs in the
New Economy are the first fishers who come and inspect new ponds. They
know better about the fishing possibilities and also about their skills than
investors who have no information from observations in the past. Conflicts
of belief can simply be the result of asymmetric information because the
party with less information has reasons to be more prudent. p > q is a stan-
dard outcome if uninformed investors "discount" the future opportunities
more heavily than better informed entrepreneurs.
The Iiterature dealing with decision anomalies 5 provides additional argu-
ments why p may be systematically higher than q in the New Economy. It
is, for example, a well-known anomaly studied in lotteries, that people
overestimate the probability of winning the greater the winner's prize
becomes. 6 If businesses in the New Economy promise "bigger deals" for
entrepreneurs than the "over-fished ponds" of the Old Economy, then this
"Iure of the jackpot" 7 can perhaps explain why entrepreneurs in the New
Economy may be particularly overconfident. Yet, why should investors be
immune against this anomaly? A second strain of literature8 on decision
anomalies may help to explain the difference between entrepreneurs and
investors. V arious experimental sturlies find out that overconfidence is
linked to the "illusion of control." People with partial influence on a
process are more likely to ignore risks than people with no influence at all.
The same holds if people feel competent in the field of endeavor. There-
fore, entrepreneurs performing the search activities may be more ready to
assume that they are "in control" of the whole discovery process or more
"competent" that outside investors.
Finally, there is a selection bias in favor of less risk-averse entrepre-
neurs in the New Economy. Again, consider our fishing metaphor. lf
fishers have a choice between fishing in known ponds with well-known
success probabilities and unknown ponds with unknown success prob-
abilities, highly risk-averse fishers will prefer to fish in known ponds while
less risk-averse fishers (entrepreneurs) will be less reluctant to try out new
ponds. Accordingly, entrepreneurs in the New economy will be less risk-
averse than entrepreneurs (or rather employees) in the Old Economy.

5 See Conlisk (1996) for an overview.

6 See Cook/Clotfelder (1993), Frank/Cook (1995).
7 See Cook/C1otfe1der (1993), p. 642, for this metaphor.
8 See for example Howell (1971), Langer (1975), Weinstein (1980),
Lichtenstein//Fischhoff/Philips (1982). For a discussion of the "competence
effect" see e.g. March/Shapira (1987), Heath/Tversky (1991).
 397

References

Bartlett JW (1988) Venture Capital - Law, Business, Strategies, and Investment

Planning, New York et al. 1988
Besanko D, Dranove D, Shanley M (2000) Economics of Strategy, 2nd edition,
New York et al. 2000
Conlisk J (1996) Why bounded rationality? Journal of Economic Literature 34:
669-700
Cook PJ, Clotfelder CT (1993) The Peculiar Scale Economies of Lotto. In: The
American Economic Review 83: 634-643
Frank RH, Cook PJ (1995) The Winner-Take-All Society, New York 1995
Franck E, Jungwirth C (2001) Warum geniale Ideen für wissenschaftlichen Erfolg
nicht ausreichen, Chair of Strategie Management Working Paper Series,
University of Zurich 2001
Howell WC (1971) Uncertainty from Internaland External Sources: A Clear Case
ofüverconfindence. Journal ofExperimental Psychology 89: 240-243
Langer EJ (1975) The Illusion of Control. In: Journal of Personality and Social
Psychology 32:311-328
Lichtenstein S, Fischhoff B, Philips L (1982) Calibration of probabilities: The
state of the art to 1980. In: Kahnemann D, Slovic P, Tversky A (eds.)
Judgement under uncertainty: Heuristics and biases, Cambridge MA, pp 306-
334
March JG, Shapira Z (1987) Managerial Perspectives on Risk and Risk Taking. In:
Management Science 33: 1404-1418
Heath C, Tversky A (1991) Preference and Belief: Ambiguity and Competence in
Choice under Uncertainty. Journal ofRisk and Uncertainty 4: 5-28
HellmannT (1998) "The allocation of control rights in venture capital contracts,"
RAND Journal ofEconomics 29: 57-76
Megginson WL, Weiss KA(l991) "Venture capitalist certification in initial public
offerings," Journal ofFinance 46: 879-903
Sahiman W A (1990) "The structure and governance of venture-capital organi-
zations," Journal ofFinancial Economics 27: 473-521
Schmidt KM (2002) Convertible Securities and Venture Capital Finance, Uni-
versity ofMunich, CESifo and CEPR, March 21,2002
Weimerskirch P (2000) Finanzierungsdesign bei Venture-Capital-Verträgen, 2.
aktualisierte Auflage, Wiesbaden 2000
Weinstein ND (1980) Umealistic Optimism About Future Life Events. Journal of
Personality and Social Psychology 39: 806-820
Japan's Venture Capital Market from an
lnstitutional Perspective·

Wemer Pascha

Stephan Mocek

1 Overview

Can Japan mobilise enough venture capital to finance its promising ven-
ture firms and to support its advance into high-tech? This is one of the
most important strategic issues facing the Japanese economy. Where are
the bottlenecks; what can be done to overcome remaining problems? In the
following section, we will first present some stylised facts to substantiate
the contention that Japan does have a problem. Afterwards, we will intro-
duce the theory-based viewpoint based on institutional economics 1 and
show why Japan still has difficulties to effectively process venture capital.
We survey changes taking place and Iook for open policy questions.
We will argue that Japan is moving towards more functional venture
capital financing 2, but that there are important institutional inconsistencies
left. In particular, the consequences of the Japanese-type network society,
understood as a set of sticky informal institutions, cannot easily be over-
come by govemment pro-activism or by the heroic establishment of arms'-
length capital markets. Rather, Japan's financial systemwill remain hybrid
in the sense of relying both on relational contracts typical of credit-based
systems and on the more explicit contracts of arm' s-length markets.

• The authors gratefully acknowledge helpful remarks from their commentator,

Dr. Martin Hemmert. An extended conference version ofthe paper is available
under http://www.uni-duisburg.de/FB5NWL/OA WI/ARBEITSPAPIERE/
AP64.pdf
1 While there are numerous sturlies on venture capital in Japan (e.g. Hamada

2002; Hosokawa and Sakurai 2000; Imai and Kawagoe 2000; Kutsuna 2000;
Storz 2000), few employ a consistent institutional approach.
2 As for relevant terms, venture businesses are understood as knowledge-

intensive and innovative, rather small companies; they need not necessarily be
young. Venture capital or risk capital earmarked for such businesses can also
encompass secured loans, in a Japanese context. Venture capital (VC) firms or
.funds are specialised entities to supply funds to venture businesses.
 399

2 The growing importance of venture capital and start-

ups in Japan - some stylised facts

From the viewpoint of growth theory, almost the only path left for
advanced economies to enjoy long-term growth is to realise gains through
the input of frontier technologies and through a rising total factor produc-
tivity (Pascha forthcoming). From this perspective, financing technology-
related endeavours is a critical pre-condition for realising intensive growth.
Empirically, while Japan has one ofthe largest ratios ofR&D input among
advanced economies, R&D efficiency still seems to be lacking, if inputs
are compared to frequently used output3•
As the introduction of new technology in a frontier economy is always
particularly risky, an important task of the financial system is to
adequately supply venture capital - both in terms of magnitudes and in
efficiency terms. Traditionally, Japan has rather relied on its major compa-
nies to introduce and divert new technologies. This is due to the legacy of
a "dual system", in which the modern, larger-scale industries introduced
technologies from abroad and modified them, whereas small and medium
scale industries offered cheap and flexible, low-tech support in hierarchi-
cally organised industrial systems. This approach may have been
extremely successful during the catching-up phase, but it has reached its
Iimits in Japan's current situation. Indeed, younger and smaller companies
have tended to grow faster in Japan (Imai and Kawagoe 2000: 117) and
one might hope to tap this potential even more. For example, the compa-
nies listed on the three Japanese stock markets for ernerging firms plan to
increase recruiting in 2003 by more than 40 percent, even in the currently
tough business situation (Nikkei 2 November 2002).
Unfortunately, the number of start-ups as related to the total number of
enterprises has shown a remarkable decline since the 1980s (Fig. 1). While
there have been fewer simple "extensions of the work bench", i.e. setting
up another low-tech establishment by a former employee, there has not
been a compensating increase in the number of dynamic, venture start-ups.

3 This is not to deny that Japan has made some considerable progress with
respect to R&D output in recent years (Hemmert 2002).
400

This development seriously undermines the benign mechanism of "crea-

tive destruction" (Schumpeter), which pushes the economy forward
through high entry and exit rates 4 • Also according to other measures, the
level of start-up activity in Japan is unsatisfactory (e.g., Reynolds et al.
2001).

7.0 -
6.0 5.9
5.6
5.0
4.0
3.5
3.0
2.0
1.0
0.0
1975-78 1978-81 1981-86 1986-91 1991 -96 1996-99

-+- tv'arkel entry rate(%) - - - tv'arket exilrate (%)

Fig. 1. Entry and exit

Note:
1. Average market entry rate = average number of new establishments in the
surveyed period I total number of companies at the beginning of the period *
100 (calculation of market exit rate analogous)
2. Calculation includes individual companies, no companies ofthe primary sector
Source: KKK 2001 a: 4
As for the reasons for this low level of new venture businesses, such
firms find it particularly hard to gain access to capital. According to a
survey, 80.6% of venture businesses consider this as their major problern
since foundation. The cost for setting up a start-up in Japan is con-
siderable: in 2001, 15.8 million Yen (142,042 Euro) were necessary on
average (KKK 2001b: 6).

4 From this viewpoint, the rise in exit rates during the latter 1990s (Fig. 1) is not
necessarily bad from a Ionger-term perspective. While such exits Iead to
unemployment and social concems in the short run, they weed out week,
unsustainable, outdated businesses. In this context, it should also be noted that
Japan has approximately an average survival rate for entries in international
comparison.
 401

We are thus led to investigate the problems of venture business

financing in Japan as a key issue for the future of the Japanese economy,
although it should be stressed that other factors like access to the labour
market or to strategic information are also critically important. To some
extent, the financial problems are straightforward. The first point is the
current recession, which has increased the risk awareness of savers and
intermediaries as well as decreased the demand and profit expectations of
business. Secondly, the current recession in Japan is peculiar, because it is
combined with a crisis in banking (bad loans, necessity of special depre-
ciation, shrinking capital base in the wake ofBasle li, etc.). Consequently,
outstanding credit as an aggregate has been declining for years. Thirdly,
new channels for venture business financing have to be sought against the
world-wide background ofa retreating market for growth stocks. However,
problems with venture financing in Japan precede the topical issues by
years, if not decades (e.g. Turpin 1986). This makes it necessary to look
more systematically, because the problems cannot be expected to vanish
once the recession is over.

3 The adequate financial system for promoting venture

businesses - an institutional issue

In a neo-classical environment with perfect markets, it is irrelevant under

which financial system companies raise funds (Albach et al. 1986: 6). This
does not hold when market imperfections like information asymmetries,
information and transaction costs are taken into consideration.
Venture businesses are particularly characterised by major insecurity,
high risk, and strong information asymmetries (Gebhardt and Schmidt
2002). Investment typically is highly specific, incorporated in the entre-
preneur and difficult to liquidate; this raises moral hazard concems and
other typica1 principal-agent issues (adverse selection, hidden intentions,
hidden characteristics).
Basically, two sets of institutions (as rules of a game) and organisations
(as players within such agame) are employed in this context: the credit-
based fmancial system operating with banks as intermediafies and the
capital market-based system, directly linking savings and investment
through stocks, bonds, etc. The perceived characteristics, strengths and
weaknesses ofboth systems are summarised in Table 1.
402

Table 1. A synopsis of the credit-based and the capital market-based financial

system
Criterion Credit-based fmancial Capital market-based
system financial system
Basic character Indirect Direct
Type of interaction Dense hybrid relations - Arms' -length markets for
banks as intermedianes stocks, bonds, etc.
Contracts Relational (plus classical) Classical: explicit/complete
Character of (Inter-) personal Anonymous
relations
Type of information Tacit Open, explicit
Risk attitude Risk averse Intermediation of risk
structures
Investment horizon Long-term Short-term
Business receiving Conventional High-tech, ernerging
favourable treatment industries
Suitable stage of Catching-up Advanced, frontier
economy
Type of socio- Network-based (frequent Individualistic (frequent use
economy use of generalised ofbalanced exchange)
exchange)
Control mechanisms Reputation, sunk costs, Rule-of-law, independent
hostage (supervisory) agencies

lt is generally argued that a capital market-based system is more

adequate for venture businesses. The principal reason is that banks will
have a risk-averse lending attitude which does not fit the necessities of
high risk ventures. Japan currently has a credit-based financial system,
which may have been very sensible for the catching-up process of the
earlier post-war period, but is inefficient for its current status as an
advanced economy in need of more venture capital.
These considerations can be summarised in two hypotheses:
Hypothesis 1 (Hl): The main reason for the Iack of venture capital
financing in Japan is the dominance of the credit-based financial
system.
Hypothesis 2 (H2): Japan, as an advanced economy, has to change- or
will eventually - transform its financial system to become more capital
market-based.
H2 is obviously based on the popular view of convergence. In a norma-
tive interpretation, Japan's traditional economic system should adopt
elements of an "Anglo-American type" economy if it wants to defend its
place among the leading nations.
 403

One might want to take issue with the convergence hypothesis (H2),
though. Financial systems are embedded in a wider institutional system,
including sticky informal (social) institutions5• Under such circumstances,
a capital market in Japan - with its peculiar set of informal institutions -
may have a different functionality as compared to another environment,
e.g. the US. Specifically, Japan's socio-economy is characterised by a far-
reaching network of relational, implicit contracts. Such contracts have
certain advantages; for instance, they allow the transfer of tacit information
and can serve as the basis of highly specific, complex investments.
However, their usual disadvantage isthat agency costs are conspicuously
high due tomoral hazard. In Japan's case, however, such costs arerather
low due to a couple of factors: (a) Society is rather homogenous which
assists the flow of tacit information; this also reduces the danger of
shirking, because such behaviour has a high probability of being detected.
(b) Defection ("exit") from relational contracts is rather difficult because
of the tightly woven social fabric. (c) The ubiquity of relational contracts
(on historical, cultural grounds) creates positive network extemalities:
investment in personal reputation can be used in a variety of social and
economic contexts.
This inclination has two consequences: (1) Relational contracts are used
under more circumstances than in other societies, in which agency
problems associated with such contracts are higher. (2) Even where
explicit contracts in arm's-length market frameworks are clearly superior
from an allocative perspective, a hybrid type of organisation (e.g. a strate-
gic partnership) may be chosen, because the gains from the market solu-
tion cannot be fully realised. This holds, because arms' -length markets
need clear and strictly enforced rules to function properly. In societies with
ubiquitous generalised relations, this is difficult to achieve (Pascha 2002).
As for a second argument, even if a (capital) market is established,
many actors will tend to choose the old ways and means of the credit-
based system, because they have already invested heavily in social rela-
tions and can easily (and cheaply) make use ofthem.
This leads us to an alternative to H2:
Hypothesis 3 (H3): Japan 's jinancial system will become more hybrid,
i.e. employing institutional and organisational elements ofboth a credit-
based and a capital market-based system.

5 Along similar lines, one could speak of a transaction atmosphere (for instance,
Picot 1991 : 148).
404

In the remainder, we will try to substantiate Hl and empirically discuss

the merits ofH2 vs. H3.

4 Making ends meet: the financial sources of venture

businesses in Japan

Private banks
Private banks are a major financing institution even for venture businesses.
Those businesses find it particularly burdensome that they need material
securities to receive such funds. In a survey, some 78% replied this was
their major problern (CKC 1997: 363). Although many banks have
founded specialised venture departments and programs, these are mainly
occupied with traditional, secured loans, not with providing holdings. We
conclude that the heritage of Japan's credit-based financial system Ieads to
the expected difficulties for venture businesses (Hl).

Se/f-help
Even in a credit-based system, there are alternatives to bank loans. One
simple option is self-help. According to a survey, 79% ofyoung entrepre-
neurs use part of their personal property to finance the start-up (KKK
2001a: 125). Almost one third ofthe start-up capital is financed that way.
The financial burden is even bigger, because many businessmen have to
use their property, real estate in particular, to secure bank loans. For
instance, this holds for 40% of venture business board members (KGSK
2000: 32). In case of a bankruptcy, businessmen thus frequently loose
(almost) all of their fortune, which makes business start-ups particularly
risky. Compared to Anglo-American societies, the "psychological cost" of
going bankrupt is also very high ("loosing one's face"). This reduces the
chances to move into risky venture businesses and is a competitive disad-
vantage vis-a-vis other frontier economies.

Business angels

Still another possibility is "business angels", i.e. private investors in early

stage companies. While in some countries such informal risk capital
investors are quite important, in Japan, only 1.4% ofthe adult population is
engaged that way (6% for the US; Reynolds et al. 2001: 24f., 42). How-
ever, there frequently is no conscious understanding in Japan of being a
venture capital "angel", and according to a survey, actually some 44% of
 405

entrepreneurs have received funds from friends and family when they
started business; only some 3.3% have accessed formerly unknown private
investors, though (CKC 2000). The share of such start-up capital from
individuals could be up to one quarter. Still, usually it is argued that the
full potential for angel investment has not been mobilised yet. Due to the
tradition of the credit-based system, there is no legacy for such direct
involvement in venture business. It is typical that so far personal relations
form the major basis for angel activity ("love money"), which does not
promise to support efficiency. By introducing an angel tax system (enjeru
zeisei) in 1997, an investment into unlisted venture businesses not older
than ten years enjoys some tax advantages (e.g. Hosokawa and Sakurai
2000). It remains tobe seen whether such isolated measures can change-
in line with H2 - the attitude and aptitude towards angel investment.

Corporate venture capital

A final approach weil in line with the traditional Japanese system is
investment by corporate venture capitalists. Due to the network character
of the Japanese socio-economy, some 10% of start-ups have received
capital from other firms, and about half of all small and medium enter-
prises state that they have already supported new firms at least once (CKC
2000).
At least three types can be distinguished. Firstly, during the intemet and
IT hype ofthe late 1990s, some companies started to invest into start-ups
of those very industries. Softbank and Hikari Tsushin have almost become
household words for such manoeuvring. lt tumed out that it is very diffi-
cult to realise synergies by combining various new economy ventures
which, in each case, have to survive the severe competition in their own
industry. It is interesting, though, that Softbank instinctively tumed to a
business model ("zaibatsu") of the overcome economic system, deeply
involved with network relationships, and not with an open, transparent,
rule-based set-up. This supports H3 rather than H2.
A second type of corporate venture investment is "in-house" ventures,
where major corporations support (former) employees with innovative
ideas (SMEA 2001: 148ff.). This type of investment is strongly associated
with tacit, informal relations between a venture entrepreneur and an estab-
lished player, which is wen in line with the traditional Japanese business
model (supporting H3).
Thirdly, larger corporations have also discovered the possibility of
investing into venture businesses not related to former employees. Often,
they do so by investing into venture capital funds. For instance, they
contributed almost 20% of the capital of such funds originating between
406

mid 1999 and mid 2000 (VEC 2001: 27). Supported companies enjoy a
tacit, trustful relationship with their knowledgeable sponsor. For them,
there is the danger of being exploited by a powerful investor, a typical
principal-agent concem in such circumstances. Forthismodel to function,
it is important that both actors are bounded by strong links discouraging
ex-post opportunism. Given such networks as a sunk cost, it is a hopeful
venture financing option in the Japanese context (supporting H3).

Venture capital firms and funds

We now turn to venture financing mechanisms usually associated with a
capital market-based system.
In the wake of a so-called third venture boom in the mid 1990s, there
was a healthy increase of venture investment by specialised firms. After
some consolidation, it has started to grow again, reaching more than one
trillion yen (10 12 ; 9.3 billion Euro) for the firsttime for the October 2000
to September 2001 period (VEC 2002: 9). Some 115 companies are active
in this field. In comparative terms, these figures are dwarfed when
compared to US and European venture capital financing, though (Fig. 2).
As for venture capital funds, some 65 were launched from 1998 onwards,
with a healthy 32 commencing in the boom year 2000. In 2001, only 14
funds were newly funded, though.

30,000
27,274

25,000

20,000
17,485

11,635
10,813

·~ ,·~ n·."" n ·-
10,000

Lllig l__[;g
5,122
5,000

919 1,015
u . n.a.CJ
0
1995 1996 1997 1998 1999 2000 2001

C U.S_ Europe D Japan

Fig. 2. Comparison of investment amounts by VCs in Japan, Europe and the U.S.

Source: VEC 2002: 19

 407

Most venture capital firms were founded by either securities firms or

banks, where they have, generally speaking, kept their conservative
behaviour (Hamada 2002: 8). According to a new survey of the Venture
Enterprise Centre, some 44% dispatch part-time executives, but few send
full-time executives or staff. Moreover, they are often only related to fund
procurement and other financial issues, not to legal or technical support
(VEC 2002: 18).
VC firms thus have still not become active business partners of venture
businesses, unlike in Anglo-American capital market-based systems. Some
change can be noticed, though, which is frequently related to regulative
measures. (Only) since 1997, pension funds are allowed to undertake
venture investrnent. By mid 2000, they accounted for 6% of all VC funds
investrnents. Since 1998, venture investrnent can be undertaken through
limited partnership (yugen sekinin toshi jigyo kumiai), which reduces risk
and is clearly more attractive for investors. Moreover, VC firms have
accepted larger chunks of venture firms. This is riskier, but also raises the
interest of the investor in his investrnent. Also due to deregulation, more
foreign firms have entered the Japanese market. By mid 2000, already
some 26% of VC funds investrnent came from abroad. New business
methods enter the Japanese market that way, including a more hands-on
approach, offering more technical and procedural advice to promising
venture businesses (Nakada 2001).
It is difficult to attribute these developments to either H2 or H3, as
actual developments may be biased because of the post-new economy
bubble and by the Japanese recession. Simply put, for H2 we would expect
radical change within a low level of activity, while for H3, little change
and, again, a low level. While a clear conclusion is not possible, we tend
towards H3. For some years, for examp1e, the percentage of funds
earmarked for younger companies increased, but as for recent figures, the
share for companies less than five years old declined once again from 62 to
55%6 •

New stock market segments

The archetypal method for capital market-based systems to supply venture
capital is through specialised stock market segments, because the listing
and publication requirements of the regular markets are frequently to diffi-

6 Figures are for periods dating from October to September 1999/2000 and
2000/2001, respectively. Of course, some structural reasons are also involved.
There was a decline for computer-related frrms, which frequently have a short
growth period (VEC 2002: 9).
408

cult for risky, young enterprises to fulfil. In Japan, there are three major
markets to take note of:
• The Japan Securities Dealers Association (JSDA) founded an over-the-
counter (OTC) market in 1963. As there was rather little trust in an
association dominated by former MoF (Ministry ofFinance) bureaucrats
and by securities dealers, responsibility switched in 2001 to the Jasdaq
Market Inc. While supposedly meant for young companies, it actually
served as a first step towards listing on the full-scale stock market
(Kutsuna 2000). The average length from foundation to an IPO on the
OTC market became more than 20 years.
• In order to attract more truly venture and young companies, the Tokyo
Stock Exchange founded a "Market of the High Growth and Ernerging
Stocks" (Mothers) in 1999, amidst the new economy-boom. The listing
requirements are comparatively easy, but are accompanied by rather
strict publication rules, although they are accompanied by few formal
sanctions. Even very immature firms can be accepted, if they fmd a
willing security house as an underwriter. At some stage, there were
reports about underworld connections (Nikkei Weekly 17 April 2000),
which of course harmed its reputation. To regain it, Mothers has intro-
duced stricter standards recently.
• As a competitor of Tokyo, the Osaka Stock Exchange opened Nasdaq
Japan in 2000, together with the American Nasdaq and heavily
supported by Softbank, the already mentioned new economy-holding.
By distinguishing a growth and a standard section, Nasdaq Japan
wanted to cater to different client needs. lt has followed a rather conser-
vative approach, but was still affected by rumours tliat Softbank
followed its own agenda. While its performance in terms of listed
companies was not too bad (see Fig. 3), Nasdaq USA finally gave up its
involvement in late 2002. The Osaka Stock Exchange has since reor-
ganised its new market segment under the brand name of "Hercules", in
analmostdesperate attempt to take a free ride on the demigod's reputa-
tion of strength.
 409

9 6 0 r - - - - - - - - - - - - - - - - - - - - - - - - . , - 110
940
100
90
920 80

900 70
80
880
50
880 40

840 30
20
820

aoo~~~~~~~~~~~~~~~~~~~~~~~J:
~~-~-~~~-~-~~~-~-~
oo oo oo oo oo oo m m m m m ~ ~ ~ ~ ~ ~ ~

- - - O T C Mart<et (left axis)

--+--Molhers (rightaxis)
- • tl· • Nasdaq Japan/Hercules (righl axis)

Fig. 3. The three "new markets" in Japan

Source: Stock market information

Very few companies currently dare to go public. For Mothers, the
number is expected to be 11 for 2002 (7 in 2001) (Nikkei 31 October
2002). Why have the stock market segments for venture businesses dis-
appointed so much? Again, it is difficult to distinguish aspects of the new
market bust and of the Japanese recession from more structural factors.
Below the surface, one finds significant changes going on. For instance,
medium, often young securities houses as underwriters as well as foreign
players beef up the dynamism of the somewhat conservative major securi-
ties companies. Also, there is rather active M&A activity among start-ups,
following a 1999 Commercial Code revision authorising corporate
purchases through equity swaps (Nikkei 21 November 2002). This could
be viewed as evidence in favour ofH2.
However, there are also problems. One major concem of stock markets
is to earn the reputation of safeguarding a suitable balance between princi-
pal and agent rights and obligations. The established rules proved not
sufficient to quieten rumours of insider trading, other misdeeds and even
criminal involvement. To compensate for this deficiency, the market
organisers tried and still try to fmd a well reputed anchor- Nasdaq, the
JSDA, the Tokyo Stock Exchange - but due to all kinds of network effects
and split interests, no agency could gain reasonable trust.
Summing up, the networking conditions of an established financial
system seriously undermined a convincing move towards a capital market-
410

based system of venture financing. The markets are still searching for an-
chors lending reputation; put differently, they want to employ relational
mechanisms in line with H3.

5 Government policies

It would be too simplistic, though, to ascribe the deficiencies of Japan's

"new markets" only to the problems of its private organisers. As elabo-
rated above, for a (capital) market to function properly one needs clear,
general and strict rules which have to be supervised by a trustworthy
agency.
Since 1992, the Securities and Exchange Surveillance Commission
(SESC) is in charge of such duties. It inspects and surveys the securities
companies; it also inspects the self-regulatory operations of the various
stock exchanges and of the Japan Securities Dealers Association (SESC
2002). However, this system suffers from considerab1e deficiencies, which
have Iead to a number of requests to increase the SESC's oversight power
(e.g. Nikkei 20 September 2002 and 2 December 2002):
• The SESC has few powers. lt can make criminal charges against a
limited scope of offences, which is usually difficult anyway, but below
this Ievel it can only recommend disciplinary action to the Financial
Services Agency (FSA), to which it belongs.
• The SESC is run by a commission, composed of three respected
individuals: a high-ranking public prosecutor as Chairman, a senior
journaHstand a senior auditor. Given the SESC's subordinated position,
though, the "outsider"-status of its heads can easily encroach on its
actual influence.
• The commission is also poorly provided with personnet Whereas the
number has increased by 50% during 2001, it is still only 182- or 364,
if inspectors of local finance hureaus are included. This compares to
3,300 staff members at the comparable US commission (SEC) (Indo and
Matsuura 2002).
• Despite the dynamism of the capital markets, the SESC has no power to
change the rules, but can only propose this to the cabinet.
• As a final point, there may be conflicts of interest between the FSA,
which is in charge of the "well-being" of the security companies, and
the SESC, which has to overlook them.
The government Council on Regulatory Reform has just decided to
recommend an enhancement ofthe SESC's powers (as of early December
 411

2002). However, despite many calls to set up the SESC as a truly inde-
pendent agency, the Council will not recommend this due to the stiff Oppo-
sition of the FSA (Nikkei 2 December 2002). In conclusion, it seems
extremely difficult to set up an independent agency to supervise such rules
despite well-founded theoretical reasons (e.g. Pascha 2002). Limited
moves into such a direction offer little help and may even be counterpro-
ductive.
Regulatory measures lik.e setting up the SESC are one important aspect
of government policies. Another aspect is the promotion of venture capital
through fiscal measures. In Japan, there is a confusing multitude of con-
cepts and measures, particularly since the so-called third venture capital
boom, a special law to promote creative small and medium enterprises
(SME) (S6z6h6 of 1995) and the revision of the Basic Law for SME,
which pinned its hopes on the dynamism of start-ups (Hamada 2000:
329f.).
So far, the performance of these programs is generally considered to
have been rather disappointing (e.g. KKK 2001a: 128f.). On a superficial
level, long and clumsy procedures, lack of co-ordination of different
programs and actors, inadequate knowledge of involved bureaucrats and
similar problems have been identified. While ex-ante loans are certainly
not the mechanism of choice, it is interesting that the authorities still rely
on it to a significant extant. This evidence also supports H3.

6 Conclusions

For the growth and development of a frontier economy lik.e Japan, funding
venture projects is one of the most important tasks for supplying adequate
production factors. We showed that many of Japan's problems to supply
venture financing are related to its legacy of credit-based financing.
More interestingly, will there- and should there- be a transformation
to a capital market-based financial system or will there ratherbe a hybrid
system? Answering this question is complicated by the fact that recent
developments are strongly influenced by the new economy bust and by the
deep recession in Japan. Still, we found a considerable amount of evidence
supporting the hybrid system-hypothesis (H3). The basic reason is that
financial systems function within an environment characterised by sticky
informal institutions. In our case, the ubiquity of relational contracts or
inter-personal networks makes it very difficult to introduce a trustworthy
arm's-length capital market based on transparent, general rules and on
their strict enforcement by impartial public or quasi-public agencies.
 412

Rather than looking for an ideal solution, Japan's actors will have tobe in-
novative and flexible to make their hybrid system respond to today's and
tomorrow's challenges.

References

Albach H et al. (1986) Finanzierung mit RisikokapitaL Poeschel, Stuttgart

CKC (Chilshö Kigyöchö) (1997) Chilshö kigyö hakusho 1997-nenpan (White
Paper on Small and Medium Enterprises 1997). Chilshö Kigyöchö, Tokyo
CKC (Chilshö Kigyöchö) (2000) Chilshö kigyö hakusho 2000-nenpan (White
Paper on Small and Medium Enterprises 2000). www.meti.go.jp/hakusho/-
chusyo/H12/01-02-03-02.html, download 22 February 2002
Gebhardt G, Schmidt KM (2002) Der Markt fiir Venture Capital: Anreizprobleme,
Govemance Strukturen und staatliche Interventionen. Perspektiven der
Wirtschaftspolitik, vol3, no 3: 235-255
Hamada Y (2002) Current Status of Japanese Venture Capital. In: VEC (Venture
Enterprise Center, Ministry of Economy, Trade and Industry) (2002) 2001
Survey of Japanese Venture Capital Investments [and] Venture Capital Fund
Benchmark Survey. VEC, Tokyo, pp 1-10
Hemmert M (2002) Der erwachende Riese. Positive Entwicklungen in der
japanischen Forschungslandschaft Japan Markt, Oktober: 22-24
Hosokawa H, Sakurai M (2000) Benchä kigyö no fainansu senryaku (Financing
strategies ofventure businesses). Sangyö Keiri, March 2000: 12-20
Imai Y, Kawagoe M (2000) Business Start-ups in Japan: Problems and Policies.
Oxford Review ofEconomic Policy, vol16, no 2: 114-123
Indo B, Matsuura H (2002), Views vary over securities watchdog. Nikkei Weekly,
21 October 2002, www.nni.nikkei.co.jp/AC/TNW/Search/
Nni20021021MQ2SESCX.htm, download 2 December 2002
KGSK (Kagaku Gijutsu Seisaku Kenkyiljo) (2000) Nihon ni okeru gijutsukei
benchä kigyö no keiei jittai to sögyösha ni kansuru chösa kenkyil (Survey on
the business situation of venture businesses and their founders).
www.nistep.go.jp/index-j.html, download 12 March 2002
KKK (Kokumin Seikatsu Kin'yil Köko Sögö Kenkyilsho) (2001a) Shinki kaigyö
hakusho 200 1-nenpan (White paper on newly founded businesses 2001 ).
Chushö Kigyö Risächi Sentä K.K., Tokyo
KKK (Kokumin Seikatsu Kin'yil Köko Sögö Kenkyilsho) (2001b): 2001 nendo
shinki kigyö jittai chösa (Survey on the state of newly founded businesses
2001). www.kokukin.go.jp/pfcj/pdf/2001sinkj.pdf, download 22 February
2002
Kutsuna K (2000) Kyilseichö benchä kigyö no haishutsu to benchä kyapitaru
oyobi tentö shijö no yakuwari (Strong growth of venture businesses and the
role ofventure capital and ofthe OTC market). Shöken Keizai Gakkai Nenpö,
Mai2000:276-2 80
 413

Nakada GG (2001) Start-ups in Japan. The Journal ofthe American Chamber of

Commerce, October: 25-28
Nikkei (various issues), www.nni.nikkei.co.jp
Pascha W (2002) Wirtschaftspolitische Reformen in Japan - Kultur als
Hemmschuh?. In: Eger T (ed) Kulturelle Prägungen wirtschaftlicher
Institutionen und wirtschaftspolitischer Reformen. Duncker & Humblot,
Berlin, pp 189-214
Pascha W (forthcoming) Perspektiven und Probleme der japanischen Wirtschaft
aus langfristiger Sicht (paper prepared for a conference of the Asia-Pacific
Committee of German Business (APA) in Berlin, 29 October 2002). In:
Draguhn W (ed) Chinas und Japans zukünftige Bedeutung fiir Ostasien und
die Weltwirtschaft. Institute of Asian Affairs, Harnburg
Picot A (1991) Ökonomische Theorien der Organisation- Ein Überblick über
neuere Ansätze und deren betriebswirtschaftliches AnwendungspotentiaL In:
Ordelheide D et al. (eds) Betriebswirtschaftslehre und ökonomische Theorie.
Poeschel, Stuttgart, pp. 143-170
Reynolds PD et al. (200 1) Global Entrepreneurship Monitor 200 1.
www .gemconsortium.org/document.asp?id=160, download 20 January 2002
SMEA (Small and Medium Enterprise Agency) (2001) White Paper on Small and
Medium Enterprises 2001. www.chusho.meti.go.jp/hakusyolh13/download/
2001eibunzennbun.pdf, download 12 January 2002
Storz C (2000) Venture Businesses in Japan: Protagonists of a New Economic
Order. www.uni-marburg.de/japanz/mitarb/Storz/venture%20busines%20
dynamics.pdf, download 12 March 2002
Turpin D (1986) Japan's Small High-tech Enterprises and Venture Capital.
Business Series No. 106, Institute ofComparative Culture, Sophia University,
Tokyo
VEC (Venture Enterprise Center, Ministry of Economy Trade and Industry)
(2001) Annual Survey of Japanese Venture Capital Investments 2000. VEC,
Tokyo
VEC (Venture Enterprise Center, Ministry of Economy, Trade and Industry)
(2002) 2001 Survey of Japanese Venture Capital Investments [and] Venture
Capital Fund Benchmark Survey. VEC, Tokyo
Structural Analysis of Multinational Network
Organ izations

Manfred Perlitz

OlafN. Rank

1 Abstract

The phenomenon of organizational networks is one of the most fashion-

able topics in management science. Accordingly, multiple books and
articles have been written about the management and the formation of
network organizations as well as about their special features with regard to
the organizational structures. Although many contributions in this field
stress the significance of the relational pattems between the network
members, the majority of empirical research sticks to traditional statistical
methods. This is especially astanishing since in related social science
disciplines network analysis has been successfully applied to the analysis
of relational pattems.
In this paper, the main characteristics of network arrangements are
discussed, leading to the conclusion that organizational network structures
cannot be adequately analyzed with standard statistical methods. In fact, an
analytical framework is necessary, which allows an integrated analysis of
both attributive and relational data. Forthat purpose we introduce network
analysis as an appropriate research method for the study of network
organizations. To exemplify the functionality ofnetwork analysis, its basic
features will be presented by using empirical network data.

2 lntroduction

The discussion about organizational networks has attracted much attention

within the field of management and corporate govemance during the last
years. Network structures, sometimes viewed as organic arrangements
(Hage 1988; Gerlach 1992; Nohria 1992), are commonly suggested to
adequately match economic changes and environmental complexity and
hence gain competitive advantage (Miles & Snow 1984, 1992; Jarillo
1988; Bovasso 1992; Gomes-Casseres 1994; Park 1996; Jones, Hesterly &
 415

Borgatti 1997; Ral11997; Oliver & Ebers 1998). Numerous reasons for the
emergence of network structures have been presented ranging from
globalization and economic devolution, shorter product live cycles, and
increasing technological complexity to increasing corporate size and a
growing nurober of different countries and markets with different cultures
and customs caused by mergers and acquisitions that force enterprises to
flexibly adjust to their competitive environment (Kutschker 1999; Perlitz
2000; Welge & Holtbrügge 1997; Hage 1988; Kutschker & Bäurle 1997;
Park 1998).
At the same time, various theoretical approaches can be observed in
order to explain the emergence of network structures both within and
between firms. Among others, transaction cost economics, resource
dependence theory, contingency theory and arguments suggested by
exchange theory are used to address the network phenomenon (Husted
1994; Oliver & Ebers 1998; Wald 2000). However, it is important to dis-
tinguish the network phenomenon and the theories being used to explain
its emergence from network analysis, a methodology to examine network
structures on an empirical basis.
In this paper, we are not addressing the issue of how to theoretically
explain the existence of organizational networks. Instead, building on a
precise definition and extensive discussion of the main characteristics of
network arrangements we are going to introduce network analysis as a
methodological framework for the analysis of network structures. By
taking all kinds of economic as well as social relationships among the ac-
tors into consideration and by integrating formal and informal coordinating
devices we suggest that network analysis is useful to empirically assess
organizational networks both among and within firms.

3 Corporations as organizational networks

3.1 The application of the network concept to organizations

The definitions of network organizations that can be found in Iiterature are

manifold and often metaphorical (Nohria 1992) and are thus not always
useful for the description of organizational network arrangements. How-
ever, the design of a structural analysis inevitably requires a precise defi-
nition of the term 'network organization'. By reviewing the existing
Iiterature we present a definition integrating the various approaches
416

whereby we consider the work of Jones, Hesterly & Borgatti (1997) tobe
particu1arly important.
Network organizations consist of a well-defined, persistent, and struc-
tured set of semi-autonomous corporate actors engaged in numerous
mutual exchange relationships in order to jointly reach the common net-
work objectives. The relationships are based on implicit and open-ended
contracts to adapt to environmental contingencies and to coordinate and
safeguard exchange processes.
The term 'well-defined' indicates that the members of the network can
clearly be identified. Although several authors assume that it is not pos-
sible to exactly define the boundaries of a network organization (Laumann,
Marsden & Prensky 1983; Thorelli 1986), we do not consider the member-
ship tobe a question of coincidence. Instead, building on Park (1996) we
suggest that the membership is based on either a conscious decision of the
individual actors or specific attributes like e.g. ownership ties or legal
contracts. Consequently, the network can clearly be distinguished from its
environment. The term 'persistent' is used to indicate that the network
organization is relatively stable over time as its members work repeatedly
together (Jones, Hesterly & Borgatti 1997). The expression 'structured' is
used to indicate that exchange processes are neither random nor uniform
but the result of a division of labor along with the assignment of strategic
roles to the individual network members (Thorelli 1986; Rank 2000).
 417

We define the actors of network organizations as separable organiza-

tional decision centers (Aldrich & Whetten 1981; Thorelli 1986; Ebers
1997). Depending on the scope of the network the actors may either be
individuals, entire corporations (which is the case for interorganizational
networks) or organizational subunits, i.e. the corporate headquarters, its
subsidiaries or other entities like strategic business units and profit centers
(for the case of intraorganizational networks) (Lincoln 1982; Ghoshal &
Bartlett 1990; Tsai & Ghoshal 1998; Oliver & Ebers 1998). The actors of
network organizations are supposed to be semi-autonomous because of the
fact that they are relatively independent from their partners as far as the
achievement of their objectives is concemed while being legally and/or
economically dependent on the rest of the network. 1
The relationships that link the individual actors mutually together
comprise streams of transactions. These relationships can be differentiated
according to their type and structure (Bonacich 1987; Ibarra 1993; Brass,
Butterfield & Skaggs 1998). They consist of exchange processes which
include but are not limited to economic transactions. Instead, all sources of
differential advantage may form the basis of exchange. As a consequence,
relationships among the actors can be found involving economic per-
formance, technological transfer, diffusion of expertise and knowledge,
development of trust, and the flow of legitimacy (Thorelli 1986; Easton &
Araujo 1989; Cliffe 1998).
Finally, the phrase 'implicit and open-ended contracts' is used to indi-
cate that means of adapting, coordinating, and safeguarding exchange
processes are not primarily derived from authority structures or from legal
contracts, although formal contracts may exist between specific dyads of
members (Jones, Hesterly & Borgatti 1997).

1 Economic and legal independence is frequently used in the course of network

definitions. However, we consider neither economic nor legal independence as
appropriate characteristics of network organizations. From an economic point
of view the actors are c1osely integrated in numerous exchange relationships
with the other members of the network. Consequently, the assumption of
completely independent market corporations within networks seems to be
unrealistic (Sydow 1992). Also the term 'legal independence' is somehow
misleading in the context of network organizations. Although the actors may be
legally independent from a formal point of view, financial linkages exist
between the network partners in a large number of organizational networks
ranging from wholly owned entities, majority, parity, and minority investments
towards different forms of cooperation and joint ventures (Bovasso 1992;
Perlitz 2000). With this respect, both economic and legal independence of the
actors are only relative.
418

3.2 Coordinating mechanisms in network organizations

It can be assumed that not all actors within a network organization are
equipped with an equal degree of power and control. Hence, by pursuing
their objectives the individual nodes will strive to increase their control.
With this respect two processes are conceivable. Firstly, control can be
increased through formalized structures and processes coming into exis-
tence through explicit contracts and codified rules. Secondly, actors may
increase their Ievel of control and power through ad-hoc and informal
coordination of activities directed towards specific objectives (Araujo &
Brito 1997). In this sense, Welge (1999) proposes that only a well-
balanced mix of both formal and informal instruments will assure the inte-
gration of the widely dispersed economic activities of firms.
As far as coordination among the network members is concemed the
used mechanisms and processes are commonly reviewed with respect to
market and hierarchy, each suggesting different coordinating forms
(Williamson 1975). Within markets the participants act basically inde-
pendent from each other, they are equipped with equal rights and are
characterized by a limited rational and opportunistic behavior. Therefore,
market-based relationships are suggested tobe elusive and rather competi-
tive. In contrast to this, instructions among mutually dependent actors are
the main coordination device within hierarchies. Hierarchical relationships
are usually applied as long-term relations and are ideally cooperative
(Macneil 1978; Sydow 1992).
But how do these two basic coordinating mechanisms actually cohere as
far as organizational networks are concemed? Within management litera-
ture, two major positions can be distinguished (Wald, Rank & Peske
2000). Academics following the intermediary position propose that net-
work organizations may be located on a continuum somewhere between
market and hierarchy (Thorelli 1986; Williamson 1991; Seibert 1991;
Sydow 1992). In contrast to this, the autarkic position suggests that net-
works form an organizational structure on its own besides markets and
hierarchies, although embodying characteristics of both (Powell 1990;
Semlinger 1993).
Whatever position one is in favor of there seems to be no doubt that the
coordinating mechanisms of network organizations consist of a synopsis of
market and hierarchy as networks relate to the fundamental coordinating
devices of both. It is suggested that the decisive factors, which determine
the kind of coordinating mechanism are formed by the Ievels of competi-
tion and uncertainty the network organization faces. With respect to coor-
dination, a positive correlation is proposed between the Ievels of competi-
tion and uncertainty and the degree to which the network actors rely on
 419

closeness and trust in contrast to market-based arrangements (Borys &

Jemison 1989). This matter is closely tied up to the question of centraliza-
tion versus decentralization. According to Perrow (1986), the decentrali-
zation of responsibilities in networks results in prudent responses to
unplanned contingencies. In contrast, centralized authority produces
relatively quick reactions based on rigid obedience.
Moreover, both formal andinformal coordinating mechanisms act side
by side depending on the number and importance of the individual activi-
ties to be coordinated (Häkansson & Johanson 1988). Bovasso (1992)
suggests that ideally, the formal organizational structure is superseded by
the informal social networks that emerge from exchange processes. In this
context, specific attention is being paid to the role of cooperation and trust
as it influences the actors' commitment towards the network relationships,
their reaction to unanticipated problems, their recourse to contractual
remedies, and their style of conflict resolution (Jarillo 1988; Husted 1994;
Ring & Van de Ven 1992; Ring 1997; De Laat 1997). In order to enhance
cooperation on common tasks, network organizations primarily rely on
social coordination and control, e.g. occupational socialization, collective
sanctions, and reputations rather than on authority or legal recourse (Jones,
Hesterly & Borgatti 1997). As a consequence, the structural embeddedness
of the actors in network organizations has to be taken into account, since it
constitutes a framework, which provides opportunities for and constraints
on action (Granovetter 1985; Stinchcombe 1986, Wasserman & Faust
1994).

4 lmplications for the analysis of network organizations

4.1 Network analysis as a methodological framework

It becomes obvious that networks are highly complex organizations.

Although a huge amount of Iiterature on the concept of network organiza-
tions can be found and different theoretical approaches have been pre-
sented in order to explain the emergence of organizational networks, the
question of how to analyze this organizational device seems to attract by
far less attention. The multitude of empirical studies on networks still
favor standard statistical methods analyzing attributive data in order to
assess network structures (Bartlett & Ghoshal 1989; Ghoshal & Nohria
1993; Taggart 1997; Kraatz 1998). The emanating question however is,
420

whether these methods are suitable in order to approach organizational

networks appropriately from an empirical point ofview.
We have argued that relations play a particular important role within
networks. Consequently, the focus of our approach has tobe put on rela-
tional data. Network data require measurements on the relationships
among the actors as weil as on the attributes of the actors, whereas statisti-
cal methods alone are not suitable to assess organizational networks
appropriately (Wasserman & Faust 1994).
Similarly, it has been shown that within network organizations formal
and informal instruments are being used to safeguard exchange processes.
In order to maintain or enhance their Ievels of control the individual net-
work actors use both devices likewise. Consequently, an analytical frame-
work has to assure that both formal and informal structures and processes
are taken into account simultaneously. There are various different relation-
ships between actors. Forthis reason, one has to make sure that the analy-
sis of network organizations covers the entire spectrum of possible rela-
tionships ranging from elusive and competitive forms as the market
element would suggest to long-term and cooperative relationships, which
can be found within hierarchical arrangements.
Moreover, in network organizations limited rational and opportunistic
behavior, as weil as behavior that is induced by the subordination of
hierarchies, is likely to occur. On the one hand, the corporate actors of
network organizations may pursue different strategic objectives. The rea-
son for this can be that differing strategie roles are assigned to the speeifie
entities as a result of the strategie process. On the other hand, the indi-
vidual actions of the network members have to be coordinated with regard
to the common network objectives. In network organizations, this coordi-
nation is predominantly achieved by stable pattems of exchange relation-
ships, whieh, once established, create trust and make individual action
more predictable. Therefore, an analytieal framework has to match a eon-
cept of human and corporate aetion, which is neither under- nor over-
socialized, but takes into account the affection of behavior by social rela-
tions. (Granovetter 1985).
 421

Finally, we have argued that the characteristics of economic and legal

independence are not suitable to determine whether a network is inter- or
intraorganizational. It can be assumed that there is no discrete point at all,
which distinguishes unequivocally interorganizational arrangements from
intraorganizational networks. It becomes obvious that the structural analy-
sis of network organizations has to be designed in a way which incorpo-
rates both inter- and intraorganizational actors dependent on the specific
research question under investigation.
The application of social network analysis to the field of business
organizations allows to adequately take all these facts into account. Hence,
in the following network analysis will be introduced as a method that is
suitable for the analysis of complex organizational network arrangements.
Being a well established research methodology in several academic disci-
plines, e.g. political science, surprisingly few applications to the analysis
of corporation's organizational structure can be found (e.g. Ghoshal &
Bartlett 1990).

4.2 Key characteristics of network analysis

Originating from sociology, sociometry, and anthropology, social network

analysis is a method to investigate social structures. It can be applied to a
wide range of topics within various academic disciplines. Generally, a
network consists of a set of actors, which are connected through different
relations or ties (Wasserman & Faust 1994). As already mentioned, the
relations linking the actors can be of manifold type. Hence, an important
precondition of every network analysis is the definition of the relations,
which have to be taken into account. Another step is the system delinea-
tion, i.e. the identification ofthe relevant actors (Pappi 1993).
Having standard social science research in mind, the key characteristics
of network analysis are that the focus is put on pattems of relations
between the nodes rather than on their attributes as suggested by standard
statistical methods. An example may illustrate this important distinction. A
research problern typically analyzed with standard statistical tools is the
question, if the profit contributions of the individual subsidiaries of a
multinational corporation (MNC) are related to the management know-
how of their respective top-executives. In contrast, the relations between
the managers, e.g. knowledge and information flows, would be the
research object when using network analysis. In this case, the related
research question would be, if the profit contributions of the individual
MNC's subsidiaries are related to the managers' position in the network of
422

information flow. 2 However, the position of a specific manager within the

network cannot be identified by simply considering his individual
attributes yet by analyzing the ties connecting him to the other managers in
the network. It needs to be mentioned that once the network position has
been identified by the means of network analytical tools, it can be treated
as an individual attribute on its own. Network analysis allows an integrated
study of both attributive and relational data as shown by Burt (1992). He
found empirical evidence that e.g. managers with a specific pattem of
relationships get promoted faster than others.
To study network organizations, network analysis serves two purposes:
Firstly, it can be used as a descriptive tool to analyze the complex rela-
tional pattems between the network actors as weil as to reduce complexity
by identifying the underlying macro-structure, i.e. dominant pattems of
relationships out of multiple networks. Secondly, network analysis can be
applied to test hypothesizes on structural properties of networks or on the
interaction of structural properties and attributive characteristics as far as
the scope of theory-guided research is concemed. Moreover, network
analysis is suitable for different organization theories such as resource
dependence theory, exchange theory, transaction cost economics or
contingency theory.
Having introduced the key characteristics of the social network
perspective, the next chapter will outline the methodological approach of
network analysis to study specific organizational networks empirically.
The structural analysis in this paper will be guided by the central research
question how exactly formal and informal structural pattems within the
network cohere in order to achieve specific organizational tasks.

2 In this context, the position of a manager corresponds to his centrality in the

information networks, i.e. the extent to which a manger has access to valuable
information via short paths in the network. However from a network
perspective, the term position is also used in the sense of the social position. In
this respect, two actors occupy the same position, if they are similarly
embedded in the network (Wasserman & Faust 1994).
 423

5 Empirical analysis

5.1 Design of the study

To illustrate the methodological approach we use network data on the

strategic planning and decision mak:ing process of BASF, the world
leading chemical manufacturer. The specification of the relations linking
the actors to a specific objective (in our case strategic planning) is neces-
sary in order to reduce the total number of relational contents to be inte-
grated into the analytical process. We operationalize the enterprise as an
intraorganizational network consisting of different corporate actors that are
linked by different types of relations.
Strategie decisions conceming the enterprise as a whole are typically
located at the top ofthe organization. Forthis reason, the set of actors con-
sists of the top two management Ievels. In the case of BASF, these Ievels
consist of the board of directors as well as the organizational divisions
("Bereiche") below them. According to the matrix type organizational
structure of BASF, three different types of divisions may be distinguished:
product divisions, geographical divisions and central service divisions like
legal, finance, and central research and development functions. lt has been
mentioned that network actors may consist of individuals and organiza-
tional entities likewise. In our case, we define all nodes of the organiza-
tional networks as corporate actors. Although the members of the board of
directors are indeed individual persons, the focus was put on their corpo-
rate function rather than on their individual or personal relations. The
entire set comprised 48 actors: 8 members of the board of directors repre-
senting the first organizationallevel and 40 second Ievel units consistingof
14 product divisions, 10 geographical divisions and 16 central service divi-
sions.3
According to the central research issue, the coexistence between formal
and informal relations within the corporate-wide strategic planning and
decision mak:ing process, data was gathered within different partial net-
works. For the purpose of our study the two expressions 'informal' and

3 It has to be mentioned that five of the central service actors formally belang to
the third organizational Ievel. Due to their relative importance with regard to
the strategic planning process, the two departments "Central Strategie
Planning" and "Central Corporate Controlling" were considered in the study.
Likewise, three organizational units of the third Ievel that directly report to the
board of directors were integrated.
424

'aetual' relations deseribe the same struetural pattems. While the network
of the formal relations eaptures the way the strategie planning proeess is
supposed to be earried out, the informal relations deseribe how strategie
deeisions are aetually made. Altogether, we gathered data within three
different partial networks. On the one hand, formal relations were defined
as needs for eooperation between eorporate aetors with respeet to the
determination and the fulfillment of strategie targets. Henee, formal rela-
tions are always reeiproeal and symmetrie. On the other hand, two net-
works eomprising the exehange of strategieally relevant information as
well as the exehange of aetive support between the organizational units
when aehieving strategie goals were eonsidered. It is obvious that both
transaetions are direeted. Beeause of this, both the information and support
networks are non-symmetrie.
While the determination of the formal relations is mainly based on eor-
porate guidelines, all data on the aetual exehange relations regarding
information and support were gathered through personal interviews with
the head of the speeifie organizational unit. In order to enhanee the
reliability of the data, only eonfirmed relations were eonsidered within the
analytieal proeess. This may be explained using short example: A rela-
tional information tie from aetor i to aetor j is only supposed to be aetually
present, if i declares to send information to j and if j at the same time
names i as a souree of information.

5.2 Empirical results and discussion

A first mean to eharaeterize the network struetures on the network level is

density. Network density deseribes the ratio between the number of aetual
relations in a strueture and the number of potentially existing relations. In a
network eonsisting of 48 aetors the number of potential relations equals
2.256. Table 1 gives the density measures for all three relational eontents.

Table 1. Density ofNetwork Structures

Number ofrelations Density

Formal Relations 1.204 53,37%
Information 771 34,18%
Support 680 30,14%
First results eoneeming the eoexistenee of formal and informal relations
ean be drawn from the density measures. It is obvious that the formal
organizationa1 strueture suggests a fairly dense system of relations. More
than half of all potential relations are supposed to be earried out within the
 425

strategie planning proeess by the eorporate units. In eontrast to this, sig-

nifieantly less relations ean be identified as far as the aetual exehange of
information and support is eoneemed. Only about one third of all possible
ties between the aetors are aetually used. Obviously, the way the strategie
planning proeess is aetually earried out differs eonsiderably from the for-
mally preseribed eooperation pattems.
It has been argued that the eorporate aetors are not likely to be inte-
grated into the network struetures in a uniform way. Instead, it ean be
expeeted that their individual number of relations varies aeeording to their
eorporate funetion within the strategie planning proeess. In order to
measure the embeddedness of aetors into the different networks, two
measures of eentrality proposed by Freeman (1979) are used. Firstly, the
aetor's degree eentrality eonsiders the number of his ties relatively to all
possible ties. Seeondly, closeness eentrality makes use of the faet that
aetors may not only be linked direetly within a strueture. A typieal feature
of network struetures is that the nodes may also be eonneeted indireetly
through some intermediary aetor or aetors. Therefore, closeness eentrality
is a ratio based on the path distanee between two speeifie aetors. 4 Tables 2
and 3 give the average results for both eentrality measures, differentiating
between the different types of aetors. 5 It has already been mentioned that
information and support ties are direeted relations. Therefore, for both
relations the sending and reeeiving of the respeetive resouree ean be
distinguished.

4 Although several different paths are likely to occur between two specific actors,
path distance equals the shortest path between them.
5 It should be noted that these results neglect the variance within the groups.
426

Table 2. Actors' Degree Centrality

Freeman's Degree Centrality
Formal Information Information Support Support
Relations Sending Receiving Granfing Receiving
Membersof
0,25 0,41 0,52 0,44 0,45
theBoard
Product
0,65 0,32 0,26 0,20 0,33
Divisions
Geographical
0,53 0,26 0,29 0,25 0,30
Divisions
Central
Service 0,58 0,38 0,35 0,35 0,20
Divisions

Table 3. Actors' Closeness Centrality

Freeman's Closeness Centrality
Formal Information Information Support Support
Relations Sending Receiving Granfing Receiving
Membersof
0,57 0,63 0,68 0,64 0,64
theBoard
Product
Divisions 0,75 0,59 0,57 0,52 0,58

Geographical
Divisions 0,68 0,57 0,58 0,53 0,56

Central
Service 0,71 0,62 0,61 0,59 0,50
Divisions

As far as the formal structure is concemed it can easily be seen from the
results that the members of the board are only sparsely embedded into the
structure compared to all other corporate units. Obviously, the board
members are supposed to rely primarily on indirect relations. As a conse-
quence, they have the highest average distance to all other nodes in the
network, which can be seen from their relatively low closeness centrality.
According to their high relevance for strategic issues, the product divisions
show the highest centrality measures in the formal structure followed by
the central service divisions. As far as the actual transfer of information
and support is concemed it can be noted that only the members of the
board increase their centrality by building and maintaining more direct
relations to other actors than suggested by the formal structure. As a result,
they are able to reduce the average distance to all other units in the strate-
 427

gic planning process. It seems to be save to say that the board members
prefer to use more direct relations to other actors rather than rely on the
indirect transfer of information and support through intermediate units. In
contrast to this, all second level units show significantly lower centrality
measures in the actual transfer networks. They have less direct relations at
their disposal than suggested by the network of formal relations and hence
their mutual distance increases. From these results it can be concluded that
second level actors deviate considerably from the formal coordination
devices as the formal relations on this organizationallevel are not fulfilled
to a comparably large extend. However, it can also be seen from the results
of the two dimensions "sending" and "receiving" that the divisions are
embedded asymmetrically. The product divisions for example send more
information to other units than they receive. When taking support into
consideration however, the picture changes. By average, the product divi-
sions receive significantly more support than theygrant.
While the centrality measures give a first indication that there are sig-
nificant differences between formal relations on the one hand and actual
information and support relations on the other hand, it is worth dedicating
a closer look to the congruence of the partial networks. We do so by using
multiplexity counts for formal and actual ties. The results from this analy-
sis can be seen in Table 4.

Table 4. Correspondence ofPartial Networks

Information Support
Actual Actual
Information Support
Relations Relations
+ - .E + - .E
+ 549 655 1204 + 495 709 1204
24% 29% 53% 22% 31% 53%
-E-:::"-'=
== - 222 830 1052 -
== =
"-'

E-::: - 185 867 1052

=-=
loo
~~
10%
.E 771
38%
1485
47%
2256
=-=
loo
~~
.E
8% 38% 47%
680 1576 2256
34% 66% 100% 30% 70% 100%
"+" indicates that a relation is present, while "-" indicates that a relation
is not present.
Along with the presurnptions based on the centrality measures it can be
seen from the nurnbers in Table 4 that formally suggested relations are not
fulfilled to a large extent. As far as information is concemed, less than half
of the formal ties can be actually identified. Within the support network,
428

the problern of unfulfillment is even more severe. However, to some

extend the units build and maintain information and support relations not
being suggested by the formal structure, although they are significantly
lower in number and hence are not able to compensate the relational unful-
fillment.
Although we do not want to judge on these differences it seems to be
interesting from the point of view of organizational diagnosis to locate
these differences more closely. While voluntary relations not being
suggested by the formal structure do certainly not cause harm to the strate-
gic planning process the unfulfillment of relations might turn out to
become a severe problem. Forthis reason, we focus on latter phenomenon
in the following analysis. Doing so we distinguish unfulfilled ties due to
their direction in vertical and horizontal relations. The results from this
count can be seen in Table 5.

Table 5. Unfulfillment ofFormal Relations

Entire Network
Information 54,40% (655)
Support 58,89% (709)

Vertical Relations
Total 1st to 2nd 2nd to 1st
Level Level
Information 7,89% (6) 10,53% (4) 5,26% (2)
Support 15,79% (12) 10,53% (4) 21,05% (8)
Horizontal Relations
Total 1st Level 2nd Level
Information 57,54% (649) 12,50% (7) 59,89% (642)
Support 61,79% (697) 0,00% (0) 65,02% (697)
Portion of unfulfilled formal relations relative to information and support
networks (absolute numbers ).
Taking the entire network into consideration the numbers show again
that more than fi:fty percent of the formally suggested relations are not ful-
filled as far as the actual transfer of information and support is concemed.
Distinguishing between the direction of the relations however reveals
interesting details. While the issue of unfulfillment of formal relations may
be almost neglected for vertica1 re1ations it turns out to be a severe
problern when looking at the horizontal relations on the second organiza-
tional level. Almost 60% of the information ties and roughly 65% of the
 429

support ties being suggested by the formal structure are actually not pre-
sent. Again, we do not want to value these differences. However, it seems
to be safe to say that by far less direct information and support ties than
formally suggested seem to be sufficient for the organizational units to
accomplish their strategic targets.

6 Conclusion

In this paper it has been argued that relations play a crucial role with
respect to the understanding and management of network organizations.
Wehave shown that a network organization cannot satisfactorily be ana-
lyzed by an examination of its members' attributes. Instead, the structural
embeddedness of the actors has to be taken into account. Furthermore,
formal and informal relations as well as their coexistence play an essential
role as far as coordination within the network is concemed. As a conse-
quence, standard statistical methods alone being designed for the analysis
of attributes, are not suitable to study the pattems of relationships as found
in organizational networks. In fact, a framework for the analysis of net-
work organizations has to enable both the study of attributive and rela-
tional data.
We have presented network analysis as a method we suppose to be
particularly well suited for a large variety of research problems in network
organizations. Although our empirical study was concemed with an
intraorganizational network there is no reason why this research design
should not be transferred to interorganizational network systems as well.
Within this paper however, we have concentrated on the basic features of a
network analytical framework illustrated by empirical examples. Due to
the immense field of potential applications of network analysis, a general
standard procedure prescribing how to design network sturlies does not
seem to be sensible. Admittedly, there are few general approaches
referring to fundamental concepts conceming system delineation, the defi-
nition of the types of networks, and the methods of data collection
(Wassermann & Faust 1994).
According to the objective of this paper we aimed to demoostrate that
the application of network analysis to organizational networks can
generate new insights into the functioning of this organizational form.
Moreover, we are confident, that in a further step these insights can be
used to derive recommendations for practitioners engaged in orgl;Uliza-
tional tasks.
430

References

Aldrich H, Whetten DA (1981) Organizations-sets, action-sets, and networks:

making the most ofsimplicity. In: Nystrom P, Starbuck WH (eds) Handbook
of organizational design, voll : Adapting organizations to their environments.
Oxford: Oxford University Press
Araujo L, Brito C (1997) Agency and constitutional ordering in networks: A case
study of the port wine industry. International Sturlies of Management &
Organization, 27(4): 22-46
Bartlett CA, Ghoshal S (1989) Managing across boarders. The transnational
solution. Boston: Harvard Business School Press
Bonacich P (1987) Power and centrality: A family ofmeasures. American Journal
ofSociology, 92(5): 1170-1182
Borys B, Jemison DB (1989) Hybrid arrangements as strategic alliances:
Theoretical issues in organizational combinations. Academy of Management
Review, 14(2): 234-249
Bovasso G (1992) A structural analysis of the formation of a network
organization. Group & Organization Management, 17(1 ): 86-106
Brass DJ, Butterfield KD, Skaggs BC (1998) Relationships and unethical
behavior: A social network perspective. Academy of Management Review,
23(1): 14-31
Burt RS (1992) Structural holes. The social structure of competition. Cambridge
& London: Harvard University Press
Cliffe S (1998) Knowledge management: The well-connected business. Harvard
Business Review, 76(4): 17-21
De Laat P (1997) Research and development alliances: Ensuring trust by mutual
commitments. Ebers M (ed) The formation of inter-organizational networks,
Oxford: Oxford University Press
Easton G, Araujo L (1989) The network approach: An articulation. In: Hallen L &
J Johanson, (eds) Networks of relationships in international industrial
marketing, Vol. 3, Greenwich: JAI Press
Ebers M (1997) Explaining inter-organizational network formation. In: Ebers M,
(ed) The formation of inter-organizational networks. Oxford: Oxford
University Press
Freeman LC (1979) Centrality in social networks: Conceptual clarification. Social
Networks, 1:215-239
Gerlach ML (1992) The Japanese corporate network: A blockmodel analysis.
Administrative Science Quarterly, 37(1): 105-139
Ghoshal S, Bartlett CA (1990) The multinational corporation as an
interorganizational network. Academy of Management Review, 15(4): 603-
625
Ghoshal S, Nohria N (1993) Horses for courses: Organizational forms for
multinational corporations. Sloan Management Review, 34(4): 23-35
Gomes-Casseres B (1994) Group versus group: How alliance networks compete.
Harvard Business Review, 72(4): 62-74
 431

Granovetter M (1985) Economic action and social structure: The problern of

embeddedness. American Journal ofSociology, 91:481-510
Hage J (1988) The pathways of evolution in organizations. In: Hage, Jerald, (ed)
Futures of organizations: Innovating to adapt strategy and human resources to
rapid technological change. Lexington: Lexington Books
Häkansson H, Johanson J (1988) Formal and informal cooperation strategies in
international industrial networks. In: Contractor F, Lorange P, (eds)
Cooperative strategies in international business: Jointventuresand technology
partnerships between firms. Lexington: Lexington Books
Husted BW (1994) Transaction costs, norms, and social networks: A preliminary
study of cooperation in industrial buyer-seller relationships in the United
States and Mexico. Business & Society, 33(1): 30-57
lbarra H (1993) Network centrality, power, and innovation involvement:
Determinants of technical and administrative roles. Academy of Management
Journal, 36(3): 471-501
Jarillo JC (1988) On strategic networks. Strategie Management Journal, 9: 31-41.
Jones C, Hesterly WS, Borgatti SP (1997) A general theory of network
governance: Exchange conditions and social mechanisms. Academy of
Management Review, 22(4): 911-945
Kraatz MS (1998) Leaming by association? Interorganizational networks and
adaptation to environmental change. Academy ofManagement Journal, 41(6):
621-643
Kutschker M (1999) Internationalisierung der Wirtschaft, Kutschker M, (ed)
Perspektiven der Internationalen Wirtschaft. Wiesbaden: Gabler
Kutschker M, Bäurle I (1997) Three + One: Multidimensional strategy of
internationalization. Management International Review, 37: 103-125
Laumann EO, Marsden PV, Prensky D (1983) The boundary specification
problern in network analysis. In: Burt RS, Minor MJ, (eds) Applied network
analysis: A methodological introduction. Beverly Hills/London/New Delhi:
Sage Publications
Lincoln JR (1982) Intra- (and inter-) organizational networks. Research in the
Sociology ofürganizations, 1: 1-38
Macneil I (1978) Contracts: Adjustments of long-term economic relations under
classical neoclassical and relational contract law. Northwestern University
Law Review, 72(1): 854-905
Miles RE, Snow CC (1984) Fit, failure and the hall of fame. California
Management Review, 26(3): 10-28
Miles RE, Snow CC (1992) Causes offailure in network organizations. California
Management Review, 34(4): 53-72
Nohria N (1992) Is network perspective a useful way of studying organizations?
Nohria N, Eccles RG (eds) Networks and organizations: Structure, form, and
action, Boston: Harvard Business School Press
Oliver AL, Ebers M (1998) Networking network studies: An analysis of
conceptual configurations in the study of inter-organizational relationships.
Organization Studies, 19(4): 549-583
432

Park SH ( 1996) Managing an interorganizational network: A framework of the

institutional mechanism for network control. Organization Studies, 17(5):
795-824
Pappi FU (1993) Policy-Netze: Erscheinungsform moderner Politiksteuerung oder
methodischer Ansatz? Politische Vierteljahresschrift, Sonderheft 23, 84-94
Perlitz M (2000) Internationales Management, 4th edn. Stuttgart: Lucius & Lucius
Perrow C (1986) Complex organizations: A critical essay, 3rd edn. New York:
Random House
Powell WW (1990) Neither market nor hierarchy: Network forms of organization.
Research in Organizational Behavior, 12: 295-336
Rall W (1997) Der Netzwerkansatz als Alternative zum zentralen und hierarchisch
gestützten Management der Mutter-Tochter-Beziehungen. Macharzina K,
Oesterle MJ ( eds) Handbuch Internationales Management, Wiesbaden: Gabler
Rank ON (2000) Rollentypologien fiir Tochtergesellschaften: Ansätze und
strategische Implikationen fiir das Internationale Management. Stuttgart:
Döbler & Rössler
Ring PS (1997) Processes facilitating reliance on trust in inter-organizational
networks. Ebers M (ed) The formation of inter-organizational networks.
Oxford: Oxford University Press
Ring PS, Van de Ven AH (1992) Structuring cooperative relationships between
organizations. Strategie Management Journal, 13: 483-498
Semtinger K (1993) Effizienz und Autonomie in Zuliefernetzwerken - Zum
strategischen Gehalt von Kooperation. In: Staehle WH, Sydow J (eds)
Managementforschung 3. Berlin/New York: de Gruyter
Seibert K (1981) Joint Ventures als strategisches Instrument im internationalen
Marketing. Berlin: Duncker & Humblot
Stinchcombe AL (1986) Stratification and organization: Selected Papers.
Cambridge: Cambridge University Press
Sydow J (1992) Strategische Netzwerke: Evolution und Organisation. Wiesbaden:
Gabler
Taggart lli (1997) Autonomy and Procedural Justice: A Framework for
Evaluating Subsidiary Strategy. Journal ofBusiness Strategies, 28(1): 51-76
Thorelli HB (1986) Networks: Between Markets and Hierarchies. Strategie
Management Journal, 7: 3 7-51
Tsai W, Ghoshal S (1998) Social capital and value creation: The role ofintrafirm
networks. Academy ofManagement Journal, 41(4): 464-476
Wald A (2000) Die Netzwerkanalyse zur Untersuchung von
Organisationsstrukturen. WiSt, 29(12): 703-706
Wald A, Rank ON, Peske T (2000) Management Beyond Hierarchies: Negotiation
Systems in Network Organizations. Working Paper No. 11, University of
Mannheim, Department of International Management, Mannheim
Wasserman S. Faust K (1994) Social Network Analysis: Methods and
Applications. Cambridge: Cambridge University Press
Welge M (1999) Informelle Steuerungsmechanismen zur Optimierung globaler
Geschäfte. Kutschker M (ed) Management verteilter Kompetenzen in
multinationalen Unternehmen. Wiesbaden: Gabler
 433

Welge M, Holtbrügge D (1997) International management under postmodern

conditions. Macharzina K, Oesterle MJ, Wolf J (ed) Global business in the
information age. Proceedings ofthe 23rd Annual EIBA Conference, Stuttgart
White HC, Boorman SA, Breiger RL (1976) Social structure from multiple
networks, I. Blockmodels of roles and positions. American Journal of
Sociology, 81(4): 730-780
Williamson OE (1975) Markets and hierarchies: Analysis and antitrust
implications. New York: Free Press
Williamson OE (1991) Comparative economic organization: The analysis of
discrete structural alternatives. Administrative Science Quarterly, 36(2): 269-
296
Portfolio Return Characteristics of Different
lndustries

Igor Pouchkarev

Jaap Spronk

Pim van Vliet •

1 Abstract

Over the last decade we have witnessed the rise and fall of the so-called
new economy stocks. One central question is to what extent these new
firms differ from traditional firms. Empirical evidence suggests that stock
returns are not normally distributed. In this article we investigate whether
this also holds for portfolios of stocks from a growth industry. Further-
more, we will compare this type of portfolios with portfolios of stocks
from a more traditional industry. Usually, only value weighted and equally
weighted portfolios are used to describe and compare portfolio return
characteristics. Instead, in our analysis, we use a novel approach in which
we use an infinite number of portfolios that together represent the set of all
feasible portfolio opportunities.

2 lntroduction

Over the last decade we have witnessed the rise and fall of the so-called
new economy stocks. The price movement of the listed growth stocks can
be characterized by boom and bust. The surge of stock prices in the late
nineties and the dramatic fall in the new millennium has fascinated many,
inside and outside academia.
The environment in which new economy fmns operate is characterized
by rapid technological change and dynamic interaction. Flexibility in deci-

• The authors like to thank the participants and discussants at the "Managing
Enterprises of the New Economy by Modern Concepts of the Theory of the
Firm" conjerence in Hagen, Germany, and Danny Vlasblom for helping them
with a part of the computational work. As usual, all remaining errors are the
responsibility ofthe authors.
 435

sion-making is necessary for survival and can be seen as a core compe-

tence. Fitted within the two-dimensional BCG matrix new economy stocks
represent the 'stars' or 'question marks', where the old economy stocks
represent 'dogs' or 'cash cows'. What these new stocks and new industries
characterize is their growth potential, which largely determines their value.
Growth potential depends on firm specific and industry specific factors:
e.g. management's capability to identify and exploit valuable growth
options, or the number of strategic alliances, and the rate of technological
change within an industry. lt remains a challenge to appropriately deter-
mine the correct value of this growth potential. Real options analysis, e.g.
(K.ester 1984; Trigeorgis 1996), could help, for it reckons with managers
flexibility to alter decisions.
Ultimately, growth potential offirms influences the risk return profile of
their cash flows. Projects or activities can be abandoned if conditions turn
out unfavorable. This Iimits downside risk. On the other band, successful
projects can be expanded, thus leaving upside potential open. Because of
this flexibility the distribution of the growth company's expected cash
flows is characterized by asymmetry. The distribution characteristics of a
firm's cash flows are of course not automatically valid descriptions of the
firm's stock return characteristics. Firstly, the market has its own percep-
tion of the firm's cash flows (e.g. due to information asymmetries).
Secondly, after interest payments only a residual ofthe cash flows goes to
the stock owners. Therefore the degree of financial leverage affects the
pay-off structure and could also introduce asymmetry in stock returns.
Thirdly the market has the possibility of diversification, which means that,
in general, not all cash flow risk is priced.
lf cash flow distributions are not symmetrical, the stock return distribu-
tions may well be. However, empirical research shows that stock returns
are not normally distributed (Fama 1965; Kon 1984). The fat tail phe-
nomenon is well documented and, in addition, there is evidence of positive
skewness in the distributions of small growth stocks (Knez and Ready
1997). Not only individual stocks returns, but also market indices are
characterized by asymmetry. Several studies (Kraus and Litzenherger
1976; Harvey and Siddique 2000) have demonstrated that systematic
skewness is priced as market risk.
The present article examines the different portfolio return characteristics
of a new economy industry versus an old economy industry. Usually indi-
ces, either market value weighted or equally weighted, are used to describe
and compare portfolio return characteristics. However, if seen as invest-
ment strategies that can actually be implemented, fund managers do have
many alternatives for tracking specific indices. Some select stocks based
on fundamental analysis or technical analysis, others follow passive strate-
436

gies. Therefore, the composition of a fund manager's portfolio is often

different from one of the usual indices. Actually, with the same compo-
nents of the usual indices, an infinite number of portfolios can be
constructed. Using a novel approach (Hallerbach et al. 2002; Pouchkarev
2004) the set of all feasible portfolios with the same components can be
simulated. We will take a closer Iook at all feasible portfolio opportunities
and examine to what extent the portfolio opportunity set of high potential
stocks differs from the portfolio opportunity set of traditional stocks.
The article proceeds as follows. In section three we will describe the
methodology and the return data of two industries, semiconductors and
mining. In section four we provide the statistics of our analyses and dis-
cuss the results. Finally section five concludes.

3 Methodology

3.1 Set of portfolio opportunities

The main idea of the approach chosen in the present article is to explore
the whole set of portfolio formation opportunities in an industry instead of
limiting to separate portfolios and industry indices. We estimate the distri-
bution of the ex post performance values (e.g. average return, variance,
semivariance etc.) of all possible portfolios that can be composed from
stocks Iisted within the industry. The development ofthe location ofthese
distributions yields a picture of the average development of the industry
over a certain time period. The development of the dispersion of these
distributions provides a picture of the development of the industry dy-
namics over time.
The single restriction we make here, by looking at portfolio opportuni-
ties in an industry, is the exclusion of short sales. Then the opportunity set
(where 'opportunity' is viewed in retrospection) consists of all portfolios
within the same industry sector with weights

L wi = 1.
n
0 ::;; wi ::;; 1. 0 , i= 1, 2, ... , n such that (1)
i=l

The number of portfolios in the opportunity set is infinite but distributions

of portfolio performance values do exist. There are several ways to calcu-
late the required distributions. In this article we use simulation to estimate
 437

the distributions. 2 (We refer to (Hallerbach et al. 2002; Pouchkarev 2004)

for further details.) The procedure is as follows:
I. In each simulation step we sample one million feasible ran-
dom portfolio weight vectors for stocks of the industry
concemed. Bach sampled weight vector defines a portfolio
and is an alternative to invest in the industry. The sampled
portfolios are uniformly distributed over the industry portfolio
opportunity set;
II. For these sampled portfolios we calculate different portfolio
return characteristics: the average rates of return, variances,
and semivariances. It is quite easy to extend the number of
characteristics, such as mean absolute deviation, Gini index,
Herfindahl index, Sharpe ratio, Treynor ratio and many others.
These statistics are estimated using 24 observations prior to
the actual evaluation step. For example, by evaluating an
industry during November 2000, the stock prices from
December 1998 to November 2000 are used;
III. We estimate the frequency distributions of the selected per-
formance measures over the whole portfolio opportunity set of
the industry;
IV. The time window is shifted one month forward and the next
simulation commences.

3.2 Data
We include the total return data of two different industries: semiconductors
and mining. The return is calculated as the monthly percentage increase of
the stock price, corrected for dividends and stock splits. The semiconduc-
tors industry is supposed to represent the new economy, whereas mining
represents an old economy industry. All US firms within the Datastream
database 3 with industrial code 936 (SEMIC) and code 04 (MNING) are
selected.
An important issue in this procedure is how to handle changes in the
evaluated industry due to new admissions, mergers, bankruptcies etc. In
case of a delisting company our strategy is to hold the company security

2 (i.e. an analytical 'full' representation in some cases, grid procedures and

simulation)
3 Datastream is apart ofThomson Financial
438

until the last month it was listed. In case of newly admitted companies we
insert a company stock into the industry opportunity set as soon as we have
24 monthly price observations. To avoid a survivorship bias all dead and
delisted firms, which have had at least 24 months of observations, are also
included in our analysis.
Our sample period ranges from December 1980 until January 2001 ,
resulting in a total of 242 months. In December 1980 the Semiconductors
sector consists of 12 firms, which number steadily increases to 90 firms in
January 2001. The number of firms in the Mining sector increases steadily
from 21 in December 1980 to 37 firms in January 2001. Figure 1 shows
the exact number of mining and semiconductors firms in each time interval
of the evaluation period.

Nwnber of stocks in • Semi::onductors Mining

semronductors, miling
100 industries, 198()..200 I

80

60

40

20

0
0 - "' "' ..... "' r- '-0 - "' ..... .... .,..
00 0'- 0 '-0.....
""
CO
- 0
00

""
CO CO 00 00 00 CO
""
00
""
00 CO
"" "" ""
0'-
"" ""
- ""- "'- -"" -"" "'- "'- "'- "'- ""- "'- "'- "'- ""- "'- ""- ""- "'- ""- """"- "' "'
""
0
0
0
0

Fig. 1. N umher of stocks in the semiconductors and mining industries from

1980 through 2001. For the sample period December 1980- January 2001,
the number of firms in the semiconductors industry increases from 12 in
December 1980 to 90 firms in January 2001. In December 1980 there are 21
firms in the Mining sector which number steadily increases to 37 firms
quoted in January 2001. Source: Thomson Datastream

4 Results

In Figure 2 we show in Panel A the distribution of the November 1985

retums that results from all the different semiconductors portfolios one
could make at the beginning of that month. Panel C shows the retums
distribution for November 1995. In addition, the retum distributions for
mining portfo1ios are given for November 1985 and November 1995 in
panelsBand D respective1y.
 439

Panel A: Semiconductors, November 1985 Panel B: Mining, November /985

: r::::::::::·::::::::::::::·::::: ::

300 •.•...... •. ...

~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~

- "'
~ =! "' -;i "' "' .,; "'
0 0 0 0 0 0
.,; "' "'
-"' "'
0 0 0 0 "! .,;
~ 0
,...: 0
,...: 0 f'i .,; ,...: 0 ,...: 0 ~ ~ 0 f'i
N
N
N "'
N N M "7

Return Return

Panel C: Semiconductors, November 1995 Panel D: Mining, November 1995

1500 • •• ••• ..••••• • • ••••••••• •• • •• • •• . 1500 • ••• • ••• • •• . • ••.• . ••• • .• ••• . • • •. •

1200 •• • .•• .•• •. •....•.• . ••.•• ..• • . • • . 1200 ....•...• ..•.• • • .••.•••. . •.• .• • •.
>. >.
g 900
}:,::::::::::::::···::::::::::::::::
.. . .... . . . ... ..
"'
::3
~@
~

300 300 ~ .. .... . .. .. ........ ..

0 0 I Ii 0 Ii Ii I Ii I I i II I I i t iO I i i t t I i 10 , I ,
;:;:: ~ ;:;:: ~ ~ ~ ~ ~ ~ ~ ~
0 0 ~ ~ ~ ~ ~ .,.,
~ ~ .,.,
~ ~ .,., 0>{:_
~
.,; "'f'i 0 "'~ -;i "' f'i .,; "'
~ 0 "' "' "'
0 0 0 0 0 0 0 0
,...: ~
~ 0
~
,...: .,; f'i 0 .-: -;i ~ 0 f'i .,;
'
Return Return

Fig. 2. Realized return distributions of semiconductors and mmmg

portfolios for November 1985 and November 1995. Panel A shows the dis-
tribution of retums in November 1985 for portfolios composed from semi-
conductors stocks (mean 14.236%). Panel B shows the distribution ofretums
in the same month for portfolios that are composed from the mining industry
stocks (mean -0.293%). Panel C and D give the distributions of retums in
November 1995 for the semiconductors portfolios (mean -2.434%) and the
mining portfolios (mean -7.753%). The bin range is 0.005% in all frequency
distributions.

Figure 3 presents in Panel A the distribution of 24-months averages of

rate ofretums in November 1985, that results from all the different semi-
conductors portfolios. Panel C shows the average retums distribution for
November 1995. In addition, the average return distributions for mining
portfolios are given for November 1985 and November 1995 in panels B
and D respective1y.
440

Panel A: Semiconductors, November 1985 Panel 8 : Mining, November /985

7500 .• . ... ... ...................... ... 7500 .... . . . ....... ...... ... ....... .... .

0 0
'#. '#. '#. '#. 0;!!. '#. ;!!. '#. '#. '#. '#. '#. '#. '#. '#. '#. '#. '#. '#. '#. ~ '#.
"' ~ "'! 0 "' ...;
9 0 "'
0 "'! <=!
"'
0
~ 0 <'i <'i
0
~
"'! <=! 0
"'
9 0 0
0<=! "'!
<'i "' <=!
N
"'
Return Return

Panel C: Semiconductors, November 1995 Panel D: Mining, November 1995

7500 ...... .... . ... ... .... .. . ......... ..

(j)X) ............. . ................... .. : r::::::::::::::::::::::::::::::::::

lnl
·~
0 0
'#. '#. '#. '#. '#. '#. '#. '#. ~ '#. ~ '#. ~ ~ '#.
0
~ '#. '#. ~ ~ '#. '#.
~ "'! ...; "'
9 0 "'
"' "'<'i
<=! 0 0 0
...; ..; "'
"' "' 0_,; _,;"' ...: "' oO "' ...; ..;
0 0 0 0
..; v; .,; ...: oO 0 <'i

Return Return

Fig. 3. Distributions of 24-months average returns of semiconductors and

mining portfo/ios for November 1985 and November 1995. Panel A shows
the distribution of average returns in November 1985 for portfolios composed
from semiconductors stocks (mean 0.007%). Panel B shows the distribution
of average returns in the same month for portfolios that are composed from
the mining industry stocks (mean -0.857%). Panel C and D give the distribu-
tions of average returns in November 1995 for the semiconductors portfolios
(mean 5.372%) and the mining portfolios (mean 0.961%). The bin range is
0.005% in all frequency distributions.

Panel A of Figure 4 graphically shows all 218 (242-24) consecutive

retum distributions of the semiconductors portfolios, similar to those in
Figure 3. Foreach period, the retum distribution is mapped on a vertical
bar. The dot on the bar represents the median value of the distribution. The
length of the fat bar represents the retum range of the middle 50% of all
portfolios. The thin bar (only shown for a few months at the beginning of
the period studied) represents the retum range capturing all generated port-
 441

folios. Panel B of Figure 4 shows the development of the semiconductors

portfolio variances. For the comparison ofthe development in this industry
with publicly available indexes, we can only use the Dow Jones US Semi-
conductors index. 4 In addition to the distribution summaries, Figure 4
graphs the 24-months moving average of returns (in panel A) and the
variances of the index (in panel B). Figure 5 gives similar graphs for the
mining industry. 5
The graphs clearly reflect several characteristics of the boom during the
late nineties. Other episodes, like the period around the 1987 crash, can be
recognized as well. In interpreting the graphs one should remernher that
for any given period, all portfolio characteristics are estimated using the 24
preceding observations. 6 The boom in the late nineties shows rapidly
increasing median returns for both sectors, be it that the pace of growth
was much higher in the semiconductors sector. Looking at the develop-
ment of the 50% middle return range, one sees that in both sectors this
range is increasing over time. This signals an increasing heterogeneity in
each of the two sectors. The 50% middle return range in the mining sector
is clearly larger than the comparable range in the semiconductors sector.
This may be explained by the fact that the mining industry is increasingly
less diversified than the semiconductors industry, notably after 1989 (cf.
Figure 1). When we compare the graphs that show the development of the
portfolio variances, we observe that the median variance is quite stable
until the mid nineties, after which they start to grow in both sectors. The
increase of the median portfolio variances in the semiconductors sector at
the beginning of 2000 is remarkable. After 1989 again, the variance in the
semiconductors sector is consistently higher than the variance in the
mining industry. In contrast, there is more heterogeneity in the mining
sector - now in terms of portfolio variances - than in the semiconductors
sector, for the entire period. Again, this may be explained by the lower
degree of diversification in mining.

4 This index consists of 81 firms in December 2002.

5 The Dow Jones US Mining index consists of only one stock in December 2002.
Thus it can be hardly used for comparison.
6 We do not use the exponential smoothing.
 442

14% ......... .... ...................................... ... ............... .. ............................................ .... lt ..

Panel A ~,1
10% ..................... .. ....... ............................. .. ........ .... ................................................ ~!

~
6% .... .~~~ ··· ..... ..................... ...... ............ .... ................. ····r··· ......................... ,....... .

~ Y'" ~litll 11t\~ '

1 111 1 1 II
2%
d I
1\ 141 ~ ~~l I ~~~ o~.t't~
. . ~ ............. 1.... .... .... 1t!i~ .... ~#, .............~H..\ ........ ......................... . . .lf.i.L........... .
1
ti I

I 11 ,~.t, ~ IL,f!l lklll. ll1t.~l ~ ~

11Ji 1tfi llll .... 'H ·q lljl

-2% ''""'''''"'''""''""'""""''"'"'"""''""''''""'''""''""'''""'"'''""'"""'"""''"""''"""'"""'"""'""""'""'"'"""""""'"'"'"""""""'"""'"""""'"""'

:::: •...,. ~
0,020 ..... ....... ........... ~ .............. .. ................................................ ~;;...... ..
~·
HIIIIH~~...................../
I

0,000 .---.-----.-r---.-----.-.----,----.-.....-~--,.---.----.--.----.----,--r---.-
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ g
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 8 ~ ~ ~ ~ ~ ~ ~

Fig. 4. Distributions ofsemiconductors portfo/ios. For the period December

1982 through January 2001 some statistics of the semiconductors portfolios
distribution are presented. Panel A shows the distribution of returns in semi-
conductors and panel B presents shows the distribution of portfolio variances
for the same period. The dot on the bar represents the median value of the
distribution. The length of the fat bar represents the retum range of the
middle 50% of all portfolios. The thin bar (only shown for a few months at
the beginning of the period studied) represents the return range capturing all
generated portfolios.
 443

14% . .. ... ..... ... .

PanetA

10% - . . . .. . .. . . .. ... .. ...... . .... . . .. ........ .. .. ....... ..

0,060

Fig. 5. Distributions of mining portfolios. For the period December 1982

through January 2001 some statistics ofthe mining portfolios distribution are
presented. Panel A shows the distribution of retums in mining portfolios and
panel B presents shows the distribution of portfolio variances for the same
period. The dot on the bar represents the median value of the distribution.
The length of the fat bar represents the retum range of the middle 50% of all
portfolios. The thin bar (only shown for a few months at the beginning ofthe
period studied) represents the retum range capturing all generated portfolios.

To gain some insight in the asymmetry of portfolios of stocks, we have

estimated both the positive and the negative semivariance of each of the
generated portfolios. The results are summarized in Figure 6.
444

Panel A: Medians ofthe negative semivariance distrihutinn."

0.01 -· - - . -- . - -· - - Mml!-S<mcxnb:Ds

0.03

Panel B: Medians ofthe positive semivariance distributions

Fig. 6. Median semivariances of semiconductors and mining portfolios. For the

sample period December 1982 - January 2001 (218 months) the negative and
positive semivariance distributions of the two industry sectors are calculated.
Panel A shows the medians of the negative semivariance distribution of semicon-
ductors and mining. Panel B gives the medians of the positive semivariance distri-
bution for both industries.

The two graphs in Figure 6 are constructed by connecting the medians

of 218 consecutive distributions, which is like connecting the dots in
Figure 4 and Figure 5 respectively 7• The first graph shows the development
over time of the median of the negative semivariance distributions for both
the semiconductors and the mining industry. The second graph shows the
development of the median of the positive semivariance distributions.

7 We do not show the complete distributions of the semivariances in order to

save journal space. Readers who require more information are kindly invited to
contact the authors.
 445

Apart from the last five years, the two sectors do not seem to behave very
differently. For the period until1995 it is hard to conclude from the graphs
whether the positive or the negative semivariance is higher on average.
However, after 1995 it is clear that the portfolios' positive semivariances
win from the portfolios' negative semivariances. This is true for both
industries.
Another way of gaining insight in the skewness of portfolio returns is to
follow particular portfolios over time. Figure 7 shows the frequencies of
the median portfolio realized returns over the entire period. When we
compare the entire period, the performance of the two industries is quite
different. The median portfolio returns distribution of semiconductors is
more asymmetrical than the mining portfolio. For the sample period
December 1982 - January 2001 (218 months) the median returns of the
two industry sectors are calculated. Panel A shows the time series median
returns of the semiconductors and panel B gives the median returns of
mining. The average median return of the semiconductors (3 .36%) distri-
bution is higher than the average of mining (2.08%). The standard
deviation in median returns of semiconductors (12.72) is also higher than
the standard deviation ofmining (9.03).
446

Panel A: Semiconductors Panel 8: Mininl?

20 December 1982- January 2001 20 December 1982 - January 2001

18 18
16 16
14 14

i :~
~ 12

1108
u. ! 8
6 6
4
2
o ~~~~~~~~
s~ '$ ~
~ ~ ~ ~
~ ~

20 December 1982- December 1998 20 December 1982- December 1998

18 18
16 16
14 14
~ 12 ~ 12
10 ii
!
~
[ 10
8 "- 8
6
4
2
0
~
:5
"r

1982-2001 (n=218) 1982-2001 (n=218)

Average 3.36% Average 2.08%
Standard deviation 12.72% Standard deviation 9.03%
Skewness 6.78 Skewness 5.24

1982-1998 (n=192) 1982-1998 (n=192)

Average 2.58% Average 2.13%
Standard deviation 9.65% Standard deviation 8.53%
Skewness 3.71 Skewness 5.82

Fig. 7. Median monthly portfolio returns of semiconductors and mining.

Panel A shows the median returns of the semiconductors sector for the entire
period of218 months (upper part ofpanel A) and for the entire period with-
out the 25 booming months, leaving 192 months (lower part of panel A). In
panel B we show the results for the mining sector.

The distribution in Panel A is more asymmetrical (Skewness: 6. 78) than

the distribution ofpanel B (Skewness: 5.24). However, ifwe leave out the
last 25 observations (the boom period), the differences between the
 447

arithmetic means and the standard deviations of the two distributions

become smaller. However, the skewness of the mining industry is higher
than in the case of 218 Observations (as shown in the lower part of panels
A and B) while the skewness of the semiconductors is much lower than
before and also considerably smaller than the skewness of the mining
industry. This underlines the importance of the size and unbalanced
changes in the semiconductors sector during the boom period.

5 Conclusions

Empirical evidence suggests that stock returns are not normally dis-
tributed. The above results suggest this is also true for portfolios of stocks.
In fact both median industry portfolio have fat tailed and positively skewed
return distributions. During the boom period (1999-2000), the semicon-
ductors industry showed an enormous increase in skewness.
The graphs in Figure 4 on the development of the performance of semi-
conductors portfolios can be compared with similar graphs of the mining
industry in Figure 5. Wemade this comparison in terms of four different
performance measures: return, variance and both negative and positive
semivariance (the latter two in Figure 6). Much to our surprise, the two
industries behaved very similar for most of the period (actually until the
beginning of the boom) in terms of all the performance measures used.
The last two years in our sample, the two booming years before the bust,
the performance of the two industries was quite different. In the mining
industry there seems to be a moderate build-up of average returns over
time and a somewhat higher variance during the last few years. In contrast,
the developments in the semiconductors industry are truly extraordinary.
The average returns show an enormous growth, as do the average portfolio
variances and both positive and negative semivariances. It is important to
note that the upside semivariance in the semiconductors industry grows
twice as fast as the downside semivariance during this period. This sug-
gests that the differences between the firms within this industry, seen
through the eyes of the investors, are getting bigger and bigger throughout
· this period.
When comparing positive and negative semivariance, the positive part
seems to beat the negative almost continuously. An important exception is
right after the 1987 stock market crash when the negative semivariance
was winning for some time. In the near future, we will further investigate
whether the millennium bust shows the same pattem as the period after
October 1987. The same holds for the minor crashes in between.
448

In this article, a new way of looking at the stock market performance of

different industries and their differences has been introduced and illus-
trated. A number of refinements and extensions can be made. One example
is to change the way the variances and semivariances are estimated.
Presently, each data entry gets the same weight in estimating these
characteristics. Because of the depreciation of the information quality of
older data, exponential smoothing of the data might be considered, next to
other techniques with similar effects. But first and foremost we will extend
our dataset. One may question whether a clear distinction between 'new
economy' and 'old economy' or between 'high growth' and 'low growth'
stocks can be defined. We tend to believe that the differences between the
industries are more gradual. We hope to shed more light on this issue by
also analyzing and comparing other industries.

References

Fama E (1965) The Behavior of Stock Market Prices. Journal ofBusiness 38:34-
105
Groenendijk A, Spronk J (1999) Portfolio Perfonnance through the Eyes of
Monkeys. In: Bonilla M et al (eds) Financial Modelling. Physica-Verlag,
Heidelberg, pp 203-213
Hallerbach W, Hundack C, Pouchkarev I, Spronk J (2002) A Broadband Vision of
the DAX over Time. (Report by Erasmus Research Institute of Management,
ERS-2002-87-F&A, available at http://www.eur.nl/WebDOC/doc/erim/erimrs
20021008171714.pdf)
Harvey C, Akhtar S (2000) Conditional Skewness in Asset Pricing Tests. Journal
ofFinance 55:1263-1295
Kester C (1984) Today's options for tomorrow's growth. Harvard Business
Review, March-April: 153-160
Knez P, Ready M (1997) On the Robustness ofSize and Book-to-market in Cross-
sectional Regressions. Journal ofFinance 52:1355-1382
Kon S (1984) Models ofStock Returns- a Comparison. Journal ofFinance 39:147-
165
Kraus A, Litzenherger R (1976) Skewness Preference and the Valnation ofRisky
Assets. Journal ofFinance 31:1085-1100
Pouchkarev I (2004) A General Framework for the Evaluation of Constrained
Portfolio Perfonnance. Ph.D. thesis, Erasmus University Rotterdam, to appear
Trigeorgis L (1996) Real Options: Managerial Flexibility and Strategy in
Resource Allocation. MIT Press, Massachusetts
Valuation of Growth Companies and Growth
Options

Markus Rudolf'

1 Abstract

The distinction between growth and value shares is a continuous issue in

the portfolio management literature. It is quite clear that growth and value
companies behave differently in terms of risk and return. This is due to the
fact that fast growing companies frequently suffer substantial initial Iosses
but fast growing revenues. This obviously must affect the way of company
valuation. Nevertheless, the development of growth company valuation
models has been much more recent. An important ingredient of the valua-
tion models for growth companies is the real options approach. This paper
addresses two aspects of growth company valuation. First, it uses the
model of Schwartz and Moon (200 I) for the valuation of NEMAX 50
growth companies. Secondly, it tries to isolate the growth option compo-
nent in the company values by using an approximation approach for the
value of options in incomplete markets.

2 lntroduction

Much has been written about the valuation of companies. Examples are
well known textbooks like Copeland, Koller, and Murrin (2000),
Damodaran (1995) or Loderer, Jörg, Pichler, Zgraggen (2000). Most com-
pany valuation models are based on discounting cash flows occurring in
the future. This requires positive expected future cash flow estimates with
reasonable estimation errors. It is typical for growth companies that they
generate negative. cash flows, possibly for many years, before they turn to
be profitable. Therefore, using cash flow based valuation models yields
negative firm values, which obviously contradicts reality.
A second weakness of cash flow based valuation models is related to
identifying the correct discount rate. According to Gordon and Shapiro

• The author is grateful to many editorial improvements contributed by Günter

Fandei and Cathrin Stammen-Hegener.
450

(1956), this dominantly depends on the cash flow growth rate. While esti-
mating growth rates implies comparatively low uncertainty for value com-
panies, growth rates in growth companies tend to be significantly higher
and reveal a much higher degree of uncertainty. In fact, revenue growth is
a key success factor for growth companies because high revenues allow
them to achieve a high market share. Excessive growth is aspired even if
this turns the cash flows negative. Therefore, it may take years until
growth companies are able to generate positive cash flows.
Hence, there are two problems associated with the classical way of
company valuation when applied to growth companies. One is the cash
flow estimation and the other is the growth rate estimation problem.
Schwartz and Moon (2001) deal with these two problems and develop a
modified technique for the valuation of growth companies. This paper pro-
vides an introduction to the firm valuation model of Schwartz and Moon
(2001) and applies it totheGerman growth company market.
Since the mid 90ies, real option models are increasingly considered as
part of the solution of the growth company valuation dilemma described
above. Trigeorgis (1996) has summarized the real option discussion in a
textbook. One of the central ideas of the real options Iiterature is that
entrepreneurial decisions imply a certain degree of flexibility. Flexibility
can be considered as an option to behave in a certain way. Examples for
these options are the option to expand an existing business (growth
option), the option to abandon an investment project, the option to wait
with investing the money, and so on. Obviously, entrepreneurial freedom
implies real options. The real options Iiterature uses pricing models for
financial options and copies them for the valuation of real options. Cer-
tainly the most important valuation model for financial options is Black
and Scholes (1973). However, this is frequently criticized, e.g. by
Ballwieser (2001). Critics assert that while real options are similar to fi-
nancial options with regard to their flexibility, there arealso important dif-
ferences violating the assumptions of option pricing theory. Particularly,
the pricing of financial options is based on the assumption that the under-
lying asset is a traded security. This is obviously violated by many real
options. For instance, the option to grow can only be priced by Black and
Scholes (1973) ifthe existing company has traded shares. This assumption
however, is frequently violated by young growth companies, which did not
yet go public. While the application of options pricing models is not possi-
ble in this case, Cochrane and Saa-Requejo (2000) provide approximation
formulas for financial options, which can be applied to the growth option
valuation problem. Applying this approximation methodology to growth
options implied in growth company values, allows separating the option
 451

component from the cash flow component in the company value. This will
be shown as second part ofthe paper.
The paper is structured as follows. The next section contains the basic
equations of the simulation models of Schwartz and Moon (200 1). Section
3 introduces and discusses the empirical results of applying the model on
the German market for growth stocks. This part is based on Keiber,
Kronimus, and Rudolf (2002). The approximation formula for growth op-
tions based on Cochrane and Saa-Requejo (2000) is then invented in sec-
tion 4. Section 5 shows how to apply this formula to growth options im-
plied in the value of a growth company. The last section concludes the
paper.

3 Simulation model

Cash burning is frequently considered as a typical property of growth com-

panies. That means that the operational business does not yet produce
revenues high enough to cover all of the expenditures. As a consequence,
profits are negative in the first years a:fter the start-up phase. Expenditures
of the operational business are financed by a stock of cash, which has been
acquired by one ore more financing rounds. Investors contributing cash are
venture capitalists, stockholders, or other business partners. The growth
company uses the initial cash balance to develop and expand the business
and to capture the market. This leads to melting cash balances but ideally
to increasing market shares. As a consequence, the cash balance function is
U-shaped, starting at a relatively high level, then a certain fraction per
period is "burned" until the turning point of the company is reached. This
is the time when the company starts to generate positive profits. Figure 1
compares 2 different cash balance functions for growth and value compa-
nies. While the value share grows at a constant rate, the cash balance of the
growth share is U-shaped reflecting the "cash burning property" described
before.
452

350
300 •
250
200
150 k X JC II
..
100
IC lt X X M

50
0
• •
+---------~
- --~~~
--.._ . ......
· -~·----------------------~

0 5 10 15

• NEMAX - stock x Va lue share

Fig. 1. Cash balance in growth and value companies compared

The U-shape of the cash balance function implies two difficulties in the
valuation of growth companies. The first problern is that the value of the
company is based on positive cash flows occurring in the future, not even
necessarily in the near future. Hence, the estimation uncertainty is con-
siderable. Secondly, while value companies have more or less constantly
increasing cash flows, growth companies have significant cash flow
growth rate volatility, again making an estimation of future cash flows
more complex. These two effects are reflected by the stock prices. Figure 2
compares 3 shares out of three segments. United Internet is the second
largest internet provider in Germany and reveals all typical characteristics
of growth companies. According to figure 2, the share price varies heavily
between 1998 and 2002. Novartis is one of the largest pharmaceutical
companies worldwide and hence a typical value share. The share price
volatility is clearly below United Internet. Finally, Qiagen operates in
similar markets like Novartis. However, it is still a young and fast growing
biotechnology company. Therefore, the price volatility is clearly higher
than for Novartis. Figure 2 compares the three price charts.
 453

Stock price in €

70 ·········--·····--··-················--·······-······-----------------------------------------------------------
60
50
40
30
20 -

10
0
c:o
"'"'"' "'
1'- 0 0 N N C')
0 0
<X)

"'"' "' "'"' "'~ 0 0 0 0 0

w
~
"' 1'- N a:i
0
N
,..;
0
N
c:i
0
N
~
0
"!
0
N
.n
0
N
N
0
N
w
0
~

0 0 0 0 0 0 0 0
ci .n ci cO .n .n c:i cO
~ 0
~
N 0 N ~ 0 "'
~ 0
~
N N

Fig. 2. Stock prices ofNovartis (value share), United Internet (growth share), and
Qiagen (consolidating share), daily data from 6 April 1998 to 26 November 2002

United Internet AG (Quarterly data)

D Cash balance
100.000 --·--------------------------------------------------------------1
80.000 I
60.000 EBIT
\JI) I
0 40.000
0
0 20.000
..!: 0
-20.000
-40.000 -+- Cash-Fiow
-60.000 (operational and
i nvestments)

Fig. 3. Cashbalance and revenue figures ofUnited Internet AG; Source: Quarterly
reports at www.united-internet.de

Figure 3 contains cash balance, cash flow, and revenue figures sup-
porting the characterization of growth companies above. Figure 3 shows a
high revenue growth rate for United Internet. Simultaneously, the cash
flows arenegative until the third quarter of 2003. Since then, the situation
seems to be stabilized, although the economic downturn in the second half
of 2002 avoids a continuous advance in cash flows. The cash balance is
not exactly like the idealistic U-shape in figure 1. However, the cash
burning effect gets quite obvious form figure 3. The cash balance starts at
a Ievel of little more than € 6 Mio in the first quarter of 1997. This sufficed
454

for 3 quarters, then the cash balance reached a Ievel of 0. This would imply
the default. However, the United Internet management managed to acquire
additional funds either by issuing fresh shares or by selling participations
in companies. These are discrete events reflected by peaks in figure 3, e.g.
in quarter 4 in 1997, quarter 1 in 1999, and quarter 2 in 2000. The major
difference compared to value companies is that the shift of the cash
balance is not due to operational profits but due to investment transactions.
Isolating these investments would produce the U-shape in the cash balance
function for United Internet. Since the second half of the years 2002,
United Internet seems to be in the transition from a growth to a value com-
pany. This is so because both, the cash flows and the eamings before in-
terest and taxes turn positive.
Capturing properties of growth stocks requires a non-traditional model
type. Schwartz and Moon (200 1) provide a mode1 appropriate for the
characterization of growth stocks like United Internet. The model is a
Monte Carlo simulation approach based on three stochastic factors. Unlike
traditional company valuation models, cash flows are modeled based on 2
separated stochastic processes, revenues and costs, which reveal imperfect
correlation. The 3rd factor is especially interesting. It models the expected
change in revenues. In other words, the difference to traditional company
valuation models is not only the distinct modeling of revenues and costs in
order to obtain the cash flows. It is furthermore assumed that expectations
are uncertain and reveal certain stochastic properties. Practically, this im-
plies that expectations vary across investors and financia1 analysts. In fact,
this is especially true for growth companies. Expectations are much more
homogenous for value companies than for growth companies. Specifically,
the following stochastic specification hold according to the Schwartz and
Moon (2001) model:
• 1st factor: This models revenue changes assumed to follow a Geometrie
Brownian motion. More specifically, if !l(t) indicates the expected reve-
nue change, cr(t) the per annum volatility in revenue growth, dzR a stan-
dard Wiener process for the revenue changes, and dt an infinitesimal
time interval, then the changes in revenues are modeled according to the
following equation:

dR(t)
R(t) =,u(t)·dt+o-(t)·dzR(t) (1)

• 2nd factor: This factor represents changes in variable costs. Together

with the first factor, this allows a determination ofthe cash flow process
over time. The variable costs process is modeledas Ornstein-Uhlenbeck
process. This implies the mean reversion property, i.e. the assumption
 455

that young companies have higher variable cost rates than more mature
companies. But for each company, the variable cost rate is assumed to
converge towards the variable cost rate of value shares. Let C(t) be the
cost function with F as the fixed cost and y(t) as the variable cost com-
ponent. Furthermore, Ky is assumed to be the mean reversion speed for
the variable costs, r
is the variable cost rate of the industry average,
q>(t) is the volatility in variable costs, and dz~p is a standard Wiener
process representing the unexpected component in variable costs. Then
the 2nd factor is represented by the following equation:

dy(t) =Kr ·(r- r(t)) ·dt + qJ(t) · dzr (t), (2)

where the total costs are fixed costs plus a percentage of the revenue as
variable costs:
c(t )= F + R(t )· r(t ), (2)

• 3n1 factor: This factor models expected revenue changes as Omstein-

Uhlenbeck process. This implies the mean reversion property for
expectations, too. The practical consequence isthat growth companies
are expected to generate revenue growth rates significantly above those
ofvalue shares. As time goes by, the revenue growth rate is expected to
go down in a non-predictable way. However, in the long run, as growth
companies have converged to value companies, the expected revenue
growth rate is at the Ievel of mature companies. Let K be the mean
reversion speed with which the expected growth !l(t) converges towards
the industry average Ji, 11(t) be the volatility in expected revenue
changes, and dzf.l a standard Wiener process representing the unexpected
component in expected revenue changes. Then we have the following
representation for the 3rd risk factor:
dp(t) =K·(Ji- p(t ))· dt + q(t )· dzJl (t) (3)

Furthermore, the volatility process for expected revenue changes con-

verges to 0, i.e. the uncertainty of revenue expectations gets 0 if the
company has become a value company:
da(t) =Ku ·(a- a(t ))· dt (4)

dqJ(t) = Kcp ·({ö- qJ(t ))· dt (5)

In (4) and (5), Ka and K~p are the mean reversion speeds ofthe revenue
change volatility cr and the variable cost volatility K~p. The two equations
456

model volatility as time dependent but non stochastic which reflects the
assumptions outlined above.
The model requires simulating those 3 stochastic processes, with regard
to the pre-specified correlation structure, for 100 future quarters. The reve-
nues - expected revenues, revenues - variable costs, and expected revenues
-variable costs correlations are represented by (6) to (8):

dzR ·dzf.i =PRp ·dt (6)

dzR ·dz1 =PRr ·dt (7)

dz1 ·dzf.i =Prf.i ·dt (8)

The simulation of the random variables yields one realization of a cash

balance in each quarter and in each simulation run. An example for the
simulation path of the revenue changes can be seen in figure 4. The erratic
line represents the simulated revenue change and the even line the
expected revenue changes according to process (3). The simulations oscil-
late around the expected path but deviations to both sides are of course
possible due to the stochastic nature of the simulation. The assumed mean
reversion speed in figure 4 is K=0.07. This implies a half-life ofmore than
9 years, i.e. the distance between the realized growth rate and the peer
group growth rate halves every 9 years. The figure is based on only 1000
simulation runs.

0,20 - . - - - - - - - - - - - - - - - - - - - - - ,
.I::. ....
~ ~ 0,15
._ ro
0"1::::1
~ ~ 0,10
c: <l>
<l> a.
> <l>
~ ~ 0,05

0 ,00
_____ .,.
-
__ __ _ _ _ _

0 5 10 15 20 25

- -- · Estimated: Kappa = 0.07 - - - - - Peergroup growth rate - - Simulated

Fig. 4. An example for the expected and simulated revenue changes

 457

Implementing the simulation approach requires to discretize the con-

tinuous time processes given in (1) to (5). As an example, discretizing the
stochastic differential equation in ( 1) allows to represent the revenue Ievels
rather than the revenue changes in each time step of length ~t. Each of
100.000 simulation steps generates a realization for the standard normally
distributed random variable zR(t). This yields one (out of 100.000) realiza-
tion for the revenue R(t) in period t. In addition to the revenues, expected
revenues, variables costs, and volatilities are discretized in similar ways so
that standard normal random variables zl' and 'Ly yield realizations for both,
expected revenue changes and variable cost Ievels. The discretized forms
of (1) to (5) can be found in Keiber, Kronimus, and Rudolf (2002).
Generating 100.000 zR,zJ.I, and 'Ly realizations is done with regular random
variable generators and with regard to the correlation requirements speci-
fied in equations (6) to (8). Reproducing the correlation matrix with the
realizations of the three random variables requires a Cholesky decomposi-
tion of the correlation matrix. The procedure is for instance described in
Rudolf (2000).

NMKX .l1037. 68 -4. 84 Index DES

At 12:09 Op 1050 .09 Hi 1051.29 Lo 1029.56
INDEX DESCRIPTION PAGE Page 11 4
NMI<X - EURO NEU MI< T BLU CHIP IX
The Neuer Markt Biue Chip Index (Pricel Index, also known as the NEMAX SO Index,
is a welghted Index of the 50 leadlng Stocks that traded on the Neuer Markt .
The Index is revised twice a year in March and September, at this time no stock
can have more than a 10/. weight . The Index was developed with a base value of
1,000 as December 30, 1997 .
l ~IP Prices Value /. Chg Net Chg No Indust ry Groups
Year_to_Date 1145.03 -9 .340 -106 .95 ~MB SO Members 112 136 ~2
~TRA 52 Weeks Ago 1876.78 -44 .688 -838 .70 9MOV Today's Movers by Index Pts
DGPU 52 Week High 2094.73 on 03/07/01 ~ QIAGEN NV +2.798
52 Week Low 638 .48 on 09/21/01 Leading n THIEL LOGISTIK +1 .073
Movers 8l SINGULUS TECH +. 938
Fundamental Information ~IM INTERNATIONAL +.738
Prlce/Earnlngs neg Ex-Dvd -.1651 lOJ MOBILCOMAG -1.170
Dividend Yield .55 on 08/01/01 Lagglnglll AIXTRON AG - 1.043
Index Information Movers l~ DAB BANK AG - .847
Currency EUR I EURDPE lD LION BIOSCIENCE -.815
Volume 19,474,080 on 03/04/02 J~N News on Today's Movers
Market Cap 39 .4SBLN No Futures Avallable
Divisor 3.72472363 No Options Available

Fig. 5. Description of the "Neuer Markt" according to Bloomberg page NMKX at

5 March 2002

After having simulated 100 quarters, a frequency distribution for

100.000 different cash balances can be created. The expected value ofthe
cash balance discounted by the risk-adjusted discountrate and adding the
458

continuation value yields the company value (RADF: risk-adjusted dis-

count factor, CV: Continuation value):
Company value =Cash Balance in quarter 100 * RADF + CV (9)
Keiber, Kronimus, and Rudolf (2002) apply the model to 46 NEMAX
50 companies. The NEMAX 50 index represents the largest companies
quoted at the Frankfurt "Neuer Markt". The most important characteristics
of that market segment can be found in the Bloomberg screen print in
figure 5. Due to a loss in credibility of many "new economy" companies
caused by balance sheet manipulations, this market segment will be dises-
tablished until March 2003. It will be replaced by other existing or newly
founded indexes. The NEMAX 50 as an index for growth companies will
continue to exist. However, there will be a major reorganization of the
DAX, SDAX (small caps), MDAX (domestic and foreign mid caps), and
NEMAX 50. Additionally, a new index named Tec-DAX collecting the 30
largest technology companies, most of them taken out of the NEMAX 50,
is established at 24 March 2003. The calibration ofthe Schwartz and Moon
(200 1) model requires specifying many parameters in addition to those
driving the stochastic processes specified above. The following 30 vari-
ables are required as input variables for the model:
• Revenue in t=O
• Initial cash balance in t=O
• Loss-carry-forward in t=O
• Plant, property, equipment in t=O
• Expected revenue growth rate t=O
• Volatility revenuegrowth rate in t=O
• Volatility expected revenuegrowth rate in t=O
• Expected variable cost changes in t=O
• Volatility of the changes in variable costs in t=O
• Peer group revenue growth rate
• Peer group revenue volatility
• Peer group expected change in variable costs
• Peer group variable costs volatility
• Correlation revenue and expected revenue growth rate
• Correlation revenue and variable costs
• Correlation expected revenuegrowth rate and variable costs
• Mean reversion speed revenue growth rate
• Mean reversion speed revenue volatility
• Mean reversion speed expected revenue growth rate
• Mean reversion speed variable costs
 459

• Mean reversion speed variable costs volatility

• Fix costs
• Depreciation rate
• Capital ratio (investment I revenue- ratio)
• Tax rate, risk less interest rate
• Market price of risk revenue
• Market price of risk expected revenue
• Market price of risk variable costs
• EBITDA multiple: 10
Classical company valuation models content themselves with expected
cash flows projected in the future, discount rates, and terminal value mul-
tiples as input variables. Hence, compared to classical company valuation
models in finance, the Schwartz and Moon (200 1) model implies an im-
mense extension of data requirement. Of course, this does increase the
problern of correctly specifying the data. But it also allows for a more ade-
quate description of the processes driving the company value. Most of the
data has been collected from all available quarterly reports until the fourth
quarter ofthe year 2000. Keiber, Kronimus, and Rudolf (2002) describe in
greater detail how each of those input variables has been identified. The
results for selected companies are presented in the next section.

4 Results

The study of Keiber, Kronimus, and Rudolf (2002) presents the results of
the simulation study for 46 NEMAX 50 companies in greater detail. Some
characteristic examples are selected here and interpretations are provided.

Table 1. Simulation results compared to capital market data for Qiagen

Risk class I
Bankruptcy risk 0%
Simulated firm value in mio. € (31 .12.2000) 3 96 1
Observed firm value in mio. € (31.12.2000) 5 488
Simulated stock price (31.12.2000) 27.88
Observed stock price (31.12.2000) 38.70
Stock price 7 March 2003 5.45
460

Table 2. Simulation results compared to capital market data for AIXTRON

Risk class 1
Bankruptcy risk 0%
Simulated firm value in mio. € (31.12.2000) 3 891
Observed ftrm value in mio. € (31.12.2000) 3 753
Simulated stock price (31.12.2000) 60.04
Observed stock price (31.12 .2000) 57.90
Stock price 7 March 2003 2.40

Table 3. Simulation results compared to capital market data for EM.TV

Merchandising AG
Risk class 2
Bankruptcy risk 6%
Simulated firm value in mio. € (31.12.2000) 2976
Observed fmn value in mio. € (31.12.2000) 3 021
Simulated stock price (31.12.2000) 5.57
Observed stock price (31.12.2000) 5.87
Stock price 7 March 2003 0.79

Table 4. Simulation results compared to capital market data for United Internet
AG

Risk class 2
Bankruptcy risk 20%
Simulated firm value in mio. € (31.12.2000) 380
Observed fll111 value in mio. € (31.12.2000) 316
Simulated stock price (3 I .12.2000) 5.54
Observed stock price (31 .12.2000) 4.30
Stock price 7 March 2003 7.37

Table 5. Simulation results compared to capital market data for BROKAT AG

Risk class 4
Bankruptcy risk 60%
Simulated firm value in mio. € (31.12.2000) 522
Observed firm value in mio. € (31.12 .2000) 843
Simulated stock price (31.12.2000) 10.70
Observed stock price (31.12.2000) 19.37
Stock price 7 March 2003 Default since 12 /200 1
 461

Table 6. Simulation results compared to capital market data for Pixelpark AG

Risk class 4
Bankruptcy risk 28%
Simulated firm value in mio. € (31.12.2000) 280
Observed firm value in mio. € (31.12.2000) 678
Simulated stock price (31.12.2000) 14.84
Observed stock price (31.12.2000) 36.10
Stock vnce 7 March 2003 0.59

Table 1 presents the results for Qiagen, a growth company belonging to

the biotechnological sector revealing positive cash flows already since
several years. The simulated firm value at December 2000 was € 3.9 bil-
lions. This implies an overvaluation of the company by more than 38%. In
terms of share prices, this implies that the observed price of € 38.70 ex-
ceeds the simulated "fair" price of € 27 .88. Although this suggests that the
stock market price was too high, the realized share price until March 2003
was € 5.45, hence only one fifth of the model price. Whether this rejects
the model or supports the hypothesis that the "Neuer Markt" over-reacted
after the hurst of the "New-Economy-bubble" since March 2000, is hard to
tell. After all, the fundamental figures of Qiagen published by the quarterly
reports suggest a value significantly above the price. Bankruptcy risk indi-
cates the percentage of defaults occurring among the 100.000 simulation
runs due to the default condition that the cash balance touches the 0-level.
While Qiagen did not have any runs ending with default, the default risk
for Brokat (see table 5) was 60%.
The second selected example is AIXTRON one of the world' s leading
manufacturers of equipment for the production of compound semiconduc-
tors. It is a spin-of of the Technical University of Aachen and operates
quite successfully since 1983. The results for Aixtron are summarized in
table 2. Revenues and costs developed well such that the bankruptcy risk
in the simulations is as low as 0%. However, the simulated stock price for
the end of 2000 is slightly higher than the observed price. Despite of the
positive result of the simulation study, the observed stock price in March
2003 is only 4% ofthe simulated level.
Table 3 contains the results for EM.TV, a media company based in
Munich. Em.tv is a supplier and co producer of animated films, a licensing
agent of marketable quality items, service provider with regard to product
development and designs as well as point of sales promotion, marketing of
mega event. The share price suffered a tremendous loss when the Kirch
Group defaulted in June 2002. Even earlier, the share price came under
pressure when two members of the executive board came on charge of
462

balance sheet forgery. Keiber, Kronimus, and Rudolf (2002) based on data
out of quarterly reports found a bankruptcy risk of only 6% (risk class 2).
This was however based on the assumption that the quarterly reports
contain valid data. The current share price is only 79 €-Cents and reflects
the difficulties the company is in.
Table 4 contains the results for United Internet, an Internet provider
based in Montabaur, Germany. Key characteristics ofthat company were
already addressed above. United Internet is a typical representative of the
"New Economy" sector. While a lot of cash has been bumed in the start-up
phase, the turn around seems now to be closer. In December 2000, the
simulation model indicated a bankruptcy risk of 20%. The simulated share
price was little higher than the observed share price. United Internet is the
only company with an increased share price compared to the time of the
hurst ofthe "New Economy" bubble.
An example for a failed "New Economy" company is Brokat, a pro-
ducer of software especially banking software. In December 2001, the
company went bankrupt. The simulation study yielded a bankruptcy risk of
60% and a simulated share price significantly below the observed stock
price in December 2000. As a matter of course, the stock price today is
zero.
The last example selected here is the e-service company Pixelpark The
cash-burning rate used to be significant and could only be abided due to
repeated help by the German media conglomerate Bertelsmann. The stock
price was continuously under pressure since the beginning of 2002.
Bertelsmann divested the former participation in December 2002. The
price in March 2003 is only 59 €-Cents.
Tables 1 to 6 contain 6 examples out of 46 NEMAX 50 cases in total.
Figures 6 depicts a regression of the observed stock prices on the simu-
lated firm values in December 2000. The results are heterogeneous. The
simulation model explains 47% ofthe firm values. The sensitivity between
observed and simulated form values is 0.6349 and is significantly different
from 0. This shows that the model explains at least a significant part of the
growth company values. However, it does not provide accurate estima-
tions. Figure 7 explains the deviations between simulated and observed
values in greater detail. A tendency of figure 7 overvalues companies
belonging to risk classes 1 and 2, whereas class 3 and 4 companies tend to
be under priced. Hence, the market considers growth companies as more
homogenous than they are. Bad and good companies in the NEMAX 50
indexarenot clearly enough separated from each other. This interpretation
of table 7 is also consistent with the closure of the "Neuer Markt" in
March 2003. The growth segment of the German exchange is replaced
particularly because of the heterogeneous quality between the companies
 463

so that it is very difficult for investors to distinguish between good

companies and bad companies. If this holds, the model of Schwartz and
Moon (2001) helps to mark-offpromising from poor companies.

IJII 140 ····················· ··-·-·····---~ ---·····-·········-··-------·--···-·············1

I
c
120
y =0,6203x + 16,687
Q)
u • •
••
·;:: 100
0..
.:"::.
u 80 R 2 =0,4669 I
0
tl 60
• • I
I
"U
Q)
c:Q) 40 •
..c 20
(/) I
0
0
0 50 100 150
Simulated stock price in €

Fig. 6. Regression analysis - simulated versus observed prices as of 31 December

2000.

Ordered according to risk classes

140 Lowrisk High risk

120
100
80
60
40
20
0

• Simulated price in E • Observed price in E

Fig. 7. Risk classes- Simulated versus observed prices as of3l December 2000.

Another key characteristic of any simulation model is the generation of

simulated frequency distributions. In contrast to valuation models based on
expected cash flows, simulating the cash flows yields information not only
about the first moment of the probability distribution but also about the
shape. The mean reversion property of the expected revenue changes as
shown in equation (2) results in skewed frequency distributions for both,
464

the company value and the revenues. Figure 8 represents 3 different reve-
nue frequency distributions for Qiagen for 3 different time horizons. Due
to increasing volatility over time, the amplitude of the distributions grows
proportionally to the time horizon. It is particularly interesting that the
distributions are skewed to the right. This implies a limited downside
potential for the revenues but there is an even higher upside margin.
Although the Schwartz and Moon (200 1) model is not a real options
approach, the skewed distribution property is typical for Options: Financial
options and also real options are characterized by an upside potential
exceeding the loss potential. The skewness in the cash flow distribution
induces a higher valuation than symmetrical cash flow patterns. The real
options Iiterature as Trigeorgis (1996) and others use the analogy of option
and company distributions to justify higher company valuations compared
to discounted cash flow models. However, there are many problems asso-
ciated with assigning option pricing theory to company valuation. The
introduction contains arguments for these problems. Obviously, these
problems do not hold for the Schwartz and Moon (200 1) model. Although
they derive skewed probability distributions, too, the model is not based on
the assumptions of the option pricing theory. Hence, the model uses real
options as component of growth company values without having to make
difficult assumptions.

0 2 3 4 5 6
Quarterly revenue in bns. of €

- - Year 1 --Year 5 --Year 10

Fig. 8. Revenue probability distributions for Qiagen.

 465

8% ~----------------------------------------~
c 7% Expected company value
~ 6%
a 5%
('Q

~4%
~ 3%
e
_Q
2%
a.. 1%
0%
0,1 1,0 2,0 2,9 3,9 4,8 5,8 6 ,7 7 ,7 8,6 9,6 10,5 11,5
Company value in bns. of €

Fig. 9. Company value probability distribution Qiagen.

Another options property revealed by growth companies as weil can be

observed in the figures 10 and 11. Similarly to financial options, increasing
volatilities enhance the company values. This is in sharp contrast to tradi-
tional company valuation where risk affects the company value in a nega-
tive way. Figure 11 shows two function graphs. The first graph indicates a
positive relationship between the volatility of expected revenue changes
and the company value. The second graph shows that there is also a posi-
tive relationship between the expected revenue volatility and the bank-
ruptcy risk. On the first glance, this seems to be counter-intuitive because
in classical capital market theory, risk is negatively associated to value. A
deeper Iook however makes clear that the skewness in the probability dis-
tributions of revenues and firm values can imply that volatility is perceived
as chance rather than as risk. This is due to the greater amp1itude of the
right hand side of the probability distribution compared to the left hand
side as it is depicted in figures 10 and 11. There is a given loss potential
regardless of the volatility. However, the more skewed the distributions
are, the higher is the possible extent of the profits. Obviously, an in-
creasing volatility increases the default risk. This is shown by figure 11.
Nevertheless, the chances outbalance the risks. In contrast to figure 11.
figure 10 shows how the volatility ofrevenue changes (not: expected reve-
nue changes) is associated to both, firm value and default risk. As one
would expect in accordance to with traditional capital market theory, this
type of volatility affects the firm value negatively. Revenue volatility
unlike expected revenue volatility refers to realized revenues and not to
fantasy associated to revenues. Therefore, revenue volatility is a classical
risk source. lt is not perceived as chance. Having the real options Iiterature
in mind allows an association of figure 11 to financial options. The chart
466

indicates that there is a real option component implied in the value of a

growth company. Unfortunately, the Schwartz and Moon (2001) model is
not appropriate to explicitly quantify the real option value component and
the fundamental value component separately from each other. An attempt
to extract these two value components can be found in the section below.
Nevertheless, Schwartz and Moon consider real options implicitly, i.e. the
company value is always a combination of the fundamental and the flexi-
bility component.

-
0,9 40%
\111
0,8 35%
0 ;::;-
ui 0,7
c: 0,6
30% :E
.0 25% ~
.5: 0,5 ---*
a.
0
Q) 0,4 > 0 : - x - - -x---*
20%
;;;J
Cö 0,3 15% §
>
10% ~
E 0,2 5% 0
ü: 0,1
0,0 0%
0% 20% 40% 60% 80% 100%
Initial vola!ility of revenue changes

Fig. 10. Firm value sensitivity with respect to different initial volatilities of reve-
nues for Intershop.

-
\.11)

0
u;
c:
1,0
0,8
35%
30%
~
:.cns
.0 .c

-
.S
Q)
0,6 25%
...0
Q.
:J 0,4 20%
ro> 0,2
)(-X-X
15%
:;
~
§ 0,0 10% c
Cll
i.L
0% 5% 10% 15%
Initial volatility of
expected revenue changes

Fig. 11. Firm value sensitivity with respect to different initial volatilities of ex-
pected revenues for Intershop.
 467

7%
6%
~
~ 5%
<U
.g 4%
a.
:!::! 3%
::>
~ 2%
0
1%
0% L-~~~~~~~~~~~~a.~~~~ . .~~

Fig. 12. Default risk over time for Intershop.

The discussion has shown that the modeling of bankruptcy risk is essen-
tial for simulation models. Risk is a rather abstract dimension in classical
firm valuation models. This is different with the Schwartz and Moon
model. Risk in the simulation model is reflected by the default probability
or at least the chance to suffer Iosses in firm values. Figure 12 addresses
the default risk as well. Different to the figures before, the figure shows
how the default risk evolves over time. The example taken there is the
software company Intershop. This is a typical "New Economy" company
with high cash buming rates and negative cash flows but high revenue
growth Ievels. Figure 12 makes clear that the bankruptcy risk is concen-
trated on the next years. The bankruptcy risk declines significantly if the
company manages to survive for the next 3 to 8 years. This reflects the
mean reversion property in expected revenue growth rates, i.e. the fact that
young companies turn from growth companies to value companies. Value
companies reveal significantly lower growth rates but also clearly lower
bankruptcy risks.

5 Growth options

Firm values are determined by the present values of future cash flows.
There is no distinction between growth and value companies in that matter.
However, future cash flows can be driven by the fundamental value of
future projects or by the flexibility to do the right thing when the time
comes. An example for this flexibility is to either expand or to shrink the
existing project. This is called a real option of an entrepreneur. In other
words, the value of a company is composed ofthe following two elements:
468

Firm value = Fundamental value + option value (9)

The difference between growth and value companies is that the
fundamental value component is significantly higher for value companies
and the option component is more significant for growth companies. It is
therefore interesting to see which part of the firm value is driven by each
of the components. The Schwartz and Moon (2001) model has an option
component implied but it does not allow to separately quantifying the two
value components. Therefore, a model, which explicitly values real options
as embedded component of the total firm value, would be helpful. This
requirement is the core of the real options literature. However, there are se-
rious weaknesses in the real options approach. The most important prob-
lern is certainly that pricing models for financial options are assigned to
real options without analyzing the underlying assumptions. A prerequisite
for the derivation ofthe Black and Scholes (1973) option-pricing model is
for instance that there is an underlying asset which is traded and which has
an observable market price. This assumption is violated with many real
options. For instance, if a growth company hasset up an Internet platform,
which allows selling books, then it might be beneficial to expand the busi-
ness and use the platform for distributing DVD's. However, if this is a
pioneering business model for DVD 's, then there is no traded share
available that could be used as a benchmark to price the option to expand.
Maybe there are companies revealing high correlations to the new project.
But option pricing requires more than that, it requires that the option value
be perfectly correlated to the value of the underlying asset.
Cochrane and Saa-Requejo (2000) provide a model allowing approxi-
mating the real option value based on the assumption that there is a traded
share that can be used as a benchmark for the project underlying the real
option. The model was not intended for the use for real options but for op-
tions in incomplete markets. This implies however the consideration of
options on non-traded underlying assets. The purpose of this section is to
provide a description of the Cochrane and Saa-Requejo (2000) approxima-
tion formula and to show how it can be applied to real options.
Incomplete markets are characterized by fewer assets than there are
states. This implies that an asset with given cash flows in the states and a
given price implies several discount factors justifying the market price. For
illustration, consider the following example in a simplified world with only
two states in the future. Let the current share price be € 1000 and the cash
flows be either € 1260 or € 840. The probabilities for the two states are as-
sumed tobe 0.5. This setting implies the following consideration for the
discount factor mup in the up state and ffidown in the down state:
0.5. In.lown. € 1260 + 0.5. mdown·€ 840 = 1000 (11)
 469

A possible solution to (11) is mup = ffictown= 111.05. A different solution is

mup = 0.8 and ffictown= 1.181. Due to the incomp1eteness of the market, there
are arbitrarily many solutions to (11). As is well known, inventing an ad-
ditional asset e.g. a call option on the existing share completes the market.
Then the call option could be priced based on the unambiguous discount
factor. Otherwise an accurate solution ofthe pricing problern is impossible
because the volatility of a not traded underlying asset cannot be measured.
But determining volatility is a key ingredient of any option-pricing model.
Cochrane and Saa-Requejo (2000) show however that an approximation
of the option value is possible even if there are arbitrarily many discount
factors. The idea is based on the assumption that even if many discount
factors are technically possible, there is only a limited set of discount fac-
tors that is economically reasonable. Restricting the volatility of the dis-
count factor does this. Even though they cannot measure the volatility of
the underlying asset and hence the discount factor, Cochrane and Saa-
Requejo (2000) recognize that there is an upper and a lower boundary for
the discount factor volatility. If the asset would be traded, then the vola-
tility could not exceed or fall below a certain boundary. Otherwise pur-
chasing the asset would not be a "good deal". Obviously, this implies an
upper and a lower price boundary for the option on the non-traded asset.
The boundaries are called good-deal bounds.
More specifically, the following optimization problems are solved: The
upper price boundary of the option on a non-traded asset is obtained by
maximizing the value of the option given the good-deal bound for the
volatility of the non-traded underlying asset. Accordingly, the lower price
boundary results from minimizing the call option value subject to re-
stricting the volatility to the upside. Cochrane and Saa-Requejo (2000)
derive the following formula for the upper and lower price boundary for a
call option on a non-traded asset V, given a traded asset S:
470

\'o·ffT
m---
X· eTJT 1Jv -r J.ls -r A
2
d rp= - - - - - · p-a (10)
2

cry·JT "v <% (~'J

a _ {1, upper bound
-1, lower bound

In (1 0), the following notation is used. C is the lower good-deal bound

for a call option on a non-traded stock, Cis the upper boundary. Equation
(1 0) expresses good-deal bounds for a call option on a non-traded asset V
revealing correlation p to the traded asset S. Hence, crv represents the
volatility of the non-traded asset and crs the traded asset volatility. Ac-
cordingly, Jlv and Jls are the expected retums on the two assets, r is the
risk-free rate of interest, T is the time to maturity of the real option, X is
the exercise price of the option, and A is a parameter associated to the
volatility bound of the underlying asset. The pricing formula in (1 0) Iooks
very similar to the Black and Scholes model. An obvious difference is the
parameter T) discounting the current value V0 of the underlying asset. T) de-
pends on the Sharpe ratios of V and S, hence it is a kind of performance
measure. If one assumes that the underlying asset V is perfectly correlated
to asset S, i.e. p=1, then T) is simply the difference between the Sharpe ra-
tios ofV and S. Ifthe Sharpe ratio ofV exceeds the Sharpe ratio ofS, then
11 is positive and hence V0 has accrued interest. Otherwise if the Sharpe
ratio of V is below the Sharpe ratio of S, then V0 is discounted by 11.
To demonstrate the use of the formula, consider the following example.
Assurne United Internet AG thinks about acquiring GeMX GmbH, which
is not exchange traded but whose cash flow dynamics are expected to re-
veal a correlation of 0.9 with the cash flows of the United Internet AG.
Furthermore assume that the volatility of the underlying asset GeMX
GmbH is 40% and the exercise price is € 100. The market price of one
United Internet AG share is € 100, the time frame within which GeMX can
be reasonably started-up is 3 months and the risk-less rate is 5%. The vola-
tility restriction is reshaped and expressed in terms of the Sharpe-ratio. The
more slack the volatility restriction is, the greater can the GeMX-Sharpe-
ratio deviate from the United Internet AG Sharpe-ratio. Here a Sharpe-ra-
 471

tio of 0.5 is assumed for the core business of United Internet. The width of
the good-deal bounds increases with the Sharpe-ratio multiple for the new
project. Figure 13 shows the upper and the lower boundary for the call op-
tion price as weil as the Black Scholes price, which would be obtained on
complete markets. The Black Scholes price for the call option is € 11.81.
The lower price boundary oscillates between € 7.40 and € 8.20. The upper
price boundary is between € 15.64 and € 16.99 and hence around 16% of
the exercise value. Obviously, the quality ofthe approximation deteriorates
with an increase of the Sharpe-ratio multiple. However, the distance
between upper and lower price boundary is amazingly insensitive with
respect to an increase in the Sharpe-ratio multiple. Thus, the Cochrane and
Saa-Requejo (2000) methodology provides a useful tool for the separation
of the real option component from the firm value.

20 --------·--------·-----·-------·-··-·----·-------------------·--·---·----------·--·1
Q)
(.)
15
·;::
a.
c: 10
0
li x- x- -.:- x- x- x- x- x- x
0 5

0
0 1 2 3 4
Sharpe-Ratio-multiple

- x- Lower bound - Black Scholes price -- Upper bound

Fig. 13. Good deal bounds for a (European) growth option

472

60%
(/)
Q)
0
..c
(.) 40%
cn
~
(.) 20%
ro Q)
äl (.)
·;:: 0%
.9 0.
c:
0 -20%
~ro -40%
>
Q)
0 -60%
70 80 90 100 110 120 130
Price in % of exercise price

Fig. 14. Good deal bounds for a (European) growth option

Figure 14 shows how the spread is related to the exercise price of the
real option. lt illustrates that deeply out-of-the-money options imply sig-
nificantly lower spreads than at- or in-the-money options. l.e. the smaller
the likelihood is that the option is exercised; the smaller will be the ap-
proximation error. This result shows that an application of the model is
quite useful, if the growth option refers to a project with a more vision like
character.

6 Conclusions

Growth companies reveal value relevant flexibilities caused by asymmetric

revenue and firm value distributions. This suggests that option-pricing the-
ory might provide insides about the fair value of growth companies. How-
ever, real options theory can frequently not be applied to growth options.
Nevertheless, if the cash flows are negative in medium term, growth op-
tions are value relevant. This is supported by the finding that uncertainty of
revenues lowers the firm value but uncertainty of expected revenues en-
hances the company value. Again, this is an option property. While real
options cannot be priced by simply using the Black and Scholes frame-
work, the Cochrane and Saa-Requejo (2000) model can approximate
growth option values by a pricing model for options on non-traded assets.
The paper uses this model for demonstrating the pricing of growth options
implied in growth companies. lt is shown that valuation errors are driven
 473

by the volatility restriction and by the exercise price of the growth option.
The narrower the volatility restriction is, the more accurate is the approxi-
mation. Furthermore, the model can more accurately price deeply out-of-
the-money options.

References

Ballwieser W (200 1) Unternehmensbewertung und Optionspreistheorie. Die

Betriebswirtschaft, vol62 no 2: 184-201
Black, Fisher, Scholes M (1973) The Pricing ofüptions and Corporate Liabilities.
Journal ofPolitical Economy 81, May-June: 637-659
Cochrane JC, Saa-Requejo J (2000) Beyond arbitrage: Good-deal asset price
bounds in incomplete markets. Journal of Political Economy 108: 79-
118Copeland T, Koller T, Murrin J (2000) Valuation - measuring and
managing the value of companies. John Wiley & Sons, 3rd edn, New York
Damodaran A (1995) Damodaran on Valuation. John Wiley & Sons, New York
Gordon MJ, Shapiro E (1956) Capital Equipment Analysis: The Required Rate of
Profit. Management Science 3: 102-110.
Keiber K, Kronimus A, Rudolf M (2002) Bewertung von Wachstumsunter-
nehmen. Zeitschrift für Betriebswirtschaft (Zffi) 72 vol7: 735-764
Loderer C, Jörg P, Piehier K, Zgraggen P (2000) Handbuch der Bewertung.
Verlag Neue Zürcher Zeitung, Zürich
Rudolf M (2000) Monte Carlo Simulation im Risikomanagement Wirt-
schaftswissenschaftlichesStudium WiSt, 29. Jg., no 7 (Juli): 381-387.
RudolfM, Witt P (2002) Bewertung von Wachstumsunternehmen. Gabler-Verlag,
Wiesbaden
Schwartz, Eduardo S, Moon M (200 1) Rational Pricing of Internet Companies
Revisited. Financial Review, vol36: 7-26
Trigeorgis L (1996) Real Options - Managerial Flexibility and Strategy in
Resource Allocation. MIT Press, Cambridge
Shareholder Value at Risk as an Instrument of
Company Valuation

Lars Schiefner

Reinhart Schmidt

1 Abstract

Various methods of determining the shareholder value have been de-

veloped to value a company. Mostly, the expected cash flows are obtained
by detailed corporate planning, yet the crucial component risk is only
broadly captured. In particular for companies in the New Economy an
elaborate consideration of risk is indispensable.
The new approach to corporate planning and valuation under risk ad-
dresses the problern of evaluating strategies in a risky environment. A
multi-period stochastic simulation model has been created. By simulating
correlated sector-specific value drivers the distribution of the net present
value of the company's cash flows is derived. The model considers the
company from the management's and different investors' points of view.
The investors are characterized by different tax rates and utility functions.
This new method is particularly applicable to volatile businesses of the
New Economy. By risk analysis and the concept of chance-constraints, this
method identifies strategies satisfying constraints with a certain minimum
probability. It is determined, which company value will be the lowest un-
der a given confidence level. Other risk measures, i.e. LPM and Expected
Shortfall, are calculated.

2 lntroduction

Discounted cash flow (DCF) analysis is a widely accepted approach to

company valuation. Future expected cash flows are discounted by appro-
priate cost of capital. However, there are problems in considering the risk
of the cash streams, which are essential for companies of the New
Economy. Thus, DCF methods are hardly appropriate for controlling the
 475

strategy in valuated businesses. Rappaport's shareholder value approach 1

discounts expected operating cash flows before interest with the weighted
average cost of capital and is, therefore, also known as the entity approach.
The equity approach discounts cash streams flowing to the shareholder
with cost of equity. In general, both of them use cost of equity determined
by an equilibrium model such as the Capital Asset Pricing Model (CAPM).
This implies the application of risk preferences of a representative inves-
tor. Individual preferences cannot be incorporated.
Furthermore, cost of capital is estimated from historical data leading to
an inappropriate consideration of risk. Different strategies may contain
different risks, but choosing another strategy does not change the cost of
capital. Business risks are entering the company's valuation without ade-
quate consideration of probabilities. Frequently, in such approaches risk is
expressed by scenarios. The use ofhistoric market data is particularly deli-
cate for New Economy companies. The time series of stock prices
available are not sufficient for an accurate estimation especially if there are
such abnormal price movements that could be observed in these markets in
the last years.
Our approach of "Shareholder Value at Risk" provides information for
generating probability distributions of the company value from different
points of view: the management's and the individual shareholder's with
different preferences and tax rates.
If the management articulates chance-constraints, the model generates
the probability of violation. Effects of changes in corporate and divisional
strategies on the probability distribution of the net present value of future
cash flows can be evaluated in return and risk dimensions. Conceming the
consideration of probabilities for valuation processes on the project Ievel
we refer to Hertz (1964) 2 , Coenenberg (1970) for company valuation by
simulation3, and Schmidt (1981) for multi-period corporate planning under
chance-constraints and the objective ofmaximizing terminal wealth4 •
The paper is organized as follows. The subsequent section presents the
concept of our model as well as a utility-based foundation of our valuation
approach. We describe the composition of a company of different divisions
and how its free cash flows are calculated. In the fourth section, the
probability distribution ofuncertain components is incorporated. We apply
a Monte Carlo simulation for investigating different business strategies
with appropriately modeled risks. A theory-based evaluation and interpre-

1 See Rappaport, A. (1986).

2 See Hertz, D. (1964).
3 See Coenenberg, A.G. (1970).
4 See Schmidt, R. ( 1981 ).
476

tation of the models results are given. The paper finishes with a conclu-
sion, which also includes suggestions for extensions and further research.

3 Model for the valuation of corporate and divisional

strategies

3.1 Concept of the model

The model aims at the overall valuation of a company based on multi-

period and multi-divisional planning under risk. A valuation independent
of individual risk preferences is achieved for the management. In a
following step, different preferences and tax rates for individual Share-
holders can be applied. The management must be able to Iook at the
company value and its probability distribution. Thus, decisions can be
made with the knowledge of the whole distribution rather than one aggre-
gate value, which ensures an adequate consideration of risk. Decision
makers can apply several risk measures to evaluate the uncertainty of
future strategies. They can also extract information about probabilities of
violating specific lower and upper bounds for key ratios of interest.
The company consists of subunits (divisions ), which buy inputs, pro-
duce, and sell. The functions of financing and taxation are assigned to the
company Ievel. Each subunit can select from a discrete set of strategies.
Inputs and outputs of a strategy are modeled via cash flows. The main
determinants of a strategy are different value drivers as they can be found
in the Iiterature on shareholder value. 5
In step 1, discounting of cash flows is executed by applying a risk-free
interest rate in order to make payments of different periods comparable
without taking risk preferences into account. A second step is the applica-
tion of a risk preference function leading to the shareholder value from the
individual investor's point of view. 6 While in the literature so far only the
final value has been used, we are in particular interested in the probability
distribution generated in step 1. By this information, the management can
evaluate the impact of alternative strategies on the company value inde-
pendent ofthe individual investor's preferences and tax rate.

5 See Rappaport, A. (1995), 54-58.

6 For this stepwise procedure see Ritchken, P.H., Tapiero, C.S. (1986),
Mittenberg, G.J.; Krinsky, I. (1987), Fraser, J.M. (1990).
 477

As far as the model's elements are stochastic, the payment of different

divisions can be correlated. 7 The user of the model can choose the type of
the distribution function and the respective parameters. Because of the
model's complexity, an analytic approach must be abandoned, therefore
simulation is used.
Our paper is related to the risk profile approach with terminal values by
Siegel8, but the method presented there is not based on utility functions.
An analytic approach to determine shareholder value related risks of banks
using the special case of normal distributions has been developed by
Prußog,9 yet it also lacks the theoretical foundation. The one-period impact
of changes in financial prices to the cash flows of a company has been
investigated in the cash-flow-at-risk approach. 10 Arealoptionapproach is
used by Schwartz and Moon, where cash flows are generated by simulated
revenues. 11

3.2 A utility-based foundation of the valuation approach

Discounted cash flow approaches basically consider the cash flows of a

period as a random variable. These methods aggregate its probability
distribution by using only the expected value, which is discounted by a
risk-adjusted interest rate 12, i.e.
(1)

with
V- company value, shareholder value
CFt- cash flow of period t
rc- risk-adjusted cost of capital
T- planning horizon
RVr- residual value in period T
The substitution of risk and return is determined by the risk premium. lt
depends on the actual method, and thus on the type of cash flow, which
risk-adjusted discountrate is applied. The entity approach uses weighted

7 See Peters , L. ( 1971) for the first application of the correlation concept
introduced by Markowitz, H.M. (1959) to real investments.
8 See Siegel, T. (1994).
9 See Prußog, C. (2000).
10 See MeVay, J., Turner, C. (1995).
11 See Schwartz, E.S., Moon, M. (2000).
12 See Fama, E.F. (1977).
478

average cost of capital 13 • The inclusion of a capital structure leads to a

circularity problern since the company value is also a component of the
debt ratio. 14 The equity approach as well as the Adjusted Present Value
(APV) approach use cost of equity, 15 where for the APV the advantage of
debt financing has to be added. These methods consider every period sepa-
rately, which implies independent periods.
The certainty equivalent approach 16 implies similar assumptions about
time independence. Given the investor's utility function U(X) for a
period's payment, the certainty equivalent CE(CF1) satisfies
U(CE(CF1))=E[U(CF1)] and is calculated to express the risk preferences of
an individual investor. These equivalent values are discounted by the risk-
free interest rate. 17
Thus, dependencies between different periods have no influence on the
result. Furthermore, the certainty equivalent method has also other nega-
tive properties making it hardly appropriate for company valuation. 18
Both approaches as well as other methods using time-additive
preferences lack an adequate consideration of intertemporal dependen-
cies.19 The assumption of time-independent cash flows in companies is a
very strict restriction.
In our model a different present value-based approach is used. 20
Thereby, not the cash flow of one period is considered but the process of
cash flows over time. The realization of the cash flow process as a path
through time is the basic component. To make payments of different peri-
ods comparab1e they are discounted by a pure time preference rate. Hence,
a realization is represented by its present value of its cash flows without
changing the risk property of the process. Thus, realizations are seen to be
equal if they have the same present value consequently realizing the usual
view in finance. As a result, we get a probability distribution of present
values of the cash flow process. This is the random variable, for which the
risk preferences of an individual are applied to. Intertemporal dependen-
cies between cash flows are properly considered. This decision rule has
already been used by Ritchken and Tabiero 21 for purchasing decisions in

13 See Rappapart, A. (1995), 58-62, Copeland, T. et al. (1994), 135.

14 See Schwetzler, B., Darijtschuk, N. (1999).
15 See Drukarczyk, J., Richter, F. (1995), 541.
16 See Ballwieser, W. (1980).
17 See Ballwieser, W. (1980), 68.
18 See Kürsten, W. (2000), 11-17.
19 See Duffie, D., Epstein, L.G. (1992), 355 for an examp1e.
20 See Fraser, J.M. (1990) for a survey on utility functions based on net present
value.
21 See Ritchken, P.H., Tapiero, C.S. (1986).
 479

inventory management as weil as by Mittenberg and Krinsky22 to evaluate

flexible manufacturing systems. Siegel (1994) used a similar approach23
although his method was not derived from decision theory with utility
functions.
To formalize the approach, let (CFJt=I, ...,oo be the stochastic process of
cash flows and ~ a time dependent time preference rate. The present value
ofa realization CP, i.e. the outcomes CPJ.CP2, ... in state s, is given by
(2)

Different realizations can yield different present values. Thus, the

present value PV itself is a random variable. Let f(y) be a risk preference
function. Given S realizations, the utility of a cash stream is calculated by
s (3)
U(CF )= Lf(PVS)P{CF=CF 8 }.
s=l

The probability of one particular present value is the sum of probabili-

ties of cash flow realizations having this present value, i.e.
(4)
P{PV=y}= L P{CF=CF 8 )}.
s with
PV 9 =y

If B denotes the number ofpossible present values equation (5) is equiva-

lent to
B (5)
U(CF)= Lf(Yb)P{PV=yb}·
b=l

The shape off determines the risk attitude of an individuaJ.24 Concavity

corresponds to risk aversion. The certainty equivalent CE is a payment
satisfying U(CF)=U(CE)=f(CE) and determines the value which an indi-
vidual is prepared to pay for this cash stream. If (CFJt=I, ...,r is the cash
flow from a company to a shareholder then the shareholder value SV is
given by the corresponding certainty equivalent. Thus, the value SV that
satisfies U((CFJ)= f(SV) is determined.
If the time preference is taken from the market as the term structure of
risk-free interest rates, then a decision support system can present aggre-

22 See Mittenberg, G.J.; Krinsky, I. (1987).

23 See Siegel, T. (1994).
24 See Fraser, J.M. (1990), 247-248.
480

gated information about the cash flows of a company in form of a

probability distribution of present values. Thus, a consideration inde-
pendent of individual risk preferences is achieved, which is a main
advantage of our model. It is worth stressing that discounting cash flows
by risk-adjusted interest rates is inappropriate for generating an unbiased
probability distribution. The application of risk adjustments implies a sub-
stitution of risk and return. Thus, the resulting distribution cannot be pref-
erence-free. 25
Similar to the approaches mentioned above we assume a planning hori-
zon T with individually forecasted cash flows. Later cash flows grow with
a certain rate g derived from the long-term economic growth rate. 26 Thus, a
realization of the cash flow process is discounted to its present value by
(6)

where the second term is a discounted residual value. The term structure of
interest rates is represented by rf,t whereas rf,r+ is an appropriate average
rate for the periods after T. Hence, we have derived a utility-based founda-
tion of our approach.

3.3 Some details of the model

The company is modeled in a MS-Excel spreadsheet. It consists of a cer-
tain number of divisions. Bach division can choose from different strate-
gies. Due to space restrictions, the example in table I considers a planning
horizon of 5 years. The planning horizon depends on the particular com-
pany. Especially for the New Economy a Ionger planning period is recom-
mendable.
The cash flows generated by the division are determined by specific
exogenous quantities. On the divisional Ievel these are the so-called value
drivers introduced by Rappaport,27 i.e. the sales growth rate, the EBITDA
margin, and the ratio capital expenditures/sales. The tax rate conceming
the operating cash flow is assumed to be the same for all divisions and is
assigned by the head office. Additional capital expenditures and divest-
ments can be determined as exogenous variables. Changes in working

25 Risk-adjusted interest rates have been wrongly used by Moser, U., 5chieszl, S.
(200 I). Thus, their derived probability distribution of a company value can not
be used.
26 See Philips, T.K. (1999), 40.
27 See Rappaport, A. (1995), 54-58. See also Copeland, T. et al. (1994), 137-144.
 481

capital depend on the capital expenditure spent. The computational results

are free cash flows for each divisional strategy.

Table 1. Determination of divisional free operaring cash flows

Division 1 Strategy 1 year1 year2 year3 year4 year5 year6-oo

growth rate of sales (%)a 3.5 3.5 3.5 3.5 3.5 3.0
sales (year 0 = 5000) 5,175 5,356 5,544 5,738 5,938 6,117
EBITDA margin (%)a 10.0 10.0 10.0 10.0 10.0 10.0
operating cash flow before taxes 518 536 554 574 594 612
- taxes on Operating cash flow -117 -111 -108 -114 -114 -134
- operating cash flow after taxes 401 425 447 459 479 478
reported: additions to tangible assets 3.0 3.0 3.0 3.0 3.0 3.0
I sales (%)a
- additions to tangible fixed assets -205 -161 -166 -272 -178 -183

+ divestments of tangible fixed assetsa 50 0 0 100 0 0

- change in working capital -16 -16 -17 -17 -18 -18
- free operating cash flow 230 248 264 270 284 276

a exogenous variables
At the company level, all quantities of the divisions are aggregated (see
table 2). Hence, sales, margins, and cash flows are summarized to one
figure. Divisional depreciation and capital expenditures are used for
purposes of tax and financial computations. Dividends are incorporated by
a payout ratio. Furthermore, there is the possibility of exogenous capital
increases. Debt is repaid due to a repayment schedule whereas fresh debt
per period is a residual variable resulting from an integrated cash flow
statement. It is also possible to model a constant debt equity ratio such that
dividends and capital increases are residual.
After the five-year planning horizon cash flows grow with a constant
rate. We use an economic long-term growth rate since in the long run
companies cannot grow faster than the whole economy. 28 Empirical studies
confirm this statement. 29

28 See Philips, T.K. (1999), 40

29 See Campbell, J., Shiller, R. (1988), Goetzmann, W.N., Jorion, P. (1993).
482

Table 2. Determination of net payments from the company to shareholders

Company Ievei year1 year2 year~ year4 year~ 'year 6-ao

growth rate of sales (%)b 3.5 4.1 4.1 5.3 4.1 3.0
25,8 26,9 28,C 29,5 30,t 31,67
sales (year 0 = 25000)b 75 36 42 37 SE 8
EBITDA margin (%)b 10.0 10.4 10.8 11.5 10J 10.1
2,58 2,80
[operating cash flow before taxesb 8 1 302E 3384 331 3198
taxes on operating cash flow -637 -684 -739 -883 -853 -734
1,95 2,11 2,2S 2,50 2,4€
F operating cash flow after taxesb 1 7 I 1 c 2,464
reported: additions to tangible
assets I sales (%)b 2.0 3.( 3.C 1.3 3.C 3.~
additions to tangible fixed assetsb -776 -80~ -841 -886 -923 -950
f+ divestments of tangible fixed
assetsb 250 c c 500 c 0
change in working ca!)italb -53 -81 -84 -39 -92 -95
1,37 1,22 1,36 2,07 1,44
I= free operating cash flo~ 2 ~ 1 6 E 1,419
interest on debt -350 -3H -299 -27t. -203 -170
re_payment of debt -500 -45E: -428 -3~9 -29C -243
f+- fresh debt 61 17~ 4? -606 -17::1 -33:-l
reported: payout ratio" 50.0 50.~ 50.0 50.0 50.~ 50.0
dividends 583 62E 676 808 78C 672
F external capital increase" 0 ~ ~ 0 ~ ~

n_et effect to shareholders 583 625 676 808 78(J 672

risk-free interest rate" 5.00 5.2~ 5.50 5.75 6.00 6.25
0.95 0.9C 0.85 0.79 0.74
~iscount factor 24 21 16 96 73
present value of a year's net effect 555 565 576 646 583
ong-term growth rate" 1.0
9,56
residual value of net payments 0
12,4
otal present value of net payments 85
a exogenous vanables
b calculated from divisional figures

According to our present value based approach the net payments to the
shareholders are discounted by the risk-free interest rate (modified equity
approach) leading to a present value of outgoing payments. For the valua-
tion by Shareholders the inflows after personal taxes per period are dis-
counted to yield a present value of the future income stream. The personal
tax rate is applied, which also influences the discount rate.
 483

4 Application of stochastic Simulation and

implementation

4.1 lncorporation of random variables

Generating probability distributions of the present value of cash flows

from the management' s point of view and the individual investor' s
perspective is one objective of our procedure. Hence, we must include
probability distributions to realize the stochastic character of the cash
stream. To incorporate value drivers as random variables is very rational.
They are not only stochastic since they are revealed only in the future but
it is reasonable that their probability distribution can be determined by the
management. Thus, we decided to take the sales growth rate and the
EBITDA margin per divisional strategy and period as random. The long-
term growth rate of cash flows is also stochastic, but with a very small
volatility.
The choice of the probability distribution and its parameters has the
biggest influence on the generated cash flows. Within the planning hori-
zon, managers are assumed not only to estimate an expected value as in
other DCF methods but also the entire distribution. They are not forced to
aggregate ideas about of the randomness of future quantities in one figure
as the expected value, but can project the uncertainty into a probability
distribution. Changes of properties of value drivers by selecting a different
strategy immediately leads to a new distribution of net present values of
the payments to the shareholders. In particular, it is possible to observe
changes in the risk dimension.
Generally, DCF-methods use cost of capital estimated from historical
stock prices to incorporate the uncertainty by a risk-adjusted discount fac-
tor. Thus, the selection of a new strategy could change the company value
only if the expected value is changed. A shift in higher moments, particu-
larly in standard deviation, is not recognized. The immediate revelation of
changes in risk is one of the advantages of our approach presented here.
To execute a realistic simulation, the consideration of correlations
between random variables is essential. There may be relations between
different divisions as well as autocorrelation of time series. The use of the
sales growth rate ensures intertemporal dependencies of payments as sales
or cash flows. At this moment, also correlations between divisions are
included whereas an incorporation of correlation between any of the
random variables is possible.
484

Even if in other DCF methods the distribution of the value drivers is

correctly determined, the use of their expected values may Iead to wrong
results. Some random variables are combined by multiplication, but we do
not have E(XY)=E(X)E(Y) in general. It is correct for independent vari-
ables but if sales growth rate s is correlated with the EBITDA margin e,
then we can have for the operating cash flow before taxes OCF for given
pre-years sales S E(OCF) =E(Sse) -::t:. SE(s)E(e).
Thus, correct determination of the expected values of the value drivers
within a company planning procedure may nevertheless Iead to wrong cash
flows and, hence, to inaccurate company values.

Table 3. Probability distribution and its parameters for one divisional strategy

value driver distribution J)arameters for selected probabil~ distribution

beta year 1 year 2 year 3 year 4 year 5 year 6-oo
growth rate of min 1 1 1 1 1 2
sales (%) mode 3 3 3 3 3 3
mean 3.5 3.5 3.5 3.5 3.5 3
max 7 7 7 7 7 4
EBITDA margin (%) min 8 8 8 8 8 9
mode 10 10 10 10 10 10
mean 10 10 10 10 10 10
max 12 12 12 12 12 11

To define the stochastic character of the value drivers we have to deter-

mine their probability distributions (see table 3). Though the simulation
software allows for 37 different distributions, we decided to choose only
three of them. The normal, the beta, and the triangular distribution are easy
to understand and help the decision maker to express the risk. Once the
manager has selected the type he determines the parameters. It may be
helpful to use elements of seenarios that are often already generated as a
basis for the probability distribution. But one should be careful with worst
and best cases. If they are in fact the bounds of an interval, then a distribu-
tion defined for all real numbers, e.g. the normal distribution, is inappro-
priate. 30 The subjective beta distribution fits to many data sets since with
the setting of the parameters a skew density function on a bounded interval
can be constructed.

° Fora negative example see Moser, U., Schieszl, S. (2001), 532.

3
 485

4.2 The Simulation procedure

The simulation is done with the package @RISK by Palisade Corporation

based on the MS Excel spreadsheet. 31 @RISK generates random numbers,
which are the realizations for the value drivers. One iteration sets these
values for all periods and divisional strategies. Then, the cash flows of all
divisions and the whole company in each period are calculated. Thus, we
have one realization for the stochastic cash flow process, which after dis-
counting generates the net present value. For further evaluations the
present values of the cash flow from the company, the present value of
payments to the shareholders, and the present value of operating cash
flows after tax per division are stored in global arrays. Furthermore, to
gather information on certain key figures, their realized values are also
stored to check violations of chance constraints. In our model these are
dividends, net profit, and the debt equity ratio. Additional key figures can
also be considered. Although @RISK allows a comprehensive evaluation
of the data, we gather the appropriate results by a Visual Basic macro.
Thus, an individual processing is assured. After the simulation we have a
vector of realizations of all key figures that we are interested in. This is the
basis for an extensive evaluation presented in the next section.

4.3 The evaluation of the output data

After the simulation we have a random sample of the cash flows' net
present values and key figure processes available. These data is used to
generate decision-supporting information. A decision maker does not have
to get familiar with an complex computer package like @RISK because all
relevant information is user-friendly presented. A graphical presentation of
the resulting probability distribution is very helpful. The shape of the den-
sity function reveals some properties of the intrinsic risk. In two diagrams
we show the present value of the cash flows from the company and to the
individual shareholder.
With the random sample available we can calculate quantities such as
means and standard deviations. Although there are some individuals who
base their decisions solely on these figures, 32 it is recommendable to apply
some risk measures to get further insight in the risk potential of the chosen
strategies. We use Value at Risk (VaR), Lower Partial Moments of first
order (LPM 1), and Expected Shortfall (ES) (see table 4).

31 For other software packages see Maxwell, D.T. (2000).

32 E.g., individuals with quadratic utility functions.
486

Table 4. Estimated mean and risk measures ofthe NPV to shareholders

Mean 12,485.8
Standard deviation 1,071.8
Value at Risk (95%) 1,657.2
Critical Value (95%) 10,828.5
LPM1(0%) 453.2
Expected Shortfall (5%) 1,990.3

In practice, managers often use Value at Risk to estimate potential

lasses. 33 In a company valuation model it is not rational to identify a lass
potential but it is reasonable to describe a negative deviation from the
mean. Thus, the Value at Risk at a confidence Ievel 1-a is the amount,
which the present value of cash flows cannot fall further below the mean
with a probability of 1-a Perhaps more informative is the a-quantile of
the distribution itself, here called Critical Value. It is the smallest present
value the company or divisions generate with a probability of 1- a If the
realizations are ordered, i.e., X 1:::;; X 2:::;; ••• XN, the Critical Value is
CVa (X)= X;a with ia =max(i Ii:::;; aN) (7)

whereas the Value at Risk is given by

VaRa(X)=X -CVa(X). (8)

Despite its practical relevance, the Value at Risk suffers from some
theoretical disadvantages. 34 The quantile describes only one point of the
probability distribution. It does not use information about the behavior in
the tails. Furthermore, it is not sub-additive. 35 Thus, in certain cases Value
at Risk contradicts the principles of diversification. Here, we also provide
some more sophisticated risk measures. The Lower Partial Moment of frrst
order describes the shortfall distance relative to some benchmark b. 36 For
our ordered realizations it is estimated by
(9)

If b is chosen as the mean, then this is a downside risk measure of

falling below the average. Furthermore, we calculate the coherent risk

33 See Jorion, P. (2001).

34 See Szegö, G. (2002), 1261-1262, Johanning, L. (1998), 53-57.
35 See Artzner et al. (1999), 216-218 for an example.
36 See Fishburn, P.C. (1977).
 487

measure 37 Expected Shortfall. lt measures the average lass in the warst

lOOdllo cases. Again, we arenot interested in a lass but in deviations from
the mean. Thus, the Expected Shortfall is estimated by 38
(10)
ESa(X) =X-~ ~Xi with ia = max(i Ii s aN).
la i=l

Hence, we present the mean and several risk measures of the cash
flows' present value from the management's point of view and the indi-
vidual investor's perspective. All these information provide the basis for
the conclusion of the decision maker.
For corporate planning it is also reasonable to evaluate chance-con-
straints of the structure
P(XsB)sp, (11)

i.e., the probability that some random variable X falls below a bound B
is at most p. 39 If the realizations of key figures in each period are stored,
we can estimate the probability of violating a lower ( or upper) bound and
compare with the given one. Key figures of interest could be dividends
(see table 5), or, for example, eamings after taxes and debt equity ratios
per period.

Table 5. Chance-constraints and probabilities ofviolation

Chance-Constraints Year1 Year2 Year3 Year4 Year5 Year6

Dividend per share

Lower bound ( €) 1.00 1.10 1.15 1.20 1.25 1.3C
Probability of violation 0.0% 4.8% 8.3% 0.2% 0.0% 0.0%

We also perform calculations to analyze the investor's perspective. By

using some special preference functions, we are able to determine the cer-
tainty equivalent of the present value of cash flows to the shareholders,
which results in a value that the investor is willing to pay for the company
(see table 6).
Furthermore, if we assume the type of preference function of the inves-
tor and the current market capitalization as a fair price for the company,
then we can calculate implicit parameters. Recalling that our model sup-
ports the management, knowledge about the preferences of the

37 See Artzner et al. (1999) for details on coherent risk measures.

38 See Acerbi, C., Tasche, D. (2002), 1494.
39 See Schmidt, R. (1981 ).
488

shareholders is very helpful. Thus, strategies could be evaluated with the

investors' preferences.

Table 6. Company values using a special preference function

parameter company value

current market capitalization 17 000.0
f(x) =1- e-ax a
0.01 15,385.2
0.001 17,580.5
0.0001 18,721.9
implicita 0.0018 17,000.0

The random sample we get from the simulation is the basis for various
estimations to extract information on the probability distributions. Further
calculations that use this sample are possible, if, for instance, another risk
measure seems to be more appropriate or if other chance-constraints are to
be investigated.

5 Conclusion

We built a new decision support system for the management to evaluate

strategies in a shareholder value context under risk. A new present value
based approach for company evaluation has been developed to make
considerations for the management possible, which are in a first step inde-
pendent of individual shareholders. In a second step, the value for indi-
vidual investors can be derived.
Our system models a company with a divisional structure that can exe-
cute different types of strategy. Operatingcash flows are determined at the
divisional Ievel, whereas payments caused by financing activities or taxes
are managed at the company Ievel.
In our model, we incorporate probability distributions of the value
drivers, which can be correlated between the divisions. Particularly for
companies of the New Economy, it is essential to model intrinsic risk of
the operating activities rather than using the risk estimates from short and
biased stock price time series. The presentation of a probability distribu-
tion ofthe company's net cash outflow and the calculation of several risk
measures, e.g., not only standard deviation and Value at Risk, but also
Lower Partial Moment of first order and Expected Shortfall, give further
insight into the risk return characteristics of the company's activities.
 489

Thus, our system provides a sound basis for the decisions to be made by
the management.
The model can also be used by extemal observers using published data.
But obviously, an intemal application with an extended data basis for the
financial planning process would produce more precise information.
Extensions of the prototype presented here concentrates on supporting the
data collection and planning procedure. Further research should be done on
estimating an implied risk preference function of the market to determine a
company value from the market's point ofview.

References

Acerbi C, Tasche D (2002) On the Coherence of Expected Shortfall. Journal of

Banking and Finance 26: 1487-1503
Artmer P, De1baen F, Eber JM, Heath D (1999) Coherent Measures of Risk.
Mathematical Finance 9: 203-228
Ballwieser W (1980) Möglichkeiten der Komplexitätsreduktion bei einer
prognose-orientierten Untemehmensbewertung. Zeitschrift für betriebswirt-
schaftliche Forschung 32: 50-73
Campbell J, Shiller R (1988) The Dividend-Price Ratio and Expectations ofFuture
Dividendsand Discount Factors. Review ofFinancial Studies 1: 195-228.
Coenenberg AG (1970) Unternehmensbewertung mit Hilfe der Monte-Carlo-
Simulation. Zeitschrift für Betriebswirtschaft 40: 793-804
Copeland T, KollerT, Murrin J (1994) Valuation- Measuring and Managing the
Value of Companies, 2nd edn. New Y ork
Drukarczyk J, Richter F (1995) Untemehmensgesamtwert, anteilseignerorientierte
Finanzentscheidungen und APV-Ansatz. Die Betriebswirtschaft 55: 559-580
Duffie D, Epstein LG (1992) Stochastic Differential Utility. Econometrica 60:
353-394
Fama EF (1977) Risk-Adjusted Discount Rates and Capital Budgeting Under
Uncertainty. Journal ofFinancial Economics 5: 3-24
Fishburn PC (1977) Mean-Risk Analysis with Risk Associated with Below-Target
Returns. American Economic Review 67: 116-126
Fraser JM (1990) Utility Functions Basedon Net Present Value. European Journal
ofOperational Research 48: 242-251
Goetzmann WN, Jorion P (1993) Testing the Predictive Power of Dividend
Yields. Journal ofFinance 48: 663-679
Hertz DB (1964) Risk Analysis in Capital Investment. Harvard Business Review
45:95-106
Johanning L (1998) Value-at-Risk zur Marktrisikosteuerung und Eigenkapital-
allokation. Bad Soden
Jorion P (2001) Value at Risk: The New Benchmark for Managing Financial Risk,
2nd edn. New Y ork
490

Kürsten W (2000) Unternehmensbewertung unter Unsicherheit oder

Theoriedefizit einer nihilistischen Kunstdiskussion über Sicherheits-
äquivalent- und Risikozuschlagsmethode. Working Paper Friedrich-Schiller-
Universität Jena 2000/09
Markowitz HM (1959) Portfolio Selection: Efficient Diversification of
Investments. New York
Maxwell DT (2000) Decision Analysis: Aiding Insight V. OR/MS Today 27: 28-
35
MeVay J, Turner C (1995) Could Companies Use Value-at-Risk? Euromoney
10/95: 84-86
Mittenberg GJ, Krinsky I (1987) Evaluating Flexible Manufacturing Systems. IIE
Transactions 19: 222-233
Moser U, Schieszl S (200 1) Unternehmenswertanalysen auf der Basis von
Simulationsrechnungen am Beispiel eines Biotech-Unternehmens. Finanz-
betrieb 3: 530-541
Peters L (1971) Simultane Produktions-Investitionsplanung mit Hilfe der Theorie
der Portfolio Selection. Berlin
Philips TK (1999) Why Do Valnation Ratios Forecast Long-Run Equity Returns.
Journal ofPortfolio Management 25:39-44
Prußog C (2000) Shareholder-Value-bezogene Abbildung von Risiken im
Wertebereich bei Banken als Basis eines Risk-Return-Controlling. Frankfurt
amMain
Rappaport A (1986) Creating Shareholder Value. New York
Rappaport A (1995) Shareholder Value: Wertsteigerung als Maßstab fiir die
Unternehmensführung. Stuttgart
Ritchken PH, Tapiero CS (1986) Contingent Claims Contracting for Purchasing
Decisions in Inventory Management. Operations Research 34: 864-870
Schmidt R (1981) Transnationale Investitions- und Finanzplanung als
Portefeuilleplanung. Rühli E, Thommen JP (Eds.) Unternehmungsführung aus
fmanz- und bankwirtschaftlicher Sicht, Stuttgart 1981: 23-44
Schwartz ES, Moon M (2000) Rational Pricing of Internet Companies. Financial
Analysts Journal 56: 62-75
Schwetzler B, Darijtschuk: N (1999) Unternehmensbewertung mit Hilfe der DCF-
Methode - eine Anmerkung zum ,,Zirkularitätsproblem". Zeitschrift fiir
Betriebswirtschaft 67: 295-318
Siegel T (1994) Unternehmensbewertung, Unsicherheit und Komplexitäts-
reduktion. Betriebswirtschaftliche Forschung und Praxis 46: 457-476
Szegö G (2002) Measures ofRisk. Journal ofBanking and Finance 26: 1253-1272
Valuation of lntangible Assets for Financial
Reporting

Martin Scholich

Andreas Mackenstedt

Markus Greinert

1 lntroduction

Analyses of successful, strongly growing businesses reveal one feature

common to practically all, namely that they possess unique intangible as-
sets. Beiersdorf, for example, is profiting since decades from the strength
of its Nivea brand; Pfizer achieves its growth through an array of patents
and know-how; and Deutsche Telekom retains undisputed leadership in the
fixed network business, despite considerable competition, with the
assistance of its customer relationships.
For management, the growing importance of brands, patents, customer
relationships and other intangible assets has given rise to a variety of
challenging issues, such as:
- How should intangible assets be managed?
- How can value be measured and communicated to investors and
lenders?
- How must or should intangible assets be recognised and valued for
financial reporting?
To deal with these issues in an adequate manner, the intangible assets
first have to be identified and valued. In the past, reservations have been
expressed concerning the valuation of intangible assets. Meanwhile, how-
ever, based on recent research and practical experience, valuation stan-
dards for intangible assets have been developed. By applying these stan-
dards, valuations can be performed in an objective manner and useful
information can be obtained for solution of the aforementioned problems.
492

2 Drivers for the valuation of intangible assets

Intangible assets, i.e., identifiable non-monetary assets that Iack physical

substance (SFAS 142 Appendix F, GAS 12 Para 7) can take a wide variety
of forms. They can be classified as follows (Arbeitskreis "Immaterielle
Werte im Rechnungswesen" 2001 pp 990 seqq):
• Customer Capital (e.g., brands, customer relationships)
• Supplier Capital (e.g., access to scarce resources, import sources)
• Investor Capital (e.g., favourable borrowing terms, high rating)
• Process Capital (e.g., organization structure and structuring of
operations)
• Location Capital (e.g., locational advantages, infrastructure)
• Innovation Capital (e.g., patents, formulas) and
• Human Capital (e.g., know-how).
How important intangible assets can be for the development of a busi-
ness is often seen when an acquisition takes place. A number of studies,
particularly those conducted recently, indicate that in many cases, intangi-
ble assets are the largest single category of assets acquired (Küting 2000 p
674, Pellensand Fülbier 2000 pp 119 seqq). The price for a business thus
cannot be paid for the value of the tangible and fmancial assets alone.
Normally, the acquirer will have to pay a price for the intangible assets as
weil, it being difficult to imitate or find substitutes for them. How valuable
intangible assets can be for an acquirer is demonstrated, for instance, by
the prices paid by Prada to acquire Jil Sander, or by Interbrew at the
Beck's takeover. The prices in these transactions paid to gain access to
unique and excellently positioned brands were above average, compared
with traditional multipliers.
Recently, a study was published on "Brand Valuation and Brand
Management Practice in German Businesses" (PricewaterhouseCoopers
and Sattler 2001 ). Two of the findings are of particular interest:
1. Businesses believe that brands are among the most important factors
affecting profits.
2. According to the average of all responses, businesses believe that Brand
Equity accounts for more than half the value of the business. For con-
sumer goods manufacturers the figure was even higher.
In light of the significance of brands and other intangible assets it is of
vital importance that they receive board attention. However, here, as else-
where, the principle applies that for a thing to be manageable it has to be
 493

measurable. Ifintangible assets aretobe managed successfully, procedures

must exist for determining the value of these assets and changes therein.
Value based management, however, is not the only purpose for which a
business needs to know the values of its intangible assets. Intangible asset
related information can be helpful to the business in communicating with
investors, lenders and analysts (Arbeitskreis Externe Unternehmens-
rechnung der Schmalenbach-Gesellschaft 2002, pp 2337 seqq, Ruhwedel
and Schultze 2002 pp 602 seqq, Achleitner et al 2002 pp 33 seq, Sirnon et
al2002 p 121). The increasing competition for capital makes it virtually a
necessity for a business to provide investors, lenders and analysts with in-
formation on the existence and value of its intangible assets. This might be
done, for instance, by means of Intellectual Property Statements (Maul
2000 pp 2009 seqq, Maul and Menninger 2000 pp 529 seqq). This can
have a positive impact on the market capitalization of an enterprise and
help reduce the gap to fair value (Fischer et al2002 pp 14 seq, Rankeretal
2001 p 278). This kind ofreporting enables a business to disclose its value
potential and communicate its strengths extemally. What is more, this ad-
ditional information reduces the uncertainty with which investors, lenders
and analysts appraise the business and its future development. They hence
might be more willing to invest in it.
The market capitalization orientation and communication with investors
should not be confined to prospective shareholders. With Basel II coming
up, business risks play an even greater role than before when bankers are
deciding whether to lend, and on what terms. Self-developed intangible
assets do not usually appear on the balance sheet, so by communicating the
existence and value of such assets a business can cause a bank to make a
more favorable appraisal of lending risk (Freidank and Paetzmann 2002 pp
1787 seqq). In this way the business is more likely to obtain a loan, and on
better terms.
Appreciation of the growing importance of intangible assets has brought
about changes in accounting for them. Considering that group financial
statements should perform the function of supplying capital market opera-
tors with information and eliminating inconsistencies between the infor-
mation available to management and that available to investors and
lenders, these changes that have been made to financial reporting rules
were urgently needed.
Among the most important of these changes are the new Statements of
Financial Accounting Standards (SFAS) 141 (Business Combinations) and
SFAS 142 (Goodwill and Other Intangible Assets) of June 2001. In a way,
these standards can be said to be the forerunner of an accounting that gives
due recognition to the importance of intangible assets. The principal ele-
ments of these standards have meanwhile been incorporated into the Ger-
494

man financial reporting standardsvia German Accounting Standard No. 4

(GAS 4) (Acquisition Accounting in Consolidated Financial Statements)
and GAS 12 (Non-current lntangible Assets). lt has also been announced
that International Accounting Standard (lAS) 22 (Business Combinations)
and lAS 38 (Intangible Assets) aretobe thoroughly revised, after which
they too will harmonize more closely with the US standards.
Specifically, SFAS 141 and 142 require:
- the Identification of Identifiable Intangible Assets,
- the Estimation ofRemaining Useful Lives oflntangible Assets, and
- the Valnation of ldentifiable Intangible Assets.
With regard to identification the new standards are precise in their
requirements. "An intangible asset shall be recognized as an asset apart
from goodwill if it arises from contractual or other legal rights ... If an
intangible asset does not arise from contractual or other legal rights, it shall
be recognized as an asset apart from goodwill only if it is separable"
(SFAS 141 Paragraph 39). This rule now makes it mandatory to recognize
as separate items intangible assets such as trademarks, internet domain
names, customer lists, order backlog, construction permits and patented
technology (SFAS 141 Paragraph A 14 seqq). lt is no Ionger permitted to
include them in goodwill, not even to simplify the reporting. These recog-
nition requirements for intangible assets will presumably be incorporated
into International Accounting Standards when lAS 22 and lAS 38 are
revised.
However precisely the recognition criteria for intangible assets may now
have been laid down, their identification remains a knotty problem. The
main difficulty lies in the attribution of the value drivers of a business to
specific intangible assets, which frequently cannot be done with the requi-
site accuracy. Looking at customer capital, for instance, it may not be easy
to split the profit contributions attributable to the brand from those
attributable to the customer relationships. This is where the valuer's
experience and inside count. His ability to use his knowledge of the
industry to identify the factors on which the success of the business de-
pends is of critical importance for correct identification of the business's
value drivers and hence intangible assets. This correct identification thus
lays the foundations for correct attribution of an intangible asset's profit
contributions.
Valnation of an intangible asset requires its remaining usefullife to be
estimated. SFAS 142 distinguishes between intangib1e assets with a finite
useful life and those with an indefinite useful life (SFAS 142 Paragraph
11 ). The useful life of an intangible asset to an entity is the period over
which the asset is expected to contribute directly or indirectly to the future
 495

cash flows ofthat entity. If no legal, regulatory, contractual, competitive,

economic, or other factors limit the usefullife of an intangible asset to the
reporting entity, the useful life of the asset shall be considered to be in-
definite (which does not mean infinite). An indefinite useful life will
probably be the exception rather than the rule.
An established brand, for instance, might be an intangible asset with an
indefinite useful life if management intends to use the brand in the future
and invests in its maintenance. Classification as an intangible asset with an
indefinite useful life means that the asset is not amortized systematically.
The carrying amount may be written down only if it exceeds the fair value.
To determine if this is the case the asset must be tested for impairment
annually, or whenever a "triggering event" occurs. The impairment test
consists in a comparison of carrying amount and fair value.

Carryi ng amou nt of a brand over time in

International R nancial Reporti ng

Previous accountlng treatment: New accountlng treatment:

• Brand was classified as an • Brand is classified as an
amortizable intangible asset non-amortizable intangible asset
• Amortization was charged over the • Brand is to be tested for impairment
expected useful life independently annually and on occurrence of
of expenditure on maintaining the triggering events
brand

Carrylng amount of brand Carrylng emount Carrylng omount C.rrylng amount

(palt of goodwill) of brand or brand aftlor of brand aftlor
on acquialtion: 100 onacqu r n: 100 10y :100 20rra: 100
,/
100 0..-------;--------r-- tooo---~~-------r-o-

arrylng amount or brand

afbtr 20 yNra: 0 ~ rw~wu~wu~wu~
~

"""""" :::~-A_n_n~_I_~Tpa_l~_nt_
·TM_t~
ol- '-----+---..:er-- "::'' '--------+------~~~~
10Y :WY 10Y :WY

Fig. 1.

This treatment of brands and other intangible assets with indefinite use-
ful lives represents a considerable advance on the traditional treatment.
Now, amortization is charged only if the value of an intangible asset has
actually diminished during the period, which from an economic perspec-
tive is logical. Management and controllers will thus find the information
conveyed by the financial reporting better suited to their purposes. lt will
496

also place investors and lenders in a position to appraise the business's

position and development more accurately.

3 Valuation approaches and methods

Until quite recently, the valuation of intangible assets was regarded with a
great deal of skepticism. Firstly, they are frequently so intimately bound up
with other intangible and tangible assets that it is difficult to identify the
intangible asset precisely (Küting 2000 p 674). Secondly, they are re-
garded as comparatively uncertain assets.
Researchers and practitioners, particularly in the United States, have
been studying the valuation of intangible assets for quite some time now,
and more attention has been devoted to it in recent years in Germany as
well. Many problems have been discussed, and it can now be said that
agreement has been reached on the basic issues involved. The knowledge
and experience gained in business valuations can be applied also to the
valuation of intangible assets. The following three approaches are
available for this purpose (International Valuation Standards Committee
2001 pp 44 seqq, Reilly and Schweihs 1999 pp 95 seqq, AICPA Practice
Aid 2001 pp 11 seqq, PricewaterhouseCoopers 2002 pp 14 seq):
• the market approach,
• the cost approach, and
• the income approach.
Under the market approach, the value of an intangible asset is arrived
at from the prices obtained for comparable intangible assets in the open
market. This is frequently done by using multiples. The market approach is
based on the premise that prices paid for comparable intangible assets in
prior transactions can provide guidance in the valuation of the subject as-
set. Obviously, it cannot simply be said that the value of the subject asset
is equal to these prices, because each of the latter was realized in a single
specific transaction. The value ofthe subject asset has tobe estimated from
the observed prices by making adjustments to reflect the individual facts
and circumstances. It may, for instance, be necessary to eliminate the
effect on the price of factors specific to the buyer or seller, or to take
changes in market conditions into account that have occurred since the
prior transactions.
However attractive and plausible the market approach may be in theory,
in practice the valuer adopting it to appraise an intangible asset is faced
with two basic problems, namely comparability, and availability of trans-
 497

action price data. Traditionally, the market approach is considered when

valuing assets such as primary products and listed securities. Commodities
such as these are largely homogeneous. Differences exist only between one
grade and another. Intangible assets, on the other hand, are typically indi-
vidual in nature, or even unique. It would be difficult, for instance, to find
a car manufacturer with a name comparable to that of Mercedes-Benz.
Patents, too, are individual in that they assure that a certain product or
process can be used only by the owner of the patent. So if a valuer wishes
to use other intangible assets for comparison purposes he has to accept that
differences between them and the subject asset will have to be ignored.
And even if the valuer is fortunate enough to find any comparable assets,
he is still faced with the problern that prior transaction prices for intangible
assets are unlikely to be available. Anybody can read the prices for basic
commodities and listed securities in his daily newspaper, but the parties to
an intangible asset transaction often agree to keep the price secret. In light
of these drawbacks, the market approach is of only limited application in
intangible asset valuations, and is, in fact, not usually the approach that is
adopted.
When the cost approach is adopted in valuing an intangible asset, the
valuer seeks to determine what it would cost to reproduce (or replace) the
subject asset. The cost approach is based on the premise that a buyer
would pay no more for an intangible asset than the amount for which it
could be reproduced (or replaced). The frrst step is to determine what it
would cost to reproduce it in new condition. The valuer must include all
the costs that this would involve, such as materials, labor, amortization and
depreciation and other costs, not forgetting opportunity costs. But this is
only the first step. The valuer must consider that usually, the subject
intangible asset is not new, and that its value has been reduced by
technical, economic or functional obsolescence. In arriving at the value of
the intangible asset, deductions must thus be made to reflect obsolescence.
The cost approach is adopted principally in valuing intangible assets the
creation of which can be followed step by step and where the cost involved
in creating it can be estimated. In the case of software, for instance, it is
frequently possible to identify the costs that would be required to repro-
duce it; the man-hours that would be needed to develop a comparable
program can be estimated with some reliability. By contrast, the cost
approach could not normally be used to value an established brand because
the establishment of a brand is a lengthy and unpredictable process the
course of which is anything but direct (Greinert 2002 pp 159 seqq). Thus
an advertising budget, for instance, provides no information on the
strength of a brand. In valuation practice, therefore, the cost approach is
similarly of little relevance where intangible assets are involved.
498

For the reasons just discussed, the income approach is the one most
widely used for valuing intangible assets. It derives from investment
theory, and estimates the value of an intangible asset by calculating the
present value of the benefits expected to be derivable from it. In simple
terms, an intangible asset is worth what it can earn.

lncome Approach

1 . ldenllfy the future cash flows an investor would expect the

subject intangible asset to generate

T Cash Flow 1
FV 1: (1 +Discount Rate)1
!=1

2. 01scount future cash flows w1th an appropriate

d1scount rate

Fig. 2.

A valuer adopting the income approach has to perform two main tasks.
He has to
• identify and project the future expected cash flows attributable to
the asset over its estimated remaining usefullife, and
• determine an appropriate discountrate.
The starting point is usually the business plan prepared for the company
or product group concerned. However, the profits forecasted for future
periods shown in the business plan cannot be used just as they are, because
they relate to the total business enterprise and are generated by employing
tangible and financial assets as weil as intangible assets. The valuer thus
needs to isolate the profits attributable to the intangible asset he is valuing.
It is safe to say that this identification is the main problern to be solved
when using the income approach (Scholich 2000 p 53). Theorists and
 499

practitioners have meanwhile come up with a variety of methods for iden-

tifying attributable profits. Theseare (AICPA Practice Aid 2001 pp 12 seq,
Reilly and Schweihs 1999 pp 159 seqq, Mard et al 2002 pp 66 seqq):
- the Relief from Royalty Method,
- the Multi-Period Excess Eamings Method, and
- the Incremental Cash Flow Method.
The Relief from Royalty Method is probably the best known and most
widely used. Under this method, the valuer determines the royalty savings
attributable to the owner of the intangible asset. These royalty savings
arise by reason of the ownership of the asset, which makes it unnecessary
to obtain a license for a comparable patent, brand, formula or other
intangible asset. The owner thus avoids paying royalties. The royalty
savings are attributable entirely to ownership of the subject intangible
asset, so this method enables the benefits of ownership to be identified.
The particular advantage of the Relief from Royalty Method, however,
is the existence of market observed royalty rates customarily demanded
and paid for intangible assets in the market. The reliability of intangible
asset valuations is increased by reference to such royalty rates, which have
been agreed between willing parties acting without compulsion on either
side.
The Molti-Period Excess Earnings Method has come to play an im-
portant role in intangible asset valuation practice. The rationale for this
method is based on the question "Would the subject intangible asset
generate the same revenues on a stand-alone basis?" The answer, of
course, is "No". Normally, an intangible asset cannot generate revenues on
its own. Intangible assets acquire meaning only as part of the enterprise to
which they belang. The valuer must thus determine what other assets are
needed to enable benefits to be derived from the subject intangible asset.
These other assets may include land and buildings, machinery and equip-
ment, an assembled workforce, working capital and others. The enterprise
does not need to own these assets, however. For the subject asset tobe able
to generate revenues, it suffices for these other required assets to be leased.
In practice, the requirements just described are taken into account by
reducing the revenues expected to be generated by use of the subject
intangible asset, in combination with other assets, by subtracting lease
charges for these other, contributory assets. These contributory asset
charges represent national Ieasing rentals for land and buildings,
machinery and equipment, assembled workforce, working capital and
others. The valuer thus arrives at profit contributions identified as
attributable solely to the subject intangible asset.
500

As will be seen from the manner of calculation, this method assumes

that the subject intangible asset is the principal profit driver in its business.
If a profit is projected over and above the cost of capital, the valuer
attributes the difference between the cost of capital and the expected
profits, i.e., the "excess earnings", entirely to the subject intangible asset.
And this is usually justified, because it is very difficult to copy or find Sub-
stitutes for intangible assets. They thus assure the enterprise of an advan-
tage over its competitors which is responsible for the excess earnings.
Another method that can be used to identify profit contributions stem-
ming from intangible assets is the Incremental Cash Flow Method.
Under this method, the valuer determines what additional cash flows, in
the form of cost savings or incremental revenues, can be generated by
using the subject asset. Cash flows resulting from cost savings are
frequently determined in valuations of process patents, formulas or
production processes that are designed to reduce costs. But there are also
product patents and formulas that enable the enterprise to realize higher
selling prices for products. A brand name also will frequently enable goods
to be sold at higher prices compared to unbranded products or price
brands. In these cases, the incremental cash flows are basically determined
by the difference in selling price between goods of the subject brand and
comparable but unbranded goods.
Whatever method is used to identify profit contributions from intangible
assets, special attention must be paid to the tax effects. A distinction must
be made between those resulting in tax expenses and those giving rise to
tax benefits. Tax expenses arise because profit contributions from intan-
gible assets are part of the business profits, and a corresponding part of the
tax charge is consequently attributable to them. The rationale for a tax
benefit is as follows: a potential buyer of an intangible asset would recog-
nize it as an asset on acquisition, and thereafter charge amortization over
its (tax-allowed) useful life. These amortization charges reduce the tax
base. The reductions in tax expenses, or "tax amortization benefit",
increase the value ofthe intangible asset (AICPA Practice Aid 2001 pp 96
seq).
Besides identifying the attributable profit contributions, a valuer
adopting the income approach needs to determine the interest rate to be
used for discounting. This discount rate is arrived at on the basis of the
expected retum on the best alternative investment available. lt thus repre-
sents the minimum return that must be obtained from the subject intangible
asset if the enterprise is not to be worse off than it would be had it chosen
the best alternative investment. The valuer seeks this best alternative
investment in the form of a comparable investment in the capital market.
 501

When arriving at the discountratetobe used, the valuer must, however,

ensure that the profit contributions from the subject intangible asset are
comparable with the return on the alternative investment, particularly as
regards risk. Risk is normally recognized by adding a risk premium to the
base rate (Wirtschaftsprüfer-Handbuch 1998 Paragraph 182 in section A
pp 60 seq). The base rate is determined on the basis of the observed return
on a (quasi) risk-free investment in the capital market. Usually, the valuer
considers the return on govemment bonds, because they are regarded as
(quasi) risk-free (IDW S 1 Paragraph 120).
The risk premium is commonly computed by use of the Capital Asset
Pricing Model, or "CAPM", which compares the return on a particular
stock with the return on the stock market as a whole (IDW S 1 Paragraph
98). There are some justified objections to the CAPM because, among
other things, it assumes a single-period planning horizon, and that all
investors will hold a risk-efficient market portfolio (Wirtschaftprüfer-
Handbuch 1998 Paragraph 192 in section A p 65). Be that as it may,
computations based on the CAPM permit an objective determination ofthe
appropriate risk premium.
The risk premium determined by use of the CAPM is the premium for
the enterprise as a whole, whereas in a valuation of intangible assets the
valuer needs to determine the risk attaching specifically to those assets.
This normally differs from the enterprise's risk. The risk premium for an
enterprise depends, for instance, on its specific capital structure, capital
intensity, capital turnover rate and other systematic risk factors. While
these risk factors do affect the value of an enterprise, they should not be
allowed to affect the valuation of its intangible assets. It cannot be said, for
instance, that the value of a patent changes just because the enterprise is
more heavily indebted. In valuations of intangible assets, therefore, the
subject enterprise's systematic risk alone is an inappropriate basis for
determining the asset risk, and should be left out of account. Here it may
be found helpful to consider the risk structure of comparable enterprises,
and to use this analysis to compute an average beta (Högl et al 2002 pp 61
seqq). Analyzing a large number of comparable enterprises has the advan-
tage that the size of the beta determined in this way is not affected by the
capital structure, capital intensity, capital turnover rate and other sys-
tematic risk factors of an individual enterprise.
502

4 Summary

Intangible assets are frequently the factor primarily responsible for a busi-
ness's success and competitive advantages. This makes it necessary for
considerable top-level attention to be devoted to the management of these
assets, and for the potential inherent in them to be communicated to
investors and lenders to assure the business of being able to obtain the
finance it needs. Forthismanagement and communication tobe possible,
intangible assets need to be measured, i.e., valued.
The valuation approach most frequently adopted in practice is the
income approach, in which the future expected profit contributions from
the intangible assets are discounted using a risk-adjusted rate. The princi-
pal problern in this approach is the identification of the profit contributions
attributable to the intangible assets. Theorists and practitioners have
meanwhile developed methods that can be used for this purpose, so that by
and large, reliable valuation of intangible assets is now possible. The
extent of the progress that has been made in this field can be seen from the
fact that the major accounting systems now require all acquired intangibles
to be identified and individually valued.

References

Achleitner AK, Bassen A, Pietzsch L, Wicheis D (2002) Effiziente Kapital-

marktkommunikation mit Finanzanalysten - Gestaltungsempfehlungen für
Wachstumsunternehmen (Efficient Capital Market Communication with
Financial Analysts- Recommendations to Growth Businesses). Finanz
Betrieb: 29 - 43
AICPA Practice Aid (2001) Assets Acquired in a Business Combination to Be
Used in Research and Development Activities, New York
Arbeitskreis Externe Unternehmensrechnung der Schmalenbach-Gesellschaft e. V.
(2002) Grundsätze für das Value Reporting (Value Reporting Principles). Der
Betrieb: 2337- 2340
Arbeitskreis "Immaterielle Werte im Rechnungswesen" der Schmalenbach-
Gesellschaft für Betriebswirtschaft e.V. (2001) Kategorisierung und
bilanzielle Erfassung immaterieller Werte (Classification and Recognition of
intangible assets). Der Betrieb: 989- 995
Eccles RG, Herz RH, Keegan EM, Phillips DMH (2001) The ValueReporting
Revolution
 503

Fischer TM, Becker S, Wenzel J (2002) Wertorientierte Berichterstattung- Ein

empirischer Vergleich der internetbasierten Geschäftsberichte von DAX 30-
und NEMAX 50-Unternehmen (Value-Based Reporting- An Empirical
Comparison of the Internet-Based Annual Reports of DAX 30 and
NEMAX 50 Companies). Kapitalmarktorientierte Rechnungslegung: 14-25
Freidank CC, Paetzmann K (2002) Auswahl und Einsatz von Datenmaterial,
Analysemethoden sowie externen Beratern zur Vorbereitung von Kreditver-
gabeentscheidungen (Selection and Use ofData Material, Analytical Methods
and External Consultants in Preparing Lending Decisions). Der Betrieb: 1785
- 1789
Greinert M (2002) Die bilanzielle Behandlung von Marken (Accounting for
Brands), LohmariKöln
Högl S, Hupp 0, Maul KH, Sattler H (2002) Der Geldwerte der Marke als
Erfolgsfaktor fiir Marketing und Kommunikation (The Value of Brands for
Marketing and Communication). In: GWA e.V. (ed) Der Geldwert der Marke,
Frankfurt am Main, pp 37- 80
IDW S 1: Grundsätze zur Durchfiihrung von Unternehmensbewertungen
(Standards for the Valuation of Business Enterprises)
International Valuation Standards Committee (2001) International Valuation
Standards
Kahle H (2002) Informationsversorgung des Kapitalmarkts über internationale
Rechnungslegungsstandards (Providing the Capital Market with Information
on International Accounting Standards). Kapitalmarktorientierte Rechungs-
legung: 95 - 106
Küting K (2000) Möglichkeiten und Grenzen der Bilanzanalyse am Neuen Markt
(Teil II) (Possibilities and Limitations of Balance Sheet Analysis in the "New
Market"). Finanz Betrieb: 674- 683
Mard MJ, Hitchner JR, Hyden SD, Zyla ML (2002) Valuation for Financial
Reporting, New Y ork
Maul KH (2000) Wissensbilanzen als Teil des handelsrechtliehen
Jahresabschlusses (The Intellectual Property Statement as a Financial
Reporting Component). Deutsches Steuerrecht: 2009- 2016
Maul KH, Menninger J (2000) Das "Intellectual Property Statement" - eine
notwendige Ergänzung des Jahresabschlusses? (The "Intellectual Property
Statement"- a Necessary Extension to the Financial Reporting?). Der Betrieb:
529-533
Pellens B, Fülbier RU (2000) Immaterielle Vermögensgegenstände in der internen
und externen Unternehmensrechnung (lntangible Assets in Internal and
External Accounting). In: Küting K, Weber CP (eds) Wertorientierte
Konzernfiihrung, Stuttgart, pp 119- 155
PricewaterhouseCoopers (2002) Shedding Light on the New Business Combi-
nation Rules
PricewaterhouseCoopers, Sattler H (200 1) Praxis von Markenbewertung und
Markenmanagement in deutschen Unternehmen (Brand Valuation and
Management Practice in German Businesses). 2nd edn. Frankfurt am Main
504

Ranker D, Wohlgemuth F, Zwimer C (2001) Die Bedeutung immaterieller

Vermögenswerte bei Unternehmen des Neuen Marktes und daraus resul-
tierende Implikationen fiir eine kapitalmarktorientierte Berichterstattung (The
Significance of Intangible Assets in "New Market" Companies and Resulting
Implications for Capital Market-Oriented Reporting). Kapitalmarktorientierte
Rechnungslegung: 269 - 279
Reilly RF, Schweihs RP (1999) Valuing Intangible Assets, New York et al.
Ruhwedel F, Schultze W (2002) Value Reporting: Theoretische Konzeption und
Umsetzung bei den DAX 100-Untemehmen (Value Reporting: Theoretical
Concept and Practical Application to the DAX 100 Companies). Zeitschrift
fiir betriebswirtschaftliche Forschung: 602- 632
Scholich M (2000) Die Bewertung der menschlichen Arbeit (The Valuation of
Human Work). In: Amold H, Englert J, Eube S (eds) Werte messen- Werte
schaffen, Wiesbaden, pp 39 - 60
Selchert FW, Erhardt M, Fuhr A, Greinert M (2000) Prüfung des Lageberichts
(Auditing ofnotes and MD&A), Bielefeld
Sirnon H, Ebel B, Pohl A (2002) Investor Marketing. Zeitschrift fiir
Betriebswirtschaft: 117 - 140
Wirtschaftsprüfer-Handbuch (1998), Band II, Institut der Wirtschaftsprüfer in
Deutschland e.V. (ed), 11th edn. Düsseldorf
Are Stars Worth their Pay?

Franz Wirl'

1 Abstract

It is often criticised that certain stars - e.g. opera singers and actors, soccer
players, CEOs, lawyers and medical doctors and 'stars' at intemet and
software companies - are not worth their pay. This paper shows within an
incentive contract that payments exceeding incremental contributions are
not necessarily an anomaly. A necessary condition isthat the agents have
some intrinsic motive to perform well even in the absence of incentives.
This in turn explains that rewards are often very high particular in those
fields (e.g. arts, sports and also in the new economy), where one expects
that incentives are less important due to an agent's intrinsic motive to per-
form well anyway. In fact, a positive incremental profit contribution of a
particular type is in generat neither necessary nor sufficient.

2 lntroduction

In quite different fields, but in particular in sports, arts and business (e.g.
benefits to CEOs, high performers at intemet, engineering and law frrms or
consultancies) it is questioned whether such 'stars' are worth the money2 •

1 I am grateful for valuable comments from the discussant, Professor

Schwalbach, and other participants at the conference.
2 According to The Economist, Nov. 9th 2002, p 69, the major European soccer
teams are in a difficult economic situation (e.g. Fiorentina was declared
bankrupt) due to the high salaries of the players and want therefore to restriet
the salaries to 70% of the turnover. On the other hand higher salaries seem to
buy success, The Economist, June 1•t, 2002, Passion, pride and profit, A survey
offootball, p 10.
506

An extemality such as weakening an opponent seems an intuitively plau-

sible explanation3• This paper, however, argues that compensations ex-
ceeding incremental profit are not necessarily an anomaly and do not even
require an extemality (and in particular no extemality of the kind men-
tioned above: they can arise from the requirement of incentive compatible
payments only. In factapositive profit contribution of a particular type is
neither necessary nor sufficient for employment within an optimal incen-
tive scheme given private information of the hired agent about his charac-
teristics, ability or work ethics. Of course, this theoretical finding does not
justify all payments to stars (including to some executives).
The paper is organized along the example of a wage contract; parts of
the derivation are relegated to an Appendix. The model is introduced and
the optimal contract is analysed in Section 3. Section 4 investigates, why
capping high rewards will not solve the problern and what conditions
foster such excess payments. Final remarks complete this investigation.

3 lncentives

The following argument is based around an incentive contract for em-

ployees, but its implication- contracting with 'efficient' types can end up
in an unavoidable loss - extends to other principal-agent relations.

3.1 Agents

Agents of different types t, which is private information of each agent,

have to carry out an observable task x. Absent incentives, the agents' pay-
offis
W(x,t) = tu(x)- K(x), (1)
where u, represents the standardized benefit that is scaled by the type t, and
K the costs associated with the action x. This specification varies the utility
across types (the multiplicative specification ensures that also marginal
benefit depends positively on the type, yet the linearity in t is not crucial)

3 For example, the recent offer to Mr. Bailack to leave Bayer-Leverkusen for
Bayern-München may consist of the expected contribution of Mr. Ballack to
Bayern-München and for a weakening of the crucial opponent for the forth-
coming national championship. Yet surprisingly, such an externality does not
warrant an excess salary in a Nash contest between 'Bayern München' and
'Bayer-Leverkusen' modelledas a Tullock lottery; see Appendix.
 507

but similar results apply if the more able types (characterized by a larger
value oft) have less costs for carrying out a task x yet identical utility. The
two functions in (1) satisfy the usual properties: u is a concave utility
function that satisfies an Inada condition (lim u'(x) ~ oo for x ~ 0)
ensuring a positive output, and K is an increasing and convex cost func-
tion. The numerical and graphical examples use the following specifica-
tions:
u(x) = -x-a = -11~ (i.e. a= Y2) andK(x) = Wrr. (2)
The utility function implies constant relative risk version (short CRRA)
and is due to this property often used in finance but also in the theory of
endogenous growth; the particular specification in (2) uses the power a =
Y2; yet despite its name, the agent as weil as the principal are assumed to be
risk neutral. The agents' autonomous actions absent a compensation by the
employer are denoted by x0(t) and must satisfy the following first order
condition, tu' = K', so that x0(t) is increasing in the type due to the implicit
function theorem4, see the example in Fig. 1. That is, agents with a }arger t
choose a larger vale of x and are thus more 'efficient' for a principal inter-
ested in large 'outputs'. This requires that the agents have some kind of an
intrinsic motivation to provide some output, here x0 > 0. Not only psy-
chologists but also some economists, in particular Bruno Frey (2000),
emphasize the importance of intrinsic motives up to the point of question-
ing whether extemal incentives crowd-out intrinsic incentives; see also the
empirical application to the NIMBY syndrome in Frey and Oberholzer-
Gee (1997). Indeed, most soccer players, opera singers and also many
professionals (of course most university professors) would perform quite
well even in the absence of incentives. In contrast, Lazear (2000a, b) stress
the importance offinancial incentives based on a case study (Safelite).

3.2 Principal

The principal contracts the observable and verifiable activity x to the agent
and accrues the gross benefit
V(x), V'> 0, V''< 0, e.g., V(x) = v0 - v/~. (3)
That is, the principal is interested in a better performance x subject to the
law of diminishing retums. The specification in (3) for numerical and

4 .io (t) = dxo (t) = u' (dt)) > 0 , dots denote total derivatives with respect to
dt tu"(x0 (t))-K"(x 0 (t))
the typet.
508

graphical purposes is based on the same type of utility functions as in (2)

for the agent and in fact the same parameter value, a = 'l'l, is assumed. As a
consequence, differences in this parameter (of 'relative risk aversion')
cannot explain that an agent might be rewarded in excess of the marginal
product.
The 'first best' choice of the activity, which is denoted by x" results
from maximizing the joint surplus V+ W that yields the first order condi-
tion: v' + tu' - K' = 0. However, this first best contract, compensating the
agent oftype t only for the incremental cost associated with the task x 1(t),
w1(t) = W(x0(t), t)- W(x 1(t), t) > 0 since x 1 > x0, is not implementable, ifthe
type t is private information of the agent.
If the efficiency of type t ( say of a hired programmer) is not observable,
the principal must provide incentives to stimulate output accounting for the
private information of the efficiency type. Let w(t) denote the incentive
payment to types with ability t carrying out the task x(t), then the principal
obtains the net profit from a type t agent:
1Z(t) = V(x(t))- w(t). (4)
The principal knows, or assumes a corresponding a priori, distribution of
types, F(t), t E kJ],f= dF/dt (density); all figures are basedonuniform
distributions. The expected profit from a contract, {x(t), w(t), t E kJ]},
i.e., rewarding the performance x(t) with the premium w(t), is
i to i
(5)
E(n) = pv(x(t ))- w(t )}iF(t)= Jv(x0 (t))f(t )dt + pv(x(t))- w(t )]f(t )dt.
! ~ ~~·----~----~
without incentives with incentives

The calculation in (5) accounts for the possibility that not all types
employed will necessarily take part in the incentive scheme but only the
sufficiently efficient types, t ~ t 0, where t 0 denotes the marginal type
working on an incentive basis. The fixed wage of those working outside
the incentive premiums, t < t0, is normalized to zero. This separation is not
only typical for many commercial enterprises but also for a number of
NGOs (e.g. Greenpeace, Global 2000) and non-profit organizations
(Caritas, the Churches, etc.) in which a large number ofpeople participate
and work due to their intrinsic motives while another group faces in addi-
tion often quite stiff incentives similar to those found in commercial enter-
prises.
 509

3.3 Constraints
An agent of type t can accrue from the principal's offer- to honour the
performance x(t) with a corresponding wage w(t) - the net gain

U(i, t) = tu(x(t))- K(x(t)) + w(t), (6)

by pretending the typei . Indeed, the agent will pretend that level of i,
which maximizes the payoff U( i, t). The familiar incentive compatibility
constraint results from the revelation principle: the principal can achieve
the maximum of the expected profit by restricting the contracts to those
that induce the agents to reveal their true type. That is, the contract must be
designed in such a way that an agent of type t carries out the assigned task
x(t). This is ensured for rational agents if
U(t) = U(t, t) ~ U( i, t). (IC)

The individual rationality constraint (IR) states that an agent enrolls into
an incentive programme only, ifthe achieveable net payoffis not less than
without incentives. That is, using the above introduced and defined func-
tion U for the agent' s payoff,
U(t) ~ R(t) = W(t, x0(t)) = tu(x0(t))- K(x0(t)). (IR)
The agent's reservation price, R defined in (IR), is the agent's payoff from
carrying out the task x0, which would be preferred in the absence of incen-
tive payments; as already mentioned the associated fixed wage component
in R is normalized to zero. The agent' s reservation price is not constant as
in most models but type dependent, in particular, R(t) =u(x 0 (t)) > 0, due
to the envelope theorem, and also convex, R(t) = u' (x 0 (t ))x0 (t) > 0 .
In analogy one can define also a reservation price for the principal,
which determines the principal' s profit without providing an incentive
7ro(t) = V(xo(t)). (8)

3.4 Optimal contract

The principal's optimal contract offer results from maxtmtzmg the

expected profit subject to the incentive compatibility and individual ra-
tionality constraints. That is, the risk neutral principal solves the following
optimisation problem:
510

max E(:r) from (5) subject to (IC) and (IR). (9)

{(x(t), w(t )},te(!,i]}

The interior solution of the target performances, the so called relaxed

program, denoted x, compare Fudenberg-Tirole (1992), is implicitly
characterized in the following way:
(10)
where h(t) =./{t)/(1 - F(t)) denotes the hazard rate, which is assumed tobe
increasing, h > 0. The above condition (1 0) reduces to
V'+ u'[t- llh] == K'. (11)

Hence v' + tu' == K' for t == t so that xr (t) = x 1(t), which is known as no
distortion at the top. However, only this type t delivers the first best,
while all other types produce less compared with the first best, x < x 1• As a
consequence, the set of types enrolled into the incentive program is never
empty since xr exceeds the autonomaus actions x 0 at least for sufficiently
efficient types since Xt > xo.
However, the relaxed program may fall below the autonomaus actions
in which case the principal prefers the autonomaus actions to those associ-
ated with the relaxed program. Consequently, the marginal type enrolled
into the incentive program, denoted t0, is determined by the intersection of
the autonomaus actions Xo with the relaxed program xr, x(to) == Xo(fo). First,
the relaxed program is meaningless for [t- llh] < 0. Second at this inter-
section, V'(x0(t0)) = u'(x0(t0))/h(t). Third and as a consequence, xr > x 0 fort
> t0 due to the assumed properties of the hazard rate and that the relaxed
program is increasing5• If no such intersection between xr and x 0 exists,
then t0 == t and all agents work on an incentive basis. Fig. 1 shows an
example of the agents' autonomaus actions, of the relaxed program and the
determination of the set of agents working on an incentive basis and those
working without explicit incentives (i.e. for a fixed wage contract). The
optimal tasks, x*(t), are then given by joining the two parts, the autono-
maus actions Xo and the relaxed program r:
(12)
* ( ) _ {x 0 (t) ! < t < t0
x t - ( )for _.
xr t to..:;, t..:;, t
 511

2.2

f
6 7
t
1.8

1.6

1.4

enrolled into incentive pro gram

1.2

without
incentives

Fig. 1. Tasks (autonomous x0 and relaxed program x') as a function ofthe typet,
k = 2, v0 = 10, v = 5, t E [ 4, 10] uniformly distributed.

cap

X
1.2 1.4 1. 6 1.8 2.2

Fig. 2. Optimal incentives, wage w as a function of the performance x, for the ex-
ample in Fig. I.

Fig. 1 shows the outcome in its direct form, i.e. output x as function of
the type, which is of course not known to the principal. Given the tasks
and the two constraints - incentive compatibility and individual rationality
- one can calculate the agents gross payoff U(t) and from that the cor-
512

responding wage w(t); for details see the Appendix. Eliminating the
dependence on the type t yields a more familiar scheme: incentive pay-
ments w ('premiums') for certain tasks x. The corresponding example in
Fig. 2 highlights that better performances require disproportional increases
of the compensation ( due to the familiar convex shape).

6.3
6.2
6.1
I
5 7 8
5. 9
5.8 1oss

Fig. 3. The principal's profit as a function of the type from the optimal contract
associated with the tasks given in Fig. 1.

This optimal incentive scheme has the following potential implications

on the principal' s incremental pro fit. An agent of type t who responds to
the incentives by delivering the net benefit n(t) = V(x(t)) - w(t) to the
principal, while offering no incentives yields the (reservation) payoff 1liJ(!)
from (8) so that
tln(t) = [V(xr(t))- V(x0(t))] - w(t) (13)
calculates the incremental profit from the incentive scheme as a function of
the (unknown) type. While the aggregate (or average) incremental benefit
must be positive, it need not be positive for each type t. First note that
tln(t) > 0 for types t sufficiently close to the marginal type t0• This follows,
because starting with types delivering a deficit is clearly sub-optimal. Thus
 513

D.n(.,t0 ) = 0 determines the marginal type 6 • Hence the claimoftbis paper- it

is optimal to hire agents and offer them rewards that result in a loss for the
principal for some types - can, if at all, happen only for 'efficient' types
but never at the bottom. Indeed, the numerical example in Fig. 3 shows
that the efficient typest> t 1 = 9.42675 ... incur a loss for the principal.
The example in Fig. 3 highlights that hiring the highly efficient agents t
> th 'asking' them to carry out the task x"(t) = x(t) and paying them the
incentive compatible wage w(t) results in a loss for the principal. The less
efficient types t0 < t < t 1 yield an incremental profit and the typet= t 1 at
least breaks even. This example seems to vindicate in particular the claim
of Bruno Frey that incentives can crowd out own efforts and can thus be
detrimental; yet it turns out that this, admittedly plausible, argument is not
correct in this case. Furthermore, similar problems of excessive payments
can arise in quite different examples, e.g. in incentives for energy conser-
vation, compare Wirl (1999).

4 Why not capping?

Given the loss incurred with efficient types, one may wonder, why the
loss-making types are not excluded in analogy to the exclusion of the types
at the bottom from participating in the incentive scheme. However, there is
an asymmetry between the bottom and the top, since 'efficient' types can-
not be excluded directly. First, their typeisnot known (this applies at the
bottom as weil) and second, types close to the most efficient type i have,
in contrast to those at the bottom, an incentive to pretend a different, more
precisely, less efficient type for any deviation from the optimal contract.
Yet it seems that the principal can avoid these loss-making instances by
replacing the original contract by a capped version: all tasks x exceeding
x(t1) will nevertheless receive the payment w{t1); a corresponding example
of such a cap is shown in Fig. 2 that avoids exactly those large premiums
for excellent performances that lead in Fig. 3 to a loss. Indeed many con-
tracts in the real world have a cap and after all there is also a cap at the
bottom since all outputs below that of the marginal type t 0 ( and thus the
corresponding types t < t 0) receive the same wage incentive as t0•

6 This condition of a positive incremental contribution to the principal's profit is,

however, not sufficient given a minimum wage ~ > 0 and no intrinsic motive,
x0 = 0, because then even the marginal type must deliver already a positive
profit contribution, see Wirl (2000).
514

Yet this (and any other kind of) capping results also in a loss and is
moreover sub-optimal. This can be explained in the following way. First,
by construction, the type t 1 break:s even, i.e., the incentive payment equals
just the incremental profit, w(t1) = V(x"(t 1) ) - V(x0(t 1)). Second the types t >
t 1 will behave like the type t 1 if confronted with the capped incentive
scheme. This follows from the concavity of the agents' utility u. In par-
ticular they definitely prefer xr(t1) to x0(t), which would be their autono-
mous choice, and this performance earns them the wage actually designed
for the type t1• That is, the capped contract induces the typest~ t1 to pre-
tend i = t1 suchthat x(t) = x"(t1) > x0(t). As a consequence, thesetypest >
t 1, which produce as the type t 1, receive also the same compensation, yet
for less incremental performance, w(t1) = V(xr(t1)) - V(x0(t 1)) > V(xr(t1)) -
V(x0(t)) since the compensation paid to type t 1 allows the principal just to
break: even and x0(t) > x0(t 1) fort~ t 1• Therefore, this capping does not only
violate the incentive compatibility constraint since the agents (here t > t 1)
choose bundles not designed for them, but incurs a loss, in fact, a larger
loss: the agents t > t 1 collect the same wage as the type t 1 that triggers yet
less incremental output. Hence it must result in deficits given the fact that
the type t 1 just break:s even.
This raises the question, what are crucial conditions such that a principal
must be prepared to expect to pay efficient types over and above their
incremental contribution? A first and usual suspect for complexities in
agency relations is here given: the type dependent reservation price of the
agent, compare e.g. Kerschbamer-Mademer (1998), Jullien (2000) and
Wirl-Huber (1997). However, this property is not the entire story and
probably not even crucial (at least directly). Second, different types per-
form different tasks absent an incentive, more precisely the efficient types
deliver a better performance even in the absence of incentives due to an
existing work ethics, professionalism, enjoying good performances (in
particular in sports and arts) and other intrinsic motives. This second
property renders immediately a third point that the 'reservation price' of
the principal is type dependent too. A further consequence of this property
is that less incremental performance is demanded from these efficient
types due to the already good performance for no incentive pay at all. Yet
incentive compatibility requires to reward high performances with higher
and higher compensations up to a point where this balance between incre-
mental performance retrieved and necessary compensation paid turns
negative.
While sufficient conditions for a situation shown in Fig. 3 are not very
informative at the modeHing stage, it is easy to obtain a sufficient condi-
tion for the opposite, that the principal profits from each employed type:
 515

L\li"(t) = V'(xr (t ))xr (t )- V'(x 0 (t ))x0 (t )- w(t) ~ 0, (14)

is sufficient that the illeremental profit is non-decreasing in the type: Sub-

stituting for w
using w = U- Wand Ü =w; yields

As a consequence, x0(t) constant for all t, i.e. all the agents perform just the
necessary minimum in the absence of incentives, ensures Ai(t) > 0 .
Therefore, x 0 > 0, i.e., some intrinsic motives in particular of efficient
types, is a conditio sine qua none that the principa/ can loose money .from
providing incentives to efficient types. The requirement - x 0(t) depends
positively on t - has the following implications of non-constant reservation
prices ofthe agent and also ofthe principal (andin contrast to the standard
model).

5 Concluding remarks

Finally an answer (restricted to optimal contracts) can be given to the

question raised in the title ofthis paper.
In the narrow sense, it can indeed be economical to reward excellent
performances (very efficient employees) and thus in particular stars like
Renaldo, Zidane and Beckham in soccer, great tenors (Pavarotti,
Domingo) and film stars (e.g., Tom Ranks, Tom Cruise, Bruce Willis)
over and above their incremental contribution.
This kind of over-paying of the most efficient parts of the labour force
is, however, an unavoidable sacrifice in order to retain the incentive com-
patibility. Therefore, taking a broader view and considering the effects on
the overall incentives the answer is: They can be worth all the money even
if the individual balance is negative for the principal. After all, the associ-
ated contract is the optimal one and these over-payments are just additional
hidden costs of contracting in world with private information.
What are likely conditions that Iead to such economically counter-intui-
tive outcomes: to pay someone in excess of the incremental contribution?
A crucial necessity for this possibility is that the agents have different
intrinsic motives to work well even absent an incentive. If all agents per-
form equally weil (not necessarily nothing) irrespective oftheir type in the
absence of incentives, the kinds of 'over' payments analysed in this paper
516

are impossible. This in turn explains the puzzle that such high salaries and
with it 'over'-payments occur in particular in fields (sports and arts) where
one expects that incentives should be of less importance given the intrinsic
motive to perform well anyway.

Appendix

A. Derivation of the optimal contract

The agents' autonomous actions for the specification (2) are given by

(Al)

The relaxed program (10) yields for the specifications in (2) and (3):
(A2)

Solving the equation x 0(t) = x'(t) determines the marginal type,

t0 =t-v. (A3)

The agents' payoffs from participating in the incentive scheme (that is

for all types t ~ t0) must satisfy the following differential equation (see
Fudenberg and Tirole (1992)):
(A4)

Integration of (A4) yields U(t). This determines the wage from w = U- W,

which for the specifications in (2) and (3) gives:
(A5)
W(t ) = k ~[ f (V + 2t - t_)_5 - -5 (-f - V )i5 - -5 (V - f-~
_!_ 3( -)i]
j5 - - V + 2t- f 5 .
4 8 8
 517

B. Externality: weakening an opponent

Consider two managers of two firms or teams (say 1 = Bayern München

and 2 =Bayer Leverkusen). The probability, p, that team 1 wins a prize P
(say the national championship) depends on the relative quality of the
teams denoted q 1 and q 2• Initially, the probability of winning the prize is
given by the Tullock lottery (see Tullock (1967)):

qO!
Po=-=-
qot +qoz
where the additional index 0 refers to the status quo ex ante. The quality of
the teams can change due to the transfer of a player, whose personal
contribution to the success of a team is v < q 20 ( otherwise the remaining
team members are a liability instead of an asset), from one team (here from
2) to the other (1 = Bayern München): q 1 = q 10 + v and q 2 = q20 - v. The
likelihood ( denoted by 2) that the sought player changes teams (i.e. moves
from '2' to '1 ') depends on the offered salary in a manner similar to win-
ning the championship; i.e., offering a higher salary increases only the
probability to acquire the player:

The two, risk neutral managers maximize the following objectives:

(B1)

(B2)

This set up ofthe objectives differs from Tullock's game since the offered
salary has to be paid only if the player joins the corresponding team, while
in Tullock (1967) the expenses are up-front and sunk. The corresponding
first order conditions for the optimal offers wi are:

Pvwz -wt(wl +2wzXqw +qzo) =O (B3)

(qw + qzoXwt + Wz)2 '
518

Pvwi -wz(2wi + wzXqio +qzo) = 0 (B4)

(qio +qzoXwi +wJ2 •

The Nash equilibrium

Pv (BS)
wi = 3(qio +qzo) = Wz
is symmetric irrespective of initial quality differences: Both managers of-
fer the same wage w and this wage is just one third of the player 's incre-
mental contribution to the prize. Therefore this extemality of stealing an
important player from a competitor does not lead to the kind of over-pay-
ment addressed in the paper, at least not for the assumed framework of
Tullock lotteries and Nash equilibria; the alternative choice of the
Stackelberg equilibrium does not alter this conclusion.

References

Frey B (2000) Leistung durch Leistungslohn? Grenzen marktlieber Anreizsysteme

für das Managerverhalten, ZtbF. Sonderheft 44: 67-95
Frey B, Oberholzer-Gee F (1997) The Cost of Price Incentives: An Empirical
Analysis of Motivation Crowding-Out. The American Economic Review 87:
746-755
Fudenberg D (1992) Tirole J, Game Theory. 2nd, MIT Press Cambridge (Mass.)
Jullien B (2000) Participation Constraints in Adverse Selection Models, Journal of
Economic Theory 93: 1-47
Kerschbamer R, Mademer N (1998) Are Two a Good Representative for Many?
Journal ofEconomic Theory 83: 90- 104
Lazear EP (2000a) Performance Pay and Productivity, Journal of Political
Economy 90: 1346-1361
Lazear EP (2000b) The Future of Personnet Economics, Economic Journal 110,
F611-F639
Olson M (2000) Power and Prosperity, Basic Books
Tullock G (1967) The Welfare Costs of Tariffs, Monapolies and Theft, Western
Economic Journal 5: 224-232
Wirl F (1999) Conservation Incentives for Consumers, Journal of Regulatory
Economics 15: 23 - 40
Wirl F (2000) Wage Bargaining and Incentive Compatibility: Is Unemployment
Optimal After All? In Dockner E, Hartl RF, Luptacik M, Sorger G (eds),
Optimization, Dynamics, and Economic Analysis, Essays in Honor of Gustav
Feichtinger, Physica Verlag, pp 200-213
 519

Wirl F, Huber C (1997) Voluntary Intemalisations (Contracts) Facing the Threat

ofa (Pollution) Tax, Working Paper, University ofMagdeburg
 Section 6

Accounting
A Tale of two Bubbles: A Preliminary Look at the
US Internet and Biotechnology Bubbles

Elizabeth Demers

Philip Joos

1 lntroduction

"Those who do not learn from history are destined to repeat its mistakes."
- Santayana
The term speculative bubble refers to a situation where prices diverge
from their so-called "fundamental values" due to investor optimism rather
than anything intrinsic to the values of the underlying assets themselves. 1
Economic historians have documented that one of the hallmarks of a
speculative bubble is "new era" thinking, or the belief that some new
price-enhancing circumstance is present in the world that somehow justi-
fies a permanently higher level ofvaluation (e.g., Galbraith (1993); Shiller
(2000)). Despite the world's long history of speculative bubbles (see, e.g.,
Kindleberger (2000) or Garher (200 1) for a detailed historical perspective),
Welch (2001) recently identified stock market "frenzies" such as the most
recent US Internet stock bubble as one of the top 10 challenges yet to be
addressed by empirical financial research. Technology market observers
and casual empiricists such as Perkins and Perkins (1999) have remarked
that the Internet stock bubble has numerous parallels with the US biotech-
nology bubble that preceded it by less than a decade. Consistent with the
"new era" thinking associated with other past speculative manias, for
example, both the biotechnology and Internet stock bubbles were charac-
terized by a belief that the new technology underlying these then-nascent
stage industries would be revolutionary in their impact on the human con-
dition. What seems most surprising is that investors in US technology
stocks did not appear to learn from the biotech bubble and we thereby
witnessed a similar, albeit more extreme, cycle of upward and ultimately
downward price spirals when the Internet industry emerged less than a

1 Shiller (2000), e.g., defines a speculative bubble as an ''unsustainable increase

in prices brought on by investors' buying behavior rather than by genuine,
fundamental information about value"
524

decade later. Surely revolutionary technological breakthroughs such as the

human genome and the Internet are likely to continue to occur periodically
now that we live in an era of unprecedentedly rapid scientific and techno-
logical advancement. This then begs the question: will we observe yet
another stock bubble in a few years' time when the next "revolutionary"
technology begins to be commercialized? If we use modern history as our
guide, the answer to this question would be "yes." As academic
researchers we are thus motivated in this study by the wisdom of the
philosopher George Santayana, and herein make a preliminary attempt at
understanding the recent tech stock bubbles so that we are not destined to
be slaves to financial markets history. In this preliminary paper we docu-
ment some of the common elements between the two most recent tech-
nology stock market bubbles. Our ultimate goal as we continue this
research will be to develop empirical models to explain the characteristics
that determine which stocks bubble, to further document the relation
between stock bubbles and the news media, and to investigate any signifi-
cant similarities or differences in bubble behavior across the two indus-
tries.
Shiller (2000) notes that the inception of the history of speculative bub-
bles is coincident with the advent of newspapers. He suggests that the
news media are fundamental propagators of speculative price movements,
and that the media may foster stronger feedback from past price changes to
further price changes. Following the theories put forth by Shiller, we make
a pioneering effort to examine empirically the relation between corporate
media mentions and stock return behavior. By undertaking our analyses
within the setting of the Internet and biotechnology industries during the
two most recent and extreme technology "bubbles," we expect to
maximize the power available to detect any underlying relations between
media activity and stock price behavior. In the current draft of our study,
we provide strong evidence of a relation between media citations and post-
IPO stock return behavior.
We also provide preliminary evidence on the association between survi-
vorship through each ofthe biotech and Internet bubbles, respectively, and
various characteristics of the firms at the time of their IPOs. We find that,
consistent with anecdotal accounts of the bubble phenomenon, firms that
go public in the early stages of each respective industry's lifecycle are
significantly more likely to survive. Biotech companies with a higher
proportion of insider directors at IPO and with higher Ievels of VC owner-
ship at IPO are more likely to survive, whereas no ownership or
governance characteristics are significantly associated with Internet com-
pany survival. Higher initial returns are a negative foreboding for Internet
 525

stocks, as higher IPO underpricing is negatively associated with survival

for these companies.
The balance of this paper is organized as follows. Section 2 provides
some background on the US Internet and biotechnology industries. Section
3 describes our sample selection process and data sources. In Section 4 we
present descriptive data related to the two technology sectors, and in
Section 5 we provide some preliminary multivariate regressions related to
bubble survivors. Section 6 concludes with a summary of the evidence that
we have documented to date and discusses our planned research for
Continuance of this study.

2 Background to the internet and biotechnology

industries

2.1 The biotechnology industry

The Biotechnology Industry Organization defines biotechnology as "the

use of the cellular and molecular processes to solve problems or make
products" (Biotech Industry Organization (2002)). Included in this defini-
tion of the industry are firms that use cells and biological molecules for
applications in medicine, agriculture and environmental management. The
biotechnology industry is very significant to the US economy. The market
capitalization of publicly traded biotech companies was approximately
$224 billion as of early May 2002. The industry has more than tripled in
size since 1992, with revenues increasing from $8 billion in 1992 to $27.6
billion in 2001. Biotech companies directly employ approximately 179,000
people and in 2001 spent an estimated $15.6 billion on research and
development (Biotech Industry Organization (2002)). Figure 1 presents a
time series chart of the value of the biotechnology index from 1980
through to the end of 2001. As is evident from the graph, the value of the
industry was gradually increasing for several decades before exploding
with other technology stocks in the late 1990s, and then ultimately
declining in value in 2000 with the general tech downtum.

2.2 The internet industry

Consistent with prior studies in the Internet sector (e.g., Hand (2001),
Demers and Lev (2001), Demers and Lewellen (2002)) we define Internet
526

companies as firms that earn the majority of their revenues as a result of

the existence of the InterneU Similar to biotech, the Internet industry is
also an economically significant sector within the US "New Economy" in
terms of both its market capitalization and wealth creation; notwith-
standing the Internet stock market correction in early 2000 (see, e.g.,
Demers and Lev (2001)), the market value ofU.S. Internetfirmsthat went
public from 1992-2001 was estimated to be over $424 billion as of
December 31 8\ 2001 (Morgan Stanley Dean Witter (2002)). As suggested
by Hand (2002) and others, Internet firms' large, rapid, and strategically
oriented spending on branding, information technology, personnel, and
R&D intangibles in many ways typify the New Economy. Internet compa-
nies are generally classified as being in either the business-to-business
("B2B") or business-to-consumer ("B2C") sector, where the nomenclature
derives from the customers to whom the Internet company markets its
products or services. Thus, one distinguishing feature of the Internet
industry relative to the biotech sector is that many Internet companies are
B2C companies, implying that they are very mass-market and consumer-
oriented. This differential in the "mass appeal" nature of the companies in
the two samples will be exploited as we investigate further the associations
between media coverage and stock bubble behavior.
Figure 2 presents a time series chart of the value of the Internet index
from 1992, when America Online went public, through to the end of 2001.
Clearly the market for Internet stocks became most exuberant in 1999
when there were many new entrants in the market (see discussion in
section 3.1.2 below), and the Internet bubble ultimately hurst in March
2000 and continued its decline through 2001. From early 2000 through the
end of 2001, the Internet index declined in value by several hundred billion
US dollars.

2 This definition was originally established by internet.com, an Internet industry

portal site, in order to distinguish between "pure play" Internet companies and
entities that would exist without the Internet generating a majority of their
revenues.
 527

3 Sampie selection and data description

3.1 Sampie selection

3.1.1 Biotechnology sample

We identify all biotechnology IPOs for the period of January 1980 through
December 2000 using the SDC New Issues database, excluding unit
offerings and ADRs. We initially select all IPOs for which the SDC bio-
technology indicator variable is equal to one. We then restriet the sample
to include only scientifically-oriented biotech firms, being those with
primary SIC codes in one ofthe following sectors: 2833 (medicinal chemi-
eals and botanical products), 2834 (pharmaceutical preparations), 2835 (in
vitro andin vivo diagnostic substances), 2836 (biological products, except
diagnostic substances), and 8731 (commercial physical and biological
research). 3 This results in a sample of385 biotechnology IPOs.
Figure 3 presents a frequency distribution of biotechnology IPOs
(including ADRs and unit offerings) by calendar year. As is evident from
the graph, there have been several waves of biotechnology IPO "hot
markets" as the IPOs exhibit dustering patterns over time that are con-
sistent with the general phenomenon documented by Lowry and Schwert
(2002).

3.1.2 Internet sample

There does not currently exist a standard SIC code or other official classi-
fication system with which to identify Internet companies, and therefore a
listing of all initial public offerings of Internet-related companies was
compiled from several sources. We began with the InternetStockList™
(provided by internet.com at http://www.internetnews.com/stocks/ list/), a
frequently cited and authoritative list of currently trading Internet
companies. Because the InternetStockList™ exhibits a survivorship bias
(i.e., only currently trading companies are included on the list), we also
referred to the Morgan Stanley Dean Witter (2002) ("MSDW")
Technology and Internet !PO Yearbook (8th ed.). The MSDW yearbook
provides a comprehensive listing of all technology and Internet IPO's for
the 1980-2000 period, including those that subsequently have been

3 This restricted classification of biotechnology firms follows that adopted by

Ernst & Young (2000).
528

acquired. The sample consists of 373 Internet companies that undertook

initial public offerings prior to the end of2000.
Figure 4 presents a frequency distribution of Internet IPOs by calendar
year. The pattern in Figure 4 exhibits some of the classical characteristics
of a bubble; the number of IPOs gradually increases, building momentum
until the height ofthe mania is reached in 1999, with a sharp decline after
the first quarter of 2000 as a result of the bursting of the Internet bubble.

3.2 Data description

Information related to the IPO deal characteristics and initial retums is

derived from the SDC New Issues Database. Daily and monthly stock
prices are obtained from the Center for Research in Security Prices (CRSP)
tapes. Data related to pre- and immediately post-IPO venture capital and
insider ownership levels, board representation, and significant share-
holder(s) are hand collected from issuing companies' prospectuses and S-1
Registration filings.
We obtain the number of media mentions from the Factiva database
provided by Dow Jones & Reuters. Specifically, we include media cita-
tions where the company is mentioned in the headline or lead paragraph of
the article for news media sources included in the "Major News & Busi-
ness- U.S." and "Top 50 US Newspapers" databases. We acknowledge
that this is a crude proxy because our measure excludes other non-print
media sources such as radio, television, and Internet media coverage.
However, we expect that our measure is merely a noisy but not biased
proxy for the underlying construct of interest (i.e., the total media coverage
related to each company) and the noise in this variable will simply reduce
the power of our statistical tests.

4 Descriptive statistics

Table 1 presents various descriptive statistics for our biotechnology and

Internet industry samples, respectively. As shown in the table, the average
proceeds raised by biotech IPOs was US$21 million, considerably less
than the average proceeds of over US$61 million raised by Internet IPOs.
While the initial retums for both samples are positive, initial retums to
Internet stocks of 82% dwarf those of the biotech sample that realized an
average 17% price change from the offer price to the first day close.
The frequency of VC backing is very similar across the samples, with
72% (73%) of Internet (biotech) firms being VC-backed. These per-
 529

centages are considerably higher than the reported frequencies of VC

backing for other non-technology stocks. 4 Consistent with the presence of
VC ownership, the levels of VC ownership are similar across the two
samples, with biotech firms once again having slightly greater pre-IPO
(post-IPO) VC ownership levels of 29% (22%) relative to almost 26%
(20%) for Internet companies. The percentage of corporate insiders (i.e.,
executives) on the boards is virtually identical across the two industries,
with 32% of board members being insiders, on average, for both samples.
There was at least one "strategic investor" in 44% (36%) of Internet
(biotech) companies. Strategie investors are defined as companies from
within the firm's own, or a closely related, industry. Pharmaceutical com-
panies are common strategic investors in biotechnology firms, for
example, while other network or computer technology firms or media
companies are typical investors in Internet firms.

5 Media activity and stock returns behavior

5.1 Media mentions around the month of IPO

Figures 5 and 6 present graphs of the average monthly media mentions for
biotechnology and Internet firms, respectively, in the calendar months
surrounding the firms' IPOs. The two industries exhibit strikingly similar
patterns with respect to media mentions over time, with the primary dif-
ference being that the average Internet firm engendered significantly
greater press coverage than the average biotech company. This is con-
sistent with the fact that many of the Internet firms included in our sample
are B2C companies, and thereby have greater mass appeal as a subject
matter to be covered in the popular press than the typical biotechnology
firm. The technical complexities and lack of consumer readiness of bio-
technology companies' products (or, perhaps more commonly, research in
process) would seem to make biotech companies less fascinating fodder
for a general public audience than their Internet industry counterparts.
As evidenced by Figures 5 and 6, the average firm in both industries
experiences a gradual increase in media attention in the months leading up
to the IPO. For the biotechnology sample, the media activity levels off
during the last month prior to IPO, whereas for the Internet sample the

4 For example, Demers and Lewellen (2002) report that only 19% of the non-
technology IPOs from January 1990 through February 2000 were VC-backed.
530

media activity actually declines during that period. This period of flat or
declining media coverage roughly corresponds to the pre-IPO "quiet
period" during which time firms are prevented from making significant
disclosures or announcements to the press. For both industries, there is a
sharp increase in media activity in the month of IPO and although the
average firm's media coverage declines significantly following the IPO
publicity-generating event month, the level of press coverage that is
sustained throughout the year after IPO is substantially higher than pre-
IPO levels of media activity.

5.2 The relation between media mentions and IPO stock

returns

Tables 2A and 2B present the partial correlations between total monthly

media mentions (ncite) and IPO initial returns for biotech and Internet
companies, respectively. Consistent with extensive prior finance literature,
we defme IPO initial returns as the difference between the closing price on
the first day of trading and the initial offer price, stated as a percentage of
the offer price. The media mention variables are calculated at the company
level of observation as the total number of media mentions within each
calendar month relative to the IPO month, with the exception of the
mediaO_2 variable, which is the sum of media mentions for a company for
the month ofiPO and the subsequent two months post-IPO.
Table 2A shows that there is a significant correlation between media
mentions in the month before IPO and biotech companies' initial returns,
however the correlations are not significant for either the month of IPO or
the month subsequent to IPO. The consumer-oriented nature ofthe Internet
industry, together with the rise of online and day trading in the late 1990s
when the majority of Internet IPOs took place, is expected to result in a
more significant role for media in the price formation process for Internet
stocks relative to biotech companies. Consistent with this expectation, and
in contrastto the findings for biotech, the evidence in Table 2B shows that
for Internet companies there is a strong positive correlation between the
levels of media coverage in each of the five months surrounding and
including the IPO month, and the IPO initial returns eamed by Internet
firms. The evidence presented in Table 2B is entirely consistent with the
presence of a "feedback loop" in which media hype initially leads to
higher IPO returns, and then these higher initial returns feed back into the
media system and result in increased press coverage in the months
subsequent to IPO. The evidence is also consistent with the media playing
 531

a more significant role in accentuating returns for the much more

consumer-oriented Internet industry.

5.3 The relation between media mentions and stock return

momentum

Tables 3A and 3B present the correlations between total monthly media

mentions and a simple measure of stock return "momentum'' for our
samples of biotechnology and Internet companies, respectively. We
calculate a proxy for the level of"momentum" in a firm's stock price using
the serial correlation coefficient between stock price changes over the 120
calendar day period subsequent to the IPO date.
Shiller (2000) suggests that the media can foster stronger feedback from
past price changes to further price changes. Table 3A shows there exists no
relation between biotechnology firms' stock return momentum and the
level of media coverage on these firms around IPO date. This finding is
consistent with the low correlations between media cites in the first months
after IPO and IPO day stock return reported in table 2A. The findings in
table 3A are in sharp contrast with these on internet firms.
Table 3B provides evidence that is strongly consistent with this feed-
back effect for Internet stocks. As shown in the table, there is a significant
positive correlation between Internet firms' stock return momentum and
the level of media coverage that the firms experience during the month of
their IPOs, and in each subsequent month. Thus, firms that receive higher
levels of media coverage have significantly higher serial correlation in
their stock returns. In other words, a positive (negative) daily stock return
is more lik.ely tobe followed by another positive (negative) daily return for
high publicity firms, whereas low publicity firms are more likely to
experience a low level of serial correlation in their daily returns (i.e., to
follow more of a random walk). In general, the evidence is consistent with
the media playing a role in accentuating stock price momentum of internet
firms, consistent with the feedback effect suggested by Shiller (2000).
However, the media does not play that role for biotechnology firms.
532

6 Preliminary multivariate analyses

6.1 Pairwise correlations between logit regression variables

Tables 4A and 4B present pairwise eorrelations between various variables

that will be used in the logit regression models diseussed in Seetion 6.2
below. A number of interesting pairwise relations emerge from these
tables, partieularly those related to the firms' ownership and governanee
charaeteristies. First, we observe that greater IPO proeeeds are raised both
by biotech and Internet firms that are venture baeked relative to those
without venture baeking, and the relation is inereasing in the Ievel of VC
ownership. We also find a negative eorrelation between the existenee of a
strategie investor and the Ievel of insider (i.e., exeeutive and direetor)
ownership Ievels for both samples. Strategie investors are also mueh less
likely to be present in venture eapital baeked Internet and bioteeh eompa-
nies, suggesting that VCs and strategie investors are substitutes in the
financing and monitoring of these high eompanies.
The variable ipovintage eaptures the time period ofthe eompany's IPO,
where firms that went publie earlier in the industry eycle have higher
values for ipovintage. Consistent with the general bubble phenomena, we
see that firms that went publie later in the lifeeyele of the industry were
able to raise signifieantly higher proeeeds at IPO and had significantly
greater initial returns and market eapitalizations than firms that were first
to go publie. These bubble traits are apparent for both industries. For
bioteeh companies, firms that went publie later in the lifeeycle tended to
have a higher pereentage of inside board members. This signifieant rela-
tionship does not exist for Internet eompanies, perhaps due to their pro-
pensity to have significant Ievels of inside board members from the ineep-
tion of the industry and to the very short life of the industry in general.

6.2 Logit survivorship regressions

Table 5 presents preliminary results from logit regressions that modelfirm

survivorship as a funetion of various ownership, governanee, and time-
related variables. The dependent variable in the model is a dummy variable
that takes the value of 0 if it has been delisted for reasons other than a
merger or acquisition, and 1 if the it was acquired or has otherwise sur-
vived through to the end of 2001. We model survivorship separately for
each of the bioteehnology and Internet industries. As shown in the table,
 533

ipovintage is significantly positively associated with survivorship for both

samples, suggesting that firms that went public early in the lifecycle of the
industry are significantly more likely to survive. This is consistent with
anecdotal evidence associated with the bubble phenomenon, wherein the
firms that IPO during the early stages of the lifecycle are presumed to have
met a higher hurdle in order to be taken public whereas later in the life-
cycle less promising companies are able to go public due to the bubble-
induced excess demand for stocks from these hot sectors. For biotech
companies, higher levels of initial returns at IPO are not indicative of
greater likelihood of survival whereas for Internet stocks greater under-
pricing at IPO is significantly negatively related to survival prospects.
Biotech companies with a greater proportion of insiders on the board at the
time of IPO are significantly more likely to survive, however this relation
does not hold for Internet firms. Similarly, for biotech companies higher
levels of venture capital ownership at the time of IPO increases the likeli-
hood of firm survival. No other ownership or governance variables are
significant for either sample. In particular, it is surprising that the presence
of a strategic industry investor at the time of IPO has no bearing on the
likelihood of survival for the Internet and biotech companies in our
samples.

7 Summary and discussion of future research

We live in an age of unprecedentedly rapid scientific advancement,

wherein "revolutionary" technologies are likely to continue to be intro-
duced over short intervals. History suggests that speculative bubbles are
associated with "new era" thinking, and that they are thereby accompanied
by the belief that some new price-enhancing innovation is present in the
world that somehow justifies a permanently higher level of valuation. This
paper undertakes a preliminary investigation of the two most recent and
economically significant technology stock bubbles in US history, with a
view to understanding this history so that the same mistakes will not be
repeated when the next revolutionary technological innovation is intro-
duced in a few years hence.
In this preliminary study we document significant parallels and differ-
ences between the two sectors that reveal interesting paths for further
investigation. We find that media is associated with stock return behavior
in a manner that is consistent with the "feedback" effect suggested by
Shiller (2000), and that the role of media is more significant for the more
mass media, consumer-oriented Internet industry than for the biotech
534

sector. Consistent with anecdotal accounts from the popular press (e.g.,
Perkinsand Perkins (1999)), we find that early entrants to both industries
are significantly more likely to survive when the bubble bursts, implying
that the hurdle to be met for going public seems to drop as the bubble gains
momentum and excess speculative demand develops for stocks in the hot
tech sector. Finally, we find that ownership and governance variables are
significantly associated with survival in the biotech sector but not in the
Internet industry, while high Ievels ofunderpricing in the Internet industry
are a foreboding of future failure.

References

Biotech lndustry Organization. (2002) Editors' and Reporters' Guide to

Biotechnology 2002-2003: 148. Washington, DC: Biotechnology Industry
Organization
Demers E, Lev B (2001) A Rude Awakening: Internet Shakeout in 2000. Review
of Accounting Studies, 6(2/3), June/September: 331-359
Demers E, Lewellen K (2002) The Marketing Role of IPOs: Evidence from
Internet Stocks. Journal ofFinancial Economics, forthcoming
Ernst & Young. (2000) The Economic Contributions of the Biotechnology
Industry to the U.S. Economy: Ernst & Young Economics Consulting and
Quantitative Analysis
Galbraith JK (1993) A Short History of Financial Euphoria. New York: Whittle
Books
Garher PM (2001) Famous First Bubbles. Cambridge, MA: MIT Press
Hand JRM (2001) The Role of Book Income, Web Traftic, and Supp1y and
Demand in the Pricing of U.S. Internet Stocks. European Finance Review,
5(3),: 295-317
Hand JRM (2002) The New Economy: Pricing ofSecurities. In: D. C. Jones (Ed.),
New Economy Handbook. San Diego: Academic Press
Kindleberger CP (2000) Manias, Panics, and Crashes: A History of Financial
Crises (4th ed.): Wiley
Lowry M, Schwert W (2002) IPO market cycles: Bubbles or sequential learning?
Journal ofFinance, 57(3), June: 1171-1200
Morgan Stanley Dean Witter (2002) The Technology IPO Yearbook: 8th Edition.
New York: Morgan Stanley Equity Research
Perkins AB, Perkins MC (1999) The Internet Bubble: Inside the Overvalued
World ofHigh Tech Stocks. New York: Harper
Shiller RJ (2000) Irrational Exuberance. Princeton, New Jersey: Princeton
University Press
Welch I (2001) The Top Achievements, Cha11enges, and Failures of Finance.
Unpublished working paper, Yale University November
 Biotech Index

)(
CD
'C
.E
ö
CD
;:,

~~ ~~ ~~ ~~ ~U ~M ~~ ~~ ~N ~$ ~H ~~ ~~ ~n ~N ~" ~~ ~V ~~ ~H ~~ ~~

Calendar Time

Fig.l.

Vl
w
Vl
536

....0
c:
....,
CO

0
0
c:
....,CO

(J)
(J)
c:
....,CO
Cl)
>< E

-
Q)
j::
"0
..5
CO
(J)
c:
...
ca
Q)
....,
CO 'C
c:
E .!
s
..5 ,..._
ca
0
(J)
c:
....,CO

<0
(J)
c:
....,CO

1.0
(J)
c:
....,
CO

-a;
c:
....,CO
xapu11aUJa}UI :JO an1eJ\
M
eil
~
 Time Series of Biotech IPOs

60

50

f/J 40
0
~
...
~ 30
GI
.c
E
:I
z 20

10

0
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989
1990 1991 1992 1993 1994 1995 1996 19971998 1999 2000 2001
Calendar Year of IPO

Fig. 3.
Ul
w
-..1
 Vl
w
00
Time Series of Internet IPOs

300

250

1/l 200
0
...!!::
~ 150
Cl)
.Q
E
:::1
z 100

50

0
1992 1993 1994 1995 1996 1997 1998 1999 2000
Year of IPO

Fig. 4.
 Biotechnology Sampie - Mean Monthly Media Mentions Around IPO Month

2,5
(II
c:
0
c:
..
GI
2
::IE
'6
"'
GI
::IE
0 1,5
.
GI
.&J
E
:I
z
c:
"'GI
::IE
0,5

0
-12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 2 3 4 5 6 7 8 9 10 11 12
Calendar Month Relative To IPO Month Vl
w
\0

Fig. 5.
 Internet Sampie • Mean Monthly Media Mentions Around IPO Month
V.
~
0
10

9
111
c 8
.2
'ECD 7
:E
ca
6
j
:E
5
0

4
-!...
E
:I
z 3
c
ca
CD
:E 2

0
-12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 2 3 4 5 6 7 8 9 10 11 12
Calendar Month Relative To IPO Month

Fig. 6.
 541

Appendix : Variable definitions

variable Definition
logproceeds Logarithm(IPO proceeds)
proceeds IPO Proceeds
Logarithm(stock market capitalization at end of 151
logmktcap
trading day)
iret Initial stock market return on IPO day
Equal to 1 if firm is Venture Capitalist backedl 0
VCdummy
otherwise
VCOWNTOT_PRE Percentage of stock owned by VCs pre-IPO period
(shortly before IPO)
VCOWNTOT_POST Percentage of stock owned by VCs post-IPO
period (shortly after IPO)
insidersprop Proportion of board members that are executives
Percentage of stock owned by all directors and
dir_exec_own_pre executives
dir_exec_own_post Percentage of stock owned by all directors and
executives
investor_ind_dummy Equal to 1 if there is a "strategic investor'11 0
otherwise
IN DOWNTOT_PRE Percentage of stock owned by strategic investor
pre-IPO period
IN DOWNTOT_POST Percentage of stock owned by post IPO-period
Number of months between IPO date and Dec 311
ipovintage
2001
Ncite 1 Ncite 2 Number of newscitations 1st monthl 2nd month
- I - resp. before IPO month
NciteO Ncite 1 N it 2 Number of newscitations duringl 1st month and
I c e 2nd month resp. after IPO month
I

Media0_2 NciteO + Ncite1 + Ncite2

Serial correlation of daily stock return in 120
Serialcorr calendar period after IP0 1 i.e. a proxy for stock
return momentum during 4 months after IPO
 V\
-~"-
N

Table 1. Descriptive Statistics for Biotech and Internet IPO Firms

BIOTECH INTERNET

variable mean sdev med mean sdev med

logproceeds 3.10 0.96 3.11 4.12 0.69 4.10
proceeds 22.2 2.6 22.3 61.5 2.00 60.3
logmktcap 18.56 1.13 18.48 6.19 1.37 6.15
iret 0.17 0.33 0.07 0.82 1.00 0.52
VCdummy 0.73 0.45 1.00 0.72 0.45 1.00
VCOWNTOT_PRE 29.45 27.31 26.50 25.92 26.67 21.40
VCOWNTOT_POST 21.81 20.55 18.06 20.27 19.46 17.40
insidersprop 0.32 0.18 0.29 0.32 0.15 0.29
dir_exec_own_pre 57.19 23.06 57.70 64.29 38.98 63.60
dir_exec_own_post 41.66 16.94 42.30 49.24 18.61 49.60
investor_ind_dummy 0.36 0.48 0.00 0.44 0.50 0.00
IN DOWNTOT_PRE 11.08 21.79 0.00 12.64 22.16 0.00
IN DOWNTOT_POST 7.48 14.38 0.00 10.70 19.24 0.00
ieovinta9e 105.86 68.51 105.00 36.99 14.23 32.00

Refer to the Appendixforadefinition ofthe variables.

Table 2A. Correlations Between Media Mentions and Initial Returns- Biotech Sampie
Variable Label ncite 2 ncite 1 nciteO Ncite1 ncite2 mediaO 2 iret

ncite_2 #newscites 2nd month before IPO month 1 -0.144 0.284 0.280 0.170 0.330 0.039
(0.007) (0.000) (0.000) (0.002) (0.000) (0.551)

ncite_1 #newscites 1 month before IPO month 1 0.249 0.258 0.299 0.337 0.131
(0.000) (0.000) (0.000) (0.000) (0.045)

nciteO #newscites during IPO month 1 0.471 0.280 0.888 0.066

(0.000) (0.000) (0.000) (0.315)

ncite1 #newscites 1 month after IPO month 1 0.324 0.748 0.092

(0.000) (0.000) (0.157)

ncite2 #newscites in 2nd month after IPO month 1 0.581 0.174

(0.000) (0.007)

media0_2 #newscites in 3 months [0,+2] 1 0.124

(0.057)

iret Initialreturn on IPO day

VI
.j::o.
w
 Vl
~
~
Table 2B. Correlations Between Media Mentions and Initial Returns - Internet Sampie
Variable Label ncite_2 ncite_1 nciteO ncite1 ncite2 media0_2 iret

ncite_2 #newscites 2nd month before IPO month 1.000 0.243 0.288 0.320 0.335 0.378 0.117
(0.000) (0.000) (0.000) (0.000) (0.000) (0.030)

ncite_1 #newscites 1 month before IPO month 1.000 0.355 0.214 0.182 0.318 0.113
(0.000) (0.000) (0.001) (0.000) (0.036)

nciteO #newscites during IPO month 1.000 0.498 0.408 0.860 0.423
(0.000) (0.000) (0.000) (0.000)

ncite1 #newscites 1 month after IPO month 1.000 0.429 0.770 0.208
(0.000) (0.000) (0.000)

ncite2 #newscites in 2nd month after IPO month 1.000 0.677 0.168
(0.000) (0.002)

media0_2 #newscites in 3 months [0,+2] 1.000 0.373

(0.000)

iret Initial return an IPO day 1.000

Table 3A. Correlations Between Media Mentions and Stock Price Momentum- Biotech Sampie
Variable Label ncite 2 ncite 1 nciteO ncite1 ncite2 mediaO 2 serialcorr

ncite_2 #newscites 2nd month before IPO month 1.000 -0.124 0.274 0.293 0.217 0.347 -0.021
(0.026) (0.000) (0.000) (0.000) (0.000) (0.713)

ncite_1 #newscites 1 month before IPO month 0.245 0.157 0.223 0.278 -0.020
(0.000) (0.005) (0.000) (0.000) (0.717)

nciteO #newscites during IPO month 1.000 0.452 0.293 0.856 0.023
(0.000) (0.000) (0.000) (0.677)

ncite1 #newscites 1 month after IPO month 1.000 0.293 0.754 0.046
(0.000) (0.000) (0.412)

ncite2 #newscites in 2nd month after IPO month 1.000 0.619 0.017
(0.000) (0.760)

media0_2 #newscites in 3 months [0, +2] 1.000 0.037

(0.505)

serialcorr Pearson serial corr 1st 4 months dail ret 1.000

Vl

Vl
"""
 Vl
Table 3B. Correlations Between Media Mentions and Stock Price Momentum -Internet Sampie
~

Variable Label nette 2 nclte 1 nciteO ncite1 ncite2 mediaO 2 serialcorr

ncite_2 #newscites 2nd month before IPO month 1 0.469 0.470 0.354 0.505 0.503 0.046
(0.000) (0.000) (0.000) (0.000) (0.000) (0.428)

ncite_1 #newscites 1 month before IPO month 1.000 0.602 0.324 0.500 0.559 0.087
(0.000) (0.000) (0.000) (0.000) (0.136)

nciteO #newscites during IPO month 1.000 0.645 0.707 0.924 0.097
(0.000) (0.000) (0.000) (0.096)

ncite1 #newscites 1 month after IPO month 1.000 0.686 0.847 0.088
(0.000) (0.000) (0.131)

ncite2 #newscites in 2nd month after IPO month 1.000 0.878 0.116
(0.000) (0.047)

media0_2 #newscites in 3 months [0,+2) 1.000 0.112

(0.055)

serlalcorr Pearson serial corr 1st 4 months daily ret

Table 4A. Biotech Survivorship- Regression Variable Correlations

Variable col1 col2 col3 col4 col5 col6 col7 col8 col9 col10 col11 col12 col13
logproceeds 1.00 0.851 0.254 -0.281 -0.062 0.034 0.297 0.174 0.182 0.085 0.069 0.115 -0.524
(0.000) (0.000) (0.000) (0.279) (0.558) (0.000) (0.003) (0.002) (0.136) (0.240) (0.055) (0.000)

logmktcap col2 1.000 0.439 -0.189 0.047 0.181 0.120 -0.034 -0.003 -0.007 -0.057 0.042 -0.486
(0.000) (0.006) (0.492) (0.009) (0.080) (0.630) (0.963) (0.915) (0.433) (0.564) (0.000)

iret col3 1.000 -0.120 -0.067 -0.046 0.018 -0.142 -0.136 -0.099 -0.004 -0.057 -0.207
(0.083) (0.338) (0.518) (0.800) (0.044) (0.057) (0.153) (0.951) (0.436) (0.003)

insidersprop col4 1.000 0.179 0.211 -0.317 -0.032 0.008 -0.095 -0.042 -0.036 0.214
(0.002) (0.000) (0.000) (0.585) (0.894) (0.098) (0.476) (0.553) (0.000)

direxecown pre col5 1.000 0.927 -0.252 0.077 0.073 -0.109 0.045 -0.030 0.182
(0.000) (0.000) (0.186) (0.214) (0.057) (0.450) (0.622) (0.001)

direxecown
post col6 1.000 -0.179 0.125 0.159 -0.117 -0.078 -0.023 0.127
(0.002) (0.034) (0.007) (0.044) (0.196) (0. 706) (0.029)

VCdummy col7 1.000 0.591 0.583 0.023 -0.211 -0.089 -0.184

(0.000) (0.000) (0.684) (0.000) (0.136) (0.001) Vl
.(:>.
-.J
 Vl
00
"""
vcowntot pre col8 1.000 0.984 -0.146 -0.276 -0.234 -0.127
(0.000) (0.012) (0.000) (0.000) (0.028)

vcowntot post col9 1.000 -0.149 -0.274 -0.228 -0.128

(0.011) (0.000) (0.000) (0.029)

Investor lnd col10 1.000 0.614 0.647 0.031

(0.000) (0.000) (0.584)

lndowntot pre col11 1.000 0.902 0.095

(0.000) (0.1 08)

indowntot post col12 1.000 0.072

(0.230)

ipovintage col13 1.000

Table 4B. Internet Survivorship- Regression Variable Correlations

Variable col1 col2 col3 col4 col5 col6 col7 col8 col9 col10 col11 col12 col13

logproceeds 1.00 0.747 0.345 -0.188 -0.037 0.072 0.234 0.162 0.190 0.209 0.115 0.150 -0.409

(0.000) (0.000) (0.000) (0.484) (0.173) (0.000) (0.002) (0.000) (0.000) (0.029) (0.004) (0.000)

logmktcap col2 1.000 0.633 -0.153 0.005 0.162 0.225 0.116 0.178 0.162 0.128 0.163 -0.299

(0.000) (0.004) (0.930) (0.003) (0.000) (0.031) (0.001) (0.002) (0.017) (0.002) (0.000)

iret col3 1.000 -0.014 -0.039 0.041 0.108 0.062 0.115 0.071 0.041 0.053 -0.222

(0.792) (0.466) (0.439) (0.038) (0.244) (0.029) (0.174) (0.434) (0.318) (0.000)

insidersprop col4 1.000 0.066 0.027 -0.317 -0.310 -0.327 -0.063 -0.045 -0.071 0.052

(0.21 0) (0.607) (0.000) (0.000) (0.000) (0.231) (0.393) (0.181) (0.326)

direxecown pre col5 1.000 0.550 -0.005 0.119 0.117 -0.132 -0.168 -0.168 0.014
(0.000) (0.928) (0.024) (0.027) (0.012) (0.001) (0.001) (0.790)

direxecown
post col6 1.000 0.081 0.159 0.179 -0.093 -0.187 -0.178 0.006
(0.126) (0.003) (0.001) (0.077) (0.000) (0.001) (0.91 0)

VCdummy col7 1.000 0.621 0.663 -0.046 -0.191 -0.177 -0.040

Vl
(0.000) (0.000) (0.375) (0.000) (0.001) (0.446)
\0
"""
 Vl
Vl
0

vcowntot pre col8 1.000 0.970 -0.177 -0.259 -0.254 -0.056

(0.000) (0.001) (0.000) (0.000) (0.288)

vcowntot post col9 1.000 -0.161 -0.267 -0.261 -0.074

(0.002) (0.000) (0.000) (0.162)

investor ind col10 1.000 0.631 0.616 -0.081

(0.000) (0.000) (0.119)

indowntot pre col11 1.000 0.979 -0.082

(0.000) (0.121)

indowntot post col12 1.000 -0.092

(0.081)

ipovintage col13 1.000

Table 5. Logistic Regression Results (Biotech and Internet Sampie)

Biotech Internet

Parameter Estimate Error Chi-S uare Pr> ChiS Estimate Error Chi-S uare Pr> ChiS

lntercept -4.0522 0.9533 18.0697 0.0001 -1.5823 0.576 7.5469 0.006

ipovintage 0.00739 0.0042 3.1005 0.0783 0.0247 0.00842 8.5849 0.0034
Ire! -0.626 1.22 0.263 0.608 -0.6078 0.1748 12.0845 0.0005
insidersprop 2.5156 1.1599 4.7038 0.0301 1.105 0.8267 1.7864 0.1814
dir_exec_own_pre 0.000719 0.0111 0.0042 0.9485 -0.00337 0.00457 0.5421 0.4616
INDOWNTOT_PRE 0.011 0.0122 0.8058 0.3694 0.00741 0.00566 1.7153 0.1903
VCOWNTOT PRE 0.0155 0.00759 4.1506 0.0416 0.00286 0.0048 0.3547 0.5515

Refer to the Appendixforadefinition ofthe variables.

Vl
Vl
Communicating lntangible Resources for the
Capital Market

Thomas W. Guenther

1 Relevance of reporting on intangibles

The importance of intangible assets like brands, customer relations,

knowledge or organisational capabilities is increasing in most westem
economies. Recent concepts like knowledge management or intellectual
capital management underline the increasing importance of these "soft"
production factors. The financial as weil as the managerial accounting is
still focusing on the ,,hard" production factors, especially the production
area with their typically physical and tangible assets and the finance and
investment area with the financial assets.
Concepts like the Skandia navigator (Edvinsson and Malone 1997), the
Intangible Assets Monitor (Sveiby 1997), the Intellectual Capital Navi-
gator (Stewart 1997), the Value Chain Scoreboard (Lev 2001) and the
Intellectual Capital Report (Austrian Research Center 2000 and Maul
2000) have been developed to find a structure for reporting on intangible
resources. Capital market research shows what indicators for intangible
resources have an impact on the capital market.
Some companies such as Skandia, Celemi International, WM-data AB,
KREAB, Jacobson & Widmark, Carl Bro als, Colaplast als or Deutsche
Bank AG started to deliver additional information added to the financial
reporting. The Austrian Research Center Seibersdorf has created a
"balance sheet for knowledge" that informs on the value of the knowledge
management activities of a research organisation. Maul suggests an intel-
lectual capital statement for universities (Maul 2000). Maul and
Menninger demand intellectual property statements as an appendix to the
financial report (Maul and Menninger 2000).
Standard setting bodies and different kind of organisations think about
expanding fmancial reporting to a more informative business reporting. In
1994 the Special Committee on Financial Reporting (often called the
Jenkins committee) submitted the Comprehensive Report demanding a re-
orientation of financial reporting on information needs of investors and
promoted a stronger future orientation and focus on non-financial items
(AICPA 1994). The Business Reporting Research Project of FASB is
 553

based on the results of the Jenkins committee and examines best practices
of voluntary disclosure of additional information like that demanded by the
Jenkins committee (FASB 2001). The FASB is currently working on a
new project "Disclosure about intangible assets". The Global Reporting
Initiative tends to develop a framework for reporting on sustainable
development integrating economic, social and environmental indicators
(e.g., GRI 2002). The Danish Agency for Trade and Industry developed a
guideline for the development of intellectual capital Statements (Danish
Agency for Trade and Industry 2000). Auditing companies like Price-
WaterhouseCoopers started initiatives on a more capital market oriented
reporting (e.g., Eccles et al. 2001). A broader reporting on intangibles is
one part of all these developments.
In Germany the workforce "Intangibles in Accounting" of the German
Schmalenbach Association started to develop concepts and approaches for
a reporting on intangibles (Arbeitskreis "Immaterielle Werte im Rech-
nungswesen" 2001). Nevertheless, reporting on intangibles so far is not a
top issue for financial and managerial accountants in Germany.

2 An understanding of intangible resources

Intangible resources are defined tobe the non-material and non-financial

resources a company can exploit for Ionger than the current reporting year
(distinguishing from current expenses or costs). "Intellectual property" are
those intellectual resources that are legally protected, like brand names,
patents or licences. Intangible resources become "intangible assets" if
they fu1fi1 the asset definition ofthe current standards (e.g., IASC Frame-
work § 49 and SFAC 6 §§ 25 and 26) and legislation (e.g. the regulations
within corporate law in Germany). From our point of view, "intellectual
capital" comprises all intangible resources of a company.
There are different approaches to classify intangible resources.
Edvinsson and Malone 1997 and Stewart 1997, classify in Human Capital,
Structural Capital and Customer Capital. Bontis 1998 uses Relational
Capital in a wider sense instead of Customer Capital and Sveiby 1997
classifies in intemal structure, extemal structure and people's competence.
The workforce "Intangibles in Accounting" of the Schmalenbach Associa-
tion suggests the seven categories human capital, innovation capital,
customer capital, supplier capital, investor capital, process capital and
location capital (Arbeitskreis "Immaterielle Werte im Rechnungswesen"
2001 ). As can be seen from the variety of classification, a dominating
framework does not exist so far.
554

For our empirical sturlies we found the classification in customer capi-

tal, human capital, innovation capital and structure or process capital help-
ful as the approach comprises all other classifications. The framework can
be associated with the four perspectives of a balanced scorecard. As can be
seen from Fig. 1 some categories of intangible resources overlap (e.g.,
technological know how, process know how, corporate culture) as they
cannot be allocated exclusively to one of the categories.

Customer Capital Human Capital

Brands Technological Know How

Customer Relations Education
Company Name / Image Process Know How
Structure of Sales & Experience
Distribution Innovations
Cooperation Adaptability
Franchise Partnerships Corporate Culture

Innovation Capital Process & Structural

Capital
Patents Information Systems
Copyrights Corporate Culture
Technological Know how Networks
Brands Locations
Proteeted Labels Investor Relations
Licences Process Know How

Fig. 1. Classification oflntangible Resources in the study

3 Empirical results on communicating intangibles

In recent years different kind of empirical research on reporting of intangi-

bles has been performed. Case studies (e.g., Johanson I Martensson I
Skoog 2001) focused on the analysis of innovative companies introducing
intellectual property statements or performance measurement systems with
focus on intangibles. Capital market research (e.g., Lev and Sougiannis
1996, Aboody and Lev 1998, Deng et al. 1999 and Lev and Sougiannis,
1999) reveals the information content and impact of isolated information
on intangibles (e.g., proved patents, listed brand names, R&D expenses
etc.) on the valuation of a company on the capital market.
 555

Our own research examines the structure and consistency of manage-

ment systems, either internal control systems or financial reporting systems
(disclosure). A study on brand management and brand valuation showed
that companies regard brands to be long-term investments, but their control
systems are short-term, profit oriented and inconsistent as a system
(Guenther and Kriegbaum-Kling 2001 and Kriegbaum 2001). Another
study on the performance measurement systems revealed that "traditional
measurement objects" like material and financial resources are widely
spread and intensively regarded within performance measurement systems.
However, for intangible resources a statistically significant difference
between (high) relevance and (low) intensity of consideration in the per-
formance measurement systems could be derived. Reasons for that are the
perception of a limited measurability, reliability and validity of intangibles
(Guenther and Gruening, 2002 and Gruening 2002).
This article is based on our recent study on voluntary disclosure of
information about intangibles of companies in the "new economy". The
objective is to examine the interaction between relevance of information
on intangibles, the structure and focus of the internal control system, the
disclosure of information on intangibles and the perceived sensitivity of
the capital market. We follow the point of view of a company to explore
objective or subjective hurdles for the disclosure of information on intan-
gibles.

3.1 Structure of population and sample

Our population consists ofal1343 companies ofthe five CDAX industries.

The structure ofthe population and the sample can be seen in Tab. 1. 24%
ofthe population responded to the investigation (response rate) and finally
54 questionnaires (return rate 16 %) could be used for the analysis. The
response rate and the returnrate are quite satisfying for this type of empiri-
cal research. A response bias due to size or industry could be statistically
significantly rejected.
Forthis article we select some results ofthe more comprehensive study
refering to the communication of intangible resources. For the statistical
tests and the resulting statistics please address to the original paper
(Guenther et al. 2003).
556

Table 1. Industry Structure ofthe Population and the Sampie

Industry Frequency Share of Frequency Share in Return Rate

(CDAX- in Popula- Population in Sampie Sampie within the
Index} tion Tndustry
Software 132 38% 16 30% 12%
Technology 92 27% 15 28% 16%
Pharma I 48 14% 9 17% 19%
Health
Media 47 14% 8 15% 17%
Telecommu- 24 7% 6 11% 25%
nication
Total 343 100% 54 100% 16%

3.2 Relevance of intangible resources for the company's

success

To meet extemal demands by the stak:eholders in the company's environ-

ment the company uses its own or acquired resources (resource based
view). The companies were asked what type of resources has what impact
on the company's business success. This approach is similar to the FASB
approach asking for critical success factors in their business reporting
research project. The resources were classified according to Tab. 2 using
the classification of intangibles shown in Fig. 1.
 557

Table 2. Structure ofResources ofthe Company in the study

Tan ible resources Machines, buildin s, inventories etc. "Classical"
~~~~~--~--7-~~~~--~~--~------~-r
Financial resources Conditions for collection of new capi- measure-
tal, e.g., rating, cost of equity etc. ment objects
Intangible resources:
Human Capital Knowledge and competence of work-
force, business climate etc.
Innovation Capital Product, service and process inno-
vation like patents, technology, pro- "Modem"
cesses etc. measure-
Customer Capital Brands, customer relations, image, Co- ment objects
operations etc.
Process I StructuraJ );> Direct value adding processes, e.g.,
Capital operations, procurement, logistics
);> Supporting processes, e.g., infra-
structure, information systems,
organisation

Using the above structure ofresources, we got the following results.

0,5
4,6

0.9
4.0

09
3.9
0.8
3,6

~ ~ ~ ~ ~ u ~ u ~ u ~
Slrength d lrlluence
(1 =ro lrtluerce; 5 =very slrcrg lrtluerceJ

Fig. 2. The relevance of intemal factors for the company' s success

Looking on the total sample, human capital is by far the most relevant
internal factor with a significant lower deviation in relation to the other
factors (Statistics: a < 0.01 for all compared factors). These results under-
line the overwhelming importance of employees for the company's per-
formance. In addition value adding processes as part of the
process/structural capital and innovation capital play a major role too
followed by financial resources and customer capital. Surprisingly the
material resources are not ranked high.
558

As financial and managerial accounting is traditionally focusing on

financial and material resources, some need for reconfiguration of the
intemal information and reporting system as weil as for the extemal
reporting can be seen.

3.3 Treatment of intangible resources within the internal

control system

In this chapter we analyse how the intemal success factors are reflected in
the intemal control system of the responding companies. To enable cam-
parisans we used the same structure (stakeholder analysis and resource
based view).
The dominance of traditional measurement systems like cost accounting
or financial accounting, that concentrate primarily on material and finan-
cial resources and directly value adding processes, can also be seen
looking on the measurement of intemal factors within the intemal control
system. Especially intangibles like human capital, innovation capital, cus-
tomer capital and supporting processes (structural capital) are dominated
by qualitative data. The differences to material and financial resources are
statistically significant. A relatively high percentage of companies does not
consider these factors at all. There are no significant influences by the type
of industry the respondents are belanging to.

Table 3. Measurement of intemal factors within the intemal control system

Information on Only Qualitative Only Not N
intemal factors quantitative and qualitative regar-
(resources) for measure- quantitative measure- ded
control purposes ment measure- ment
ment
Material Resources 75% 0% 9% 15% 53
Financial Resources 74% 8% 13% 6% 53
Human Capital 15% 6% 60% 19% 53
Innovation Capital 15% 2% 57% 26% 53
Customer Capital 17% 4% 53% 26% 53
Primary Processes 47% 6% 28% 19% 53
upporting 53
Processes 30% 6% 34% 30%
(modus value in bold characters)
 559

We conclude that companies see and have measurement problems for

intangible resources. Companies have problems to report on intangibles
for capital markets because they do not know how to measure intangibles
adequately and because they do not measure intangibles at all.

3.4 External reporting

In this chapter we will discuss the companies' attitudes towards a public

disclosure of information on intangibles in the financial reporting
(disclosure). As standards are given by nationallegislation (e.g., German
commercial law) or international standard setting bodies (e.g., SFAS
141,142 or lAS 38), we concentrate on the voluntary disclosure of
information in addition to legal requirements.

3.4.1 Relevance of general accepted accounting principles for

the voluntary external disclosure
Stating on the relevance of general accepted accounting principles of
financial reporting for the voluntary disclosure of information on
intangibles, a strong confmnation of all five accounting principles with
high means and low standard deviation can be concluded (Tab. 4). The
accounting principles had been derived from the German GAAP
Framework (e.g., Coenenberg 2000, pp. 59). This underlines that the
companies prefer to have the same standards for voluntary information as
for mandatory information. There seems to be less space for more
subjective information (e.g., using indicator models with indicators for
softer aspects like customer satisfaction, company image etc. ), partial
disclosure (e.g., focusing on the needs and requirements of every
company) or differently defined indicators (e.g., the different possibilities
to define innovation rate or percentage of new customers). We doubt
whether there might be decision useful information on intangibles if the
strict traditional accounting principles are just transferred to the voluntary
disclosure on intangible resources.
560

Table 4. Relevance of General Accepted Accounting Principles for Voluntary

Disdosure on Intangibles
Relevance of Accounting Principles Mean Standarddeviation
Trueness I Reliability 4,8 0,4
Fair presentation 4,6 0,5
Completeness 4,3 0,8
Consistency 4,3 0,7
Materiality 4,2 0,8
(Scale: 1: not relevant to 5: very relevant)

3.4.2 Disclosure of information on intangible resources

Looking on intemal factors, the resources a company uses, most
companies disclose information only at the corporate Ievel. Information on
intangible resources like human capital, innovation capital, customer
capital and process capital is not disclosed at all by a significant share of
the companies (Tab. 5). This conflicts with the stated relevance of these
intangibles for the company' s success (Fig. 2).
Consistent with the chosen research method, the content and the
intensity of the disclosure was not examined as the focus of the study is on
the structure of the information and its consistence with the relevance and
the intemal control system. Information on material resources was not
regarded in this question as financial reporting is traditionally concen-
trating on material resources.

Table 5. Level of disdosure of internal factors

Disdosure on intemal Only at At corporate Only at No

factor: corporate and segment segment disclosure
Ievel Ievel Ievel at all
Financial Resources 73% 13% 2% 12%
Human Capital 58% II% 4% 26%
Innovation Capital 64% 9% 4% 23%
Customer Capital 52% 12% 8% 29%
Primary Processes 56% 8% 8% 29%
SuEEortins; Processes 47% 8% 4% 41%
(modus value in bold characters)

Examining the scale of the data given on the corporate Ievel, we found
that quite understandable information on financial resources is delivered
quantitatively. However, reporting on intangibles is mostly qualitatively, if
information is given at all. These differences between financial resources
and intangible resources is statistically significant. More than 50 % of the
 561

companies report only in qualitative terms on human capital, innovation

capital and customer capital. For primary and supporting processes 19 %
and 12 % of the respondents give quantitative information, whereas 44 %
and 43 % give qualitative data (Tab. 6). This not only underlines that
voluntary reporting on intangib1es is poor, butthat the quality ofthe data is
primarily qualitative, which means nominal descriptions or some ordinary
data like "customer satisfaction has increased". lt is quite obvious that this
is not adequate for a further processing of the data and a thorough analysis
of this resources that were ranked with a high relevance by the companies.
An influence of the industry on the results neither holds statistically for the
level and the scale of disclosure.

Table 6. Scale of disclosure of intemal factors on corporate Ievel

Scale of disclosure on Quantita- Quantitative Only No
intemal factor: tive and qualitative disclosure
infor- qualitative information at segment
mation information Ievel
Financial Resources 60% 8% 19% 13%
Human Capital 11% 2% 57% 30%
Innovation Capital 19% 2% 53% 26%
Customer Capital 6% 2% 56% 37%
Primary Processes 19% 0% 44% 37%
Supporting Processes 12% 0% 43% 45%
(modus value in bold)

Statistical tests show that the hypothesis of independence of perceived

relevance and disclosure on resources can not be rejected. To put it
bluntly, relevant information especially on intangibles is not collected
within the intemal control systemandin additionnot disclosed adequately.
We again conclude to have an information gap for extemal reporting.
Postulating the management approach for the structure of extemal
reporting, an independence of the scale of disclosure within extemal
reporting from the measurement of the companies' resources in the
intemal control system can be rejected statistically significantly. Looking
on the level of disclosure (corporate or segment level) the relationship with
the intemal control system is significant for human capital, primary
processes and supporting processes. Material resources are not regarded as
the level and scale of extemal reporting is legally determined. For
reporting on resources and especially for reporting on intangibles the
extemal reporting seems to follow the data available for intemal control
purposes. Looking on the descriptive statistics this Ievel can be assessed to
562

be poor, resulting in a low Ievel of reporting within the internal control

system as weil as for external disclosure (data constraints).

3.4.3 Hurdles for the disc/osure of information on resources

To examine the hurdles for the limited structural disclosure of information
on resources, companies were asked for the major hurdles. Additional
information on financial resources was divided according to the structure
of the capital in costs of equity and the debt rating of the company. The
latter influences the cost of debt capital. Primary and supporting processes
were regarded together.
There seem to be no major hurdles for disclosure of additional
information on cost of equity or debt ratings. However, intangible
resources like human capital, innovation capital and customer capital have
to face severe hurdles due to missing measurability, harm on the
competitive position and objectivity. This conflicts with the high relevance
of these factors for the companies' success. A tremendous injormation
gap may result. For processes the hurdles are seen as weil but with lower
percentages. This might be due to the fact that processes have been the
target of process management tools like business process reengineering or
activity based costing etc. for more than a decade now. This resulted in a
fond of information on processes with is available within the internal
control system (see the results in Tab. 7).

Table 7. Hurdles for the expansion ofvoluntary disclosure on internal factors

Argument again t the
lnno-
expan ion of disclosure Cost of Debt Human f Customer Proces-
va wn .
on the specific internal equity rating capital .tal cap1tal ses
cap1
factor

Missing Relevance
0
(7%)
0
(9%)
0
(9%) (9%)
0
(II%)
• (18%)
•
Missing Measurability
0
(9%) (17%)
• •••• ••• (38%)
••• •••
(45%) (38%) (30%)

Might harm competitive

eosition
0
{9%) {17%)
• •• ••• •{53%}
••• ••
{23%) {30%) {26%)

Problems with
Objectivi~
0
{9%)
0
{9%)
••• •••
•••• ••• {30%}
{52%) {33%} {37%}
No adequate processing
b~ users of information
0
~2%~
0
~2%~
•• • (11 %)• ••
~20%~ ~11%~ ~23%~
 563

Legend:

Percentage of respondents Symbol

0 up to less than 10 percent 0
10 up to less than 20 percent
•
20 up to less than 30 percent ••
30 up to less than 40 percent
•••
40 Q_ercent and more
••••
3.5 Information processing on the capital markets

The last step in the flow of information is the use of information by

addressees. One of the major addressees are the current or potential
investors, which represent the capital market. The question is whether or
not voluntary disclosure of information on intangibles can support the
information processing of the extemal capital market.
With regard to the sensitivity of information on intemal factors
(resources), we got the following pattem (Tab. 8):

Table 8. Sensitivity ofthe capital market on information about extemal factors

The reaction ofthe capital market on information on

the specific intemal factor is ...

lntemal factor Reaction can To low Adequate To strong

not be seen

Material Resources 31% 10% 55% 5%

Financial Resources 6% 19% 62% 13%

Human Capital 33% 35% 30% 2%

Innovation Capital 27% 39% 27% 7%

Customer Capital 33% 29% 33% 4%

Primary Processes 43% 24% 33% 0%

Supporting Processes 50% 20% 30% 0%

(modus value in bold characters)

For material and financial resources, the sensitivity of the capital market
is regarded tobe quite adequate with high percentages of 55% for material
564

resources and 62 % for financial resources. N evertheless, for all

intangibles resources a high percentage of the companies assess a not
existing or to low reaction on information on human capital (68% = 33%
+ 35 %), innovation capital (66% = 27% + 39 %), customer capital (62%
= 33% + 29 %), primary processes (67% = 43% + 24 %) and supporting
processes (70% = 50% + 20 %). The differences between material I
financial resources and intangible resources again are statistically
significant. An industry bias is only significant for the customer capital.
We again tested the interaction between the perceived sensitivity of the
capital markets and the perceived relevance ofinformation on a company's
resources. The independence can be rejected for "financial resources" and
"primary processes". For these two factors a correlation between relevance
and sensitivity of capital market reaction can be deducted. Nevertheless,
for all other resources, and especially for intangible resources, a rela-
tionship between perceived relevance and perceived sensitivity on capital
markets can not be derived. We conclude, that the capital market seems to
process other information as these seen to be relevant by the companies
(information processing gap). Furthertests showed that the level and the
scale of disclosure on intangibles resources is not related with the
sensitivity of the capital markets on that items.

4 Questions to answer

lnterpreting our results we can derive two groups of questions which have
to be solved to promote any reporting on intangibles.

4.1 Questions concerning measurement within the internal

control system

• What are intangible resources ?

Whereas the recognition of intangible resources as "intangible assets" is
defined by nationallegislation and standards (e.g., lAS 38 or SFAS 50, 53,
86, 141, 142 etc.) and guided by GAAP, the understanding of intangible
resources that are not recognized in financial statements is not so well
defined. Is the brand value of BMW due to the label BMW, the excellent
product and production know how, the good customer relationship or the
quality of the product ? A separation of intangible resources like assets and
a separate measurement and valuation seems not to be realistic. If we leave
clear definitions of intangibles to recognized intangible assets, we can still
report on intangibles on an indicator level, even if we will not be able to
 565

identify single intangible resources (in the sense of assets) that caused that
effects.
• How to classify intangibles ?
If we have problems to identify intangible resources, are we able to
classify intangibles in groups ? Different approaches of classification have
been developed by researchers which overlap. Some are more
comprehensive than others, some are overlapping. For interpreting
information on intangibles, it may be helpful to have a standard
classification scheme.
• How should intangible resources be measured ?
There are numerous different approaches of measuring and valuing
intangible resources. They can be classified in DCF approaches, multiple
methods, capital market derived methods, real option approaches, excess
return approaches, the cost approach and indicator approaches (Guenther
et. al. 2002). As can be seen from the perceptions ofthe respondents in our
study valuation (= monetary measurement) will be able for some selected
intangible resources. For brands sophisticated valuation approaches based
on DCF and multiple methods have been developed (e.g., PwC/GfK,
Interbrand or Nielsen). Tobe precisely and adequate, these methods have
to be complex, sophisticated, long-term and data-driven. For financial
accounting purposes the cost approach is requested added by the DCF
approach for the impairment test. Due to problems in identifying and
classifying intangibles indicator models seem to be more efficient and
comprehensive.

4.2 Questions concerning disclosure of information on

intangibles

• Should we change financial statements or use add-ons ?

Only if intangible resources fulfil the asset definition, they can be
integrated in the financial statements. For Germany, the workforce
"intangibles in accounting" of the Schmalenbach Association, demanded
the cancellation of the § 248 II HGB, restricting the recognition of intan-
gible assets to purchases only (Arbeitskreis "Immaterielle Werte im Rech-
nungswesen" 2001 ). A tremendous change of financial accounting stan-
dards to integrate intangible resources that are not assets, seems not to be
on the way as the current standards developed now over centuries. For all
other intangible resources a reporting as part of the management report or
as an appendix to the financial reports (like Skandia and others) seems to
be feasible.
566

• Do we need standards for reporting on intangible resources ?

A phase of voluntary disclosure on intangible resources will show what
kind of information is measurable, relevant, complete, valid, auditable,
applicable and efficient. If information on intangibles is decision useful in
the long-term it may be helpful in a second phase after gaining some
experience with this different kind of information to have frameworks and
standards for the disclosure on intangibles. This might deliver a uniform
format for analysis by addressees and standard information delivery by the
companies.
• Can we transfer GAAP to voluntary reporting on intangibles ?
According to our study, companies want to simply transfer the GAAP
they are used to to the reporting on intangibles. Due to the missing
experience with such kind of data and due to the different scale of
information, the perception of this kind of data has to change. From my
point of view, information on intangibles will never be as reliable,
complete, consistent and objective as information on material and financial
resources. If we are honest, also in the traditional areas reliability, fair
presentation, completeness, consistency and materiality leave room for
further development.
• Do we need a general format for voluntary disclosure ?
In accounting standards and national legislation balance sheet and
income statement have to deliver information within a given format. F or
intangibles different kind of formats, especially for indicator models, are
discussed. Refering to the two phases of further development of disclosure,
if may be helpful not to have a fixed format for disclosure on intangibles in
the first phase to gain experience what's the most information efficient
one. F or the second phase a proven format that is generally accepted might
be helpful to support comparisons and decision making.

5 Frameworks for communication on intangibles

In the last decade different kind of frameworks for the disclosure on

intangibles have been developed and are currently discussed. Most of the
frameworks are based on indicator models using different kind of
measures combined with an inner logic to disclose information on a
company's intangible resources. The most prominent examples for such
frameworks are e.g. the Skandia Navigator (Edvinsson and Malone 1997),
the Liahona- H.O.M.ES Model (Bomemann et al. 1999), the Intangible
Assets Monitor (Sveiby, 1997), the Intellectual Capital Navigator (Stewart,
1997), the Value Chain Scoreboard (Lev 2001 ), the Reporting Framework
 567

of the workforce "Intangibles in Accounting" of the Schmalenbach

Association (Arbeitskreis "Immaterielle Werte im Rechnungswesen"
2003) and the Intellectual Capital Report (Austrian Research Center 2000
and Maul, 2000).
In this article the Value Chain Scoreboard (VCS) and the Framework of
the workforce "Intangibles in Accounting" will be further presented as
formats for the voluntary disclosure of information on intangibles.

5.1 The value chain scoreboard (VCS)

The Value Chain Scoreboard (VCS) developed by LEV takes an approach

similar to the balanced scorecard by incorporating non-financial and
financial measures of different perspectives (Lev 2001 ). Measurements are
classified into nine different groups (Fig. 3). Each ofthe nine measurement
groups belong to one of the corporate value chain's three stages: dis-
covery, implementation or commercialisation.
The VCS provides a broad selection of measurements for different types
of intangibles which can be easily adapted through the selection of a subset
to the company's specific requirements (management approach).
Within the first phase ofthe value chain "discovering and learning" Lev
distinguishes between internal renewal, acquired capabilities and
networking as major sources of new ideas for products, services, or
processes. The second stage, "implementation ", considers measurements
referring to the transformation of new ideas into working products,
services or processes. Readiness for commercialisation reduces an
intangible's risk and allows for better estimation ofthe future development
of a "new idea". The third phase, "commercialisation ", presents past and
potential financial results for the new ideas' outcome.
568

Dilicovery aod learoiog I mplemeotatioo Commerdalization

l . l.ntern al rt:ne"' ' 4. lnlt:ll~rual propeny 7. CutomH'!

• R.....-ch and developm<nt • Patems. tradem811cs. and · Marketing alliana:s

. WOO. fon:e tnt.inins Utd COpyrights · Bnmd wlues
dcvelopment • Litcnsing agrccments · Customer chum and va lue
• Organilllliooal apiml. • Coded kno~<'-how • Online Saleo
proccsses
8. Performance
5. Tccbnological feasibill ty
2. A<qul.-..1 C.pabilitleo - Re~ues., eam ings. and
• Clinical tcsts, Food and Drug m.vltct share
• Tcdmology purchase - lnnovaa.ion revenu
Administrat ion approvals
- piiiO\U utiliZDtion · Patent and know-how royulties
• Beta tesu, "'t'king pilots
- Capital expa1ditures • Kno\\1odge eami ngs and asscts
. Fi.rSl mover

9. Growtb P""p«U
3. Nd,. or1d ng 6. 1nrr.md
• Product pipeline and launch
- R&D allian<:cs and joint • Thrcshold traff.e dales
ventures • Online pu"'hase • Expected efl't<iendes and savings
• upplict and Customer · Major lntemet alliances - Planned initiatives
integration • Expected breakevm and
- Communities ofpracaice bum rate

Fig. 3. The Value Chain Scoreboard by Lev, 2001

The VCS is helpful for both intemal decision making and

communicating with investors. To assuremaximal usefulness Lev suggests
that the specific indicators should be quantitative, standardised and
confirmed by empirical evidence as relevant to users.
Quantitative indicators can be completed by qualitative indicators in an
annex to the scoreboard. Standardisation of the figures enables
camparisans across companies. Empirical sturlies - executed on a large
scale by Lev and his colleagues - assure that the chosen indicators are
linked to a company's value and that these linkages are scientifically
robust.
The VCS was developed to enhance the information content of financial
reports by looking at all intangibles, rather than to serve as a specific tool
for measurement or valuation of a chosen intangible. lndicator approaches
assess the economic benefits from an intangible in a comprehensive
manner. Lev for example, focuses only on indicators with "confirmation
through empirical studies". Consequently, the indicators are limited to
those constructed from accessible data. Moreover, the Scoreboard shows
measures at different stages of the value chain. This could allow the
Observation of an intangible's development in addition to cause and effect
relationships.
 569

5.2 Suggestion of the workforce "intangibles in accounting"

of the Schmalenbach association

The workforce "Intangibles in Accounting" of the German Schmalenbach

Association is currently working on a framework for reporting on
intangibles (Arbeitskreis "Immaterielle Werte im Rechnungswesen" 2003).
The workforce consists of representatives of companies and of university
professors working in financial and managerial accounting. The
framework will be independent from the recognition of intangibles as
"intangible assets" according to international standards or national
legislation. It's supposed tobe an add-on to the financial reports presented
as part of the management report. The primary objective of the additional
voluntary disclosure on intangibles is to promote a reporting on the general
strategy of the management of intangibles and the identification of relevant
value drivers. The framework is supposed to be consistent with German
Accounting Standard DRS 12.
The workforce "Intangibles in Accounting" uses a seven category
classification of intangibles (innovation, human, customer, supplier,
investor, process and location capital). For every category a minimum
catalog of indicators is suggested. Following the management approach the
company can individually decide on the selection and structure of the
reported indicators. Generally accepted accounting principles like
consistency, trueness or reliability have to be used for the voluntary
reporting. A further detailed reporting for the segments of the companies
according to segment reporting is recommended.
The following presentation of the framework is restricted to the
structure and the catalog of indicators for the categories innovation capital
and human capital (# 3 and # 4 in Tab. 9).

Table 9. Suggestion for a voluntary reporting on intangibles by the Schmalenbach

Association (AK Immaterielle Werte im Rechnungswesen, 2003)
Category 1: Innovation Capital

lndicator Explanation I Differentation

R&D Expenses Indication of R&D Expenses, R&D Expenses in

relation to sales, information on deviation and
concentration of R&D Expenses
Portfolio of patents and umher, structure and (residual) useful life of
similar intellectual pro- property rights and patents
perty rights Presentation similar to the fixed assets (initial value,
increases, decreases, changes, final value), only units
no valuation intended.
570

Patents and similar in- Number and structure of intellectual property rights
tellectual property rights and patents filed for application
filed for application

Pending suits with patents Number and importance of current pending suits
and intellectual property with patents and intellectuaJ property rights
rights

Innovation rate Sales of products introduced within the last three

;":ears in relation to total sales

Category 2: Human Capita/

Indicator Explanation I Differentation

Demography of Classification in age groups (in years):

employees [<25] [25-39] [40-54] [>54]
Affiliation with the Classification of length of affiliation with the
company company (in years)
[<5] [5-15] [>15]
Fluctuation Nurnber of employees leaving the company within
the reporting year in relation to total number of
employees
Qualification of Breakdown as a percentage of total number of em-
employees ployees:
• Vocational Training inhouse I with other
companies
• University Degrees
Training Expenses per employee
Number oftraining days (per employee)
Employee satisfaction Explanation of chosen method

Absence tim Nurnber of days per employee

Value Added (Value Added per person - Personnet costs per

person) x umber of employees

The number of employees is already part of the notes (according to

German GAAP); in addition, the number of employees per segment should
be part of segment reporting.
 571

5.3 Comparison of the two frameworks

The Schmalenbach Framework is based on the conceptual analysis of

frequently discussed indicators for intangible resources and tries to deliver
a first suggestion for a reporting framework. It's not driven by empirical
capital market research like the value chain scoreboard of Lev as public
information on some indicator is limited.
Comparing the two approaches similarities of categories and indicators
(e.g., for patents and brands) but also differences can be observed (stronger
focus on networks and intemet in the value chain scoreboard). The two
examples also relate to the questions raised in the previous chapter. A
clear, uniform format or standard for disclosure of information on
intangibles is not insight. There is some time need to gather experience
with different kind of frameworks and indicators. After that phase
additional research is needed to condense and harmonise the reporting
frameworks. '

6 Conclusions

The structural changes in westem economies result in an increasing

importance of intangibles for the short term and long term success of
companies. The communication on intangible resources is one of the big
challenges for modern management in the "new economy". For the state of
the art and for the further development of reporting on intangibles the
following conclusions can be drawn:
• Our empirical research shows that human capital is connected with high
relevance for the success of the company followed by process capital,
innovation capital, financial resources and customer capital.
Surprisingly, material resources are perceived to have only medium
relevance for the company's success.
• Looking on the intemal control systems ofthe companies in our study, it
is no surprise that material and financial resources are measured
quantitatively. For intangible resources qualitative approaches are
dominating. Only for primary processes like production or operations
quantitative data is used. We confirm an information gap on the
resource side which is caused by measurement problems for most of the
intangibles.
• For an voluntary extemal disclosure of information on intangibles,
companies tend to transfer the same general accepted accounting
principles used for fmancial reporting. We doubt if this will be possible,
572

as a complete and fair presentation of all intangible resources might not

be realisable, especially considering materiality. This perception of the
companies might be an hurdle for a broader voluntary reporting on
intangibles as the intended standards for disclosure might not be
realistic.
• Information on resources is primarily disclosed on the corporate level.
Information for financial resources is dominated by quantitative data
whereas information on intangibles is qualitative. We confirm, that here
relevant data is not disclosed (information gap) and that extemal
reporting seems to be restricted by the poor level of data available for
intemal control purposes.
• Hurdles for the disclosure of information on resources are the missing
measurability, a possible negative impact on the competitive position
and limited objectivity.
• The reaction of the capital market on additionally disclosed information
is generally to low or not existing. This holds especially for all
information on intangible resources. A relationship between relevance
of information and sensitivity can not be derived. Also the level and
scale of disclosed information is not correlated with the sensitivity. Only
for the intangibles we get high correlations with the extemal reporting.
But here, this means that a low level of voluntary disclosure results in a
low or not existing sensitivity (information processing gap).
• The reporting on intangibles is still in development. Many questions
remain difficult to answer and we doubt whether an adequate solution
will be possible at the moment. It might be helpful to gather experience
in a first observation phase to condense and harmonise reporting later on
in a second phase.
• Several frameworks for reporting on intangibles exists which are all
based on indicator models. A dominating model is not insight at the
moment as experience is missing.
To promote communication on intangibles the existing hurdles for a
voluntary disclosures have to be realised and reduced. The poor level of
voluntary disclosure is primarily caused by the poor level of intemal
control systems on intangibles and by perceived inconsistencies in the
reaction of the capital markets. There is much space left for additional
research on feasible indicators for intangible resources.
 573

References

Aboody D, Lev B (1998) The Value-Relevance of Intangibles: The Case of

Software Capitalization. J Accounting Research, Supplement 1998, pp 161-
191
AICPA (1994) (ed) Improving Business Reporting- a Customer Focus, Meeting
Information Needs oflnvestors and Creditors, New York
Arbeitskreis "Immaterielle Werte im Rechnungswesen" der Schmalenbach-
Gesellschaft für Betriebswirtschaft e.V. (2001) Kategorisierung und
bitanzieHe Erfassung immaterieller Werte (Classification and Recognition of
intangible resources). Der Betrieb, 54:989-995
Arbeitskreis "Immaterielle Werte im Rechnungswesen" der Schmalenbach-
Gesellschaft für Betriebswirtschaft e.V. (2003) Freiwillige externe
Berichterstattung über immaterielle Werte (V oluntary Disclosure on
intangible ressources). Working paper, Cologne
Austrian Research Center (2000) Intellectual Capital Report 1999, research report
ofthe Austrian Research Center. Seibersdorf
Bontis N (1998) Intellectual capital: An Exploratory Study that Develops
Measures and Models. Management Decisions, 35:63-76
Bornemann M, Knapp A, Schneider U, Sixl KI (1999) Holistic Measurement of
Intellectual Capital, International Symposium, Proceedings Measuring and
Reporting Intellectual Capital: Experience, Issues and Prospects.
http://www .oecd.org/Idsti!sti/industryI indcomp/act /ams-conf/symposium-
old.htm#holistic. Amsterdam
Coenenberg AG (2000) Jahresabschluss und Jahresabschlussanalyse (Financial
Reporting and Financial Statements Analysis). 17th edn, Modeme Industrie,
Landsberg/Lech
Danish Agency for Trade and Industry (2000) A Guideline for Intellectual Capital
Statements. Copenhagen
Deng Z, Lev B, Narin F (1999) Science & Technology as Predictors of Stock
Performance. Financial Analysts Journal, 1999: 20-32
Eccles RG, Herz RH, Keegan EM, Philipps DMH (2001) The Value Reporting™
Revolution- Moving Beyond the Eamings Game. WileyNCH: New York
Edvinsson L, Malone MS (1997) Intellectual Capital: Realizing Your Company's
True Value By Finding Its Hidden Brainpower, HarperCollins Publishers,
NewYork
Financial Accounting Standards Board (FASB) (ed) (2001) Business Reporting
Research Project
Global Reporting Initiative (GRI) (ed) (2002) Sustainability Reporting Guidelines,
Boston
Gruening M (2002) Performance Measurement Systeme als Management-
werkzeug (Performance Measurement Systems as a Management Tool),
Deutscher Universitaetsverlag, Wiesbaden
Guenther T, Kriegbaum-Kling C (2001) Brand Valuation and Control: An
Empirical Study. Schmalenbach Business Review, 53:263-294
574

Guenther T, Beyer D, Menninger J (2003) Hurdles for the Voluntary Disdosure of

Information on Intangibles - Empirical Results for "New Economy"
Industries. Working paper, Dresden University ofTechnology
Guenther T, Gruening M (2002) Performance Measurement - The German
Experience. FSR Forum, special edition on Performance Measurement,
October 2002, University ofRotterdam
Guenther T, Kirchner-Khairy S, Zurwehme A (2001) Methods for the Valuation
and Measurement of Intangible Assets. Dresdner Beiträge zur Betriebswirt-
schaftslehre, Nr. 51/2001, 1:24
Johanson U, Martensson M, Skoog M (2001) Measuring to understand intangible
performance drivers. European Accounting Review, 10:407-437
Kriegbaum C (200 1) Markencontrolling (Brand controlling). Vahlen, Munich
Lev B (2001) Intangibles: Management, Measurement, and Reporting. Brooklyn
Institution Press, Washington D.C.
Lev B, Sougiannis T (1996) The Capitalization, Amortization and Value-
Relevance ofR&D. J Accountingand Economics, 107-138
Lev B, Sougiannis T (1999) Penetrating the Book-to-Market Black Box: The
R&D Effect. J Business, Finance and Accounting, 419-449
Maul KH (2000) 'Die Wissensbilanz als Teil des handelsrechtliehen Jahres-
abschlusses - Wissensbilanzen dargestellt am Beispiel des Jahresabschlusses
von Hochschulen (The knowledge balance sheet as part of the financial
Statements - knowledge balance sheets presented as an example for the
financial statements ofuniversities)'. Deutsches Steuerrecht, 47: 2009-2016
Maul KH, Menninger J (2000) Das "Intellectual Property Statement" - eine
notwendige Ergänzung des Jahresabschlusses? (The "Intellectual Property
Statement" - a necessary appendix to the financial reports ). Der Betrieb,
53:529-533
Stewart TA (1997) Intelleemal Capital: The New Wealth of Organizations.
Nicolas Brealey Publishing, London
Sveiby KE (1997) The New Organizational Wealth: Managing and Measuring
Knowledge-Based Assets. Berrett-Koehler Publisher, San Francisco
Business Valuation in the New Economy - Back
to the Basics

Dirk Hachmeister

1 lntroduction

Cash is king. This quote from Copeland, Koller and Murrin also applies to
the valuation of young enterprises with high growth opportunities. From
this perspective there is no difference as far valuation issues are concemed
in comparison to the old economy. From this perspective I will look at
special problems in valuing enterprises in the new economy. Three years
ago my perspective would have been old-fashioned, but today it seems that
the market has now remernbered this fundamental business rule.
Although this quote - cash is king - belongs to the generally accepted
principles of valuation there are a lot of characteristics of young enter-
prises with growth opportunities which make valuing these firms a chal-
lenge. In this presentation I am going to look at these characteristics and
their consequences for business valuation in the new economy.

2 Characteristics

Generally enterprises in the new economy can be described using the fol-
lowing characteristics. 1
a) On the one band they have diverse growth opportunities, on the other
band they need a lot of capital to realize these opportunities. The sources
of this capital will change during the life cycle of the enterprise. In the
early stages these companies need a lot of risk capital from the owners or
shareholders; in the maturity stage the projects will be financed by credi-
tors to a large extend. 2 Therefore there is only a small leverage at the be-
ginning, later on the leverage will be much higher.
b) A lot of capital is invested in intangible assets such as research and
development (R&D) or brand names. Because recognising these expendi-

1 Schwall (2001), p. 28-75.

2 We have a reversed LBO situation with a low level of debt at the beginning and
an increase of debt over time.
576

tures on the balance sheet is not permitted, these companies will not only
have negative free cash flow but also a negative taxable income. It may be
complicated to integrate a loss carry-forward into the discounted cash flow
valuation formula.
c) Drawing up a business plan in innovative and dynamic industries is
highly complicated. Firstly, it is unclear which standardwill be established
and which enterprises will survive. The cash flow profile will be very
asymmetric because there is a chance that the company will be the "Micro-
soft" ofthe industry, however it is more likely that it will be the "Commo-
dore" or "Atari". Secondly, indicators from the past are not available for
predicting future cash flow. Making future cash flow predictions seems
like gambling.
d) The management and a Iot of the employees will have a large, maybe
contingent, stake in the company. They not only receive salary, they also
usually have a Iot of stock options. On buying a company with outstanding
stock options, these stock options should be integrated into the valuation
formula. Otherwise the dilution effect from options will not be taken into
account. 3
I will now take a Iook at how to integrate the first three characteristics
into the solution structure of the discounted cash flow and the challenges
arising out of these characteristics.

3 Discounted cash flow

Using the discounted cash flow technique you can choose between three
different valuation formulas- the weighted average cost of capital
(WACC), flow to equity (FTE) and adjusted present value (APV):
Using the weighted average cost of capital formula (WACC) the free
cash flow is discounted with the weighted average cost of capital of the
shareholders and the creditors. The free cash flow is defined as the cash
flow without taking financing activities into account; an alternative defini-
tion is the cash flow of a wholly equity fmanced company. When cal-
culating the weighted average cost of capital the cost of capital for the
shareholders and the creditors will be needed as weil as the target capital
structure of the company. The cost of equity capital will be influenced by
operating and financing risks. In order to estimate the value for the share-
holder we must deduct the market value of the liabilities from the enter-
prise value. The WACC formula is also known as the textbook formula.

3 This problern will not be investigated in this paper. Several authors have ad-
dressed this issue Damodaran (200 1), Bachmeister (200 1) or Soffer (2000).
 577

Ifwe use the flow to equity, we will discount the cash flow to the share-
holders with the cost of capital for the shareholders. The cash flow to the
shareholders is the free cash flow without the cash flow to the debt holder.
Due to the fact that the cost of capital for the shareholders is influenced by
operating and financing risks, we also need to know the planned capital
structure. The flow to equity formula is the most popular valuation formula
in Germany (Ertragswertmethode) although we do not explicitly differen-
tiate between operating and financing risks.
The adjusted present value formula is similar to the WACC formula.
Just like WACC the adjusted present value formula discounts the free cash
flow and calculates an enterprise value. In order to reach the shareholder
value the market value of the liabilities must be deducted from the enter-
prise value. The adjusted present value formula divides the enterprise
value of the firm into two components. The first component - the operat-
ing value - reflects the value of a wholly equity financed company, be-
cause the free cash flow is discounted with the cost of capital which only
reflects the operating risk. The tax benefits from debt financing are calcu-
lated as a separate component. Although text books recommend this for-
mula for special valuation scenarios4 - such as leveraged buy-outs- it is
seldom used in practical valuation situations.
All three formulas are based on the same idea and under ideal conditions
they would give the same results, although it will be difficult to achieve
the same result in practical valuation situations. However, these formulas
are not different valuation methods, they only differ on how to implement
debt and tax effects into the valuation formula.

4 Challenges

4.1 Debt and tax effect

As our example shows, the company's free cash flow is negative due to the
high research and development (R&D) expenditure or the establishment of
a brand name. Due to the fact that these expenditures must be expensed
there is also a negative earnings before interest and taxes (EBIT) and (to
make it less complex) a tax reimbursement at the end of the period. We do
not have to factor a loss carry-forward into the formulas. The negative free

4 Brealey/Myers (2000), pp. 555-564, Ross/Westerfield/Jaffe (2002), pp. 489-

493.
578

cash flow shows the capital requirement, which is to be financed by share-

holders or creditors.

Table 1. Free cash flow trend

2003 2004 2005 2006 2007

Sales 21,000.0 28,100.0 35,410.0 44,410.0 44,410.0
Cost of Sales - 19,025,0 -25,045.0 -31,120.0 - 37,120.0 - 37,120.0
R&D -9,000.0 - 7,000.0 - 6,000.0 -3,000.0 - 1,000.0
EBIT - 7,025.0 -3,945.0 - 1,710.0 4,290.0 6,290.0
Tax (40%) 2,810.0 1,578.0 684.0 - 1,7 16.0 -2,516.0
OPAT -4,2 15.0 -2,367.0 - 1,026.0 2,574.0 3,774.0
Net Working 539.0 -416.0 - 699.0 0.0 0.0
Capital
Gross In- - 8,000.0 - 5,540.0 - 4,760.0 - 3,750.0 - 3,750.0
vestment
Depreciation 3,025.0 3,375.0 3,750.0 3,750.0 3,750.0

Free Cash -8,65 1.0 -4,948.0 - 2,735.0 2,574.0 3,774.0

Flow

The assumptions about the financing policy determines which formula

is correct. If we take an equity-to-debt-ratio of 9 to I, cost of debt 6.0%,
operating risk adjustrnent 4.0% and financing risk adjustment 0.3% the fol-
lowing valuation table will arise.
 579

Table 2. WACC calculation

Free cash Cost of

Year Present value
tlow capital
2003 - 8,651.0 1.0960 - 7,893 .2
2004 - 4,948.0 1.2012 -4,1 19.2
2005 - 2,735,0 1.3165 -2,077.4
2006 2,574,0 1.4429 1.783.9
2007 3,774,0 1.5814 2,386.4
Continuing value 39,312.5 1.5814 24,858,7
Enterprise value 14,939.2
Liabilities - 1,493.9
Equity value 13,445,2

Estimating a target capital structure is one issue, but we have to make

sure that the company has enough capital to finance the projects. In order
to reach the equity-to-debt-ratio of 9 to 1 we need the following sources of
capital.

Table 3. Transactions with shareholders and debt holders

2003 2004 2005 2006 2007

Credit raising - 1,008.5 - 735 .0 - 584.3 - 109,5 0

Transaction with -7,703.0 - 4,314.3 -2,281.8 2,528.7 3,614.8

the shareholder
Is it possible to model the financial strategy of a young and innovative
enterprise within a simple target capital structure? How will the capital
structure be influenced, if it is impossible to receive debt capital during
2003 due to the early stage of the projects? How will the capital structure
be influenced, if we finance all of the projects with debt capital in 2004
and later on. In reality it may be impossible to change the financing
sources in this way. In this example we integrate a more typical financing
strategy for young enterprises with growth opportunities over the life cycle
of these enterprises. Implementing this financial strategy in our business
plan the following structure of cash flows to shareholder and debt holder
will arise:
580

Table 4. Transactions with shareholders and debt holders under the new financing
strategy

2003 2004 2005 2006 2007

Credit raising 0 - 5,008.5 - 2,998,3 0 0

Transaction with -8,711.5 0 0 2,189.2 3,989.2

the shareholder
In my opinion, this financing strategy is more typical for enterprises in
the new economy. If we assume this strategy as the planned strategy for
the future, we will have to face two challenges: Firstly, we do not know
the capital structure. Secondly, due to the varying capital structure, the cost
of equity will change over time; we have to calculate the net present value
with time-specific discount rates. 5 However we cannot adjust the cost of
capital, because the ratio of debt to equity is not known in advance. We
must be aware of the results of the calculation in order to be able to adjust
the cost of capital; however if we know the results, it will not be necessary
to know the cost of capital. lt is possible to break this circle using the
backward induction method, but in practical valuation situations, it will be
more convenient to solve the problemsinan other way. 6
In order to solve the problems with varying capital structures the ad-
justed present value formula separates the flow to equity in different com-
ponents. After this operation, the separated cash flow is discounted at the
risk equivalent discount rates, which are independent of the capital struc-
ture. This is the main difference in comparison to the flow to equity for-
mula or the weighted average cost of capital formula which require the
knowledge of the capital structure to estimate the discount rates. If we
separate the flow to the shareholder (C 1) the following components will
arise:

(4.1)

with Cu,t =EBII;(1-s)-t1FA1 -t1NWC1

5 Without taxation we do not have any problems with the WACC formula, be-
cause the capital structure is irrelevant for the weighted average cost of capital.
The cost of capital does not decrease with a higher leverage because we substi-
tute expensive equity with eheaper debt. The cost of capital will decrease be-
cause the benefit of debt financing will be higher.
6 Richter (2002), p. 133.
 581

Cu.t: Free Cash Flow without financing activities

TaxSht Tax benefits of debt financing
~~: Credit raising (credit repayment would be negative)
rsBt_,: Interest to debt holders
MAt: Net investment (after depreciation and amortization)
ilNWC1: Increase or decrease in net working capital

We divide the value-relevant cash flow to the shareholders into the free
cash flow ( Cu,t), the tax benefit of debt financing (TaxShr) and the cash
flow to creditors (rnBt-l - LJBt).
To discount the different components we need the following discount
rates. Because of the fact that the free cash flow is defined before financing
activities, we can use a discount rate integrating only the operating risk of
the cash flow (in the example 10%). Adjustment to the leverage risk is not
necessary. In this example the tax benefit of debt financing and the cash
flow to debt holder is without any risk, because the amount of debt is
planned in advanced. In order to use a risk free discount rate (in the exam-
ple 6%) for the tax benefits we have to make sure that there will be no re-
striction in the tax loss carry-forward. 7
Although the target capital structure is unknown, meaning that we can-
not discount the cash flow to shareholder, we are able to discount the sepa-
rate components of this cash flow with risk equivalent discount rates. Us-
ing this knowledge the following valuation table arises:
The adjusted present value appears to be a good formula to use for vary-
ing capital structures because the capital structure does not have to be
known in advance. The flow to equity formula and the weighted average
cost of capital formula require this information. Therefore difficulties arise
when valuing young enterprises with growth opportunities and varying
capital structure.

7 Richter (2002), p. 136.

582

Table 5. Adjusted present value

Year Free ca h Cost of Present Tax bene- Cost of Present

flow capital value fitfrom debt value
debt capital
2003 - 8,651.0 1.100 - 7,864.5 29.1 1.0600 27.5
2004 -4,948.0 1.210 - 4,089.3 29.1 1.1236 25.9
2005 - 2,735,0 1.3310 -2,054.8 126.8 1.1910 106.5
2006 2,574,0 1.4641 1.758.1 185.3 1.2625 146.7
2007 3,774,0 1.6105 2,343.4 185.3 1.3382 128.4
Continuing 37,740,0 1.6105 23,433,6 3,087.7 1.3383 2,307.3
value
13,526.4 2,752.4
Enterprise 16,278.8
value
Liabilities - 1,493.9
Equity value 14,784,9

4.2 Loss carry-forward

The present value of the tax paid to the govemment is influenced by the
way tax reimbursements are received. If we do not receive a reimburse-
ment in the period in which the loss is recognized, we will have a tax loss
carry-forward, which we would have to factor into the formula.
Using the weighted average cost of capital or the adjusted present value
we estimate the free cash flow, which is defined without taking financing
activities into account. The tax benefit from debt financing is integrated
into the formula by means of an adjustment to the cost of capital (WACC
formula) or aseparate component of cash flow (APV formula).
Dividing the cash flow into different components, which is characteris-
tic of the adjusted present value formula, will become more complex with
a tax loss carry-forward. This results from the fact that in some periods the
EBIT will be positive although the EBT will still be negative. 8 Due to this
fact we have to analyse tax loss carry forwards and tax benefits from debt

8 Richter (2002), p. 162.

 583

financing, because they are no Ionger independent. We have to Iook at the

benefits from debt financing if there is a tax loss carry-forward and we
must also analyse the allocation of the amount of loss carry-forwards,
which result from debt financing, over future periods. 9 The following ex-
ample assumes a taxrate of 50% and interest expenses of 50 EURO.

Table 6. Tax loss-carry-forward and tax benefit from debt financing

2 3 4 5 6 7 8
EBJT, -100 - 100 -100 100 200 200 300 300
LC-FEBIT,t 0 -100 -200 -300 -200 0 0 0
TaxEBtT.t 0 0 0 0 0 100 150 150
EBT, - 150 -150 -150 50 150 150 250 250
LC-FEBT.t 0 150 300 450 400 250 100 0
TaxEBT.t 0 0 0 0 0 0 75 125
TaxSh, 0 0 0 0 0 100 75 25

lt is only in period 8 that we are able to calculate the tax benefits using
the standard formula sr8 Bt-I. During the previous periods (6, 7) the ad-
vantages will be more pronounced because we do not have to pay any tax
although we had calculated tax expenditure in our business plan. The dif-
ference results from the fact that since period 6 there has not been any
more loss carry-forward on EBIT, however there is a tax loss carry-for-
ward on EBT. The sum of the tax benefits in periods 6 and 7 corresponds
to the sum of the tax benefits over periods 1 to 7 in a tax system with tax
reimbursement at the end of period.
Loss carry-forwards do not make it impossible to calculate an adjusted
present value, 10 but the procedure will be more complex because we have
to estimate both the tax expense without any financing activities and the
tax expense with financing activities in order to calculate the tax benefit of
debt financing. In a WACC formula it will be nearly impossible to factor
the tax loss carry-forward into the valuation formula, because we only
have information on the Ievel of debt financing, not on the amount.
So far we have arrived at the following results: On the one hand the use
of WACC and the flow to equity formula will become more complicated,
if we cannot assume a single target capital structure. The adjusted present

9 Richter (2002), p. 196.

° Copeland/Koller/Murrin (2001), p. 150, did not draw attention to the problems
1

in estimating tax benefit from debt financing.

584

value formula is more flexible in integrating varying capital structures over

time. On the other hand we realise one advantage of the flow to equity
formula in integrating tax loss carry-forwards, because it is moresimple to
build-in the non-linearity of taxation with loss carry-forwards. Adjusted
present value and the weighted cost of capital formula must estimate cash
flow of a wholly equity financed corporation before they value the benefit
from debt financing in a complex way.
Therefore the WACC formula is not to be recommended for the valua-
tion of young enterprises with growth opportunities. Flow to equity and
adjusted present value have different advantages and disadvantages in in-
tegrating the characteristics of enterprises in the new economy. Both for-
mulas have difficulties estimating the cost of capital (with or without fi-
nancing risk) and discounting negative, value-relevant cash flows. This
point will be discussed in the next section.

4.3 Discounting negative cash flow

In the above mentioned formulas and examples we discounted expected

cash flow with a risk-adjusted discount rate. If we discount negative cash
flow using this approach, we will arrive at contra intuitive results: If we
assume a range of cash flow - illustrated through two seenarios with the
probability of0.5- of(- 120) and (- 100) at the end ofthe period, we will
calculate a mean cash flow of (- 110). Using a risk-adjusted discountrate
of 10 % we calculate a net present value of (- 100). We should come to the
same result discounting the mean minus a risk discount (RD) at the risk
free discount rate of 6 %.
In order to calculate the same result, we must assume an implicit risk
discount of (+4!). In this situation the obligation to pay 120 or 100, both
with the probability of 0.5, has the same value as an obligation to pay 106
in both scenarios! An alternative description isthat the danger of paying
14 EURO is compensated by a saving of 6! If we use risk adjusted dis-
count rates to estimate the present value ofnegative cash flow, we will not
use a risk discount, we will calculate a risk premium. This calculation is
not appropriate for risk averse people.
We will have the sameproblern ifwe assume a zero cash flow. The pre-
sent value of the expected cash flow is zero. In order to arrive at the same
result as when using risk-adjusted discount rates we implicitly calculate a
risk discount of zero. This means, the obligation to pay 1 million Euro with
a probability of 0.5 is compensated by a benefit of 1 million with an prob-
ability of 0.5. In this calculation we implicitly assume a risk-neutral per-
son.
 585

5 A relevant valuation formula for young enterprises

with growth opportunities

To develop an appropriate discount formula I implement a tracking port-

folio approach. In order to value the cash flow of an enterprise we track a
portfolio with the same results. If it is possible to replicate the cash flow,
we will know the price of the enterprise, because it will be the current
price of the portfolio. The prices of the enterprise and the portfolio must be
the same otherwise there will be some arbitrage opportunities. As it is
nearly impossible to track a portfolio that is totally equivalent, we only
have to track the value-relevant items. In order to define value-relevant
items we need a model, a theory explaining which items are relevant. 11 In
our example we assume the market only considers the mean and the stan-
dard deviation of a cash flow distribution.
We assume the following model: 12 Company and tracking portfolio (the
mix of market portfolio and the risk-free security) are equivalent
=
z1 x · {1 + rM) + y. (1 + i) if
E[z1] =E[x ·(1 + rM )+ y ·(1 + i)]

p · S[z1] = s[x ·(1 + rM) + y ·(1 +i)]

with E[ z1] = Mean ofthe value-relevant cash flow
S [z1] = Standard deviation of value-relevant cash flow
p Correlation coefficient ofthe value-relevant cash flow with the
return from the market portfolio
p =Cov[z1,rM ]/(S[zt]·aM)
x Amount invested in the market portfolio
y Amount invested in the risk-:free security

In order to calculate the expected return and the standard deviation of

the tracking portfolio we also need to know the expected return and stan-
dard deviation ofthe market portfolio (f.JM,aM) and the risk-free rate of
return (i).

11 Grinblatt/Titman (1998), p. 362 f.; Spremann (2002), p. 322 f.

12 Spremann (2002), p. 324 f.
586

Solving the above equation to x and y, we arrive at the following equa-

tions:
E[zJ} = x·(I +rM )+ y ·(I+ i)

From this step we arrive at x=S[zJ]·(pfO'M ). Putting this equation

into the first one, we arrive at:
E[zJ}- (I+ JlM) · p·S[zJ}
y = - - - - - -O'M
-'-"------
I+i
The sum of x and y is the current value of the portfolio. Therefore we
know
E[ Z]-] - (JIM-i) • p · s[-]
Z]

zo=----~-----
O'M
I+i
The risk discount is the product of the market price of risk -
(JiM - i)/ O'M - and the systematic risk- p · S[ zt]. This correlation is in
principle reliable for all periods. F or discounting we only have to use the
return ofthe risk free security. The value ofthe equity results from the fol-
lowing valuation formula:
(5.1)

This formula has different advantages in comparison to traditional dis-

counted cash flow formulas:
• With this formula we directly discount the cash flow to the shareholders.
Therefore we do not have any difficulties factaring in the tax lass carry
forward.
• Varying capital structure combined with financing risk does not have
any influence on the valuation formula, because we use the risk free rate
of return for discounting. Therefore we do not have to divide different
parts of cash flow nor do we have to discount these parts with risk-
equivalent discount rates.
 587

• We do not have these difficulties using risk adjusted discount rates with
negative cash flow, because we directly calculate a risk discount using
market information.

6 Abandon option

It is relatively simple to integrate the possibility of bankruptcy or an aban-

don option. In a simple version of the formula we assume that we can es-
timate the probability ofbankruptcy ofthe project in every period. We will
arrive at the expected cash flow in t= 1 with the probability of (1-tq) .
With the probability of tq the expected cash flow is zero (there is no obli-
gation to pay the liabilities from private capital). We can only arrive at pe-
riod 2 if the company is not closed down in period t= 1. 13 We will arrive at
the following valuation formula:
(6.1)

In the formula we factor the possibility that we can reach the next period
with a given probability. The value-relevant cash flow of a period is only
calculated at the ex ante probability of reaching this period. If this formula
is used, the independence of the default risk and the other systematic risk
will be assumed. There will be no difficulties in valuing the bankruptcy or
abandon option with the more sophisticated option pricing model. 14
There are also examples in the Iiterature which integrate a simulation
model into the valuation formula. 15 Using a simulation approach is espe-
cially helpful for the valuation of young, innovative companies with
growth opportunities. With a simulation approach we calculate a range of
possible prices for a company. The simulation model does not provide a
decision however it will present possible prices for the company, which
are to be valued.

13 Spremann (2002), p. 364.

14 Grinblatt/Titman (1998), pp. 412-428; Ross/Westerfield/Jaffe (2002), pp. 212-
215.
15 Schwartz/Moon (2000a); Schwartz/Moon (2000b); Loderer et al. (2002), S.
810-822; Keiber/Kronius/Rudolf(2002); Richter (2002), pp. 177-380.
588

7 Results

1. I have mentioned some of the problems involved in valuing firms in the

new economy. In my opinion the special characteristics of these firms
challenge the traditional textbook formulas, creating issues which have
to be addressed before the problems in prediction future cash flows
should be discussed.
2. The above mentioned valuation formula does not make the mistake of
traditional valuation formulas. These formulas may be appropriate for
companies with positive cash flows without any tax loss carry forward
and a known level or amount of debt. However these traditional formu-
las cannot take into account the characteristics of young and innovative
enterprises with growth opportunities and negative cash flow. Inte-
grating varying capital structures and a tax loss-carryforward as well as
discounting negative free cash flow under uncertainty would not be a
prob lern.
3. In order to get this formula we assume that it is possible to replicate the
cash flow of a company with a tracking portfolio having the same value-
relevant characteristics. In this paper we assume that only the mean and
the standard deviation of a cash flow distribution are the relevant items
in valuing cash flow. It is possible to integrate a more ambitious model
of the capital market.

References

Böhmer C (200 1) Valuation of dot.coms, Working paper University of St. Gallen.

Brealey R, Myers A, Stewart C (2000) Principle of Corporate Finance, 6th edn,
Boston (Mass.) et al.
Copeland T, KollerT, Murrin J (2001) Valuation, 3rd edn, New York et al.
Damodaran A (2001) The Dark Side ofValuation, Upper Saddle River
Grinblatt M, Titman S (1998) Financial Markets and Corporate Strategy, Boston
(Mass.) et al.
BachmeisterD (2001) Integration von Aktienoptionsplänen in den DCF-Kalkül,
Working paper University of Leipzig
Keiber K, Kronius A, Rudolf M (2002) Bewertung von Wachstumsunternehmen
am Neuen Markt. Zeitschrift fiir Betriebswirtschaft, vol 72: 735-764
Loderer C et al. (2002) Handbuch der Bewertung, Zürich, Frankfurt am Main
Richter F (2002) Kapitalmarktorientierte Unternehmensbewertung, Frankfurt am
Main
Ross SA, Westerfield RW, Jaffe JF (2002) Corporate Finance, 6th edn, Boston
(Mass.) et al.
 589

Schwall B (200 1) Die Bewertung junger, innovativer Unternehmen auf der Basis
des Discounted Cash Flow, Frankfurt am Main
Schwartz ES, Moon M (2000a) Rational Pricing of Internet Companies, Financial
Analysts Journal, vol 56, no 3: 62-75
Schwartz ES, Moon M (2000b) Rational Pricing of Internet Companies Revisted,
Working paper University of California, Berkley, September
SofTer LC (2000) SFAS No. 123 Disdosures and Discounted Cash Flow Valua-
tion, Accounting Horizons, vol14, no 2: 160-189
Spremann K (2002) Finanzanalyse und Unternehmensbewertung, München, Wien
New Financial Accounting Standards for the New
Economy ? - Some Remarks on the Ongoing
Debate-

Christoph Kuhner

1 Abstract

Numerous empirical studies reveal that accounting numbers - equity as

well as earnings' measures - have lost value relevance, i.e. explanatory
power for stock market capitalisation and/or abnormal returns, during the
last decades. This effect is of particular significance after the rise of the
New Economy in the early 90thies and at firms belonging to 'new
economy' industries. On the grounds of this observation, commentators
argue in favour of a change of the current accounting model. Proposed
amendments include:
(i) widening of intangible asset recognition criteria, including the capi-
talization ofR&D, advertising and human resource expenditures;
(ii) measurement of intangibles at fair value;
(iii) a new set of 'revenue accounting' provisions, due to the observa-
tion that traditional revenue recognition concepts fail to capture the
critical events of the value creation cycle at new economy firms.
From an industrial economics perspective, the focus of these proposals
is on the balance sheet treatment of competitive advantages of the firm.
From a normative viewpoint, the arguments in favour of these proposals
are founded on the observation of a change in competitive environments
due to the transition from old economy to new economy. The paper analy-
ses whether these changes should imply a change in the current accounting
system. Competitive advantages and their financial accounting treatment
are analysed conceptually on the grounds of economic theory. The insights
gained give rise to a certain scepticism conceming the necessity of a
change ofthe current accounting model.
 591

2 Most, but not all intangible assets are information

goods.

In accounting literature, there is a great amplitude of different definitions

conceming intangible assets. In this paper, we will define intangible assets
along the following three criteria: 1
- They lack of physical existence.
- They are not financial instruments.
- They are long term in nature.
SF AS 141, A 14 provides a list of examples of intangible assets, which
may be apt to be recognized separately from goodwill. The list is non-
exhaustive, but includes the most significant items. 2 It is possible to assign
the assets included in this list to three different categories:
(i) Contractual rights concerning the use and the delivery of tangible
assets and services
Examples are customer contracts, lease agreements, etc. Financial
accounting for those items will not incur any special problems apart from
those connected with the underlying positions.
(ii) Contractual and legal positions concerning competitive advan-
tages due to the legal or contractual establishment of market
entry barriers
Examples are non-competition agreements, licenses and royalty.
(iii) Positions concerning the use and the delivery of information
This category constitutes the bulk of the items mentioned in SFAS 142,
Appendix F. In the terminology of Lev, 3 intangibles, knowledge capital
and intelleemal capital are interchangeable terms. More carefully, it can be
argued that most intangible assets are information goods. At least four
categories can be distinguished.
- Information goods for consuming purposes: all artistic related
intangible assets
- Information goods for purposes of technical production of physical
assets: patented and unpatented technology, trade secrets, in process
R&D.

1 Kieso/Weygandt/Warfield, p. 600; similar in: SFAS 142, Appendix F.

2 See Appendix.
3 Lev (2001), p. 5.
592

- Information goods, which concem the technical production, trans-

mission and transformation of information itself: Computer software,
software for information transmission via the world wide web and via
electromagnetic waves.
- Information goods which concem the market transmission and trans-
formation of information (i.e. the release of information asymmetries
and the lowering oftransaction cost): Trademarks, trade dress, inter-
net domains, "brand name capital", reputation capital, customer rela-
tionship capital, employee relationship capital, market design of
trading platforms, contract design, etc.
Consumption, production, transmission, the economic use and the tech-
nical use of information are most heterogeneaus activities. It might be
expected that the balance sheet treatment of these activities will reflect this
heterogeneity.

3 Information goods cannot straight forwardly be

categorized as "private goods" or "privately owned
economic resources", nor as "public goods".

As it is weil known, it might be difficult or even impossible to assign,

enforce and protect property rights conceming information. Pieces of art or
Iiterature might be protected by copyrights; patents or trade secrets might
prohibit the use and the knowledge of technical innovations. However, not
all information goods are apt to be protected (e.g. human capital, reputa-
tion), and protection even if founded on a legal title will in many cases be
far from perfect. 4 Non-excludability of the use of information corresponds
to non-rivalry: Transmission, technical use and consumption of informa-
tion goods is non-rivalling: information coded in digital data files can be
used by arbitrarily numerous people without its quantity being diminished
or impaired.
On the grounds of non-rivalry of consumption and technical use, infor-
mation is often qualified as a public good. This qualification is not
perfectly correct, because the market use of information follows substan-
tially different rules: In competitive markets, the economic use of
information by competitors may be considered as hyper-rivalrous, because
first-mover effects are most significant: Consider only stock exchanges

4 A case in point is NAPSTER. Moreover, e.g., anecdotal evidence indicates that

many New Economy firms failed because they did not succeed in protecting
their newly developed technology by globally enforceable patents.
 593

where even small quantities of relevant information may give rise to

tremendous trading gains if privately hold; information which is common
knowledge, however, has no value for the players. A piece of information
may even be of negative value if the players are not conscious that it is, in
fact, common knowledge of the market and, therefore, is embodied in the
market prize. Similar phenomena may be observed at markets on which
innovative products are traded and where the first mover "takes it all":
research and development, branding and advertising expenditures of the
second and third movers may be irrecoverably lost.
It might be expected that financial accounting treatment will reflect
these enigmatic qualities of information goods.

4 The importance of intangible goods as corporate

value drivers has risen dramatically in recent years.

The importance of information for economic development has risen

dramatically during the last decades: 'industrial society' is transforming
into 'information society' and information as an economic resource is at
the core of our perception ofthe New Economy. 5 There are several causes
for these fundamental changes:
(i) Due to digitalisation and computerisation, the cost of coding,
transmission and copying have declined towards almost zero; the
technical quality of transmissions and copies has risen in the
sameway.
(ii) Together with the accumulation of wealth, consumer goods are
becoming more and more sophisticated; including more infor-
mation components, software components as weil as components
of product and brand design.
(iii) For the same reason, research and development is becoming
more important.
(iv) Due to deregulation and privatisation, many long-term oriented
research and development activities which traditionally used to
be carried out by state institutions and non-profit organisations
are nowadays privately organized by profit maximizing firms.
Therefore, the perception of the public that even long-term R&D
projects are privately owned and privately traded commodities
has spread.

5 For an overview of different perceptions ofthe New Economy see e.g. De Long
/Summers (2001).
594

(v) In recent years, economic science as weil as real world experi-

ence is discovering that one of the main sources of economic
wealth is the release of information asymmetries between market
players and the decline of market transaction cost. 6 Transaction
cost are declining e.g. because of the design of innovative con-
tractual schemes (e.g. compensation schemes, market micro-
structure design, auction design, mechanism design), because of
signalling activities by corporate players (advertising, branding,
investing in reputation) and because of the Internet as a new and
global trading platform.

5 Competition in markets for information goods of the

new economy has its own rules leading to an increase
in unsystematic risk.

Research on competition and business strategies in the New Economy is at

an early stage; however, some general observations on the changing rules
of competition can be made: 7
- Rising importance of sunk costs in early stages of product life cycle:
Particularly, R&D, branding and advertising expenditures play a
dominating role in the early stage of a New Economy product's life
cycle, thus enhancing the asymmetry between cash outflows in ear1y
stages and cash inflows in later stages. Early stage investing expen-
ditures may be usually qualified as sunk, because in the case of fail-
ure, items such as R&D or brand building will not be recoverable.
- Increasing returns to scale because of non-rivalrous usage of
information goods: Because of digitalisation, the cost of reproducing
an information good for usage by costumers are in most cases negli-
gible. Together with huge amounts of fixed cost, this will result in
tremendous returns to scale in the course of production and mar-
keting.
- Network effects: 8 Network effects arise if the consumer value of a
commodity is rising with the number of its users. Network effects are
omnipresent in the New Economy, as different technical standards
(e.g. Netscape vs. Internet Explorer); different global trading plat-

6 See on1y the pioneering work of Jensen/Meckling (1976), pp. 305-60.

7 See, e.g. Lev (2001), pp. 21-49; ShapiroNarian (1999); Varian (2001).
8 For the ana1ysis ofnetwork effects see most influentially: Economides (1996);
for application to the New Economy: Varian (2001).
 595

forms and different global communication platforms compete for a

global public and the attractiveness of adopting a standard and of
trading and communicating on a platform will increase as the number
of its users increases.
- "Tipping markets" and the occurrence of "winner-takes-all"-competi-
tion: Tipping markets and winner-takes-it-all competition are the
direct consequences of network effects and increasing retums to
scale: It is a well documented fact that in many New Economy mar-
kets (take only for example the markets supplied by Microsoft), only
few or even one supplier will survive and, consequently, benefit
exclusively from his or her monopolistic position.
The above-mentioned points imply that investments of individual firms
in New Economy projects will be far more riskier than investments intra-
ditional industries; at least, unsystematic risk will spread in comparison
with the 'tangible economy'; volatility of discounted cash flows will rise.

6 The proposition that, in general, investment in

intangible assets is riskier than investment in tangible
assets is weil rooted in economic theory of market
equilibrium and competition.

New Economy projects will in many cases reveal a real option-like, even
negative NPV-cash flow projection. Knowledge assets therefore will be, a
priori, far more riskier than physical or financial assets. However, this
viewpoint is not uncontested. Lev argues:
"Surely, uncertainty about the future benefits of a clinically proven
drug is not larger than the uncertainty associated with the expected
benefits of commercial property in a newly developed area or of a
loan granted to an enterprise operating in a developing country,
which are both recognized as assets by GAAP." 9
It is obvious that the riskiness of different projects, intangible or physi-
cal, can only be determined by analysing them individually. However,
some heuristic conclusions can be drawn if the different functions of
physical and intangible goods, which they perform in the course of the
corporate value chain, is taken in account: Lev himself, in his definition of
intangible assets, focuses on such a functional approach:

9 Lev (2001), p.124.

596

"( ... ) intangible assets are non-physical sources of value (claims to

future benefits) generated by innovation ( discovery), unique
organizational designs, or human resource practices.
The emphasis oftbis definition is on discovery, innovation, entrepreneu-
rial creativity and uniqueness. In other words: on competitive advantage in
comparison to other market players. Valuing intangible assets, therefore,
means valuing the competitive advantages of a firm, i.e. to estimate the
expected present value of the economic rent (quasi-rent) associated with
the existence of competitive advantages.
Competitive advantages result from deviations from perfect competitive
equilibria; they represent options to capture economic rents by taking
advantage from market imperfections. Although general equilibrium with
perfect competition is a hypothetical state which, in reality, virtually never
will be obtained, in market economies, in general, there will be a trend
towards it. We can therefore generally characterize competitive advantages
as ephemeral in nature, although their life period may be indeterminable ex
ante: In the - hypothetical - state of neoclassical general equilibrium, no
value at all will be attached to intangible assets which in substance consti-
tute competitive advantages. In contrast, tangible assets will not lose value
as markets move towards competitive equilibrium.
Acquisition of competitive advantages is at the core of entrepreneurial
activity, 10 and, in an equal way, valuation of competitive advantages is an
entrepreneurial task.

7 Capitalization rules may be based on the paradigm of

fair value measurement of assets and liabilities or on
the performance-measurement-paradigm (matching of
revenues and expenses).

Generally spoken, standard setters used to be reluctant to opt for recogni-

tion of intangible goods as assets on the balance sheet. To analyse this
traditional reluctance, it might be useful to remernher what are the ultimate
reasons for capitalisation of assets: Why and when should an asset be
capitalized in the balance sheet? This question was subject of most influ-
ential debates in accounting theory in the last century. Two, partly oppo-

Io See, for example: Kirzner (1973).

 597

site, partly complementary viewpoints have emerged as results of these

debates: 11
1) The sum of capitalized assets should reflect as exact as possible
the fair value of the marketable fortune of an enterprise.
If based on this paradigm, a capitalization rule for intangible assets
would imply identification of all identifiable intangible assets and valua-
tion at fair value.
2) Capitalized assets should reflect cash outflows, which are not
designed to generate revenues in the current period but in future
periods.
Following the last concept, i. e. the matehing principle concept ("let
expenses follow revenues"), capitalisation rules should guarantee a sepa-
ration between operative and investment expenditures of a corporate entity
for the purpose of timely performance measurement; investing activities
should be identified with sufficient reliability. Assets on the balance sheet
should reflect the amount of capital expenditures invested in positive
present value projects.

8 Measuring the fair value of intangible assets implies

fair value measurement of competitive advantages.
The rise of the fair value paradigm as financial
accounting concept will Iead to a substantial change
in the role of financial accounting and auditing.

Fair value of an asset represents the amount at which that asset could be
bought or sold in a current transaction between willing parties, that is,
other than in a forced or liquidation sale. 12 In the absence of quoted market
prizes, fair values have to be estimated e.g. by discounted cash flows or by
option pricing models.
At least in the US, in recent years there has been evidence of a turn-
around towards a more fair-value based measurement of assets and liabili-
ties in financial accounting. Fair value was introduced in SFAC No. 7 as

11 See, e.g., Upton (2001), p. 13. The two competing paradigms were also at the
core of the at the time most influential debates about static vs. dynamic
interpretations of fmancial accounting in German literature, see seminally:
Schmalenbach (1919), pp. 1-60.
12 See, e.g.: SFAC No. 7 (g1ossary).
598

measurement objective for present-value-based accounting measurements.

Intangible assets which are acquired together with corporate entities are to
be measured with their fair values according to SFAS 141, fair values are
the benchmark for measuring the value of intangible assets with indeter-
minate lives and of acquired goodwill according to the impairment test
SFAS 1411142. Thus, for many balance sheet items, acemal accounting via
direct fair value measurement replaces more and more acemal accounting
by amortization of expenditures over the estimated useful lifetime of an
asset. The key argument in favour of this treatment is that, for many intan-
gible assets, usefullife is not determinable. 13
Measurement of fair values for positions which cannot straight-
forwardly be replicated by items traded in capital markets implies entre-
preneurial judgment: lt has been pointed out that the huge majority of
intangible assets consists of competitive advantages. In opting for fair
value measurement of intangible positions as an element of GAAP, stan-
dard setters initiate a fundamental change in the role of certified public
accountants and auditors: In the future, auditors will be more and more
required to testify entrepreneurial judgments by testifying fair values of
information goods and competitive advantages. Also on the grounds of the
recent experiences, 14 one might be sceptical about whether such a change
in the role of the accounting and auditing profession will strengthen the
quality and integrity of fmancial accounting.

9 From a matching-principle viewpoint, capitalizing

expenditures for intangibles as delayed charges
would imply exact measurement of changes of the
firms' intangible capital

From a matehing principle point of view, capitalization of intangible assets

requires the identification of the investing character of cash outflows with
sufficient reliability. Consider for example expenses for branding and
advertisement: Not at all obvious is the answer to the question whether
they generate surplus cash flows in future periods or whether they are
spent in order to maintain the existing brand name capital of the corpora-
tion. With respect to tangible assets, e.g. property and equipment, this
question can be answered in a much more Straightforward way.

13 See, e.g.: SFAS 142, par. B79-B94.

14 See, e.g., Coffee (2001).
 599

Consequently, capitalization of intemally generated information goods

would imply judgments about value impairment or value enhancement of
the information capital stock in a period as a whole.
Example: Consider an entity, which in 2002 spent 1.000.000 € for
purposes of advertising their existing brands. The advertising
expenditures are intended to sustain the public's long-term percep-
tion of the brand name. Should an intangible asset be recognized
and valued at € 1.000.000,- ?
Answer. lt depends. And it depends not only on the question whether the
advertising activity is successful or not. It depends also on the question
whether due to the advertising campaign, the value of existing brand name
capital of the firm is enhanced, i.e. whether the resources spent on adver-
tising and branding are sufficient to counter the periodical impairment of
brand name capital.
Example: Imagine that our corporation, indeed, had spent
1.000.000,- € on branding and advertising. But, however, to main-
tain the achieved Ievel of public attention in recent periods, an
expense of 3.000.000,- € would have been necessary. Should the
corporation recognize a liability of 2.000.000,- ("provision for
omitted branding expenditures")?
US GAAP excludes generally the recognition of such provisions as they
do not match the definition of a liability: They do not imply a present ob/i-
gation, that means: a legal or economic commitment to deliver cash or
assets or services to a third party. German law, for example, allows recog-
nition of similar provisions under certain circumstances. 15
The example demonstrates that, from a capital maintainance perspec-
tive, a rational capitalization rule for items such as branding and adver-
tising expenditures, human capital expenditures or research and develop-
ment expenditures would imply more or less precise measurement of the
annual changes in the firm' s brand name capital, human capital and R&D
capital, a task which, if performed or supervised by public accountants,
might qualify as an abrogation ofknowledge in the Hayekian sense. 16

15 § 249 (2) Commercial code: "Provisions may be recognized for present or

form er years' expenses which are precisely determined by their type and are
probable or certain at the balance sheet date but uncertain in respect of the
amount or timing of the corresponding cash outflows." However, in the
example presented, recognition would be not allowed because of the
uncertainty of future cash outflows.
16 For the concept of arrogation ofknowledge see e.g. Hayek (1974).
600

10 Accounting numbers have lost value relevance. This

phenomenon is frequently attributed to the
incomplete recording of intangible value drivers in
financial statements. Standard setters' reluctance to
change capitalization rules for intangible assets is,
however, weil founded on economic grounds.

Accounting rules, so the well-known complaint, have ignored in many

respects the dramatic changes of economic environments and remain at an
inferior stage of development. 17 The complaint is supplied by numerous
empirical studies which (i) indicate that the book to market ratios have
declined dramatically during the last years and thus, the balance sheet
equity numbers are failing more and more to represent the real value of a
company, and which (ii) indicate that accounting numbers and particularly
earnings figures have lost 'value relevance', i.e. predictive power for the
explanation of abnormal returns. 18
However, standard setters remain reluctant towards a radical change of
capitalization rules for information goods. The rule that intemally
generated intangibles shall be capitalized only under very restrictive
circumstances was modified partly 19 but is not challenged substantially, up
tonow.
The reason for this is, after all, the lack of reliability2°: Following the
conceptual framework of the FASB, capital expenditures should not
qualify for capitalization if the future cash flows they generate are too un-
certain. Uncertainty of the future cash flows can be measured by their
expected volatility (variance/standard deviation). ReHability of book
values as a signal for fundamental value is declining with rising volatility
of the discounted cash flows (i. e. fundamental value), they generate. 21
Following the FASB, recognition of an asset requires a certain minimum
threshold of reliability, i.e. a maximum threshold for the variance of the

17 See, e.g., Lev (2001) pp. 79-103; Lev/Zarowin (1999), pp. 353-385.
18 See, for example, Francis /Schipper (1999), pp. 319-352.
19 Restrietions goveming the recognition of intemally generated assets are, e.g.,
part of financial accounting according the German Commercial Code (HGB, §
248(2)), ofUS-GAAP: SFAS 142.10, ofiAS 38,39-52.
20 ,,Reliability is the quality of infonnation that assures that infonnation is
reasonably free :from error and bias and faithfully represents what it purports to
represent." SFAC No. 2, glossary.
21 Fora theoretical foundation see, e.g., Kirschenheiter (1996), pp. 43-60.
 601

expected value of future cash flows, it generates. 22 Thus, certain investing

expenditures may not qualify for capitalization, even if they generate posi-
tive NPV and even they can be classified as identifiable assets on the
grounds of identifiable future benefits they generate.
Economic characteristics of intangible assets are, among others, non-
perfect excludability of usage, enhanced importance of sunk cost, econo-
mies of scale and network economies. After all, the most significant argu-
ment in favour of an accounting treatment of intangible assets differing
from tangibles is their characteristic as competitive advantage: They
vanish with the move towards perfect competitive equilibria and, thus,
reveal a fundamental difference in comparison to tangible economic
"resources". Accounting for intangible assets is therefore an entrepre-
neurial task.

References

Coffee JC (2001) The Acquiescent Gatekeeper: Reputational Intermediaries,

Auditor Independence and the Govemance of Accounting, Working Paper,
Columbia Law School. At: www.ssm.com
De Long JB, Summers L (2001) The "New Economy": Background, Historical
Perspective, Questions, and Speculation, conference paper, Economic Policy
for the Information Economy. A symposium sponsored by the Federal
Reserve Bank of Kansas City, Jackson Hole, Wyoming, August 30 -
September 1
Economides N (1996) The Economics of Networks, International Journal of
Industrial Organization, vo114: 670-699
Financial Accounting Standards Board (1980) Statement ofFinancial Accounting
Concepts no 2: Qualitative Characteristics ofFinancial Accounting, Norwalk
Financial Accounting Standards Board (2000) Statement of Financial Accounting
Concepts no 7: U sing Cash Flow Information and Present Value in
Accounting Measurement, Norwalk
Financial Accounting Standards Board (2001) Statement ofFinancial Accounting
Standards no 141: Business Combinations, Norwalk
Financial Accounting Standards Board (2001) Statement ofFinancial Accounting
Standards no 142: Goodwill and Other lntangible Assets, Norwalk
Francis J, Schipper C (1999) Have Financial Statements Lost Their Relevance ?,
Journal of Accounting Research, vol37: 319-352

almost everyone agrees that criteria for formally recognizing elements in

22 "( ... )

financial statements call for a minimum Ievel or treshold of reliability of

measurement that should be higher than is usually considered necessary for
disclosing informationoutside financial Statements(... )." SFAC No. 2, Par. 45
602

Hayek v. FA (1974) Friedrich August von Hayek- Prize Lecture, Lecture to the
memory of Alfred Nobel, December 11
Jensen MC, Meckling W (1976) Theory of the Firm: Managerial Behavior,
Agency Costs and Ownership Structure, Journal Of Financial Economics, vol
3: 305-60.
Kieso DE, Weygandt JJ, Warfield TD (2001) Intermediate Accounting, 10. edn,
New York etc.
Kirschenheiter M (1996) Information Quality and Correlated Signals. Journal of
Accounting Research, vol35: 43-60.
Kirzner I (1973) Competition and Entrepreneurship, Chicago, London
Lev B, Zarowin P (1999) The Boundaries of Financial Reporting and How to
Extend Them, Journal of Accounting Research, vol37, Supplement: 353-385
Lev, B (200 1) Intangibles - Measurement, Management and Reporting, Brookings
Institution, Washington D. C.
Schmalenbach E (1919) Grundlagen dynamischer Bilanztheorie. Zeitschrift fiir
betriebswirtschaftliche Forschung, vol 13: 1-60
Shapiro C, Varian HT (1999) Information Rules. Boston
Upton WS (2001) Business and Financial Reporting, Challenges from the New
Economy, Financial Accounting Series- Special Report, FASB
Varian HT (2001) HighTechnology Industries and Market Structure, conference
paper, Economic Policy for the Information Economy. A symposium
sponsored by the Federal Reserve Bank of Kansas City, Jackson Hole,
Wyoming, August 30 - September 1

Appendix : List of intangibles to be recognized

separately from goodwill according to SFAS
141, A 14

a. Marketing related intangible assets

(1) Tademarks, tradenames
(2) Service marks, collective marks, certification marks
(3) Trade dress
(4) Newspaper mastheads
(5) Internet domain names
(6) Noncompetition agreements
b. Customer-related intangible assets
(1) Customer lists
(2) Order or production backlog
(3) Customer contracts and customer relationships
(4) Noncontractual customer relationships
 603

c. Artistic related intangible assets

(1) Plays, operas, ballets
(2) Books, magazines, newspapers, other literary works
(3) Musical works, such as compositions, song lyrics, advertising
jingles
(4) Pictures, photographs
(5) Video and audiovisual material, including motion pictures,
music videos, television programs
d. Contract based intangible assets
( 1) Licensing, royality, standstill agreements
(2) Advertising, construction, management, service or supply
contracts
(3) Leaseagreements
(4) Construction permits
(5) Franchise agreements
( 6) Operating and broadcast rights
(7) U se rights, such as drilling, water, air mineral, timher cutting,
and route authorities
(8) Servicing contracts, such as mortgage servicing contracts
(9) Employment contracts
e. Technology based intangible assets
(1) Patented technology
(2) Computer software and mask works
(3) Unpatented technology
(4) Databases including title plants
(5) Trade secrets such as secret formulars, processes, recipes
Controlling the Assets of the New Economy ... and
not only the New Economy

Rainer Strack

1 lntroduction

In the "old economy," business success depended primarily on strategic

investment in capital goods-properties, manufacturing equipment,
inventories, and so forth. Today, a company's performance is more a result
of its investment in a new set of assets: customers and employees (Siegert
2000). The ability to find, retain, and develop the best of these assets has
suddenly become the leading indicator of success.
Are companies adapting weil to this change? After all, new kinds of
assets imply a need for new systems of control. When assets were pri-
marily capital-intensive, the balance sheet and profit-and-loss statement
were used to husband investment capital and measure rewards to share-
holders. That approach made sense for Rockefeller's Standard Oil and
Sloan's General Motors. The new economy, which depends more on good
ideas, customer relationships, and quick decision-making than on physical
or financial capital, requires a very different approach. The consequence:
Neither management nor controlling is currently in a position to determine
effectively how company value should be managed. This hurts both the
quality of strategic decisions and the ability to adequately measure com-
pany performance internally and externally. This is where WorkonomicsTM
and CustonomicsTM come in-The Boston Consulting Group's new human
resources and customer-oriented Controlling systems (Strack and Villis
2001; 2002) The underlying concept offers a set of straightforward,
quantitative and qualitative key metrics that form a mirror image of classic
controlling and financial metrics.

2 Workonomics™: the concept

The Boston Consulting Group's WorkonomicsTM concept aims not to

replace classic capital-based value management concepts, but to comple-
ment them with human capital controlling. Therefore, it will first be neces-
 605

sary to briefly review traditional, capital-based value management con-

cepts.

2.1 Classic value management

Many businesses today have traded in the static capital return metrics
(ROI, ROE) for dynamic, value-oriented measures of success that are theo-
retically based on estimated company value (Rappaport 1999). Besides the
capital value method, the economic profit or residual income methods, as
they are called-such as Stern and Stewart's EVA1M concept (economic
value added; Stewart 1990) and The Boston Consulting Group's CVA
concept (cash value added; cf. Stelter 1999, or Strack and Villis 2001}--
are the most widely known and used. While static return metrics such as
return on investment (ROI) measure only company unit efficiency, value-
oriented economic profit metrics also take profitable growth into account
as an essential source ofvalue creation. This is their central advantage.
Economic profit (EV A/CVA) is here defined as profit over and above
the cost of capital as determined by the capital markets. lt can also be
expressedas the difference between return on investment (ROI) and cost
of capital (CoC) multiplied by invested capital (IC).
EVA/CVA = profit - cost of capital x invested capital ( 1)
= (ROI - CoC) x IC
A positive profit is not enough: from the shareholder perspective,
companies must eam at least the cost of capital for the invested capital.
Altematively, shareholders can invest in other businesses that guarantee
coverage of the cost of capita1 in the return. On1y that which is over and
above the cost of capital is real surplus income and a contribution to value.
What are the options for raising EVA/CVA?
• lmproved retum on capital (ROI) andlor
• Profitable investment (IC) growth, i.e., growth with a given ROI that is
above the cost of capital (CoC).
The variables of the EVA/CVA concepts deal with capital (return on
capital, cost of capital, invested capital), which are measured and used for
steering things like capital allocation. The main levers for increasing value,
such as higher capital return and profitable capital growth, are also
oriented purely on capital. Therefore, this perspective is called the "capital
view." Employees are not explicitly accounted for. Only personnel costs
appear as a deducted variable in ROI or EVA. In these dassie controlling
606

systems, human resources are viewed as pure cost factors-and costs are
usually supposed to be minimized.

2.2 Workonomics™: the metrics

The goal of Workonomics TM is to bring a level of transparency and struc-

ture to the human factorthat is comparable to the one brought by capital-
based systems to the factor of investment capital, in a combination of two
systems: the Workonomics TM system provides a complete set of simple,
quantitative, HR-oriented metrics, forming a mirror image of dassie cont-
rolling and fmancial metrics. (Strack and Villis 2001; 2002) For the new
HR-side metrics, EVA/CVA is expressedas follows.
R: revenues
MC: material cost plus other expenses and adjustments
PC: personnel cost
D: depreciation
P: number of employees

EVA = (ROI - CoC) IC (2)

= Profit- CoC x IC

=R-MC-PC-D-CoC x IC

The following new metrics are introduced

R-MC-D-CoCxiC
VAP = - - - - - - - (3)
P

VAP: (Value added per person)

ACP= PC (4)
p

ACP: (Average personnel costs per person)

With these new definitions we end up with
 607

EVA= (VAP-ACP)P (5)

As seen in figure 1, the only other way to express EVA or CVA, if not
with capital metrics, is with three metrics relevant to human resources
(Strack and Villis 2001, 2002; Strack 2002):
• Value added per person (VAP): This is the average value added of
the employees. The analysis of individual employees is not possible
here. Value added is somewhat more widely defined than usual
(revenues- material expenses; see also Wundererand Jaritz 2002).
Furthermore, employees must generate depreciation and the cost of
capital for invested capital. The broader definition is essential, as
otherwise, if investments are substituted for employees, the value
added per person (V AP) always rises (even if the substitution is inef-
ficient). Value added per person can be interpreted as the produc-
tivity of the employees.
• Average cost per person (ACP): This is the average HR cost per
employee.
• Number of employees (P) is the total number of people employed.
EVA or CVA is positive if the value added per person (V AP) is high er
than the average cost per person (ACP). As stated above, this perspective
is the human resources view.

-
Ca~ HRview

Formula ilWI · CoCl iC - i\ \1'- \CI'll'

Return on Investment
(ROI)
-- Value added per person
(VAP)

-
Drivers Costs of Capital Average cost per person
(CoC) (ACP)

lnvested capital Number of people

(IC) (P)

-
ROI VAP
CoC ACP
Chart

IC p

Fig. 1. The Workonomics™ metrics

608

The factor of personnel to capital-related costs determines which of the

two views is more important for the respective company. The high er a
company's personnel costs, the more relevant the Workonomics TM metrics
are. For large German corporations (DAX 30) the factor of personnel to
capital-related costs is almost always larger than one--even for dassie
power and industrial companies (Strack et al. 2000). Therefore,
WorkonomicsTM is highly relevant not only for the new, but also for the
old economy.
Both the capital and the human resources views are anchored in EVA or
CVA. Furthermore, the equations have the same form in both views, so
that direct analogies and correspondences are possible:
• Average cost per person (ACP) corresponds with the cost of capital
(CoC),
• value added per person (VAP) corresponds with capital return (ROI),
and
• number of people (P) corresponds with invested capital (IC), i.e., the
deployment of resources in the human resources view corresponds to
the deployment ofresources in the capital view.
Thus the balance sheet of the capital view becomes, in the human
resources view, a sort of human balance sheet. The two measures of effi-
ciency are return on investment (ROI) for capital employed and value
added per person (VAP) for people employed. The two growth metrics are
invested capital (IC) and number of people (P). In this way, capital and
employees, as essential company resources, are put on equal footing and
integrated in a single value management approach.
How can value be raised in the Workonomics TM view-that is, how can
a positive .:1EVA or .:1CVA be achieved? Here, too, the analogy with
dassie value management holds.
 609

lmprovement of value Profitable employment Reduct1on of personnel

added growth costs

Fig. 2. Options for increasing value in the human resources view

Higher value can be achieved in the following ways (see figure 2):
• Raising the value added per person (VAP). Actions here are process
improvements, and above all, human resources development measures
such as training, recruiting, etc.
• Another possibility for increasing value is profitable workforce expan-
sion, i.e., growth with a value added per person (V AP) that is above the
average cost per person (ACP). Possible measure? Hiring employees
whose potential value added is higher than their HR costs. This brings in
a whole new dimension of value creation, namely the explicit lever of
profitable workforce expansion. Shareholder value and workforce
expansion do not necessarily have to be at odds; profitable workforce
expansion leads to an increase in value. But since a larger workforce
creates value only if the value added per person (VAP) is higher than
the average cost per person (ACP), the connections between produc-
tivity, wages, and workforce size are direct and clear: If VAP is lower
than ACP in certain company units, staff cutbacks will effect an increase
in value.
• Finally, reducing average cost per person (ACP) also acts to increase
value, such as by moving production operations to lower-wage countries
or-from an economic standpoint-lowering non-wage costs. Addi-
tionally, the linkage of ACP to VAP can form the basis for new, vari-
able compensation models.
Lowering human resources costs is the usual strategy, but the first and
second levers demoostrate new options for raising value. The focus shifts
610

from simple cost reduction to increased productivity and profitable staff

expansion. Workonomics ™ metrics can also be calculated as averages for
companies or company units, thus opening new, additional perspectives on
value generation for the respective company. The Workonomics™ concept
is thus perfectly suited for the strategic control of human capital or for
operative human resources controlling.
At many companies, strategic HR controlling is pure reporting with a
content focus on personneI costs and the size of the workforce. Sometimes
a few additional metrics from the balanced scorecard's Iearning quadrant
are brought in, but human capital is not systematically managed. So how
can Workonomics™ be used for strategic control?
First, the quantitative Workonomics™ metrics for various company
units can be calculated-both historically and for planning. Especially in
planning, enormous productivity increases are often anticipated, but not
backed up with concrete human resources measures or projects. With
Workonomics™, the human resources department has a quantitative con-
trolling instrument that is directly Iinked to the company's key ratio. And
Workonomics ™ metrics can be further broken down; average cost per
person (ACP), for example, can be divided into compensation components
for human resources cost controlling. Or the workforce can be broken
down into employee groups for workforce controlling. The company's
management philosophy determines the degree of detail; decentralized
companies generally stick to a few quantitative metrics.

Quantitative
targets

Qualitative
targets

Process
control

Fig. 3. The levels of strategic human resources Controlling

 611

But purely quantitative Controlling usually does not suffice (see figure
3). Generally, HR management measures do not directly improve quanti-
tative targets, but often indirectly influence them via individual employee
behavior. To account for this complex interplay of factors in the HR
controlling concept, the quantitative results Ievel must be expanded to
include qualitative targets and the process control perspective (Strack et al.
2000). There are three central qualitative goals in HR management that
significantly influence the quantitative metrics. It must be ensured that
employees
• have the required skills
• use their skills-with support from supervisors-in a goal-oriented way,
and
• are motivated to use their skills.
Consequently, precise targets must be formulated for the three dimen-
sions of skills, management quality, and employee motivation, and their
attainment systematically monitared with appropriate metrics. To accu-
rately represent individual strategic emphases, it makes sense to further
differentiate the three metrics (e.g., international skills as a subcriterion for
skills). The company determines the desired Ievel of detail. Even if a direct
functional connection between quantitative and qualitative targets can
sometimes be difficult to determine, their interaction can be derived on the
logical Ievel. The linkage of qualitative and quantitative targets not only
directly couples HR management goals with those of the company as a
whole, but makes the comprehensive control ofhuman capital possible.
Measures on the HR process side work to improve results in these
quantitative and qualitative metrics. A human resources development
process, for example, directly influences employee skills and thus value
added per person (VAP). And staffing processes affect hiring and
dismissals (P). HR processes can thus be consciously used to effect change
in quantitative and qualitative targets. Ultimately, these metrics must be
managed in the context of a controlling process-that is, goals must be
derived, corresponding metrics determined, and planning backed up and
substantiated with measures. Finally, the actual attainment of goals must
be reviewed. A human resources controlling process thus conceived and
implemented is a quantum leap as compared to classic personnel cost and
staff reporting.
612

3 Custonomics ™: the concept

In the Custonomics™ perspective (Strack and Villis 2001, 2002), customer

capital moves into the foreground. CVA or EVA is expressed solely with
metrics relevant to customers, rather than with capital or human resources
metrics. CVA or EVA can be expressed as the difference between value
added per customer (VAC) and the sales and marketing costs per customer
(ACC), multiplied by the number of customers (C)'. Since here, too, CVA
or EVA is the common anchor, and the form of the equations is identical,
analogies and correspondences similar to those in the Workonomics ™ per-
spective can be drawn (cf. figure 4).

Customer view

!KO I - (<{ l IC I \ ' \ C- \ C'C I C

Return on investment ........ Value added per customer

........
(ROI) (VAC)
Cost of Capital Acquisition and retention cost
(CoC) per customer (ACC)

lnvested capital
(IC)
........ Number of customers
(C)

ROI VAC
·I
CoC
........ ACC

IC c
Fig. 4. The Custonomics™metrics

In this view, value (a positive /1CVA or !:!.EVA) can be created by

increasing the value added per customer (VAC), with cross-selling, for

1 Exact definitions in the CVAIEVA framework are: VAC = VA* /C with VA* =
R-MC*-PC*-D*-CoC·IC* and ACC = SMC/C. Abbreviations: VA*: value
added; C: number of customers; R: revenues; MC*: material costs plus other
expenses plus adjustments without sales and marketing share; PC*: personnel
cost without sales and marketing share; D*: depreciation without sales and
marketing share; CoC cost of capital; IC* : invested capital without sales and
marketing share; SMC: sales and marketing costs. (Strack and Villis 2001)-
 613

example; by lowering sales and marketing costs (ACC), with measures

such as variable sales costs and effective media selection; or by profitably
growing the customer base (C), such as with special customer retention or
acquisition programs. With this, we have an appropriate quantitative con-
trolling instrument for numerous customer-oriented companies (e.g., Inter-
net companies, retailers, telecommunications firms, insurers, banks). The
concept can be used for managing companies, company units, customer
segments and sales locations, and even, in extreme cases, to guide deci-
sions on individual sales staff members. Custonomics1M is more than a
pure controlling tool, because in the process of customer analysis and
segmentation, it touches on many strategic questions. As in
Workonomics1M, the Custonomics1M metrics can be broken down in a
value driver tree. The resulting operative metrics are then quantitatively
linked to EVA and CVA, and sensitivity analyses and seenarios can be
easily performed.

4 Conclusions

These two perspectives can be rounded out with a third: the supplier view
(Supplynomics). Integrating all four perspectives (see figure 5) gives us
the integrated RAVE1M (real asset value enhancer) approach, or balanced
value management that focuses on real assets (Strack and Villis 2002). In
the RAVE1M concept, classic capital-based value management, with ratios
like ROI, is complemented with three crucial dimensions. Workonomics1M
measures and controls human capital, Custonomics1M measures and con-
trols customer capital, and Supplynomics1M measures and controls supplier
capital. All three are anchored in a universal controlling metric, EVA or
CVA. Thus, the single-sided, capital-oriented controlling instruments em-
ployed at many companies are not replaced, but completed, with three im-
portant new dimensions. These dimensions reflect the real assets of most
oftoday's companies. More than a simple system ofmetrics, RAVE1M rec-
ognizes these assets as centrat sources of value creation and enables the
quantitative, value-oriented strategic and operational control of their value
potential. Hence, RAVE1M could be ideally connected to the classic
Balanced Scorecard approach with a quantitative connection between the
central controlling metric and the four dimensions.
614

Cap/la/ vl ew

Supplier Customer
vlew vlew
(Supplynomlcs) (Custonom /cs "')

HRview
(Worlconomlcs "'-)

Fig. 5. The RAVE™ approach

References

Rappapart A (1999) Shareholder Value: Ein Handbuch fiir Manager und

Investoren. 2nd edn, Schäffer-Poeschel, Stuttgart
Siegert T (2000) Entwicklungstendenzen der wertorientierten Geschäftsfeld-
Steuerung. Die Zukunft der diversifizierten Unternehmung. In: Hinterhuber
HH (ed), München
Stelter D (1999) Wertorientierte Anreizsysteme fiir Führungskräfte und
Mitarbeiter. In: Bühler W, Siegert T (eds), Unternehmenssteuerung und
Anreizsysteme, Stuttgart
Stewart GB III. (1990) The Quest for Value-A Guide for Senior Managers. In:
HarperCollins
Strack R, Franke J, Dertnig S (2000) Workonomics™: Der Faktor Mensch im
Wertmanagement Zeitschrift fiir Führung und Organisation 69, no 5: 283-288
Strack R, Villis V (200 I) RA VE™: Die nächste Generation im Shareholder Value
Management. Zeitschrift für Betriebswirtschaft, 71, 1: 67-84
Strack R, Villis V (2002) RA VE™: Integrated Value Management for Customer,
Human, Supplier and Invested Capital. European Management Journal 20, no
2: 147-159
Strack R (2002) Workonomics: Wertorientierte Steuerung des Humankapitals. In:
Klinkhammer H (ed), Personalstrategie, Neuwied, pp71-90
Wunderer R, Jaritz A (2002): Unternehmerisches Personalcontrolling: Evaluation
der Wertschöpfung im Personalmanagement 2nd edn Neuwied
The Valuation of lntangibles in New Economy
Firms

Peter Witt

1 lntroduction

The valuation of quickly growing start-ups in New Economy industries

like E-commerce, mobile commerce, biotechnology, medical technology
and others has always been a challenge. In particular, it has tumed out to
be difficult to apply traditional valuation techniques, e.g. asset valuation,
discounted cash flow valuations or discounted revenue valuations. Market
valuations - if they were available at all - have not been overly reliable
either as stock prices of New Economy firms have fluctuated strongly in
the last years. 1
In the New Economy (as much as in the old economy), a valuation
based on the assets in place is completely misleading because the firms
under consideration typically have little or no tangible assets in place. The
focus of industries that belong to the New Economy is on innovation and
on intellectual capital. The term intellectual capital depicts knowledge that
can be converted into (positive) future cash flows. 2 Knowledge companies
might own very few tangible assets and still be enormously profitable. 3
Therefore, quantifying intangible value drivers like human capital, tech-
nology know-how, social networks, customer loyalty, and speed of growth
are ofutmost importance in the valuation ofNew Economy firms.
The existence of intangibles is not only a challenge for the valuation of
companies, it is also (and foremost) an accounting problem. Accountants
do not treat knowledge assets or intellectual capital as assets. Therefore,
they are not being capitalized like physical or financial investments.
Although intangibles surpass physical assets in many firms, both in value
andin contribution to growth, they are routinely expensed in a company's
financial reports and hence remain absent from balance sheets. 4

1 See Rudolf!Witt (2002), p. 201-203.

2 See Sullivan (2000), p. 5 and 25.
3 For examples see Stewart (1997), p. 33.
4 See Lev (2001), p. 7.
616

Recently, there have been suggestions to change accounting practices in

order to make intangible assets more visible in a company's reports. 5 In
2001, the FASB (Federal Accounting Standards Board in the U.S.) issued
two new statements touching intangibles. 6 They state that companies
which acquire other companies will no Ionger have to amortize goodwill
because the intangibles that make up goodwill have infinite lives. The new
F ASB rules also call for greater identification and separate accounting for
intangible assets that were formerly treated as goodwill. But to be put in
the balance sheet, these intangibles need to arise from a contractual or
other legal right or they need to be capable of being separated and sold to
qualify as accountable assets. To give examples, this holds for noncom-
petition agreements, secret biotech formulas, unpatented technology and
the like, but not for human capital, first mover advantage or brand value.
Intangible assets that do not comply with the FASB's requirements arenot
allowed to show up in U.S. firms' balance sheets.
German accounting standards are even tighter. Intangibles may only be
treated as assets if they were acquired separately. For all other forms of
intangibles there are no balance sheet entries, no matter if they are capable
of being separated and sold. This directly follows from the German
accounting tradition of protecting creditors and avoiding balance sheet en-
tries for assets that are difficult to liquidate and prone to subjective judg-
ments. Even U.S. advocates of accounting rules that require firms to show
more intangibles on the balance sheet all the time, i.e. regardless of an ac-
quisition, point at significant measurement and definition problems for in-
tangible assets. 7
In this paper, I will first identify growth opportunities, high risks, and
new business models as the major value drivers in the New Economy
(chapter 2). Chapter 3 Iooks at the accounting problern of intangibles , i.e.
at individual knowledge assets in New Economy firms, especially in E-
commerce and biotech industries. Such intangibles to be analyzed in more
detail are innovation capabilities (chapter 3.1), customer binding capabili-
ties (chapter 3.2), organizational capabilities (chapter 3.3), and social net-
working capabilities (chapter 3.4). In this section of the paper, I will also
try to indicate how these individual intangible assets may be evaluated.
Chapter 4 investigates the effect of intangibles on a firm's value, i.e. the
problern of how to value New Economy companies that dispose of one or
many forms of relevant intangibles. The three valuation techniques that

5 See e.g. Thak:er (200 1).

6 FAS 141 on "Business Combinations" and FAS 142 on "Goodwill and Other
Intangible Assets".
7 See Sveiby (1997), pp. 151-162.
 617

will be analyzed in more detail are relative valuations (chapter 4.1 ), cus-
tomer valuations (chapter 4.2) and real option methods (chapter 4.3).
Chapter 5 summarizes and draws some conclusions.

2 Value drivers in the new economy

The New Economy is characterized by a high degree of uncertainty and

fast development. In what some observers have come to call the "informa-
tion age", knowledge and communication have replaced natural resources
and physical labor as the fundamental sources of wealth. 8 Obviously,
knowledge assets can never replace physical assets completely. But their
importance has increased strongly. To give a few quantitative indicators:
Services and knowledge-related high-tech products permanently increase
their share of macroeconomic value creation. Capital spending for infor-
mation machines has risen stronger than any other form of investment.
R&D expenditures contribute significantly higher to the productivity and
the output of firms than other expenditures. 9
An important characteristic of the New Economy is the increasing im-
portance of human capital. For the purpose of this paper, the term depicts
the present value of expected cash flows from employees' contributions to
the company they are working for. Examples ofthe importance ofhuman
capital in the New Economy abound: Biotech firms largely depend on the
qualification and the creativity of their research teams. E-commerce com-
panies are faced with labor markets on which talented specialists like web
programmers and software experts are in short supply.
Human capital is a corporate asset, "but people cannot be owned". 10
Therefore, a central intangible asset of a company is the ability to translate
the tacit and the explicit knowledge of employees into organizational
knowledge and to organize optimal conditions for learning, i.e. the creation
ofnew knowledge. Another intangible asset is a firm's ability to recruit the
most talented and motivated employees which (amongst others) depends
on career and learning opportunities, the reputation or "brand" of the com-
pany as an employer, and the working environment in terms of fun and

8 SeeStewart(1997),p.3-17.
9 For a report on various empirical studies on the effects of the information age
see Lev (2001), pp. 51-77.
Jo Stewart (1997), p. 84.
618

challenges that it offers. Finally, an important intangible asset is a firm's

ability to keep and motivate its best employees. 11
As the discussion so far proves, many New Economy firms largely rely
on intangibles to create value. They have shown much higher market-to-
book ratios than firms from more traditional, old economy industries. 12 But
they have also shown much more volatility in terms of market valuations
and in terms of survival rates. Many of the New Economy firms are start-
ups. They use new, unproven business models, are being financed by
private equity like venture capital, and are typically highly focused. These
firms need to realize high growth rates to reach the minimum efficient size,
to utilize increasing returns to scale and to create value. They pursue high-
risk strategies with accordingly large probabilities of corporate failure. 13
The applicability of traditional, cash flow-oriented valuation techniques
- e.g. discounted cash flow method or discounted revenue method- to New
Economy firms is limited for a number of reasons. Firstly, the estimation
of a time series of cash flows is frequently difficult respectively more or
less impossible. Many business models in industries like biotech, E-com-
merce and medical technology are characterized by huge upfront invest-
ments in customers, R&D, or technology. They produce large losses for a
number of years before positive cash flows show up. Thus, it is hard to
forecast future cash flows and their growth rates. Much of a typical New
Economy firm's value stems from the ''terminal value", i.e. cash-flows oc-
curring in the more distant future that does not lend itself to detailed plan-
ning any more.
The large risk involved with many business models in the New Econ-
omy also makes the selection of an appropriate discount rate difficult. In
traditional valuation techniques, higher risks require higher discount rates
which, ceteris paribus, produce lower valuations. Although the procedure
is perfectly justified for the evaluation of investment projects in large
firms, it is a systematically wrong treatment of risk when it comes to the
valuation of small and quickly growing firms. The cash flow structure of
New Economy firms is asymmetric. Chances for huge future returns have
to be compared with limited downside risk. As a matter of fact, a start-up
can only go bankrupt once. Managers of New Economy firms have the
flexibility to expand, reduce, or terminate investments made earlier. A high
risk of the business model actually increases value instead of reducing it.

11 For a more detailed survey on the management of human capital in New

Economy fmns see Stewart (1997), p. 86-106.
12 See Sveiby (1997), p. 5-8.
13 Fora discussion ofbusiness mode1s and their risks in E-commerce and biotech
firms see Rudolf/Witt (2002), pp. 121-176.
 619

The downside risk is limited whereas the upside potential is not. Real op-
tion theory, which I will look at in chapter 4.3, captures this notion of
managerial flexibility and asymmetric pay-offs from growing, high-risk
firms.

3 The value of individual intangibles

This chapter addresses the valuation of intangibles from the accounting

perspective, i.e. investigates potential ways of evaluating individual
knowledge assets. The analysis is restricted to four (to my mind important)
categories of intangibles: innovation capabilities, customer binding capa-
bilities, organizational capabilities, and social networking capabilities. In
selecting these four categories, I deliberately deviate from the focus of
existing research on the subject. 14 It further needs to be mentioned, that the
paper does not intend to produce a comprehensive list of knowledge assets
in New Economy firms and their value. 15 It will also make no contribution
to the regulatory problern which intangibles should be included in the
balance sheet and which ones should not.
The disclosure of intangibles in company reports may be beneficial for
shareholders because it helps them to better judge the company and its
value driver. But it also raises a number of important legal and economic
questions. First of all, the information revelation principle of economics
suggests that there are enough incentives for voluntary disclosure of intan-
gibles and their value if there is demand for this type of information. Thus,
there would be no need for regulation and mandatory disclosure. 16 From a
game theoretic perspective, the information revelation principle could fail
if intangible assets represent intellectual secrets such that disclosure would
benefit competitors. From a legal point of view, the disclosure of intangi-
bles (mandatory or voluntarily) increases the danger of shareholder law-

14 Sveiby (1997, pp. 8-11) distinguishes three categories of intangibles: intemal

structure, extemal structure, and employee competence. Stewart (1997) divides
intangibles into the three classes: human capital, structural capital, and
customer capital. Sullivan (2000, pp. 17-18) also uses three, although slightly
different categories: human capital, intellectual assets, and intelleemal property.
15 In the U.S., the FASB has tried to do so. It exhaustively lists intangibles in
seven categories: technology based values, customer based values, organization
based values, market based values, employee based values, contract based
values, and legally based values.
16 See Lev (200 1), p. 86.
620

suits in cases when the subjective evaluation of intangibles later proves to

have been misleading.

3.1 The value of innovation capabilities

All New Economy firms operate under market environments that force
them to innovate. Therefore, they need innovation capabilities like R&D
capacity, information on customer preferences, and technological know-
how. Some ofthe process or product innovations that firms in New Econ-
omy industries create are patent protected. Patent protection conveys the
right to exclude other firms from unauthorized use of the intelleemal assets
contained in the patent but also to generate revenues from the exclusive
use of a technology. 17 Some innovations do not get patent approval
although the company filed one. In other cases, the innovating company
did not try to file a patent application although it could have tried to do so.
Finally, some technologies simply are not patentable and need to be pro-
tected by other means.
Patents are difficult to value directly because they are production factors
(input) and need to be translated into cash flow equivalents. To give an
example, a patent on a biotechnologically produced substance may lead to
a blockbuster drug after clinical trials and approval, but it is still hard to
pin down the net asset value at the time of patent approval. In practice,
four indicators for the value of patents have been developed: the amount of
R&D spent to generate the patent, the patent's replacement cost, the
patent's perceived market value, and the net present value of the income
from the patent.
Whereas the first two measures are input-related and thus not directly
linked to future economic benefits from the patent, the second two directly
try to measure contributions to firm value. This is a difficult task as the
time between patent approval and market entry with a related product may
be long. 18 When the patent protects a production process, its benefits may
need to be measured not in additional revenues but in cost savings. Indirect
effects on future cash flows, e.g. know how generated from a patent that
never produced any new product or process but nonetheless made other
innovations possible, are also difficult to incorporate in the valuation.

17 See Sullivan (2000), p. 52.

18 Austin (1993, p. 256) empirically studied the relation between patent approval
and stock price increases for U.S. biotechnology firms and found that patents
readily identifiable with end products tend to be more valuable than average
patents.
 621

Although patents are certainly an important outcome of innovative

activity, the valuation of patents alone may not correctly depict the inno-
vative capabilities of a firm. Some technologies and services are not
patentable, e.g. web design, tumor vaccination, or consulting services.
Others could no doubt be filed for patent application, but firms deliberately
decide not do so because the technology development is too fast, too much
know-how would leak out when filing the application, or the cost of the
procedure is deemed too high. Under these circumstances, the innovation
activities of the firms produce no measurable output in the sense of an
intellectual property right. Therefore, the innovation capabilities of a firm
need to be also measured by other indicators, e.g. the proportion of new
products in annual sales. The same type of measure could be applied to
new processes and their share in a firm's production technology.
Admittedly, a number of valuation problems remain. It is difficult for
outsiders to make a judgment on the share of new products and processes.
lt is also hard to distinguish between innovation output of different quality.
To give an example, only a few of all the new substances of an innovative
biotech company may turn outtobe successful and to add value. Finally, it
may also not be possible to find a reliable indicator for the innovative
know-how that was derived from failed projects, from unsuccessful intro-
ductions of new products, and from very basic research.

3.2 The value of customer binding capabilities

For all companies, customers are the source of revenue. The more cus-
tomers a firm has and the more these customers buy, the higher is a firm's
va1ue (given that it operates with positive margins). Therefore, the ability
of a company to attract customers and to make them repeat customers, i.e.
to lock them in, or in more friendly terms, to bind them, is a value gen-
erating capability. One of the most important intangible assets that help a
New Economy firm to attract and to retain customers is its brand. Brands
confer economic benefits like pricing power, distribution reach, and the
ability to launch new products. Those benefits produce higher returns on
tangible assets than unbranded competitors can realize. Existing studies
have tried to put a monetary value on a company's brand, mainly by esti-
mating the after-tax excess returns (in comparison to competitors from the
same industry) that a firm earns on its tangible assets per year and then
calculating a net present value of these excess returns over a forecasting
period. 19

19 See Stewart (1997), pp. 227-228 for an example.

622

In some industries, especially in markets for network products, being a

first mover and growing the customer base more quickly than competitors
is more important than actual customer numbers and current revenues per
customers. The reason is that customers stay with a firm once they have
made a few purchases and that new customers prefer to go to the largest
competitor because it offers the biggest benefit in terms of installed cus-
tomer base. Thus, a network product is characterized by the fact that the
value of buying and using this product for one individual customer is in-
creasing in the number of other users. Typical examples of firms from the
New Economy offering network products are intemet auction sites like
eBay.com and trading platforms for used cars like mobile.de. As the
installed base of customers of such companies grows, more and more users
find the service attractive and also become customers. The key challenge
for the company is to be the first mover and to obtain a critical mass of
customers. 20
If products exhibit network properties first movers will have advantages
over early and late followers. This is especially true if leapfrogging, i.e.
surpassing competitors that started earlier and currently have more cus-
tomers, is difficult or impossible. Therefore, the first mover will de-
liberately invest huge amounts in new customer acquisition and lose
money in the short run. The strategic goal is to recoup these investments
from sales to a very large group of customers in later stages of the com-
pany's development. In their most extreme form, markets for network
products are "winner-take-all-markets" in which the first mover van-
quishes all competitors and becomes a natural monopolist. With respect to
company valuation, this has important implications. In the early phases of
growth, customer acquisition costs are high, and prices need to be low (or
even zero). Therefore, there are only little or no cash inflows in the early
periods and increasingly large cash flows in later stages of the company's
development, but only for the winners.
In practice, New Economy firms and analysts have so far only tried to
evaluate single facets of a firm's customer binding capabilities. Apart from
the examples of brand valuation mentioned earlier, measurement efforts
focused on customer satisfaction, customer acquisition cost, and customer
loyalty. 21 Network effects have been very well analyzed in theoretical stud-
ies,22 but there have not been too many examples of winner-take-all-mar-
kets in the New Economy. eBay.com is the only crystal clear role model I
am aware of. It has shown exceptional growth rates in the past, triggered

20 See ShapiroNarian (1999), pp. 13-17.

21 See Stewart (1997), pp. 240-243 and Lev (2001), pp. 67-72.
22 See ShapiroNarian (1999), chapter 7.
 623

by starting to offer its service for free, reached a market share in online
auctions of more than 80 percent in a number of countries world-wide, and
is currently one of the few profitable E-commerce companies.

3.3 The value of organizational capabilities

Although human capital is an important component of many capabilities

and intangible assets in New Economy firms, its influence is especially
pronounced on the efficiency of the organization. Large parts of what
firms can and cannot do stem from their organizational culture. The con-
cept itself is difficult to measure empirically but few observers deny its
importance for firm value. Organizational capabilities in the sense of work
processes, hierarchical structures, and corporate culture are difficult to
copy and transfer to competitors. They are "tacit" in the sense that these
capabilities cannot be written down or communicated easily to others. The
reason is that individuals accumulate tacit knowledge mainly through
direct "hands-on" experience. 23
A number of existing studies on intangibles have made general sug-
gestions for the evaluation of human capital and employee competence. 24
Firstly, there is an empirically significant relation between high morale
amongst employees and superior financial performance of a firm (although
the causality could go both directions ). There is a similar relation between
employees' attitudes and customers' attitudes, i.e. evidence of a direct
interplay between the two intangible assets human capital and customer
capital. 25 Positive effects on human capital and employee competences
were also found for total quality management programs, teamwork train-
ing, and profit-sharing systems. 26
Besides these general concepts, there are also proposals to calculate
very concrete ratios and to translate them into an aggregate valuation of a
firm's organizational capabilitiesY On the input side, training and educa-
tion costs measure a firm's investmentsinhuman capital. The number of
years in the profession and the level of education serve as proxies for an
employee's individual qualification. The average age and the turnover of
employees are potential indicators for the stability of employees' compe-
tences. Similarly, the "rookie ratio" determines the share of people with

23 See Nonak:a (1994), p. 21.

24 See e.g. Sullivan (2000), pp. 173-205 and Stewart (1997), pp. 84-106.
25 See Stewart (1997), pp. 229-235.
26 See Lev (2001), p. 74.
27 See Sveiby (1997), pp. 168-184.
624

less than two years' employment and tak:es it as a proxy for low efficiency
and little stability.
In rapidly changing environments, keeping competitive advantages not
only means protecting the organizational structures and processes. lt also
requires the development of new competences and the implementation of
organizational learning. In particular, New Economy firms need respon-
siveness to rapidly changing market and technology requirements, an
intangible asset that has been labelled as a "dynamic capability". 28 The
term "dynamic capabilities" describes a company's ability to integrate,
build, and reconfigure intemal and extemal competences to address rapidly
changing environments. Firms mainly develop dynamic capabilities by
recombining their current capabilities. 29 Flexible organizational forms have
found to be supportive for employees' leaming and innovation skills. 30
Although the concept of dynamic capabilities is certainly a relevant and
broadly accepted theoretical framework, attempts to develop instruments
for practical measurement are still in their infancy. This is mainly due to
the fact that the valuation of static capabilities and intangible assets is
already dif:ficult and not yet well established. Innovative and reactive
learning capabilities are even more dif:ficult to measure. Furthermore, the
valuation of dynamic capabilities goes beyond an analysis of the firm un-
der consideration. It has to tak:e changes in the competitive environment
and extemal technological developments into account as well.

3.4 The value of social networks

Quickly growing New Economy firms, in particular start-ups, face two

competitive disadvantages. They are new to the market and lack reputation
with customers, suppliers, and potential employees (liability of newness ).
They are also small, i.e. typically do not dispose of an ef:ficient minimum
size (liability ofsmallness). Despite these generalliabilities, some start-ups
grow successfully to become profitable and well-established firms in their
markets. One explanation for this phenomenon is the theory of social
interaction and social networks in business that has been formulated by
social science and entrepreneurship research. The theory postulates that
some firms are better equipped than others with a special intangible asset,
and that is social networks.

28 See Teece/Pisano/Shuen (1997) and Eisenhardt/Martin (2000).

29 See Kogut/Zander (1992).
°
3 Fora comparison of different flexible organizational designs see Witt (1997),
pp. 379-393.
 625

The "network success hypothesis" 31 assumes a positive relation between

the networking activities of founders, the size of their personal social net-
work and their start-up's success. The rationale behind this hypothesis is
the theory of socially embedded ties that have advantages over market
relations that people could engage in without the need for any prior social
relation. Network contacts, i.e. socially embedded ties to other people,
allow entrepreneurs to get resources eheaper than they could be obtained
on markets. Examples are spouses that work in the company for less than
market wages and consulting services by experienced managers that the
entrepreneur gets for free. Furthermore, networks also enable entrepre-
neurs to secure resources that would not be available on markets at all, e.g.
reputation from famous board members, customer contacts from business
friends etc.
Personal social networks are intangible assets that founders own and
utilize for their companies. Over time, the personal networks of entrepre-
neurs are transformed into social networks of the firm in the sense of or-
ganizational communication links to other institutions and authorities.
They are difficult to determine and to visualize, but directly contribute to a
firm's value. They do so by giving a company the opportunity to get valu-
able resources like information on new markets and technologies that can-
not be bought. A typical example is the co-operation of biotech start-ups
and university institutes where the founders used to work before they
started their own business. Networks also enable firms in the New Econ-
omy to get resources at lower costs than market prices, e.g. if an E-com-
merce start-up gets advertisements in large newspapers for free because
one of the investors is a publishing house.
Recent work on the social networks of entrepreneurs and small firms in
New Economy industries has tried to develop analytical approaches to
value personal networks and their effects on firm performance. 32 Starting
out with a "map" of the personal network of the lead entrepreneur, one
needs to evaluate the relevance and the accessibility of individual network
partners. Both factors determine potential monetary benefits from the net-
work. In addition, one needs to estimate the cost of acquiring new network
partners and keeping existing ones. The difference between the benefits
and costs of social networks then indicates their value for the firm. If there
is more than one founder, the social networks of the other team members
must be evaluated in the same way. Overlaps in networks of different per-
sons may be eliminated if more than one individual contact has zero mar-
ginal value for the company. The theory on how social networks impact

31 See Brüderi/Preisendörfer (1998), p. 213.

32 See Witt (1998) and Witt/Rosenkranz (2002).
626

firm value is admittedly in its infancy. But first empirical studies indicate
that the intangible asset "social networks" is important for the success of
companies in the New Economy and thus also for their valuation. 33

4 Valuation models for firms using intangibles

In recent years, some new or "modern" valuation techniques have been

proposed for New Economy firms. 34 They all realize the importance of
intangibles for a firm's value but focus on different value drivers and
choose very different methods to quantify the effects of intangibles on a
company's future cash flows. Therefore, not all new valuation techniques
are equally well suited for all kinds of New Economy frrms. Besides, not
all valuation methods classified by their authors as "new" or "modern" live
up to their promises when being analyzed in their methodological details.
In the following, I will focus on only three "modern" approaches to corpo-
rate valuation. 35

4.1 Relative valuations

In a relative valuation, better known as multiple method, the value of a

company is derived from the market prices of comparable companies, so
called peer groups. Market prices may be stock prices of stock exchange
listed companies from the same industry or prices being paid in transac-
tions with comparable firms. The valuation takes place by comparing
known values for variables such as eamings, revenues, or more technical
indicators like patent portfolios, drug development pipelines, or page
impressions. The assumption behind this comparison is that other firms in
the industry are comparable to the firm being valued and that the capital
market prices these firms correctly. 36
Classical multiples like the price-eamings ratio, the price-sales-ratio, or
the market value to replacement value ratio do not fit New Economy firms
very weil. As discussed in chapter 2, many quickly growing firms in
industries like E-commerce and biotech do not have any sales, le:ft aside

33 See Witt/Rosenkranz (2002), pp. 95-98.

34 For an overview of traditional and modern valuation techniques for growing
firms see Rudolf/Witt (2002).
35 The reason is simply that I deem them to be better suited for the valuation of
New Economy frrms than other "modern" techniques.
36 See Damodaran (1994), p. 15-16.
 627

any eamings. Due to their reliance on intangibles, the replacement values

are marginal. So the classical multiples simply arenot applicable. None-
theless, such companies may have high market values as the example of
the German start-up Alando.de shows. The company was acquired in 1999,
six months after foundation and still in the pre-revenue, pre-eamings
phase, by eBay. com for several million Euros.
To be able to value New Economy firms in early stages of their de-
velopment, other multiples have been suggested. 37 They use variables that
are more or less closely related to a company's financial success and easy
to measure. For growing E-commerce firms without sales and profits,
"webmetrics", i.e. specialmultiples like page impressions per month, "new
eyeballs", or reach indicators have been developed. For biotech compa-
nies, there are also special multiples that try to measure easy to obtain per-
formance indicators of the business model. Examples are product pipe-
lines, patent portfolios, and the number of licensing agreements.
Relative valuation techniques are so enormously attractive for practitio-
ners because they appear to be easy to handle. They are also attractive
theoretically, because they use the fact that (information efficient) capital
markets translate all available information on companies and their intan-
gibles into stock prices. lntangibles do not have tobe evaluated separately,
accounting information does not matter. From a methodological point of
view, relative valuations are by no means "modern". It can be shown that
common multiples like the price-eamings-ratio are nothing more than sim-
plified versions ofthe discounted revenue method. 38
But the major deficiencies of relative valuation methods stem from their
assumptions. One may not be able to fmd comparable firms to calculate
the multiples. In fact, the more innovative a firm or its industry is, the less
likely we find comparable companies with market prices. In strict theoreti-
cal sense, no firm is comparable at all, simply because they all have dif-
ferent employees, different entrepreneurial teams, and different corporate
cultures. It is also unclear theoretically, if capital markets are information
efficient or not. lf they are not, valuation errors in the market (over- and
undervaluation) will be transferred to the valuation ofNew Economy firms
from the same industries. Finally, the New Economy specific multiples
frequently show too much variance and too little relation to financial suc-
cess measures like cash flows to be reliable indicators of corporate value.

37 See Rudo1f!Witt (2002), pp. 179-199.

38 See Damodaran (1994), pp. 198-202.
 628

4.2 Customer valuation models

Customer valuation is directly based on the notion of existing customer

bases as intangible assets. The method has mainly been developed for the
valuation of E-commerce firms that heavily invest in the acquisition of
new customers. They try to utilize network effects and lock in customers,
i.e. prevent them from switching suppliers.
In a simple version, the customer valuation method relates all future
cash flows, costs, and revenues to customers. lt then separates three differ-
ent sources of value: the present value of sales to existing customers, the
present value of future sales to new customers, and the value of the real
option to sell other products and services to those customers that were at-
tracted by the existing product range. A well known example is the intemet
retailer Amazon.com that was established in 1995 to sell books online. The
company later expanded to also sell compacts disks, electronics, and other
products on its web page. Given the acquisition cost per customer, the net
profit margin per purchase, the average purchase volume per year, the
retention rate per customer per year, and an appropriate discount rate, we
can calculate the lifetime (net present) value per customer. It is also possi-
ble to see how increases in tax rates, marketing costs, retention rates etc.
would affect the company's value.
In a next step, the evaluator tries to determine how many new customers
Amazon.com can acquire per year given its marketing strategy, the strength
of its brand, and the recommendations from existing customers. This
figure Ieads to a second (net present) lifetime value, that for new cus-
tomers. The higher the company's capability to attract new customers and
to bind old ones, the higher its corporate value. Finally, one needs to
evaluate the value of the option to sell new products and services to cus-
tomers and thus to increase the average cash contribution per customer per
year. In the case of Amazon.com, evaluations were aiming at values
generated from extending the firm' s business model from books to music,
electronics, and other products. Economically speaking, the intangible as-
set under consideration here is the firm's capability to profitably cross-sell
its products to the existing customer base.
An alternative procedure to calculate customer value are binomial
models. 39 They capture the idea of asymmetric pay-off structures for
quickly growing New Economy firms by modeling the possible future de-
velopment of the customer base in a binomial tree including the bank-
ruptcy option. The binomial model has only two possible developments of
a variable from one point in time to another: up or down. Assuming given

39 See Krafft!Rudolf/Rudolf-Sipötz (200 1).

 629

probabilities for up and down movements of the number of customers and

assuming that the cash flow contribution per customer depends on the size
of the installed customer base, i.e. that network effects prevail, the bino-
mial tree produces different seenarios for the cash flow generation over a
number of periods. The seenarios are not symmetric because after a num-
ber of downward movements, the number of customers is zero. That means
bankruptcy so the number cannot decrease any further in the future. On the
other hand, after a number of upward movements for the number of cus-
tomers, the cash contribution increases exponentially. A backward solution
of the tree and discounting the respective cash flows Ieads to the expected
firm value.
Just like relative valuations, customer valuation methods contain no
"new" methodology. The only innovative thing they do is to categorize all
estimated future cash flows into different sources, in particular the value
from existing customers, the value from future customers, and the option
value extracted from selling new products to present and future customers.
All three components are based on discounted cash flow calculations.
Therefore, the customer valuation technique described above is formally
equivalent to traditional discounted cash flow-valuations. Its main advan-
tage is to make it easier to forecast future cash flows because all those
estimates are being related to more concrete indicators like the number of
customers, the average revenue per customer per purchase etc. From a
strict methodological standpoint, a customer valuation cannot give any
insights on the problern of valuing intangibles that traditional valuation
methods could not come up with as well.

4.3 Real option valuation models

Real option theory takes into account that managers of quickly growing
firms typically do not simply wait for their investment decisions to turn
into cash flows. They can manage risk. They also have different asset
Options to influence the development of investments and to increase the
value of their company. The notion that managerial flexibility from asset
options has a value in its own right is the theoretical core of real option
models for company valuation. In this paper, I cannot give a complete and
comprehensive overview of the various valuation techniques for real
options. Instead, I will try to point out the main potential and the Iimita-
tions ofreal option thinking in the valuation ofNew Economy firms, espe-
630

cially those firms that largely (or completely) derive their value from
intangibles. 40
Real asset Options give a company's management various types offlexi-
bility. Important examples are abandonment options, options to defer
investment, options to expand, and switching options. 41 In biotech firms,
the option analogy is most obvious. The biotechnological drug develop-
ment process proceeds from substance identification over pre-clinical and
clinical testing to approval from government authorities, e.g. the U.S.
Federal Drug Administration (FDA). The successful completion of one
step in the process gives a company the option to further pursue own de-
velopment activities with existing capacity, to license the project to a large
pharmaceutical company, or to invest in new R&D capacity. In cases of
expected major legal changes (e.g. with respect to the admission of stem
cell research), biotech firms have the option to wait with new R&D pro-
jects until uncertainty resolves.
Real option valuation is important as a qualitative concept to indicate
the potential effects of managerial flexibility on corporate value. But due
to a number of serious methodological problems conceming the existence
of an underlying with market prices, exclusive option rights, and the inter-
action of multiple options, the real options approach will not be applicable
quantitatively for some New Economy firms. This is especially true for
companies in the early phases of their development process.

5 Summary and conclusion

This paper has investigated intangibles in New Economy firms, especially

in the E-commerce and the biotech industry. Four major groups of intangi-
bles were identified: innovation capabilities, customer binding capabilities,
organizational capabilities, and social networking capabilities. From an ac-
counting perspective, the problern is to evaluate these knowledge assets
individually and to decide whether or not they should be entries in balance
sheets. 42 From a shareholder's or manager's perspective, the more interest-
ing question is perhaps how the business models in the New Economy and

4° For an in-depth theoretical analysis of real options see Dixit/Pindyck ( 1994).

An easier to read practitioner's guide to real options is Copeland/Antikarov
(2001). For an application ofreal option valuation to New Economy firms see
Rudolf/Witt (2002), pp. 203-224.
41 See Copeland/Koller/Murrin (1994), pp. 474-477.
42 The most comprehensive source for answers to this problern is Lev (200 1).
 631

the intangibles needed for their realization affect corporate valuation. 43

This paper has taken a closer look at three "modern" valuation techniques
that aim to do just this: relative valuations, customer valuations, and real
option valuations.
A first conclusion from this paper is that traditional valuation techniques
like the discounted cash flow method remain important tools in the New
Economy. They are essential components of many modern methods like
the customer valuation technique and the multiple method. A second con-
clusions relates to market values. Whenever capital markets are able to
translate information on companies, their business models, and their intan-
gibles (at least partly) into stock or transaction prices, corporate valuation
can benefit from the existence of this information about the market price of
comparable companies. A third conclusion is that option analogies are
fruitful for the analysis of quickly growing companies in the New Econ-
omy. In particular, the real option approach captures a very important
value driver, namely that managerial flexibility leads to asymmetric cash
flow expectations. I personally subscribe to the notion, that "at the begin-
ning of the new millennium, the application of real options is just begin-
ning to take off." 44
A final conclusion from the analysis above is that there are no clear-cut
valuation techniques that are applicable to all quickly growing New Econ-
omy Firms. We need to distinguish industries, business models, and strate-
gies. The main challenge in this arena is to use the theory of the firm (in
Gutenberg's andin modern versions) to understand business models, iden-
tify intangibles as value drivers, and to use valuation techniques that
incorporate the effects of both on future cash flows.

References

Austin D (1993) The Va1ue of Intangib1e Assets. An Event-Study Approach to

Measuring Innovative Output: The Case of Biotechno1ogy. American Eco-
nomic Review 83:253-258
Brüderl J, Preisendörfer P (1998) Network support and the success of new1y
founded businesses. Small Business Economics 10: 213-225
Copeland T, KollerT, Murrin J (1994) Valuation. Measuring and Managing the
Value ofCompanies, John Wiley & Sons, New York et al.
Copeland T, Antikarov V (2001) Real Options. Texere Publishing, New
York/London

43 For more detailed answers to this question see Rudolf/Witt (2002).

44 Copeland/Antikarov (2001), p. 23.
632

Damodaran A (1994) Damodaran on Valuation. Security Analysis for Investments

and Corporate Finance. John Wi1ey & Sons, New York et al.
Dixit A, Pindycke R (1994) Investment under Uncertainty. Princeton University
Press, Princeton
Eisenhardt K M, Martin J A (2000) Dynamic capabilities: Wbat are they? Strategie
ManagernentJournal21: 1105-1121
Kogut B, Zander U (1992) Knowledge ofthe firm, combinative capabi1ities, and the
replication oftechnology. Organization Science 3: 383-397
Kraffi M, Rudolf M, Rudolf-Sipötz E (200 1) Valuation of customers in growth
companies- a scenario based model. Working Paper WHU, Vallendar
Lev B (2001) Intangibles. Management, Measurement, and Reporting. Brookings
Institution Press, Washington
Nonaka I (1994) A Dynamic Theory of Organizational Knowledge Creation.
Organization Science 5: 14-37
RudolfM, Witt P (2002) Bewertung von Wachstumsunternehmen. Gabler-Verlag,
Wiesbaden
Shapiro C, Varian H R (1999) Information Rules. A Strategie Guide to the Net-
work Economy. Harvard Business School Press, Boston
Sveiby K (1997) The new organizational wealth. Berret-Koehler Publishers, San
Francisco
Stewart T (1997) Intellectual Capital. The new wealth of organizations. Double-
day/Currency, New York et al.
Sullivan P H (2000) Value-Driven Intellectual Capital. John Wiley & Sons, New
York et al.
Teece D, Pisano G, Shuen A (1997) Dynamic capabilities and strategic managernent.
Strategie Management Journal 18: 509-533
Witt P (1997) Reorganisations to increase entrepreneurial flexibility. Jornal of
Enterprising Culture 5: 375-402
Witt P (1999) Information networks of small and medium-sized enterprises. Journal
ofEnterprising Culture 7: 213-233
Witt P, Rosenkranz S (2002) Netzwerkbildung und Gründungserfolg. Zeitschrift für
Betriebswirtschaft 72, Special Issue 5: 85-1 06
ValueReporting TM
New Trends in Corporate Reporting 1

Joachim Wolbert

1 The earnings game

Events in the US over the last year will have a long Iasting effect on corpo-
rate behaviour and the accounting profession. A major part of the problern
clearly stems from what many analysts, investors and indeed managers
themselves believe is an undue focus on short-term earnings (the so called
"Eamings Game"), which can Iead to manipulation and divert attention
from sustainable value creation. The rebuilding of public trust has already
started, but the concem today is that we are overlooking the root of the
problern - and that is keeping the corporate reporting model focused pri-
marily on short term financial information and, in so doing, providing an
opportunity for the earnings game to resurface unchecked during the next
bull market.
The debate therefore needs to focus on the adequacy of today' s
reporting model. We should be questioning how investors, customers and
employees can make an objective assessment of how a company is being
run as weil as its future prospects.
What is the shape of the information that they need in order to be able to
differentiate good management from bad, Juck from skill? In a recently
published book, "Building Public Trust", PricewaterhouseCoopers sets
forth its vision of the corporate reporting model. The Three-Tier Model of
Corporate Transparency offers a new vision of corporate reporting.
• Tier One: Traditional financial Statements prepared in accordance
with globally accepted accounting principles (Global GAAP)
• Tier Two: Industry-specific standards for measuring and reporting
performance, consistently applied
• Tier Three: Company-specific information including strategy, plans,
risk management practices, compensation policies, corporate
govemance and performance measures unique to the company.

1 This article is based upon the publications of PricewaterhouseCoopers

indicated under the "References" listed at the end.
634

1.1 Tier one: global generally accepted accounting principles

Today, every country with a public capital market uses its own set of
generally accepted accounting principles (GAAP) or one based on other
national or international standards. The quality of these standards varies
widely from country to country, as does how well they are applied by
companies and audited by independent accountants.
The creation of a set of global generally accepted accounting principles
- Global GAAP - would allow companies to access the world' s capital
markets more easily and with less cost. If Global GAAP existed, investors
could more easily and accurately compare the historical performance of
any company, in any country, thus broadening their range of investment
choices. Global GAAP would require addressing interpretation of, and
compliance with, standards on a global basis.
Even Global GAAP standards would come under criticism, since they
are based for the most part on an historical cost model. Today's complex
business environment challenges the relevance of historical cost informa-
tion and is putting strain on this model.
In order to meet the informational needs of investors, companies should
supplement GAAP with other financial and non-financial information
regarding past performance and future prospects of value creation.

1.2 Tier two: industry-specific standards

Investors need the ability to compare companies in any given industry in

ways that go beyond required financial reporting. In order to meet these
information needs, companies should supplement GAAP with other, value
based financial and non-financial information regarding past performance
and future prospects. Companies also need to compare their own per-
formance with that of industry peers. Therefore, it is in their own best
interest to voluntarily and collaboratively develop industry-wide standards
based on critical value drivers unique to their industries.

1.3 Tier three: company-specific information

Even if investors had all the information provided at the first two tiers,
they would still need additional information specific to individual compa-
nies. This information would include management' s view of the competi-
tive environment, company strategies, unigue company value drivers and
the company's commitments to other stakeholders. Such company-specific
information forms the basis of Tier Three. While well-defined extemal
 635

Standards cannot be developed for Tier Three content, general guidelines

for content, as weil as extemal standards for the format of reporting such
information, certainly could be developed.
These three tiers should not be viewed as separate, discrete reporting
Ievels. Rather, they should serve as an integrated model for improving
corporate transparency overall. There is, however, a particularly close link
between Tier Two and Tier Three.

2 Contents of ValueReporting™

In attempting to define ValueReporting™, it is worth to say what it is not.

ValueReporting™ is not advocating that companies:
• Place a value on intangible assets (eg customers, people, innovation),
putting them in a balance sheet.
• Use the framework to reconcile or explain the difference between net
assets and market capitalization.
• provide forward-looking information in the form of financial projec-
tions.
ValueReporting™ is trying to create a bridge between management
information used intemally to manage the business and the information
which drives extemal investment analysis.

3 Structure of ValueReporting™

To achieve greater transparency, a challenge awaits management: re-

porting, in a logical and organized way, all the information it has decided
to disclose to shareholders and stakeholders. Stakeholders are not data
miners. They require and deserve clear, logically presented information.
The ValueReporting™ Framework (Exhibit 1) offers a solution that
companies can use for organizing the information they report to Stake-
holders. A codification of PricewaterhouseCoopers' extensive global
industry research, it provides a comprehensive means of structuring
intemal and extemal reporting.
636

3.1 Exhibit: The ValueReporting™ framework

• Competilive • Goalsand Objectives • Cuslomeß • Financial Position

Environment • OrganisationalOe ign • Peop l • Risk Prolile
• Regulatory • Govemance • lnnovalion • Economic Perlormance
Environment • Brands • Segmental Analysis
• Macr<H<:OnOmic • Supply Chain
Environment
• Environmenllll, Social
and Ethica l

Fig. 1.

The Framework presents four basic categories of information. Together

they create a coherent and complete medium-term picture of a business,
against which short-term financial performance can be explained. The
Framework builds on a number of underlying principles, primarily the
notion of transparency.
The four categories of the Framework link to and build upon one
another. Forthis reason, the Framework should not be viewed as a static
medium for presenting discrete bits of information. Used properly and
integrated into internal management processes, it becomes a dynamic tool
for assessing and monitoring all key aspects of performance and for com-
municating publicly their contribution to value creation. Few, if any, com-
panies have fully adopted all ofthe ValueReporting™ concepts. However,
a small but growing number of companies have begun to embrace its prin-
ciples, and some instructive examples within a specific area of the Frame-
work can be found worldwide. Many examples appear in the ValueRe-
porting™ Review, published annually by PricewaterhouseCoopers. Of
course, transparent reporting about a particular piece of information is not
the same thing as a high overall Ievel of transparency. Nor is it the same
thing as creating value for shareholders, what is ultimately based on
performance, not transparency per se.
 637

3.2 Market overview

In this category, management reports its assessment ofthe competitive and

regulatory environment, including the broader macro-economic environ-
ment in the geographies where it does business. This type of information
varies by industry and by individual company.
While most of this information comes from sources outside the com-
pany, management must shoulder responsibility for ensuring its accuracy
and specifying sources. When the information comes from within the
company and represents management's assumptions and calculations, this
too should be made clear. Naturally recognizing that assumptions and
judgment are involved, stakeholders want to know management's view so
that they can go on to form their own opinions about the extemal environ-
ment and whether management has been too optimistic or pessimistic
about it.

3.3 Value strategy

Against the context of this defined competitive landscape, management

need tobe able to articulate its strategy for competing within this market-
place, ideally supported by quantified medium-term targets and relevant
milestones. The strategy should be the blue print that determines the
chosen mix of risk, retum and growth which management has chosen for
the business. The mix ofrisk (eg cost of capital) and retum (eg cash flow)
should be understood and visible for each business segment, as weil as for
the corporate centre and the group as the whole. Furthermore, management
should be clear to distinguish the portion of their growth strategy that will
be achieved organically versus the portions requiring acquisitions.
lnextricably linked to the strategy are the underlying organisation and
govemance structures that determine how the company is managed and
how the board exercises its responsibility.

3.4 Value creating activities

Having communicated a well-defined strategy that includes the goals and

objectives for those activities that drive value and provide competitive
advantage, management should clearly communicate at a more detailed
Ievel the actions to invest in, and manage, these key activities.
Businesses today must carefully manage a complex web of relationships
to ensure their success. This includes the ability and commitment of their
people to deliver against strategy, the degree to which customers trust their
638

products or services, their relationships with a wider stakeholder group,

and their capacity to share knowledge, to learn and to adapt. By
understanding and reporting on how it is enhancing these areas of business
activity, and by providing a sense of the company's health through the
metrics and information provided, management can help the investment
community assess both current performance and likely future outcomes.
The assumption on which the ValueReporting™ model is based is the
objective of management of delivering long-term growth in shareholder
value. The key to the model, however, is to look beyond the high level
measures, such as Total Shareholder Return (TSR) or EVA® (Economic
Value Added®), to the activities within the company that create value. The
ValueReporting™ framework encompasses value-based measures re-
garding innovation, brands, customers, supply chain efficiency and people.
The key is not only to drive the strategy by an understanding of these ac-
tivities, but to manage the business in a consistent way. The net result
being that the communication of value to the investment community
becomes a consistent and logical extension ofthis intemal process.
lntemal information about the real value drivers of a company serves as
the basis both for effective management and for appropriately transparent
extemal reporting. Management should measure and manage those things
that create value and help the companies to meet the legitimate expecta-
tions of its stakeholders. In practice, this elementary advice can be difficult
to follow across the full complexity of a company. Doing so requires clear
articulation of a company 's strategy at both the corporate and business unit
levels. It also requires identifying the value drivers that must be measured
and managed. The concept of value drivers and the balanced scorecard has
been around for some time; yet, however compelling in concept, it remains
in practice challenging to implement. It is our experience that the informa-
tion appearing in the majority of boardrooms remains predominantly
financial in nature. Without information on these value creating activities,
management are typically flying blind - when financials tell them there is
a problem, management have already missed the optimal point for taking
appropriate corrective action.
A deep understanding of how value is created is clearly important for
executives. A PricewaterhouseCoopers Management Barometer Survey
asked 156 executives about their interest in 15 business topics for future
research. The first choice was business performance measures for building
models that incorporate financial and non-financial information. 2 This is
not a trivial finding. Quite a few companies have already attempted to

2 Unpublished Survey conducted by PricewaterhouseCoopers in the first quarter

of2001.
 639

trace value creation from cause to effect - from value driver to measurable
value result. In another survey conducted by PricewaterhouseCoopers in
the US about Executives' perceptions of different value drivers, 69 percent
of the executives have attempted to demonstrate empirical cause-and effect
relationships between the different categories of value drivers and both
value creation and future financial resuits. Less than one-third feit that they
had truly completed the task. This reflects the difficulty of the task. Still
more challenging, however, is to combine numerous cause-and-effect
relationships into an overall business model that maps the holistic relation-
ships across a set of different value drivers. Here 61 percent had made at
least a modest attempt, but only 10 percent feit they had really succeeded.
Establishing cause-and effect relationships helps enormously when
attempting to construct a business model. The survey mentioned above
compared companies that reported progress on causal linkages and estab-
lishing holistic business models to companies that reported making little or
no progress in this regard. The former group of companies reported a
higher five-year historical growth rate in revenues. Companies that have
grasped how to measure and manage their value drivers do a better job of
delivering top-line revenue growth.

3.5 Financial performance

This category of the ValueReporting™ Framework is, of necessity, a key

focus for management and for investors. Whether or not a company is able
to deliver financial performance in line with expectations, however, is a
function of the success with which management has understood its market,
developed and executed its strategy, managed performance and ultimately
communicated all the information in the ValueReporting™ Framework to
the market.
Put another way, it is this category of the Framework where the ele-
ments of risk, retum and growth come tagether in the form of financial
performance. Here management should be commenting on cash flows, the
issues around funding and working capital management, and - critically -
the interaction between of the key value-creating activities (highlighted in
the previous category), and financial performance.
Clearly the majority of information needed is provided by the traditional
financial reporting model: however, additional information is needed to
explain the implicit retum to shareholders, over and above dividend pay-
ments, and to address the prospects of future financial performance.
640

4 Benefits of corporate transparency

The ValueReporting™ Framework can provide a systematic method of

identifying the types of information that are required, industry by industry,
to evaluate the quality and sustainability of corporate performance. But,
codifying best reporting practice from industry experts is not enough to
change practice. For companies to assume the additional expense of an
improved disclosure strategy, they need to understand why such enhanced
disclosure might produce economic benefits. Indeed, any meaningful
information is likely to have associated with it substantial collection and
dissemination costs. Moreover, it is possible that in some situations, some
companies might feel that becoming more transparent would incur some
proprietary costs ... in other words, they might argue that they will be
giving away some strategically sensitive information. These are
commercial realities.
It is important to note that the benefits of enhanced transparency may
come through in a variety of different guises. For many companies, the
most immediate economic benefit has come through in the ability to attract
and retain the best and brightest employees. Equally, the willingness of
customers to enter into a transaction with a transparent company has been
subject to some interesting case study analyses. For many metals and
mining companies, the dominant benefit of becoming more transparent
takes the form of the ability to attract joint partners on favourable terms or
the success rate ofbids for new licences.
Often of greatest interest, however, is the relationship between the
quality of reporting and the cost of both equity and debt. Basic finance
theory tells us that the greater the confidence that investors have in the ac-
curacy of their analysis of the future potential of the company, the lower
will be the returns that they demand for participating in that stock and thus
the lower the cost of capital the company faces.

5 lmplementation

Needless to say, any decision to change a reporting strategy will need the
ongoing support by the CEO and the Board. Given that support exists,
however, how can a company set about changing its corporate reporting
model? We argue that there are three essential steps in implementing
ValueReportingTM.
The first is to ascertain the complete information needs of its important
institutional value-oriented long-term investors. This can be done through
 641

surveys and interviewso These data can be collected directly by the com-
pany or with help of outside advisorso This "information list" should then
be supplemented by finding out what information sell-side analysts are
interested in and what senior managers think the company should be
reportingo
The second step has two partso The first is to assess the quality of the
company's current reporting model. How weil is it communicating the
critical information required by the investment community? How does its
reporting compare against its major domestic or global peers? This analy-
sis allows management to identify the greatest gaps in the information that
they provideo The second part of the assessment phase is to ascertain the
company's readiness to reporto Do they have access to the information that
is demanded by the user in a routinised fashion? The result of this as-
sessment phase will be the identification of any obvious "quick wins" -
areas of the reporting model where data would be highly valued by the
investor and which is easily disseminated by the companyo Provided that
there are no strategic costs associated with this data, the company may
reap sizeable rewards with little effort by leveraging this information set
into the public domaino
The third and final stage of the process is to design a "blueprint" for the
company's reportingo A communication strategy cannot be changed over-
night. Given the time and money involved in developing intemal informa-
tion systems, it is important to understand where the company wishes to be
in 3 to 5 years time and to develop an actionplan accordinglyo
Milestones for implementation will clearly need to be developed and
monitored at this stageo These should not be underestimated in their sig-
nificance as they may imply deep-rooted changes for the companyo Given
that "what gets measured gets done", the reporting model that is deployed
throughout the firm is likely to have consequences that go far beyond the
corporate dashboardo Cultural and training implications must be reviewedo
A constant re-assessment of the informational needs of the user and the
quality of reporting of peers will also be requiredo Technological enablers
for data gathering and dissemination need to be evaluated, and so ono In
short, this is a continuously evolving challenge for management today; a
challenge that will require the vision and support ofthe leaders ofindustryo
The rewards for the standards setters in this arena are real and substan-
tial.. and the cost of not participatingo weil, ask the shareholders of
0 0 0

Enron, Worldcomoooo
642

References

ValueReporting Review 2003; Transparency in Corporate Reporting. This annual

publication is available at www.valuereporting.com
DiPiazza Jr. SA, Eccles RG (2002) Building Public Trust, The Future of
Corporate Reporting. New York, N.Y., USA
Eccles RG, Herz RH, Keegan EM, Phillips DMH (2001) The ValueReporting
Revolution, Moving Beyond the Earnings Game. New York, N.Y., USA
List of Contributors

Albach, Horst, Waldstr. 49, D-53177 Bonn

Albers, Sönke, Christian-Albrechts-Universität zu Kiel, Institut fii.r
betriebswirtschaftliche Innovationsforschung, Westring 425, D-24098 Kiel
Blaga, Steffen, FernUniversität Hagen, Lehrstuhl fii.r Betriebswirtschaft,
insb. Investitions- und Produktionstheorie, Universitätsstr. 41, D-58084
Hagen
Bogaschewsky, Ronald W., Bayerische Julius-Max:imilians-Universität
Würzburg, Lehrstuhl fii.r Betriebswirtschaftslehre und Industriebetriebsle-
hre, Sanderring 2, D-97070 Würzburg
Bonaccorso, Luca, University ofCatania, Faculty ofEconomics, Corso Ita-
lia, 55, I-95129 Catania
Braßler, Axel, TU Ilmenau, Fachgebiet Produktionswirtschaft und Indus-
triebetriebslehre, Postfach 100565, D-98684 Ilmenau
Coltman, Tim, University ofWollongong, 2522 NSW, Australia
Demers, Elizabeth, Wm. E. Sirnon School of Business, University of
Rochester, Rochester, NY 14627, USA
Demgenski, Caroline, Institut fii.r Mittelstandsforschung Bonn, Max:imil-
ianstr. 20, D-53111 Bonn
Devinney, Timothy, Australian Graduate School of Management, Univer-
sity ofNew South Wales, Sydney 2052 NSW, Australia
Dietl, Helmut, Universität Paderborn, Lehrstuhl fii.r Organisation und In-
ternationales Management, Warburger Str. 100, D-33098 Paderborn
Fabel, Oliver, Universität Konstanz, Lehrstuhl fii.r Betriebswirtschafts-
lehre, insb. Unternehmenspolitik, Postfach 144, D-78457 Konstanz
Fandel, Günter, FernUniversität Hagen, Lehrstuhl für Betriebswirtschaft,
insb. Investitions- und Produktionstheorie, Universitätsstr. 41, D-58084
Hagen
Franck, Egon, Universität Zürich, Lehrstuhl fii.r Unternehmensführung,
Plattenstr. 14, CH-8032 Zürich
Fritz, Wolfgang, TU Braunschweig, Institut fii.r Wirtschaftswissenschaften,
Abt-Jerusalem-Str. 4, D-38106 Braunschweig
644

Gensler, Sonja, Johann Wolfgang Goethe-Universität, Lehrstuhl für Elec-

tronic Commerce, Mertonstr. 17, D-60054 Frankfurt am Main
Gössinger, Ralf, Universität Kaiserslautern, Lehrstuhl für Produktionswirt-
schaft, Gottlieb-Daimler-Str. 42, D-67663 Kaiserslautem
Greco, Salvatore, University of Catania, Faculty of Economics, Corso
Italia, 55, I-95129 Catania
Greinert, Markus, PWC Deutsche Revision, Marie-Curie-Str. 24-28, D-
60439 Frankfurt
Guenther, Thomas, TU Dresden, Fakultät Wirtschaftswissenschaften,
Mommsenstr.13, D-01062 Dresden
Haase, Michaela, FU Berlin, Institut für Marketing, Otto-von-Simson-Str.
13-15, D-14195 Berlin
Hachmeister, Dirk, Universität Leipzig, Professur für Allgemeine Be-
triebswirtschaftlehre, Marschnerstr. 31, D-041 09 Leipzig
Joos, Philip, Wm. E. Sirnon School of Business, University of Rochester,
Rochester, NY 14627, USA
Kay, Rosemarie, Institut für Mittelstandsforschung Bonn, Maximilianstr.
20, D-53111 Bonn
Kayser, Stefan, WHU Otto Beisheim Graduate School of Management,
Burgplatz 2, D-56179 Vallendar
Kleinaltenkamp, Michael, FU Berlin, Institut für Marketing, Otto-von-
Simson-Str. 13-15, D-14195 Berlin
Kuhner, Christoph, Universität zu Köln, Seminar für Allgemeine Betriebs-
wirtschaftslehre und für Wirtschaftsprüfung, D-50923 Köln
Kuivalainen, OBi, Telecom Business Research Center, Lappeenranta Uni-
versity ofTechnology, P.O.Box 20, FIN-53851 Lappeenranta
Mackenstedt, Andreas, PWC Deutsche Revision, Marie-Curie-Str. 24-28,
D-60439 Frankfurt
Matarazzo, Benedetto, University of Catania, Faculty of Economics, Corso
Italia, 55, 1-95129 Catania
Midgley, David, INSEAD, Boulevard de Constance, F-77305 Fontaine-
bleau
 645

Mocek, Stephan, Gerhard-Mercator-Universität Duisburg, Institut für In-

ternationale und Regionale Wirtschaftsbeziehungen, Mülheimer Str. 212,
D-47048 Duisburg
Okubayashi, Koji, Graduate School of Business Administration, Kobe
University, 2-1, Rokkodai-cho, Nada, Kobe, 657-8501, Japan
Paffrath, Rainer, Universität Lüneburg, Lehrstuhl für Produktion und Wirt-
schaftsinformatik, Scharnhorststr. 1, D-21335 Lüneburg
Pascha, Werner, Gerhard-Mercator-Universität Duisburg, Institut für In-
ternationale und Regionale Wirtschaftsbeziehungen, Mülheimer Str. 212,
D-47048 Duisburg
Perlitz, Manfred, Universität Mannheim, Lehrstuhl für Internationales
Management, Schloß (L 4.1 ), D-68131 Mannheim
Platania, Pietro, University of Catania, Faculty of Economics, Corso Italia,
55, 1-95129 Catania
Pouchkarev, Igor, Erasmus Universiteit Rotterdam, P.O. Box 1738, NL-
3000 Rotterdam
Rank, Olaf, Universität Mannheim, Lehrstuhl :fiir Internationales Manage-
ment, Schloß (L 4.1), D-68131 Mannheim
Reese, Joachim, Universität Lüneburg, Lehrstuhl für Produktion und
Wirtschaftsinformatik, Scharnhorststr. 1, D-21335 Lüneburg
Reiß, Michael, Universität Stuttgart, Betriebswirtschaftliches Institut,
Abteilung II- Organisation, Keplerstr. 17, D-70174 Stuttgart
Rudolf, Markus, Otto Beisheim Graduate School of Management,
Dresdner Bank Stiftungslehrstuhl für Finanzen, Burgplatz 2, D-56179
V allendar
Saggau, Björn, Universität Lüneburg, Lehrstuhl für Produktion und
Wirtschaftsinformatik, Scharnhorststr. 1, D-21335 Lüneburg
Schiefner, Lars, Martin-Luther-Universität Halle-Wittenberg, Lehrstuhl für
Finanzwirtschaft und Bankbetriebslehre, Universitätsplatz 8, D-06099
Halle
Schmidt, Reinhart, Martin-Luther-Universität Halle-Wittenberg, Lehrstuhl
für Finanzwirtschaft und Bankbetriebslehre, Universitätsplatz 8, D-06099
Halle
Schneider, Herfried, TU Ilmenau, Fachgebiet Produktionswirtschaft und
Industriebetriebslehre, Postfach 100565, D-98684 Ilmenau
646

Scholich, Martin, PWC Deutsche Revision, Marie-Curie-Str. 24-28, D-

60439 Frankfurt
Skiera, Bernd, Johann Wolfgang Goethe-Universität, Lehrstuhl für Elec-
tronic Commerce, Mertonstr. 17, D-60054 Frankfurt am Main
Spronk, Jaap, Erasmus Universiteit Rotterdam, P.O. Box 1738, NL-3000
Rotterdam
Stadtler, Hartmut, TU Darmstadt, Fachgebiet Fertigungs- und
Materialwirtschaft, Hochschulstr. 1, D-64289 Darmstadt
Stammen, Markus, AlP-Institut GmbH, Hochstr. 13, D-58084 Hagen
Steven, Marion, Ruhr-Universität Bochum, Lehrstuhl für Angewandte
BWL I, Universitätsstr. 150, D-44801 Bochum
Strack, Rainer, The Boston Consulting Group, Stadttor 1, D-40219
Düsseldorf
Taalikka, Sanna, Telecom Business Research Center, Lappeenranta Uni-
versity ofTechnology, P.O.Box 20, FIN-53851 Lappeenranta
Vliet, Pim van, Erasmus Universiteit Rotterdam, P.O. Box 1738, NL-3000
Rotterdam
Winter, Stefan, Universität Würzburg, Lehrstuhl für Personal und Organi-
sation, Sanderring 2, D-97070 Würzburg
Wirl, Franz, Universität Wien, Lehrstuhl für Industrie, Energie und Um-
welt, Brünnerstr. 72, A-1210 Wien
Witt, Peter, WHU, Otto Beisheim Graduate School of Management,
Lehrstuhl für Unternehmertum und Existenzgründung, Burgplatz 2, D-
56179 Vallendar
Wolbert, Joachim, PWC Deutsche Revision, Arnulfstrasse 25, D-80335
München
Zwick, Thomas, ZEW Mannheim, Postfach 103443, D-68034 Mannheim
 647

Scientific Committee

Horst Albach, Bonn, Germany

Sönke Albers, Kiel, Germany
Uschi Backes-Gellner, Zurich, Switzerland
Hans Corsten, Kaiserslautern, Germany
Günter Fandel, Hagen, Germany
Wolfgang Kürsten, Jena, Germany
Christoph Kuhner, Cologne, Germany
Bemd Rudolph, Munich, Germany
Manfred Schlüter, Hamburg, Germany
Joerg E. Staufenbiel, Cologne, Germany
Stefan Winter, Würzburg, Germany

Organizing Committee

Uschi Backes-Gellner, Zurich, Switzerland

Günter Fandel, Hagen, Germany
Manfred Schlüter, Hamburg, Germany
Joerg E. Staufenbiel, Cologne, Germany