THE COORDINATIONOF MANUFACTURING

INTERDEPENDENCE IN MULTINATIONALCOMPANIES

BRIANCEMASCARENHAS*
Rice University

Abstract. Manycompanies have recently adopted global manufacturingrationalizationpro-

grams. In order to realize the full benefit of these internationallyintegrated manufacturing
strategies, however, companies need to overcome the problem of coordinating operations
where interdependence exists among subsidiaries. This article, which is based on a field
study of 25 multinationalcompanies, provides insights on how manufacturinginterdepen-
dence among country subsidiaries is coordinated. Specifically, the study examined the
relationships between increasing manufacturinginterdependence withthe use of 4 coordina-
tion modes-impersonal methods, use of organizationmembers who have a system-sensitive
outlook, executive compensation systems based on a subsidiary's performance, and per-
sonal communication-and with the use of 3 communication patterns-communication
between managers of different country subsidiaries, communicationwith top management,
and communicationwitha central manufacturingstaff group. Findings suggest that although
there is an extensive use of impersonal methods of coordination,with increasing interdepen-
dence there is a change in coordination mix towards greater use of system-sensitive
members and personal communication. Increasing interdependence is also significantly
associated with greater communication among country-subsidiary managers and greater
communicationwith a central manufacturingstaff group. Although the relationship between
increasing interdependence and communication with top management through the formal
hierarchy was not found to be statistically significant, interviews with managers suggested
the importanceof this channel when communicationamong country-subsidiarymanagers and
communicationwith the central manufacturingstaff group are inadequate for managing the
manufacturinginterdependence. Implicationsare drawnfor companies planning to increase
the internationalintegrationof their manufacturingoperations.

* Many companies have recently adopted programs of global manufacturing INTRODUCTION

rationalization. Instead of multiproduct and multistage plants autonomously
serving a national market on a local-for-local basis, it has become feasible and
economically attractive to develop plants that manufacture only one product, or
are involved in only a few stages of the production process, for the worldwide
market [Doz 1978]. Firms that have embarked on such rationalizationprograms
attempt to tap sources of efficiency through economies of scale and through
capturing and combining the comparative advantages of differentcountries.
There are several forces that have pushed companies to embark on global
rationalization schemes: the lowering of trade barriers; the existence of unex-
ploited economies of scale beyond the size of national markets; the rise of global
competition, raising the need for a global stategy; competition from low-cost
producers; concerns about increasing efficiency while at the same time maintain-
ing employment, rather than simply transferring production to other countries;
improvements in communications and transportation; and technology changes
that have increased the optimal economies of scale in manufacturing.
However, although it has been relativelyeasy to diagnose analyticallythe need for
global manufacturing rationalization, its implementation has been difficult, and
many companies have not been successful in their pursuit of rationalization[Doz
1978]. There are several important barriers to international rationalization.Host
country interests, in their attempts to gain greater control over their environment,
are increasingly trying to limit the autonomy of managerial decision making in
multinational companies [Doz and Prahalad 1980; Doz, Bartlett, and Prahalad

*BrianceMascarenhasis AssistantProfessorof InternationalBusiness and Business Policyat the

Jesse H. Jones GraduateSchool of Administration, Rice University.His teaching and research
interestsare inbusiness strategyand international
management.

Journal of InternationalBusiness Studies, Winter1984 91

Palgrave Macmillan Journals

is collaborating with JSTOR to digitize, preserve, and extend access to
Journal of International Business Studies ®
www.jstor.org
 1981]. Similarly,geographical distances, differences in cultures and time frames,
varying local market conditions, the occurrences of particularsituations arising in
one subsidiary but not in others make international manufacturing integration
difficult.In fact, many top executives feel that the major problem in conducting
internationalbusiness is how to coordinate and manage effectively the diversity
inherent in international operations [Davidson and Haspeslagh 1982]. Although
many companies have adopted a global product division structure, it has been
discovered that formal reorganizations in themselves do not guarantee effective
integration of company resources [Davidson and Haspeslagh 1982]. Moreatten-
tion should be paid to coordinative processes used to achieve international
integration [Bartlett 1983; Davidson and Haspeslagh 1982].
This article reports on a study undertaken to obtain insights about practices used
by multinational companies to coordinate the manufacturing interdependence
between their subsidiaries. First, the literatureon the coordination of work flow
interdependence is reviewed. Then, hypotheses derived from this review are
tested using data generated by a field study of multinationalcompanies. Finally,
the results are analyzed and implications for practice are drawn.

THE Subunits of organizations usually play a dual role. (The term "subunit" is a general
LITERATURE
term commonly used in the literatureto referto a part of an organization such as a
work-unit, department, division, or subsidiary.) On one hand, each subunit is a
quasi-separate entity with its own tasks and objectives; on the other, each subunit
also needs to be linkedwith other parts of the total organization. Hence, it is often
necessary to consider its interdependence with other subunits. Lawrence and
Lorsch [1967] have suggested that effective organizations are ones that differen-
tiate their subunits in response to a dynamic and diverse environment but, at the
same time, maintaincoordination and integration among them. Thompson [1967]
postulated the existence of 3 types of work flow interdependence-pooled,
sequential, and reciprocal-that, respectively, required coordination by rules and
procedures, plans, and mutual adjustment on the part of subunits.
Empirical work has tended to support Thompson's propositions. In laboratory
experiments Baumler [1971] has shown that control and coordination methods
relying on impersonal methods such as rules and procedures were used often
when interdependence was low, but were used less frequently when there was
extensive subunit interdependence. In a study of work flow interdependence
within subunits, Van de Ven, Delbecg, and Koenig [1976] found that there was a
positive association between the use of coordination mechanisms in general and
the level of interdependence, but that the use of impersonal coordination
mechanisms was invariantwith greater interdependence. Tushman [1979b] has
shown that coordination methods to manage interdependence within units were
different from those methods used to handle interdependence between units.
Substantial intra-unit interdependence appeared to be coordinated through
lateral communication among unit members, while substantial inter-unitinterde-
pendence tended to be coordinated through the supervisors of the separate
subunits. In fact, it has been suggested that a 2-step informationflow, first up to
the supervisor and, second, between supervisors, may be a more effective way of
transferring communication across organizational boundaries than widespread
direct communication across all levels between subunits [Tushman 1979b].
Cheng [1983] has demonstrated that organizations which coordinate their work
flow interdependence tended to outperformorganizations without this coordina-
tion.
In the context of international business, a number of related studies have
emerged, although the relationship between manufacturinginterdependence and
coordination has not been specifically measured. For example, the control and
coordination of the marketing and R&Dfunctions have received attention, al-

92 Journalof InternationalBusiness Studies, Winter1984

though the emphasis in these studies has been primarilyon the question of
centralization or decentralization of decision making [Weichman 1974; Picard
1977; Fischer and Behrman 1979]. Franko [1971] has pointed out that when
activities of subunits are interdependent, decision making cannot easily be
decentralized without the likelihood of a system suboptimization problem and,
hence, increased cost. Ithas also been suggested that the internationaltransfer
of managers can serve an informalcoordination and control function through a
personal socialization process that results in the internalizationof common goals
and procedures [Edstrom and Galbraith 1976, 1977]. Studies of communication
patterns in multinational companies have shown that verbal and interpersonal
modes of communication tend to be used more frequently than impersonal modes
in reporting and understanding problems [Keegan 1974; Picard 1980]. Finally,
Brandtand Hulbert's study of communication patterns between corporate head-
quarters and overseas subsidiaries has pointed out the obvious-that increasing
the quantity of informationflow without regard to quality will not necessarily
improve the relationship between the 2 entities. Moreattention is needed to what
is being transmitted and to the mode of transmission [Brandt and Hulbert 1976,
p. 63].

The units of analysis were 25 U.S. subsidiaries of 25 different multinational DESIGNOFTHE

companies covering a broad range of industries and headquartered outside the RESEARCH
U.S. All were at least 50 percent owned by a non-U.S. parent company. Thirty STUDY
subsidiaries listed in the Directory of Foreign Manufacturersin the U.S. [Arpan
and Ricks 1979] that were geographically accessible to the researcher were
contacted through a letter sent to the president of each subsidiary. The letter
informed the executive of the issues being investigated and requested the
company's participation in the study. Although the presidents of 5 of the
companies contacted agreed in principle to participate in the study, they kept
postponing the interview for reasons of either time or travel and, therefore, were
not included in the group of companies actually studied. In-depth interviews with
top executives, each lasting about one hour and 30 minutes, were finally
conducted at the premises of the 25 companies. The titles of the executives who
were interviewed were "president" or "vice-president."
There may be limitations on the extent to which the findings of this study are
generalizable to all companies operating in differentcountries. Because of limited
resources, the study focused only on subsidiaries that were involved in some type
of manufacturing or assembly work in the U.S. Subsidiaries that were exclusively
service-oriented, and those involved only in trade (that is, importing or exporting)
were not included. Furthermore, the analysis was conducted only from the
perspective of the U.S. subsidiary, not from that of other subunits with which it
may have been interacting. Finally,only non-U.S. companies were examined in
this study. It is therefore possible that the results of this study are situation-
specific and are not generalizable to differentconditions. For example, because of
the larger size of the U.S. market relative to the markets in other countries ["Note
on F.D.I. in the U.S." 1981], the U.S. subsidiary of a foreign company may
characteristically assume greater importance to the firmthan a foreign subsidiary
of a U.S. company. This difference in relative importance may dictate a higher
degree of control and coordination over the U.S. subsidiary of a foreign company
than for a foreign subsidiary of a U.S. company.
It was felt that a cluster of case studies was the most appropriate method of
researching the interdependence-coordination question. Because the concept of
interdependence is ambiguous and complex, it needed to be defined and clarified
forthe respondents. At the same time the researcher needed to capture as much
informationas possible about the nature of each subsidiary's operations in order
to check the applicability of the research instrument to the company being

Journal of InternationalBusiness Studies, Winter1984 93

 studied. Moreover, a sufficiently large number of cases had to be studied to
undertake rigorous statistical testing of the hypotheses [McClintock, Brannon,
and Maynard-Moody 1979]. To reduce a potential drawback with personal
interviews-that of introducingdifferingbiases when interviews are conducted by
different people-only one interviewer was used.
The interviews began with the interviewer informing the executive of the issues
being investigated. The executive was then asked to comment on the nature of
the subsidiary's operations (the inputs, production process, and outputs), on the
production flow interactions with other subunits of the company, and on how this
interdependency was coordinated-that is, how the subsidiary tried to ensure the
on-time arrivaland delivery of parts and products of the required quality and
specification. Ineach instance an attempt was made to identify current practice
and to spot problems encountered with existing systems. The executive was then
asked to fill out a questionnaire designed to measure interdependence and
coordination mode. A finalcheck was made forthe face validityof the research in-
strument by asking the respondent if there were any problems encountered in
fitting the questionnaire to the subsidiary's operations. No significant problems
were reported.

OPERATIONALI- Interdependence has been defined as a condition that exists. where actions taken
ZATIONOF within one subunit affect the actions and work outcomes of another subunit
VARIABLES[McCann and Ferry 1979]. The measurement of interdependence requires a
discussion of the concept of increasing interdependence. Thompson identified 3
Interdependence types of interdependence-pooled, sequential, and reciprocal-that respectively
are increasingly difficultto coordinate because they contain increasing degrees of
contingency [Thompson 1967]. Pooled interdependence, also called "indirect"
interdependence [Miller1959], exists where actions of one subunit have only a
circuitous or second-order impact on activities of another subunit [Thompson
1967]. Pooled interdependence exists when 2 or more subunits depend on a
common resource, such as the same market or sources of supply [Miller1959].
For example, if the sales of one subsidiary of a multinationalcompany to a third-
country market deprives other subsidiaries of sales, their pooled interdependence
is said to exist. Sequential interdependence exists where actions taken in subunit
A have a direct impact on subunit B but the reverse is not true [Thompson 1967].
Reciprocal interdependence exists when actions taken in subunit A directly affect
subunit B and vice versa. This study's measure of interdependence focused only
on direct sequential and reciprocal interdependence in the manufacturing func-
tion.

Coordination Coordinationof subunits may be defined as the function of insuring that subunits'
behaviors are properly interwoven, sequenced, and timed so as to accomplish
some joint activity or task completion [Tuggle and Saunders 1979]. Inthis study,
coordination practices were examined in 2 stages. Inthe first stage, the methods
of achieving coordination are measured and examined. The second stage in-
volves an analysis of the patterns (actors and direction) of communication which
is one of the methods for achieving coordination.
According to Marchand Simon there are 2 general ways of achieving coordina-
tion: by programming behavior and by communication and feedback so that an
adjustment can be made [Marchand Simon 1958]. The objective of programming
is to provide coordination by fostering predictability. This is achieved by pre-
specifying the activities of individualsubunits so that behavior in one subunit is
synchronized with and does not create unexpected contingencies for the activi-
ties of another subunit. It should be pointed out that the term "programming"
used in this context is not restricted to a rigid, mechanistic approach such as

94 Journal of InternationalBusiness Studies, Winter1984

requiringparticularcomponent specifications from a subunit, but also involves a
broader, more subtle approach that guides and shapes the behavior of the
organization members. Common ways of programming behavior are through
employee socialization (which is a function of job rotation, trainingprograms, and
long periods of indoctrination) [Edstrom and Galbraith 1977] and impersonal
control methods such as standard operating procedures, deadlines, and work
schedules. Similarly,the type of reward and incentive system exerts a powerful
influence on the behavior of organization members [March and Simon 1958].
Incentive plans that are tied to a subunit's performance may encourage intra-
subunit cooperation but are likely to create inter-subunit competition [Galbraith
1977] and are most useful where little cooperation among subunits is required
[Salter 1973]. Moreover, financialrewards tied closely to a subunit's performance
are not likely to evoke in members the often spontaneous, uncertain, and
ambiguous task of cooperating across subunits [Galbraith 1977]. This broad
package of programming serves to reduce idiosyncratic behavior that might be
dysfunctional to activities in other subunits.
The second general way of achieving coordination is through communication and
feedback. This involves interactive communication between members of different
subunits about actual conditions and about deviations from plan so that adjust-
ments can be made. Interactive communication thus reduces uncertainty with
regard to workflow contingencies, promotes joint decision making, and facilitates
the taking of corrective action.
The following common coordination methods that reflect the basic March and
Simon classification of programming and communication were measured in the
study using the scales shown in Appendix B.
Impersonal Methods-the use of legitimate methods universally distributed
in the company that do not require verbal interaction between individuals.
Examples are: standard operating procedures, deadlines, regular reports,
budgets, predetermined work plans, and schedules.
System-Sensitivity-the ability of subunit members to foresee the impact of
other subunits of actions taken in one subunit and thereby to undertake
appropriate behavior that avoids system suboptimization problems. This
system-sensitivity tends to be acquired through a process of socialization of
the members of the organization, and is a function of the recruitment
process, training programs, and internationaltransfers [Edstrom and Gal-
braith 1976, 1977].
Compensation System-the use of a reward system that does not encour-
age key subunit members to pursue the parochial interests of their own
subunit at the expense of the rest of the organization.
Personal Communication-the use of communication and feedback involv-
ing verbal interaction between individuals, such as group meetings, tele-
phone conversations, and face-to-face communication.
Because this conceptual breakdown of coordination methods does not give any
insights about the actors involved and the direction of communication, a further
analysis involving communication patterns is in order. Ithas been suggested that
workflow interdependence across subunits may be coordinated through commu-
nication between managers of different subunits [Tushman 1979b; Doz, Bartlett,
and Prahalad 1981], through communication with a central manufacturing staff
group [Skinner 1964], or through communication with a higher level line manager
that has jurisdiction over several subunits [Doz, Bartlett, and Prahalad 1981].
These basic actors and directions of communication in the context of a multina-
tional company are shown in Table 1 and were measured by asking the respon-
dents to rate on a 7-point scale the extent of use of each of the communication
patterns to coordinate the manufacturing interdependence of their subsidiary.

Journal of InternationalBusiness Studies, Winter 1984 95

 HYPOTHESESGreater interdependence creates increased workflow uncertainty, which, in turn,
may create a need for greater coordination [March and Simon 1958; Tushman
1979a,b]. Thompson has posited that when interdependence is low, coordination
is likely to be achieved by impersonal methods such as rules and procedures.

TABLE1
Common Patterns of Communication

1. Directcommunicationbetweenmanagers ofdifferent
countrysubsidiaries.
2. Communication witha manufacturing
staffgroup.
3. Communication witha linemanagerintheformalhierarchyhigherthanthe countrysubsidiary
manager (forexample,a regional
managerorglobalproductmanager).

With increasing interdependence, however, impersonal coordination methods

alone may be insufficient to integrate operations, so there is a need for other
coordination methods [Thompson 1967]. Interdependence between subunits is
likely to be coordinated by communication between managers of different
subunits rather than by direct communication between members of different
subunits below the level of manager [Tushman 1979b]. It has been suggested
that to avoid suboptimization problems when interdependence exists between
subsidiaries of a multinationalcompany it may be importantto centralize decision
making [Franko 1971; Doz 1980; Prahalad and Doz 1981]. Finally, it has been
posited that when a subunit is interdependent with others, it is difficultto evaluate
its members based on the performance of the subunit because of impacting
contingencies that may arise outside theirdirect control in other subunits [Tuggle
and Saunders 1979]. In light of these propositions the following hypotheses were
set forth and tested:

H: Greater subsidiary manufacturing interdependence is associated with

a) a greater use of total coordination;
b) a greater use of the personal communication method;
c) a greater use of the system-sensitivity method of coordination;
d) a smaller use of a compensation system based on a subsidiary's perfor-
mance;
e) a greater use of communication among managers of different country
subsidiaries;
f) a greater use of communication with top management in the formal hierar-
chy above the level of the country subsidiary manager;
g) a greater use of communication with the central manufacturing staff group.
RESULTSThe results of this study should be viewed with caution because control was not
achieved for a variety of independent variables, and the results are based on a
small sample size and involve the pooling of subjective ratings across different
raters where comparabilitymay not exist. Descriptive statistics of the companies
that were studied are shown in Tables 2 and 3. Table 2 shows the companies
broken down by national origin, and Table 3 shows the median and the range for
the company characteristics of size and number of subsidiaries around the world.
Table 4 shows the means and standard deviations for all the variables. Fromthe
table it can be seen that the use of communication between country-subsidiary
managers, the use of impersonal methods, and the use of personal communica-
tion are higher than the use of other coordination methods and communication
patterns. To examine how the differentcoordination methods and communication
patterns varywith increasing interdependence, however, one must analyze Table5.

96 Journal of InternationalBusiness Studies, Winter 1984

 TABLE2
National Origin of Companies

Country Number

England 5
West Germany 5
Holland 3
Norway 2
Sweden 2
Switzerland 2
Japan 2
France 2
Yugoslavia 1
Belgium 1

TOTAL 25

TABLE3
Company Size and Proliferationof Subsidiaries

Range Median

Company'ssize (numberof employees*) 140-20,000 2,000

U.S. subsidiary'ssize (numberof employees*) 28-1,500 370
Numberofsubsidiariesownedbyparentcompany 2-100 7

*Numberof employees is the most frequentlyused measureof organizationsize [Kimberly1976].

TABLE 4
Means and Standard Deviations for AllVariables

Standard
Mean Deviation

CoordinationMethods
1. Impersonalmethods 4.18 2.25
2. System-sensitivity 3.70 1.81
3. Compensationsystem 3.35 2.01
4. Personalcommunication 4.05 1.61
5. Totalcoordination 3.82 1.28
Communication Patterns
1. Communication withtop management 3.20 1.88
2. Communication withthe centralmanufacturing
staffgroup 3.95 2.72
3. Communication amongcountry-subsidiarymanagers 4.40 1.82
Interdependence 255.0 72.8

In Table 5, hypotheses a, b, c, e, and g are supported. That is, greater

manufacturing interdependence is significantly associated with greater use of
total coordination (0.44), with greater use of personal communication (0.58), with
greater use of system-sensitive employees (0.57), with greater use of communica-
tion among country-subsidiary managers (0.64), and with a greater use of
communication with a central manufacturing staff group (0.45). Compensation
systems (0.03), impersonal methods (0.33), and communication with top manage-

Journal of InternationalBusiness Studies, Winter1984 97

 TABLE5
Zero-OrderCorrelations of Interdependence
with Methods of Coordinationand Communication Patterns

Methods
Coordination
1. Impersonalmethods 0.33
2. System-sensitivity 0.57**
3. Compensationsystem 0.03
4. Personalcommunication 0.58**
5. Totalcoordination 0.44*
CommunicationPatterns
1. Communication withtop management -0.09
2. Communication withthe centralmanufacturing
staff group 0.45*
3. Communication amongcountry-subsidiarymanagers 0.64**

at the 0.01 level

**Significant
at the 0.05 level
*Significant
ment (-0.09) were not found to be significantly related to the level of interdepen-
dence, although it should be mentioned that there is a positive trend between
interdependence and the use of impersonal methods that might have been
statistically significant with a larger sample size.
Although hypothesis d posited a negative relationship between interdependence
and the use of an executive compensation system based on a subsidiary's
performance, no significant relationship was found. Perhaps the reason for not
observing the expected relationship may be that reward and incentive systems,
which are usually set in favor of national subsidiary autonomy before the move
towards international manufacturing integration, are difficultto change abruptly
because of vested interests [Doz and Prahalad 1981].
The results in Tables 4 and 5 suggest that while impersonal methods are used ex-
tensively they do not vary the most with increasing interdependence. This
evidence supports the notion that rigid rules and procedures may be capable of
managing interdependence in a stable and fixed environment [March and Simon
1958], but may be less suitable for managing tensions in a complex and highly
variable international business environment [Doz, Bartlett, and Prahalad 1981]
because no organization can specify, in advance, a rule for every possible future
contingency [Ouchi and Price 1978]. Discussions with executives indicated that
impersonal methods have a limited informationprocessing capacity and tend to
be slow, even when quick action may be needed.
Personal communication, on the other hand, which was found to be strongly
related to interdependence, may be a versatile method for balancing the internal
manufacturing interdependence between subsidiaries and the external depen-
dence of a subsidiary on its local environment. Interviews suggested that personal
communication is an effective tool for handling tensions and solving problems
because it facilitates discussion of subtle aspects of a problem, because it
permits interactive and critical evaluation, because it provides forthe synthesis of
different points of view, because it enables joint decision making, and because it
facilitates timely action. Personal communication in the formof telephone calls or
meetings highlights the urgency and importance of a problem, smoothes differ-
ences of viewpoint, permits a firsthand understanding of the peculiar operating
conditions of another subsidiary, and fosters camaraderie among members of
different subsidiaries which, in turn, furtherfacilitates coordination.
The study uncovered evidence that multinational companies also use system-
sensitive members in managing interdependence. Discussions with executives
indicated that system-sensitive members can facilitate coordination in subtle
ways. Their knowledge of a variety of areas within the company helps them

98 Journalof InternationalBusiness Studies, Winter 1984

 envision work flow repercussions on other areas of actions taken in one area and
may lead to taking precautionary actions. Or, system-sensitive members can rely
.on contact in other parts of the company to help manage their subsidiary's
dependencies. They may be in a position to play a coordinating and mediating
role between local interests and parent company interests since their experience
in a variety of subunits provides a common frame of reference. The socialization
process may instill company-wide goals as opposed to parochial subsidiary
interests. This, in turn, may reduce adversarial attitudes and may foster coopera-
tion in dealings with members in other areas of the company [Doz, Bartlett, and
Prahalad 1981].
In the light of the propositions about the need to manage interdependence
through centralization of decision making [Franko 1971; Doz 1980; Prahalad and
Doz 1981], it is surprising to note that communication with higher level line
management (-0.09) is not significantly associated with interdependence. This
may be because management through the formal hierarchy usually results in
overlytight control, low motivation among members, mistrust, extreme specializa-
tion, and an overdependency of subordinates on superiors which may have
negative effects both on the effectiveness of the organization and on the
psychological well-being of its members [Argyris 1964; Likert1967]. Discussions
with executives provided insights forthe apparent low use of communication with
top management in coordinating work flow across subsidiaries. If top manage-
ment is not familiarwith the local environment, it may be better to have the
country subsidiary manager who is closer to the situation handle the coordination.
The manager is in daily contact with and can gauge the importance of local
interest pressures that may create contingencies for the work flow integration
across national boundaries. If the entire coordination responsibility were to be
handled at the top management level, the process might be slow and top
management might quickly become overloaded with operational duties. Top
management usually is more involved in business strategy formulation and
evaluation, in gathering and evaluating informationabout competitors, in evaluat-
ing and approving requests for capital expenditures, and in personnel evaluation
than in coordinating operations.
There are several reasons why communication takes place with the central
manufacturing staff group (0.45). The group often is in charge of allocating export
orders to different subsidiaries [Doz 1980]. Interviews with managers revealed
that the central group sometimes plays the role of a neutralarbitratorin resolving
conflicts between subsidiaries when agreement between them cannot be
reached. The center may provide subsidiaries with information on production
programming, quality control standards, product specifications, and the legiti-
mate forecast of worldwide demand on which subsidiaries base their production
activities. Communication with the central group may also be fostered by staffing
it with well-respected experienced country subsidiary managers who have exten-
sive contacts in the field.
The use of inter-manager communication across subsidiaries is also significantly
associated with greater interdependence (0.64). Organizations are rational and
seek to reduce theircommunication burden [Marchand Simon 1958]. By commu-
nicating first to managers and then across managers, the cost of communication
may be reduced if intra-unitcommunication is less costly than inter-unitcommuni-
cation [Tushman 1979b]. Doz has pointed out that when a company that was
originallyoperating with autonomous nationalsubsidiaries moves towards greater
international manufacturing integration, it may be politically necessary to give
country-subsidiary managers some influence in the process instead of having
total centralization of decision making [Doz 1978]. Interviews highlighted addi-
tional reasons for the relative importance of inter-managerial communication
between subsidiaries. In many cases the locus of decision-making power is at the

Journal of InternationalBusiness Studies, Winter1984 99

 managerial level. Lower level employees often communicate with their subsidiary
manager because he/she may have greater knowledge of, better contacts with,
and more influence over other subsidiaries. Inter-managerial communication
involves negotiation and joint decision making which, in turn, facilitate implemen-
tation of the decision. A subsidiary of a multinationalcompany that is located in
the U.S. and that may be important to the parent may be staffed with superior
managerial talent that is capable of assuming the coordination responsibility itself
instead of passing it on to top management.
The interviews and the literature provided an understanding of the dynamics of
communication patterns by suggesting that coordination may be first attempted
by communication between subsidiary managers; then, if conflicts are still
unresolved, by communication with the center group; and, as a last resort, by
communication with top management [Doz, Bartlett, and Prahalad 1981]. First,
country managers may try to reach agreement among themselves on their own
since they are closer to the local problems and in order to help the implementation
of the agreement. Ifthis fails, the central manufacturing staff group in the role of a
neutralarbitratormay attempt to resolve the conflict. As a last resort, the conflicts
pass on to top line management who use their formal authority to make and
implement a decision.
This dynamic behavior sheds light on the Table 5 correlations of interdependence
with different patterns of communication which show that interdependence is
associated in descending order of magnitude with communication among coun-
try-subsidiarymanagers, with communication with the central manufacturing staff
group, and with communication with top management. Moreover, this behavior
also helps to understand why, in Table 6, communication among country-
subsidiary managers is strongly related to communication with the central
manufacturing staff group (0.45) but not with communication with top manage-
ment (-0.04), while communication with the central manufacturing staff group is
strongly related to both communication with top management (0.43) and commu-
nication among country-subsidiarymanagers (0.45). Table 6 also shows that there
is a high degree of intercorrelationbetween some of the other variables. Some of
this interrelationshipwould be expected because of definitionaldependency-for
example, total coordination is defined to be the average of the 4 coordination
methods.
It is interesting to note, however, the high correlations between impersonal
methods with all the other coordination methods except compensation system
and with the communication patterns. Discussion with executives and the
literature suggested that when impersonal methods of coordinating work flow
between subsidiaries are inadequate for coping with a contingency, managers
resort to other methods of coordination-for example, personal communication
[Galbraith1977]. Once agreement is reached through these other means on how
to cope with a particularcontingency, revisions are made in impersonal methods,
such as changes in schedules, which are then used in their revised form
[Galbraith 1977]. This may explain the high correlations between impersonal
methods and some of the other variables. Also noteworthy is that communication
among country-subsidiary managers is significantly associated with personal
communications (0.44).
In order to obtain insights about the individual relationship of each coordination
method with interdependence, it was important to disentangle the interrelation-
ship among coordination methods. This procedure was accomplished through
partial-correlationanalysis which first removes the portions of a particularcoordi-
nation method and of interdependence that are accounted for by the independent
variables for which control is desired and then conducts the correlation analysis.
The third-ordercorrelation coefficients of each of the coordination methods with
interdependence while controlling for the remaining 3 coordination methods are

100 Journal of InternationalBusiness Studies, Winter1984

shown in Table 7. The variable "total coordination"was omitted in this step of the
analysis because it was composed of the 4 coordination methods that were
included.
As can be seen by comparing Tables 4 and 6, there is littledeteriorationin the rela-
tionships between interdependence on one hand and system-sensitivity and
personal communication on the other, while controlling for other independent
variables. There is, therefore, a reduced chance of a spurious relationship of
interdependence with system-sensitivity and personal communication. Itis inter-
esting to note that the relationship of interdependence with impersonal methods
turns negative while controlling forthe other coordination methods. This supports
TABLE6
Non-ParametricCorrelations Among Independent Variables
1 2 3 4 5 6 7 8
Methods
Coordination
1. Impersonalmethods
2. System-sensitivity 0.54**
3. Compensationsystem -0.05 -0.04
4. Personalcommunication 0.68** 0.30 0.14
5. Totalcoordination 0.85** 0.62**0.59** 0.84**
Communication Patterns
1. Communication withtop
management 0.44* 0.33 0.26 -0.28 0.17
2. Communication withthe central
manufacturingstaffgroup 0.43* 0.17 0.11 -0.13 0.13 0.43*
3. Communication amongcountry-
subsidiarymanagers 0,47* 0.13 0.31 0.44* 0.21 -0.04 0.45*

*Significantat the 0,01 level.

at the 0.05 level.
*Significant

TABLE7
Third-OrderPartialCorrelations Between Interdependence
and Methods of Coordination

Impersonal methods -0.351

System-sensitivity 0.45*
Compensationsystem 0.04
Personalcommunication 0.59**

1. Forexample,shouldbe interpretedas the relationshipbetween impersonalmethodsand interde-

pendence when controllingforpersonalcommunication,system-sensitivity,and compensation.
at the 0.05 level.
*Significant
at the 0.01 level.
**Significant

the propositions noted above about the limited use of impersonal methods for
coping with increasing interdependence.

The findings of this study suggested that although the use of impersonal methods IMPLICATIONS
of coordination is extensive in multinationalcompanies, with increasing manufac-
turinginterdependence there appears to be a shift in coordination mix toward the
use of system-sensitive members, personal communication, communication with
a central manufacturing staff group, and communication between country subsid-
iarymanagers. The shift toward these coordination methods and communication
patterns suggests the importance of socializing organization members to build a

Journal of InternationalBusiness Studies, Winter1984 101

 system-sensitive outlook. Executives who plan to push their companies towards
greater international manufacturing integration need to develop processes that
will build a system-sensitive outlook and that will develop personal relationships
across subsidiaries. This, however, is a subtle, time-consuming, and difficult
process: relationships among members develop only gradually; existing norms
and attitudes may be entrenched; often only changes that draw unanimous
support from all power bases can be implemented at the outset; and successful
change requires top management's ongoing support [Doz and Prahalad 1981]. To
be successful, however, the strategy of the company must fit its administrative
capacity.
APPENDIXA MEASUREMENTOF MANUFACTURING INTERDEPENDENCE
Executives were asked to respond to the followingquestionnaire which measured
the degree of manufacturing dependency of the U.S. subsidiary or subunit on
other subunits for the sourcing of its inputs or the sale of its outputs:

Pleaseindicatewhatpercentof the totalworkflow(100%)of yoursubunitis describedby each of the

followingworkflowsituations(Cases A-D).Sumofallcases shouldnotexceed 100%.Pleaseexamine
allcases beforeassigningpercentages.

CaseA: Wheresourcingof inputs,production,and sale of out- Your

put arecarriedout by oursubunitindependentlyof othersub- INPUTS subunit
unit(s).
PRODUCTION

OUTPUT
%of workflow

Case B: Whereinputsare sourcedexternallyfromothersub- Other

unit(s)but productionand sale outputarecarriedout indepen- INPUTS subunit(s)
dentlyby yoursubunit.
PRODUCTION Your
subunit
OUTPUT
%of workflow

Case C: Wherethe sourcingof inputsand productionare un- Your

dertakenindependentlyby yoursubunitbut outputis sold exter- INPUTS subunit
nallyto othersubunit(s).
PRODUCTION Other
subunit(s)
OUTPUT
%of workflow

Case D: Whereyoursubunitundertakesproductionbut is de- Other

pendenton othersubunit(s)forinputsand sale of output. INPUTS subunit(s)

Your
PRODUCTION Your
subunit
OUTPUT Other
I
subunit(s)

%of workflow

Based on the percentages that were allocated to the different cases-A, B, C,

D-a score for the U.S. subsidiary's manufacturing interdependence was com-
puted by allocating the following weights to the different cases and multiplying
them by the respective percentage.

102 Journalof InternationalBusiness Studies, Winter1984

 Case Weight

A 1
B 2
C 3
D 4

Following the assumption that, all other things being equal, firms with an input
dependency are more vulnerable than firmswith an output dependency [Thomp-
son 1967, p. 54; Kiggundu 1981], Case C was assigned a higher weight than Case
B. For example, ifa respondent had allocated the percentages, 30, 10, 25, and 35,
respectively, to the cases A, B, C, and D, the interdependence score for the firm
would be 265, computed as follows.

Case Weight Percentile Product

A 1 x 30 - 30
B 2 X 10 =20
C 3 X 25 75
D 4 x 35 =140
Total 265

Inorderto test the validityof this interdependence measure, the interdependence

score was correlated with the mean of the responses to the following 2 questions:

1. Howoften is yoursubunitinvolvedin productionflow interchangeswithother subunit(s)of the

company?
1 2 3 4 5 6 7
Very
Never Often

2. To whatextent are productionactivitiesof yoursubunitaffected by how other subunit(s)of the

companydo theirjobs?
I 2 3 4 5 6 7
Production
Othersub- activitiesare
unit(s)have substantially
no effecton affectedby
production othersub-
activities unit(s)

The correlation coefficient between questions 1 and 2 was 0.81, and the correla-
tion coefficient between the interdependence score and the mean of the 2
questions was 0.72. Mohr's Interdependence Index was not used to check for
validitybecause the index applies to interdependence of tasks withina subunit as
opposed to between subunits [Mohr1971].

MEASUREMENTOF COORDINATION APPENDIXB

The following 4 methods that bring about coordination were measured: imper-
sonal methods, system-sensitivity, compensation system, and personal commu-

Journalof InternationalBusiness Studies, Winter1984 103

 nication [Van de Ven, Delbecq, and Koenig 1976; Brandt and Hulbert 1976;
Edstrom and Galbraith 1977; Thompson 1967]. Using a 7-point scale, enclosed at
the ends by "never" and "very often," respondents were asked to rate how often
each of the following mechanisms was used to coordinate the workflow between
their country subsidiary and other country subsidiaries of the company. The
scores of the items were averaged to obtain the frequency of use of each method.

Coordination
Method MeasuringItems

Impersonalmethods use of formalpoliciesand procedures

use of workplans and schedules1
System-sensitivity throughan understandingof whatthe repercussionson workflowof
othersubunitsmay be of actionstakenin yoursubunit
Personalcommunication throughgroupmeetings
throughverbalcommunication
Compensationsystem Respondentswereasked to rate:to whatextentarethe keyemployees
of the subsidiaryrewarded(forexample,by commissionsand
bonuses) on the basis of yoursubsidiary'ssales and profits?2

1Although one canwonderhowfamiliartopsubsidiarymanagementis withimpersonalmethodsinthe

organizationandthe extentof theiruse especiallyby lower-levelemployees,the interviewsgenerally
suggested thatforthe companiesstudiedtopmanagementhada detailedknowledgeofthe use ofim-
personalmethodsbecause they hadrisenthroughthe ranksand theyoftenhadto intercedewhenthe
use of impersonalmethodscouldnot providethe needed answers.
2Theempiricalrationaleforthismeasureis based on Ouchiand Maguire'sstudy[1975]whichshowed
thatan importantcontrolmechanisminorganizationsis based on the measurementof outputs(output
control).

REFERENCES Argyris,C. Integratingthe Individualand the Organization.NewYork:JohnWiley,1964.

Arpan,J., and Ricks,D.A.Directoryof Foreign Manufacturersin the U.S. Atlanta,GA:Publishing
Services Division,GeorgiaState University,1979.
Bartlett,C.A. "MNCs:Get Offthe. ReorganizationMerry-GoRound."HarvardBusiness Review,
March-April 1983.
Baumler, J . V. "DefinedCriteriaof Performancein Organizational
Control."AdministrativeScience
Quarterly,September1971,pp. 340-50.
Brandt,W.K., and Hulbert,J. M."Patternsof Communicationsin the MultinationalCorporation: An
EmpiricalStudy."Journalof InternationalBusiness Studies, Spring1976.
Cheng, J. L.C. "Interdependenceand Coordinationin Organizations:A Role-SystemAnalysis."
Academy of ManagementJournal,March1983,pp. 156-162.
Child,J. "Predictingand UnderstandingOrganizationalStructure."In OrganizationalStructure:
Extensionsand Replications,edited by D.S. PughandC. R. Hinings.Farnborough, Hants,England:
Saxon House, 1976.
Davidson,W.H., and Haspeslagh,P. "Shapinga GlobalProductOrganization." HarvardBusiness
Review,July-August1982.
Doz, Y.L. "ManagingManufacturing Rationalization
WithinMultinational
Companies."Columbia
Journalof WorldBusiness, Fall1978.
. Strategic Managementin Multinational Companies."Sloan Management Review,
Winter1980.
Doz, Y.L.; Bartlett,C. A.; and Prahalad,C. K. "GlobalCompetitivePressures and Host Country
Demands."CaliforniaManagementReview,Spring1981.
Doz, Y.L., and Prahalad,C. K. "HowMNCsCope with Host GovernmentIntervention." Harvard
Business Review, March-April 1980.
. HeadquartersInfluenceand StrategicControlin MNCs."Sloan ManagementReview,
Fall1981.

104 Journal of InternationalBusiness Studies, Winter1984

 Transferof Managers:Some ImportantPolicyImplica-
J. R. "International
Edstrom,A., and Galbraith,
tions."ColumbiaJournalof WorldBusiness 11 (1976),p. 100-12.
_____.. "Transferof Managers as a Coordinationand Control Strategy in MultinationalOrganiza-
tions."AdministrativeScience Quarterly,June 1977.
CommunicationsSystems and the Decision Process." Management
Ference, T. "Organizational
Science, October1970 (B),pp. 83-96.
Fischer,W.,and Behrman,J. "TheCoordination of ForeignR&DActivitiesby Transnational
Corpora-
tions."Journalof InternationalBusiness Studies, Winter1979.
Franko,L.G.Joint VentureSurvivalin MultinationalCorporations.NY:PraegerPublications,1971.
Galbraith,J. OrganizationDesign. Reading,MA:AddisonWesley,1977.
Horowitz,J. A. "ManagementControlin France,GreatBritain,and Germany."ColumbiaJournalof
WorldBusiness, Summer1978.
Keegan, W.J. "MultinationalMarketingControl."Journalof InternationalBusiness Studies, Fall
1972,
"MultinationalScanning: A Study of the InformationSources Utilized by Headquarters
ExecutivesinMultinationalCompanies."AdministrativeScience Quarterly,September1974.
Kiggundu, M.N. "TaskInterdependenceandthe Theoryof Job Design."Academy of Management
Review,July1981,pp. 499-508.
Kimberly, J. "Organizational
SizeandtheStructuralistPerspective:A Review,Critique,and Proposal."
AdministrativeScience Quarterly,December1976,pp. 571-597.
and Integrationin ComplexOrganizations."
Lawrence,P., and Lorsch,J. "Differentiation Adminis-
trativeScience Quarterly,June 1967, pp. 1-47.
Likert,R. TheHumanOrganization:Its Managementand Value.NewYork:McGraw-Hill, 1967.
March,J. G., and Simon,H.A. Organizations.NewYork:JohnWiley,1958.
McCann,J., and Ferry,D. "AnApproachforAssessing and ManagingInter-Unit Interdependence."
Academy of ManagementReview,January1979, pp. 113-119.
McClintock,C.; Brannon,D.; and Maynard-Moody, S. "Applyingthe Logic of Sample Surveys to
QualitativeCase Studies:TheCase ClusterMethod."AdministrativeScience Quarterly,December
1979.
Mclnnes,J. M. "FinancialControlSystems for Multinational Operations."Journalof International
Business Studies, Fall1971, pp. 11-28.
Miller,E. "Technology,Territory,and Time:The InternalDifferentiation of Complex Production
Systems."HumanRelations,August1959, pp. 243-272.
Mohr,L. "Organizational Technologyand Organizational Structure."AdministrativeScience Quar-
terly,December1971, pp. 444-59.
Mott,PaulE. TheCharacteristicsof EffectiveOrganizations.NewYork:Harperand Row, 1972.
"Noteon F.D.I.in the U.S." HarvardBusiness School, IntercollegiateCase ClearingHouse, Case
#9-381-031, 1981.
Nunnally,J. PsychometricTheory.NewYork:McGraw-Hill, 1978.
Ouchi,W., and Maguire,M.A. "Organizational Control:Two Functions."AdministrativeScience
Quarterly,December1975.
Ouchi,W.,and Price,R. "Hierarchies,Classes, and TheoryZ: A New Perspectiveon Organization
Development."OrganizationalDynamics, Autumn1978.
Picard,J. "HowEuropeanCompaniesControlMarketingDecisionsAbroad."ColumbiaJournalof
WorldBusiness, Summer1977.
Structuresand IntegrativeDevices in EuropeanMultinational
"Organizational Corpora-
tions."ColumbiaJournalof WorldBusiness, Spring1980.
Prahalad,C. K., and Doz, Y.L. "AnApproachto StrategicControlin MNCs."Sloan Management
Review,Summer1981.
Salter,M.S. "TailorIncentiveCompensationto Strategy."HarvardBusiness Review, March-April
1973.
Skinner,C.W. "Managementof International Production."HarvardBusiness Review, September-
October1964.
Snadowsky,A. "Communication NetworkResearch."HumanRelations25 (1972),pp. 283-306.
Thompson,J. D. Organizationsin Action. NewYork:McGraw-Hill, 1967.
Tuggle, F.D., and Saunders,C. B. "Controland Its OrganizationalManifestations:A Propositional
Inventory." Reviewof Business and Economic Research, Spring1979.

Journal of InternationalBusiness Studies, Winter1984 105

 Tushman,M. "ManagingCommunicationNetworksin R&DLaboratories."Sloan Management
Review,Winter1979a.
. "WorkCharacteristics and Sub-unit Communication Structure: A Contingency Analysis."
AdministrativeScience Quarterly,March1979b.
Van de Ven, A.; Delbecq, A.; and Koenig, R. Jr. "Determinantsof CoordinationModes Within
Organizations." AmericanSociological Review,April1976.
Wiechman,V.E. "Integrating Multinational
MarketingActivities."ColumbiaJournalof WorldBusi-
ness, Winter1974.
William,R. "Organizational Size,Rules,andSurveillance."
InOrganizations:Structureand Behavior,
Vol.1, edited by J. Litterer.NewYork:JohnWiley,1969.

106 Journalof InternationalBusiness Studies, Winter1984