The Politics of Service Delivery in Democracies: Better Access For The Poor
The Politics of Service Delivery in Democracies: Better Access For The Poor
The Politics of Service Delivery in Democracies: Better Access For The Poor
Delivery in Democracies.
Better Access for the Poor
Edited by Shantayanan Devarajan and Ingrid Widlund
ISBN: 978-91-976600-2-0.
Contents
Ingrid Widlund
Starting points
Clean drinking water, primary education, preventive and curative health faci-
lities are all fundamental for human development and well-being. Yet, as the
World Development Report 2004 demonstrated, the provision of such vital
services in developing countries is typically skewed in favour of the non-poor.
The better-off have not only the means to rely on private alternatives, but a
higher percentage of public spending accrues to the richer sections. Further-
more, public services intended to benefit the broader citizenry tend to be
fraught with problems of corruption, absenteeism and low quality. Funding is
misappropriated, service providers do not report to work, buildings are in need
of repair, basic materials (like medicines and teaching devices) are missing, and
so forth (World Bank, 2003).
This is not an acceptable situation in any context, but the dismal performan-
ce of democratic regimes is particularly problematic and noteworthy. Democra-
cies have the institutional mechanisms for toppling leaders who govern badly.
Rulers can formally be held accountable for their performance, and citizens
have a decisive influence on who is to govern on what grounds. As we all know,
this is in many contexts not enough to ensure an equitable distribution of
public services, or making improved educational and health outcomes a politi-
cal priority. Why not? And what can be done about it?
These are questions for governments and citizens in the countries concer-
ned, but they are also questions for donors of official development assistance.
How to best strengthen democratic procedures and institutions so that public
decision-making also benefits people in poverty underpins the very rationale
for democracy promotion in development cooperation – to ensure influence
for each and every (adult) citizen over his or her life situation. There is further-
more an increasing move from supporting single projects to untied general bud-
get support and sector-wide approaches. The core commitments of the Paris
Declaration – harmonization, alignment and ownership – follow from the ana-
lysis of the WDR 2004 to strengthen domestic systems and processes. The-
se commitments have also resulted in an increased focus on Public Finance
Management, i.e. methods and routines for the efficient use of public resources
in policy planning and implementation.1 New forms and priorities for official
development assistance raise new challenges. Will for example direct budget
support to central governments risk undermining the authority and room for
manoeuvre of local governments? And what are the appropriate instruments
for democracy promotion when moving away from project support?
The main insight of Making Services Work for Poor People, is that the pro-
1
See for example Sida’s manual (Sida, 2007).
2
2
See Appendix for a summary of the workshop discussions.
3
This is not a water-proof distinction as governance reforms obviously alter the relationships
between actors on the political arena, but it nevertheless serves to identify a relevant difference in
emphases between the contributing authors.
3
tan Coelho and Nílian Silva (chapter 11) analyse health service reform in two
different contexts, the Cabo de Santo Agostinho municipality in the Northeast,
and the São Paolo metropolitan area respectively. In Cabo, significant achieve-
ments were made on a number of crucial health indicators, largely thanks to
the opportunities the system created for a committed local government that
took office in the late 1990s. The main feature of decentralised governance that
comes through in these chapters, is that it allows citizens to participate in the
management of services, and thereby improves government performance. Both
Nelson’s and Duarte’s accounts indicate that merely shifting resources and de-
cision-making powers to local levels do not necessarily generate large benefits.
Complementary measures are needed.
Omar Azfar (chapter 8) studies whether decentralisation leads to impro-
vements in preference matching. The better the matching, the more respon-
sive can and will the government be. Are local level government officials more
knowledgeable about local level preferences than officials higher up in the
hierarchy? Survey data from Uganda and the Philippines suggest that this is
indeed the case, although such knowledge was still fairly limited in both
countries. In James Manor’s analysis (chapter 7), democratic decentralisation
can facilitate constructive influence of local on higher levels in the system. In
most developing societies there is, in his view, a gap between intermediate (di-
strict or departement) and local levels, to the detriment of service provision,
among other things. How to bridge this gap is the main theme of Manor’s paper,
and he points above all at the neglected role of local ‘fixers’ and the potential
of demand-driven programmes. But decentralisation has a role as well. When
local service providers, like teachers and health professionals, are made accoun-
table to local elected councils, absenteeism, for example, can be curbed. On the
other hand, responsiveness to the electorate may not always improve services,
as there is often a more immediate demand for small construction works, in
irrigation and other infrastructure for example. Similarly and curiously, Azfar’s
results from Uganda showed a very low demand for immunization. Adapting to
local demand in this respect would be counter-productive, and by implication,
responsiveness, regardless of the service, is not necessarily a good thing.
Arrangements to enable and promote local citizen participation in the ma-
nagement of social services follow from decentralised governance and have also
been regarded very favourably, and hence supported, by international donors.
This is also a case of governance reform. According to Joan Nelson, client par-
ticipation has been a widely accepted reform in health and education in Latin
America, intended to strengthen both responsiveness and accountability. In her
assessment, it is quite a challenge to make school and (especially) health coun-
cils effective, especially in serving the interests of the poor. There are indeed
successful cases, but also many examples of the opposite.
The participatory features of Brazil’s new health system are given importan-
ce in Duarte’s account of how health improved in a Northeast municipality, in
particular by involving the inhabitants in the introduction of local health teams.
Based on review of experiences from Brazil, India and Bolivia, Azfar relatedly
suggests that various participatory arrangements, such as participatory bud-
geting, is a promising avenue for improving the communication of information
between citizens and public officials, and hence the matching of preferences.
5
In addition, participation can also impact on how preferences are formed. The
demand for immunization in Uganda could have been higher if citizens had had
the opportunity to learn from health experts in a participatory process. Both
Duarte and Azfar thus convey a largely positive view on the impact of this type
of governance reform for improving government responsiveness.
Peter Spink (chapter 12), in contrast, raises a word of caution about the ‘the
new institutional forms’ of governance at sub-national level in Latin America,
reforms that gather various civil society actors and community representatives
into local planning bodies, advisory councils, and so on. Although there are ex-
amples of arrangements having a positive impact on active citizenship and, as
is being implied, public services for poor people, the author sees clear risks that
these participatory arenas may also reproduce inequities. Similar concerns are
raised by Nelson.
From another angle, Anuradha Joshi (chapter 6) expresses some reservations
about the perceived benefits of ‘institutionalised participation’, as well as of de-
centralisation. In particular, she holds that poorer citizens are unlikely to be able
to secure greater responsiveness or hold service providers accountable on an
individual basis. Collective action is required to overcome the various hurdles
for doing so. Some of the recommendations of the WDR 2004, e.g. to increase
choice of service providers, have in her view therefore clear limitations. In ad-
dition, the state’s ability to respond to these demands cannot be taken for gran-
ted, as the role of public agencies partly are redefined with citizen participation
(along with pluralisation and decentralisation) to entail more complicated tasks
of negotiation and public engagement. The message here is that broader politi-
cal constraints may overrule the advantages of governance reforms.
Redirecting priorities
Institutional arrangements matter, but Duarte’s paper especially, suggests that
the commitment and priorities of bureaucrats and elected leaders matter as
much, or more. They do so both for taking public service provision seriously,
and for establishing and making operational the institutions which, to judge
from the Cabo case, may contribute to better development outcomes.
Allocating more resources is one factor for increased access and quality. From
the Latin American experience, Nelson points out that low-income groups do
not have the bargaining power of the more vocal middle-classes, and it therefore
takes a strong pro-poor commitment to reallocate resources to the benefit of
people in poverty. One example is conveyed by Marcus Melo (chapter 9) who
describes a recent school reform in Brazil (Fundef), which in five years led to a
seven-percent increase of primary school enrolment in the country. The reform
combined an increase of resources, with institutional measures to encourage the
country’s municipalities to take a greater responsibility for primary education.
Rather than a matter of ‘commitment’, the inadequate prioritisation for basic
services is in the volume analysed in terms of the logic of the political market.
Philip Keefer (chapter 4) argues that the incentives of a politician are contin-
gent upon his/her being believed by potential voters. Politicians who cannot
credibly appeal to broad segments of the population will not care much for
6
policies that benefit the many, but concentrate on measures that can easily be
targeted to select groups. Linked to this are problems of information. It can be
difficult for people to assess in what ways political choices impact on the health
and education of their family. Hence, they do not hold their politicians accoun-
table for real inadequacies. As a consequence, reforms in service delivery are
more challenging than in any other policy area. They require politicians with
the right incentives, a public administration that works, and citizens informed
enough to give credit to the political representatives who actually deliver.
Stuti Khemani (chapter 5) proposes an idea for a campaign to give health
and educational outcomes an increased political salience in developing demo-
cracies, and thereby real incentives for politicians to improve service provision
in these areas. Khemani notes that more information by itself, through greater
media coverage for instance, is unlikely to help. It is rather a matter of encoura-
ging more public attention to basic development parameters, something which
journalists are not trained to assess. By regularly generating and disseminating
information about specific developmental outcomes, citizens would be in a bet-
ter position to evaluate government performance on such outcomes, which in
turn could trigger political competition around them.
An underlying assumption (or conclusion) in a ‘political market approach’ is
that change must occur through changed incentives, politicians’ incentives in
particular. Khemani’s idea of an outcome-oriented information campaign, for
example, ultimately aims at making quality service delivery be in the self-inte-
rest of political competitors, as adequately informed citizens exercise pressure
via the ballot box. The explicit recommendation to donors is to help finance
the required collection of data, and/or condition aid upon its production and
public dissemination. This suggestion has bearing upon a larger discussion on
conditionality, more specifically outcome-based conditionality, where it is being
argued that international donors and development banks are placing too much
importance on certain policy indicators, such as tariffs on trade and interests
rates, rather than on actual outcomes.4
Khemani’s specific proposal is the idea of improving social services via chan-
ged perceptions and expectations amongst the poorer voters. The call for donor
conditionality concerns neither a particular outcome or a particular policy as-
sumed to result in a particular outcome, but a measure that enables outcomes
like child mortality and literacy to assume centrality in a democratic public
debate on political performance. The suggestion is different from both support
to independent media, a classic device for promoting democracy, and projects
to publicise the allocation of resources for a certain purpose to counter the
misuse of public funds.5 The distinctive feature is that it seeks to overcome
existing broad-based political hurdles to the better provision of social services.
It constitutes a new way of thinking about relevant ‘conditionalities’, and a new
method to address social development and democratic governance in an inte-
grated manner.
With his emphasis on political credibility as a basic factor, Keefer recom-
mends that international donors support policies that allow politicians to in-
4
See for example Kanbur, 2005.
5
See for example Reinikka and Svensson, 2003.
7
crease their credibility. In most countries this implies raising awareness and
strengthening the ability to monitor, all in line with Khemani’s ideas. In fragile
states, however, a strategy for the long-run is to support government efforts to
make and fulfil whatever promises their capacity allows, if only in the form
of food baskets, which as he notes, is a strategy ‘that donors universally abju-
re’. In parallel, donors can pursue the often chosen strategy to support service
provision through non-governmental organisations, but this, he implies is only
a constructive method in the short- or medium term. One can note that in
Keefer’s analysis, politicians’ incentives follow from their credibility with voters
and supporters only. Any free-standing commitment, on ideological grounds for
instance, is not really conceived of here.
sentatives, is on the increase. As noted, also Coelho and Silva suggest that the
involvement of civil society – within a broader frame of participatory measures
– may have played a significant role for the progress made in Brazil.
Brazil’s reform in primary education, as shown by Melo, did not have that
type of background, but this experience indicates that associations of public
servants – in this case municipal and provincial education secretaries – can be
important supporters of service delivery reform. Although not ‘civil society’ in
regular development parlance, such associations are clearly not ‘state’ either. In
a more general discussion, Joshi similarly points out that professional associa-
tions and public sectors unions can be important actors for or against service
delivery reform. Where state capacity for regulation and monitoring is weak,
such organisations can moreover improve the quality of services, and become
more responsive, through a concern for professional norms and standards. Along
with Melo’s example, this implies that it might be useful for donors to have a
broader conception of civil society than is commonly the case, and by implica-
tion, extending some support, to those types of organisations as well.
Peter Spink’s (chapter 12) presents lessons and experiences from a systema-
tic effort to identify and learn from innovative practices in service provision
amongst sub-national governments in Brazil. One major lesson is that the con-
cerns of poor people are best voiced by their own organisations – no matter
how loosely organised or otherwise odd to the outsider. In addition, none of
the successful experiences started out as clearly defined development projects
or policies. They were usually created by public officials and social activists in
close interaction to deal with specific local problems. Both these lessons (and
others) have, however, been difficult for domestic and international aid agencies
to absorb.
This might be an important observation for donors. Although it is most
certainly easier to interact with a conventional NGO, or a membership-based
organisation of the familiar type, people in poverty sometimes organise in oth-
er ways – in ways that benefit innovation in public management. A broader
implication is that a role for civil society organisations as advocates or providers
in a certain sector cannot be entirely separated from their role as players for
the long-term benefit (or detriment) of democracy, from Spink’s perspective
through the effect on how citizenship is experienced. NGOs at both ends of the
donor-recipient relationship could do well to bear that in mind.
presented in Duarte’s paper, for example, suggests that local participatory ar-
rangements can have a real impact on improving services for the poor. However,
Duarte also points out that these arrangements function poorly in small muni-
cipalities. The lesson may be that donors need to have realistic expectations and
learn more about what conditions are required for reforms to work well.
The other approach, geared at the broader political arena, appears in at least
two varieties here. Joshi emphasises for example, the importance of citizens en-
gaging collectively vis-à-vis the state, as well as the importance of state capacity
to negotiate competing demands from collective actors. Keefer and Khemani
focus on the relationship between voters and elected representatives, and how
the expectations and incentives of both set of actors can be modified through
other or better information. Also Spink’s analysis of local innovations and the
adjoining argument for considering the ‘citizenship effect’ of public services
conveys an interest in the broader political arena, in contrast to specific go-
vernance arrangements. It goes without saying that the policy implications of
these varieties differ from one another, and neither they, nor the distinctions, are
clear-cut either. Suffice it to say that donors, to the extent the new aid agendas
will allow for it, may find reason to place greater emphasis on the factors of the
broader political arena, and in addition try to bridge the sector gap between de-
mocracy promotion on the one hand, and efforts to promote health, education
and the like, on the other hand.
Note also that two different conceptions of citizen impact in democracies
can be discerned. As Keefer himself notes, the emphasis of his (and it could
be added: Khemani’s) approach is on citizens as monitors of government per-
formance. If the ability to monitor is strengthened, the argument goes, so is
accountability and by implication the reach and quality of basic services. In
the second conception, citizens can have an impact through active, in-between
election, participation, whether as innovators like in Spink’s example, or other
forms of active engagement. Participation does not preclude monitoring, but
the conceived role for citizens is nevertheless of a different kind. Both avenues
for impact and a strengthened democratic citizenship are clearly important, but
the focus and emphasis in a strategy for change can be different.
What unites the contributions in this volume is the interest in and sensiti-
vity to the often intricate political context for shaping realities on the ground:
Is there a local school, is there money for teaching materials, does the teacher
show up? The new concern amongst donors for public finance management
makes perfect sense: handling public money well is a key concern from any
perspective. But to the extent the volume has a main message, this is to never
forget that public finance is always managed (or mismanaged) in a particular
political milieu of competing interests, conflicting demands, and incentives gui-
ded by forces other than economic and administrative rationality.
10
References
Kanbur, Ravi, 2005, “Reforming the Formula: A Modest Proposal for Introdu-
cing Development Outcomes in IDA Allocation Procedures”, January, 2005,
http://people.cornell.edu/pages/sk145/papers.htm
Reinikka, Rittva and Jakob Svensson, 2003, “The Power of Information: Evi-
dence from an Information Campaign to Reduce Capture”, World Bank: Wash-
ington D.C.
World Bank, 2003, World Development Report 2004. Making Services Work for
Poor People. Washington D.C.
11
Anirudh Krishna
1
Interview with Mothulal Vaishnava, Kailashpuri village, Udaipur district, May 22, 2002.
13
2
“A recent report by the National Council for Applied Economic Research, based in New Delhi,
forecast that the number of ‘crore-patis’, Indian society’s rough equivalent of a millionaire, rose by
two-and-half times in the last three years to an estimated 53,000 households nationwide” (Somini
Sengupta, New York Times, February 28, 2006).
15
In addition, respondents’ mothers are also mostly well educated, though not on
average as well educated as their fathers. 52 percent of respondents’ mothers
have a Bachelors degree or better, while another 31 percent have high school
diplomas. It is this combination of relatively highly educated fathers and mothers
16
Two conclusions follow from the analysis in this section, which must be ad-
ded to the conclusion reported in the previous section:
(b) Poor service provision is part of a larger problem of exclusion and disconnec-
tedness. Poor people in poor communities, even when they acquire education, are still
unable to advance economically as far as their talents permit.
(c) Lack of information (about alternative career opportunities, pathways and
preparation) is an important explanatory factor. Lack of high-quality education
together with absence of information severely restricts the opportunity for upward
mobility.
The first row of this table shows that among a nationally representative sample
of 3,000 rural households in India, studied by Bhide and Mehta (2004), 14 per-
cent escaped from poverty between 1970 and 1982. During the same 12-year
period, another 13 percent of households fell into poverty. Carter and May’s
(2001) study of 1,171 households in KwaZulu-Natal, South Africa, shows
similarly that over a five-year period, 1993-1998, ten percent of households
moved upward out of poverty, but more than twice as many households, 25
percent, fell into poverty during the same time period.
A host of studies – undertaken by different scholars, in diverse countries, and
using different methods – show commonly that a falling tide operates alongside
18
a rising tide everywhere. A subgroup of these studies goes further and probes
household-level reasons responsible for escaping poverty and falling into po-
verty.
Between 2002 and 2006, colleagues and I undertook a series of studies span-
ning 223 communities and over 25,000 households in diverse areas of Kenya,
Uganda, Peru and India, using the Stages-of-Progress methodology, which
allows us to examine poverty dynamics from the perspective of the commu-
nities surveyed.6 We found that escaping poverty and falling into poverty are
not symmetric in terms of reasons. Poor people escape from poverty as a result
of one set of reasons, but people fall into poverty on account of a different set
of reasons. Ill health and high healthcare costs are overwhelmingly the most
important reason for households’ descents into poverty.
Health and health-related expenses were mentioned as important reasons
associated with nearly 60 percent of all descents recorded in villages of Ra-
jasthan, India, 74 percent of all descents examined in Andhra Pradesh, India,
and as many as 88 percent of all descents studied in villages of Gujarat, India.
In communities of Kenya, Uganda and Peru that we studied, respectively, 76
percent, 71 percent, and 67 percent of all descents were associated with ill-
health and health-related expenses. Not only does ill-health reduce the earning
capacity of a household’s members; in the absence of affordable and easy-to-
access healthcare facilities, it also adds considerably to the household’s burden
of expenditure, thereby striking a double blow, which quite often results in
tragedy. The resulting dependence of survivors, including orphans, upon other
households contributed further to descent in many cases. The story is not dissi-
milar in richer countries. More than half of all personal bankruptcies in America
are attributed to high medical costs (Himmelstein, et al., 2005).
High-interest private debt compounds the ill effects of health. Scholars have
computed that a very large part of debt incurred by poor families in India and
elsewhere arises on account of large healthcare expenses (Dilip and Duggal,
2002). In rural Vietnam, 60 percent of poor households were found to be in
debt, with medical expenses as the main reason for indebtedness (Ensor and
San, 1996).
6
Results from these studies are reported in Krishna (2004, 2006c), Krishna, et al. (2004, 2005,
2006a, 2006b). Working papers and a full description of the Stages-of-Progress methodology are
available at www.pubpol.duke.edu/krishna
19
7
Affordable and effective healthcare services provided by a local NGO have staved off descents
into poverty within one exceptional community.
20
8
One notable example is that of the right-to-information movement in India. See http://www.
righttoinformation.info
21
achieve) will also help show the way.9 When one person from a poorer commu-
nity succeeds in becoming a software engineer others like them will also aspire
to and plan for better things.
One should not assume that education alone will suffice. “He is good at mat-
hematics and he wants to be an engineer, but no one here can guide him about
what he needs to do”, the father of a gifted twelve-year old boy told me in a
village of Andhra Pradesh, India. I have heard similar experiences narrated by
countless other children and parents as I investigated poverty and its causes in
diverse parts of India, Kenya, Uganda, and Peru. No one guides young people
toward suitable career pathways because no one within these communities
possesses the required information and no one from outside these communities
has been made responsible for filling this gap. My own two children, atten-
ding high school in the United States, were privy to a barrage of career-related
information, with access not only to dedicated counselors at school, but also to
several college guide publications, to libraries and internet sources, and to nu-
merous friends and their parents, who opened their eyes wider to the available
alternatives.
The children whom I meet in rural communities in developing countries
have none of these avenues available to them. They live on the wrong side of
a very wide information divide. The information-poor environments in which
they live severely limit their capacities in the political and the economic realm.
Libraries and internet access are hardly available in these settings, and news-
papers have only recently started penetrating remote rural areas. As one newly
recruited software engineer told me in Bangalore, “Yes, I actually knew that
there probably are many job opportunities, but I didn’t have any access. Even
now we don’t get newspapers in my village.”
Reducing the gap in information remains a key task of development. More is
being learned about how this task can be accomplished. The media can become
a powerful counter-force to lack of information. In the Indian state of Kar-
nataka, for instance, the provision of information about local-level democracy
through the means of public radio was received enthusiastically by listeners in
far-flung villages, and it enabled poorer individuals to participate in numerous
activities associated with democratic governance at the local level.10 A televi-
sion program informing people about their legal rights and remedies quickly be-
came the most popular show aired in Armenia, as reported by the World Bank
(2005: 157). Providing newspapers and building libraries in rural schools should
further help reduce persistent information gaps, along with career counseling
centers and education at school about democratic rights and responsibilities.
Regular radio or television programs – “How to become a nurse” (or a physician
or mechanic or software engineer) – can also help considerably.
Individuals who have suffered themselves from the non-availability of useful
information are quick to realize the critical importance of information provi-
9
I am supporting an effort to put together a how-to manual in a group of villages of north India
where I work frequently. Compiled from the experiences of village youth who have made it to bet-
ter-paying positions, this manual would demonstrate to other village youth in a step-by-step fashion
how they, too, could attain a similar position if they tried.
10
A.P. Kripa and G.S. Ganesh Prasad, “The Message of Panchayati Raj”, Frontline, April 15-28,
2000.
22
sion. Here, for example, is what one young software engineer hailing from rural
area told me.
Question: Suppose you had a pot of money, and you wished to help
kids in rural areas. What is the first thing that you would do?
Answer: “I don’t need [much] money for doing this. I would just give
information to the people: this is what you need. I need to create the
balance, the urge to become whatever they want to become. [You] just
have to make it visible to them – what they can achieve if they take
this approach – and confidence that they themselves could do it. I have
to sell that concept to them. That would be enough.”11
11
Interview with Padmanabha Naidu, Bangalore, December 21, 2005.
23
References
Bhide, Shashanka and Aasha K. Mehta, 2004, Correlates of Incidence and Exit from
Chronic Poverty in Rural India: Evidence from Panel Data, Working Paper 15, New Delhi:
Indian Institute of Public Administration and Chronic Poverty Research Centre.
Booth, John and Mitchell Seligson, 2006, Inequality and Democracy in Latin America:
Individual and Contextual Effects of Wealth on Political Participation, paper presented at
Workshop on Poverty and Democracy, Duke University, February 17-18, 2006. Available
at www.duke.edu/~elb13/Workshop.html.
Bratton, Michael, 2006, Poor People and Democratic Citizenship in Africa, paper pre-
sented at Workshop on Poverty and Democracy, Duke University, February 17-18, 2006.
Available at www.duke.edu/~elb13/Workshop.html.
Carter, Michael and Julian May, 2001, “One Kind of Freedom: Poverty Dynamics in
Post-apartheid South Africa”, World Development, vol. 29, no. 12, pp. 1987-2006.
Dilip, T.R. and Ravi Duggal, 2002, Incidence of Non-Fatal Health Outcomes and Debt
in Urban India, working paper, Mumbai: Center for Enquiry into Health and Allied
Themes (CEHAT).
Ensor, Tim, and Pham Bich San, 1996, “Access and Payment for Health Care: The
Poor of Northern Vietnam”, International Journal of Health Planning and Management,
vol. 11, no. 1, pp. 69–83.
HDR, 2005, Human Development Report, 2005, New York: United Nations Develop-
ment Programme.
Himmelstein, David, Elizabeth Warren, Deborah Thorne, and Steffie Woolhandler,
2005, “Illness and Injury as Contributors to Bankruptcy”, Health Affairs, February 2.
Krishna, Anirudh, 2004, “Escaping Poverty and Becoming Poor: Who Gains, Who Lo-
ses, and Why?”, World Development, vol. 32, no. 1, pp. 121-36.
Krishna, Anirudh, 2006a, “Poverty and Democratic Participation Reconsidered: Evi-
dence from the Local Level in India”, Comparative Politics, vol. 38, no. 4, pp. 439-458.
Krishna, Anirudh, 2006b, Do Poor People Care Less for Democracy? Testing Individual-
Level Assumptions with Individual-Level Data from India, paper presented at Workshop
on Poverty and Democracy, Duke University, February 17-18, 2006. Available at www.
duke.edu/~elb13/Workshop.html.
Krishna, Anirudh, 2006c, “Pathways Out of and Into Poverty in 36 Villages of Andhra
Pradesh, India,” World Development, vol. 34, no. 2, pp. 271-88.
Krishna, Anirudh and Vijay Brihmadesam, 2006, “What Does it Take to Become a
Software Engineer? Educated Parents, Information Networks, and Upward Mobility in
India”, Economic and Political Weekly, Bombay, India, July 29, 2006.
Krishna, Anirudh, Mahesh Kapila, Mahendra Porwal, and Virpal Singh, 2005, “Why
Growth is not Enough: Household Poverty Dynamics in Northeast Gujarat, India”, Jour-
nal of Development Studies, vol. 41, no. 7, pp. 1163-1192.
Krishna, Anirudh, Patti Kristjanson, Maren Radeny, and Wilson Nindo, 2004, “Esca-
ping Poverty and Becoming Poor in 20 Kenyan Villages”, Journal of Human Development,
vol. 5, no. 2, pp. 211-226.
Krishna, Anirudh, Patti Kristjanson, Judith Kuan, Gustavo Quilca, Maren Radeny,
and Alicia Sanchez-Urrelo, 2006b, “Fixing the Hole in the Bucket: Household Poverty
Dynamics in Forty Communities of the Peruvian Andes,” Development and Change, vol.
37, no. 5, pp. 997-1021.
Krishna, Anirudh, Daniel Lumonya, Milissa Markiewicz, Firminus Mugumya, Agatha
Kafuko, and Jonah Wegoye, 2006a, “Escaping Poverty and Becoming Poor in 36 Villages
of Central and Western Uganda”, Journal of Development Studies, vol. 42, no. 2, pp. 346-
370.
Milly, Deborah, 1999, Poverty, Equality, and Growth: The Politics of Economic Need in
Postwar Japan, Cambridge, MA: Harvard University Press.
24
Munshi, Kaivan and Mark Rosenzweig, 2005, “Traditional Institutions Meet the Mo-
dern World: Caste, Gender and Schooling Choice in a Globalizing Economy”, available
at www.econ.brown.edu/fac/Kaivan_Munshi/bombay12.pdf.
Sen, Amartya, 1999, Democracy as Freedom. New York: Random House.
Sen, Binayak, 2003, “Drivers of Escape and Descent: Changing Household Fortunes
in Rural Bangladesh”, World Development, vol. 31, no. 3, pp. 513-534.
World Bank, 2003, World Development Report, 2004: Making Services Work for Poor
People, Washington, DC: The World Bank.
World Bank, 2005, World Development Report, 2006: Equity and Development, Wash-
ington, DC: The World Bank.
Yadav, Yogendra, 1999, “Electoral Politics in the Time of Change: India’s Third Electo-
ral System, 1989-99”, Economic and Political Weekly, Mumbai, August 21, 1999.
25
Joan M. Nelson
Introduction
This paper seeks to deepen our understanding of the politics of improved edu-
cation and health services for the poor, by unpacking the concept of “reform”
and exploring the varied political challenges of different categories of reform.
The discussion is based mainly on evidence from Latin America, but much of
the argument is probably relevant in other regions.
In the early l990s, improving economic conditions and (re)democratization
in much of Latin America prompted hope for rapid progress in reducing in-
equality and poverty. More specifically, less stringent fiscal constraints and
stronger electoral incentives were expected to encourage improved education
and health services. Fifteen years on, social service budgets and coverage have
indeed increased. But despite improvements in some countries and certain re-
spects, poor quality and egregious inefficiencies persist, and bear particularly
harshly on the poor. More broadly, poverty remains widespread and inequality
may have worsened. There is rising malaise in much of the region regarding the
performance of (more or less) democratic governments.
Theory argues that democracies should do better than authoritarian regimes
in providing broad-based, high-quality social services. Some two dozen cross-
national quantitative studies conducted since the late l990s do indeed show
correlations between indices of democracy and expenditures on social services
and transfers. However, links between democracy and social outcomes are much
weaker and less consistent. Better outcomes depend not only on increased spen-
ding, but on reallocation of resources and improved institutional incentives and
competence—that is, on reforms. A wide range of case and comparative studies,
opinion surveys, and other evidence suggest that the effects of democracy on
reform are far more complex, contingent, and variable than its impact on spen-
ding.1 Electoral pressures are only one among several sets of causal factors at
work.
Reform is often discussed as if it was a coherent entity. In reality, reform is
an umbrella label for a range of diverse actions, having in common only the
intent to increase equity, quality, and/or efficiency of service provision. Different
kinds of reforms present different political challenges. They arouse different in-
terests and generate different alliances. They engage ideologies and principles in
varying degrees. Some reforms are based on long experience and strong expert
consensus about what really improves performance; other categories reflect cur-
rent conventional wisdom, but only flimsy empirical evidence of what works.
This paper explores the politics of three categories of reforms: reallocation
1
For references and fuller discussion of these points, see Nelson, 2005.
26
2
There are exceptions: for instance, attempts to modernize Mexican text books in the mid l990s
triggered strong nationalist reactions.
3
Equipos Bàsicos de Atenciòn Integral de Salud (Basic Teams for Comprehensive Health Care).
27
4
However, many teachers and health professionals resist assignment to facilities serving poor
rural areas or urban neighborhoods.
28
patterns described are fairly universal, the interesting questions concern the
contexts and strategies reformers utilize to shift allocation in favor of the poor.
5
Such subsidies are seldom large enough to counterbalance the ability of wealthier districts to
raise local revenues; therefore, the gap between wealthier and poorer states and cities generally
widens as a result of decentralization.
29
Even in more normal times, links to goals broader or more urgent than im-
proved education or health sector performance can increase the priority of re-
forms, including pro-poor components. Colombia’s l993 health sector reform
is a clear example of such linkage. The Gaviria government (l990-l994) sought
to deepen domestic capital markets, as part of a broader agenda of economic
reforms. To this end, it sought pension reform along Chilean lines. Key legisla-
tors refused to approve pension reform unless the second major element of the
social security system, public health insurance, was greatly broadened. The l991
constitution had mandated health sector reforms, but the government had not
been able to agree on the best model of reform, until linkage with the high-
priority pension issue (and changes in government personnel) prompted plans
for a sweeping overhaul of health sector institutions and finance with the goal
of providing universal coverage.
As noted earlier, rising revenues clearly facilitate increased funding for ex-
panded coverage, scope, and demand subsidies. A strong fiscal position reduces
the motivation for top macro-economic officials to resist increases, diminishes
their influence relative to spending ministries, and eases competition for fun-
ding among and within sectors.
Changes in the budget process may also encourage pro-poor allocations. The
most obvious example is the introduction of participatory budgeting in dozens
of cities in Brazil and in the Indian state of Kerala. These experiments have been
in place only a few years. Research assessing their effects, including pro-poor
impact, suggests wide variation reflecting local political traditions and configu-
rations (Avritzer, 2006).
Program design also affects the politics of launching and sustaining pro-poor
programs. A substantial literature and much experience argue that broad targe-
ting is more politically sustainable than narrow targeting (Nelson, 2003). But
some targeted purposes and groups are more politically appealing than others.
Poor children are a more appealing target than the poor in general. Outside of
the context of disaster relief, support for investment by the poor and measures
that require self-help efforts (like CCTs) may win wider approval than support
for consumption.
Budget allocations are only the first step in a long process of implementation.
Funds allocated to pro-poor activities are often diverted to different uses, inclu-
ding outright corruption. Fairly simple measures to track and tighten financial
transfers, such as Uganda’s well-known decision to post the sums allocated to
each public school, may offer as much or more potential for increased funding
of pro-poor programs as efforts to influence priorities and choices earlier in the
budget process.
6
Pay increases are often adopted not as a means to motivate better performance but as sweeteners
to win acceptance of other measures service providers dislike, including governance reforms.
31
Pre- and in-service training is probably the most widely supported means to
improve service providers’ performance.7 However, such training rarely focuses
on the special problems of motivating and communicating with poor and mi-
nority students or patients. More broadly, in much of Latin America the quality
of training itself is poor. Training institutes can become political patronage: in
Argentina and Peru, politicians pressed to establish institutes in hundreds of
towns. Assessments show little or no improvement in quality resulting from in-
service training (Navarro, 2002; Vegas and Umansky, 2005).
Targeted stronger supervision and technical support in principle could have a
major impact. In the early l990s, Chile launched programs to improve qua-
lity and equity in rural primary schools, combining increased inputs (buildings,
supplies, lunches) with intensive supervision and support. Small rural schools
were grouped in clusters; multi-grade teachers in those schools met twice a
month to receive training, discuss problems and share solutions. Teachers in
schools serving the poorest 10 percent of the primary school population met
weekly in two-hour workshops, with external technical support, to diagnose
problems and discuss remedies. Related programs targeted secondary and urban
schools in poor areas (Delannoy, 2000: 16-19).
The politics of increased inputs focused on providers’ incentives and capa-
city is broadly similar to the politics of added inputs to expand coverage and
scope of services. Dispute does not focus on principles or general desirability,
but on budget priorities. But unlike most conflict regarding expansion, unions
may well take direct action regarding salaries; their demonstrations or strikes
can close down a system.
7
Indeed, Navarro states that many education improvement programs avoid components dealing
directly with teachers, because they are politically and ideologically charged; the exception, how-
ever, is investment in teaching training (Navarro, 2000: 3).
32
Indirect links between reward and performance: new approaches to finance and
competition
A less direct approach to strengthening incentives for efficient and responsive
social services is change in funding systems so that money follows clients. Most
public schools, clinics and hospitals in Latin America are mainly funded by
annual budgets, based largely on the previous year’s expenditures. In contrast,
voucher systems, charter schools, and capitation and related payment systems
for clinics and hospitals reimburse providers on the basis of services actually
provided. In areas where parents have a choice of schools and patients have a
choice of care providers, these payment systems are intended to motivate pro-
viders to try to attract clients by offering quality services. Advocates of this ap-
proach argue it broadens consumers’ choices and stimulates better performance
by providers.
The sole nation-wide experiment with school vouchers in Latin America was
introduced in Chile in l980, under Pinochet. Chile had a strong tradition of pri-
vate schools; the voucher system sharply expanded private school enrollment.
The effects have been intensively analyzed. Yet experts passionately disagree
on the results – a striking demonstration of the difficulty of assessing education
sector outcomes even where data are unusually good. Studies during the l990s
generally confirm that private subsidized (voucher) schools performed better
than municipal schools, controlling for student backgrounds (Fiszbein, 2005:
31-32).
8
The recent U.S. No Child Left Behind Act emphasizes collective sanctions rather than rewards.
As is well-known, NCLB has provoked widespread bitter opposition, as well as intense debate and
disagreement among education experts regarding its intended and unintended effects.
33
However, most analyses conclude that the voucher system did not improve
education for most poor children. Those in rural areas often had no alternative
to the sole public school serving the area. In urban areas where students and
their parents have more choice, many poor parents did not shift their children
out of mediocre (or worse) public schools, whether through lack of interest,
poor information, or perhaps concern about an unwelcoming ambiance in pri-
vate schools. There is little evidence that competition from the private schools
has had a positive impact on the public schools.
vel backing in democratic settings usually reflects linkage with higher priority
goals.
Decentralization
Decentralization of social services has been a powerful trend in and beyond
Latin America. However, it has mainly been a by-product of broader gover-
nance objectives including deepened democracy, heightened efficiency, and in
some cases reduced reliance on central funding. Perhaps because decentrali-
zation is so often driven by forces and objectives larger than the education or
health sectors, the process has often been hasty and poorly prepared, without
adequate attention to financial, legal and regulatory details at the sector level
and with little preparation or capacity-building in state and local governments.
The actual impact of decentralization on service provision, in general and for
the poor more specifically, varies tremendously among and within countries.
Decentralization’s potential for bringing government operations closer to
those served prompts support in principle from both left (seeking responsive-
ness) and right (seeking efficiency). But stakeholders within the education and
health sectors are often skeptical or hostile to decentralization. In traditionally
centralized systems, national providers’ unions fear dilution of their power, and
prefer to negotiate with a single central authority rather than multiple state and
local governments. So may major contractors in some cases: private hospital
associations accustomed to dealing with the national Social Security adminis-
tration were among the most intense opponents of decentralization in Brazil’s
constitutionally mandated health sector reforms. Teachers and health workers
worry that pay and personnel policies vary among states and municipalities,
and often fall short of federal standards. National ministries and agencies are
35
reluctant to cede authority to sub-national units. Less obviously, state and local
officials in education and health departments often are not enthusiastic about
increased authority and responsibility. They fear the national government will
not transfer adequate funds in a timely manner, and they are reluctant to deal
directly with feisty teachers’ and health workers’ unions. In the late 1980s,
Venezuela adopted a decentralization program that permitted but did not
require state governments to assume responsibility for education and/or health
care delivery. Few states opted to take formal responsibility for either sector,
though many launched some initiatives (Navarro, 2000: 209).
Decentralization does not eliminate the politics of reform: it simply shifts
much of the politics to the state or local level. State and local politicians, like
their national counterparts, face similar obstacles and have only weak incentives
to pursue reforms that alter institutions or link rewards more closely to perfor-
mance. However, education and health usually consume much larger parts of
state and local than national budgets. Therefore, governors and mayors may pay
more attention to these sectors, both as potential sources for winning support
and as important determinants of fiscal balance or imbalance. Political parties
and citizens’ groups are also more likely to mobilize around problems of social
service delivery at the local than the national level. Many of the most interes-
ting innovations in social services have been introduced at the level of states
or cities. The Brazilian state of Ceara introduced changes in monitoring health
care delivery, now imitated elsewhere in Brazil. Minas Gerais and several other
Brazilian states promoted major education reforms, as have San Luis in Argen-
tina and Bogota, Colombia. Participatory budgeting in Brazil is the product of
municipal initiatives.
Local leaders’ values and commitment, the local party system, and social and
economic structure and traditions shape the priority placed on pro-poor mea-
sures and on better social services in general. The central government may also
influence pro-poor efforts at provincial and local levels, if central authorities
themselves give this goal high priority. Central ministries have important roles
even in highly decentralized systems: they must manage financial transfers in
timely and predictable manner, develop ways to subsidize poorer districts to
counteract intrinsic tendencies for gaps to widen, develop and assess standards
and collect and analyze data. Beyond these basic functions, they can encourage
reforms through information, networking and clearing-house activities, and
contingent funding. Despite highly decentralized education and health sectors
in Brazil, the federal Ministries of Health and Education have effectively used
contingent transfers to press state and local governments to encourage preven-
tive public health programs and efforts by municipal schools to enroll more
students (Arretche, 2004; Draibe, 2004).
threatening. Principals can often guide and limit parents’ and community par-
ticipation. Hospital directors have less control over capitation and related fun-
ding changes used to strengthen clinic and hospital accountability.
ding authority to hire (from a pre-screened short list) and fire school directors.
When the governor who had championed this reform left office, his patronage-
oriented successor reversed many of the changes.
Citizen councils in the health sector are less widespread and may well be
less effective on average than school councils. Ordinary people are likely to feel
even less competent to assess clinic or hospital performance than school per-
formance, and clinic and hospital directors are probably still more intensely op-
posed to what they view as lay interference in management. Moreover, school
councils represent a set of concerned parents renewed each year as new cohorts
of children move through the schools. There is no analogous self-renewing pool
of interested users for clinics and hospitals. In l998, as part of a push towards in-
creased hospital and clinic autonomy, Costa Rica mandated community health
councils for each of the hospitals and larger clinics in the country. Council
members are elected for two year terms, and have substantial responsibilities
including overseeing budget execution, drafting performance contracts, and se-
lecting directors. But at least in initial years, surveys indicate that three-quarters
of the councils played little role in administration of their hospitals or clinics.
The biggest obstacle they faced was refusal of the local health authorities to
cooperate (Clark, 2004: 208-209).
Colombia’s complex health sector reform of l993 established citizen coun-
cils from local to national level; these have mostly exercised little initiative. The
CLAS clinics9 of Peru are a striking exception to the pattern of ineffective user
participation in the health sector. Launched in l994 on a pilot basis, the sys-
tem spread gradually at the discretion of regional health authorities. Although
seriously threatened in the late l990s (Ewig, 2004: 234, 237), by 2006 CLAS
councils co-managed 35 percent of all primary health facilities (Altobelli,
2006).
Are citizen councils effective pro-poor advocates? Where clients are mainly
poor, councils could be powerful instruments for more pro-poor service. In
more mixed communities or neighborhoods, experience suggests that coun-
cils are likely to be dominated by the somewhat better-off. To the extent that
interests of the not-so-poor diverge from their poor neighbors’ concerns, the
councils will be less effective pro-poor advocates, though they may improve the
over-all quality of service (and thereby also benefit the poor).
9
Comités Locales de Administración en Salud (Local Health Administration Committees).
39
Summing up
Expanding the scope and coverage of social services has long been recognized
as much easier, in political terms, than institutional reforms. This paper tries
to probe more deeply into the varied politics of different kinds of social ser-
vice reforms, both those that are explicitly pro-poor and broader changes that
potentially can benefit the poor.
Regarding increased coverage and scope: the challenge is to raise the bargain-
ing power of the poor or their advocates, relative to middle class pressures on
budget allocations. Possible strategies include: changing perceived priorities by
publicizing data on inequality; bundling targeted pro-poor programs with broa-
der expansion; participatory budgeting or other changes in budget procedures;
targeting broadly enough so that beneficiaries become a political constituency
(as with CCTs).
Like higher spending, changes in governance are not automatically pro-poor.
The challenge is to include pro-poor features. Decentralized service delivery is
more likely to be pro-poor if national governments find ways to make central
funding partially contingent on poor-poor measures. Increased autonomy for
schools or hospitals can become pro-poor if linked to effective user oversight.
However, low-income users often need technical aid and support to exercise
oversight effectively.
Reforms designed to link providers’ incentives to better performance have
proved extremely difficult to adopt and implement, and with rare exceptions
have had limited impact. Yet providers’ performance is crucial to better quality.
The issue is so intensely political, and political circumstances vary so widely,
that seeking “best practice” may be futile. Sustained political commitment plus
imagination, informed by knowledge of good and bad local and international
experiences, may be as much of a “formula” as is possible on this topic.
40
References
Altobelli, Laura, 2006, Peru Country Director, Futuras Generaciones Peru, personal
communication to the author, March 24.
Arretche, Marta, 2004, “Toward a Unified and More Equitable System: Health Re-
form in Brazil,” in Robert R. Kaufman and Joan M. Nelson (eds.), Crucial Needs, Weak
Incentives: Social Sector Reform, Democratization, and Globalization in Latin America,
Washington D.C.: Woodrow Wilson Center Press; Baltimore and London: Johns Hop-
kins University Press.
Avritzer, Leonardo, 2008, Participatory Institutions in Democratic Brazil, Washington
D.C.: Woodrow Wilson Center Press, forthcoming.
Boylan, Delia, 1996, “Taxation and Transition: The Politics of the l990 Chilean Tax
Reform,” Latin American Research Review, vol. 31, no. 1, pp. 7-31.
Clark, Mary, 2004, “Reinforcing a Public System: Health Sector Reform in Costa
Rica,” in Robert R. Kaufman and Joan M. Nelson (eds.), Crucial Needs, Weak Incentives:
Social Sector Reform, Democratization, and Globalization in Latin America, Washington
D.C.: Woodrow Wilson Center Press; Baltimore and London: Johns Hopkins University
Press.
Center for Education Reform, 2005, Charter Schools Still Making Waves, available at
www.edreform.com,
Delannoy, Francoise, 2000, Education Reforms in Chile, l980-98: A Lesson in Pragma-
tism, Country Studies, Education Reform and Management Publication Series, vol. 1, no.
1, June, Washington, D.C.: World Bank.
Draibe, Sonia M., 2004, “Federal Leverage in a Decentralized System: Education Re-
form in Brazil,” in Robert R. Kaufman and Joan M. Nelson (eds.), Crucial Needs, Weak
Incentives: Social Sector Reform, Democratization, and Globalization in Latin America,
Washington D.C.: Woodrow Wilson Center Press; Baltimore and London: Johns Hop-
kins University Press.
Economic Commission for Latin America and the Caribbean (ECLAC), 2005, CE-
PAL News. Vol. XXV, No. 6, June.
US Department of Education, 2003, Challenge and Opportunity: The Impact of Char-
ter Schools on School Districts, available at www.ed.gov/rschstat/eval/choice/ summary.
html
Ewig, Christina, 2004, “Piecemeal but Innovative: Health Sector Reform in Peru,” in
Robert R. Kaufman and Joan M. Nelson (eds.), Crucial Needs, Weak Incentives: Social
Sector Reform, Democratization, and Globalization in Latin America, Washington D.C.:
Woodrow Wilson Center Press; Baltimore and London: Johns Hopkins University Press.
Fiszbein, Ariel (ed.), 2005, Citizens, Politicians, and Providers: The Latin American Ex-
perience with Service Delivery Reform, Washington, D.C.: The World Bank.
Foray, D., Richard Murnane and Richard Nelson, 2007, “Randomized Trials of Educa-
tion and Medical Practices: Strengths and Limitations”, Economics of Innovation and New
Technology, vol. 16, forthcoming.
Gershberg, Alec Ian, 2004, “Empowering Parents while Making Them Pay: Autono-
mous Schools and Education Reform Processes in Nicaragua,” in Robert R. Kaufman and
Joan M. Nelson (eds.), Crucial Needs, Weak Incentives: Social Sector Reform, Democrati-
zation, and Globalization in Latin America, Washington D.C.: Woodrow Wilson Center
Press; Baltimore and London: Johns Hopkins University Press.
Grindle, Merilee, 2004, Despite the Odds: The Contentious Politics of Education Reform,
Princeton, N.J., Princeton University Press.
Huber, Evelyne, Thomas Mustillo and John Stephens, 2004, Determinants of Social
Spending in Latin America, paper prepared for the meetings of the Society for the Ad-
vancement of Socio-Economics, Washington, D.C., July.
41
Lewis, Maureen, 2005, “Governance and Corruption in Public Health Care Systems”,
Washington D.C., Center for Global Development, draft manuscript, November.
Lloyd-Sherlock, Peter, 2004, “Ambitious Plans, Modest Outcomes: The Politics of
Health Care Reform in Argentina,” in Robert R. Kaufman and Joan M. Nelson, (eds.),
Crucial Needs, Weak Incentives: Social Sector Reform, Democratization, and Globalization
in Latin America, Washington D.C.: Woodrow Wilson Center Press; Baltimore and Lon-
don: Johns Hopkins University Press.
Lowden, Pamela S., 2004, “Education Reform in Colombia: The Elusive Quest for
Effectiveness,” in Robert R. Kaufman and Joan M. Nelson (eds.), Crucial Needs, Weak
Incentives: Social Sector Reform, Democratization, and Globalization in Latin America,
Washington D.C.: Woodrow Wilson Center Press; Baltimore and London: Johns Hop-
kins University Press.
Melo, Marcus Andre, 2004, The Politics of “Hard Wiring”: Social Policy and the Trans-
formation of Brazilian Federalism, paper prepared for presentation at the Latin American
Studies Association (LASA) meeting, Las Vegas, October 5-7.
Murnane, Richard and David Cohen, l986, “Merit Pay and the Evaluation Problem:
Why Most Merit Pay Plans Fail and a Few Survive”, Harvard Educational Review, vol. 56,
no. 1, pp. 1-17.
Navarro, Juan Carlos, 2000, “The Social Consequences of Political Reforms: Decen-
tralization and Social Policy in Venezuela”, in Joseph S. Tulchin and Allison M. Garland,
Social Development in Latin America, Boulder, CO.: Lynne Rienner with the Latin Ame-
rican Program of the Woodrow Wilson International Center for Scholars.
Navarro, Juan Carlos (ed.), 2002, Quienes son los Maestros? Carreras e Incentivos
docents en America Latina, Washington, D.C.: Inter-American Development Bank.
Navarro, Juan Carlos, 2005, “Two Kinds of Education Politics,” in Ernesto Stein, Ma-
riano Tommasi, Koldo Echebarría, Eduardo Lora and Mark Payne (eds.), The Politics of
Policies: Economic and Social Progress in Latin America, 2006 Report, Washington D.C.:
Inter-American Development Bank.
Nelson, Joan M., 2005, Democracy, Social service Expansion and Institutional Reforms
draft paper, November.
Nelson, Joan M., 2003, ”Grounds for Alliance? Overlapping Interests of the Poor and
Not So Poor,” in Peter P. Houtzager and Mick Moore (eds.), Changing Paths: International
Development and the New Politics of Inclusion, Ann Arbor, MIT.: University of Michigan
Press.
Ramirez, Patricia, 2004, “A Sweeping Health Reform: The Quest for Unification, Co-
verage and Efficiency in Colombia”, in Robert R. Kaufman and Joan M. Nelson (eds.),
Crucial Needs, Weak Incentives: Social Sector Reform, Democratization, and Globalization
in Latin America, Washington D.C.: Woodrow Wilson Center Press; Baltimore and Lon-
don: Johns Hopkins University Press.
Saliba, Alcyone (former Secretary of Education, Parana, Brazil), 2005, “Parents’
Associations and School Report Cards in Parana, Brazil”, Presentation at a World Bank
conference, Washington DC, January 13-14.
Stasavage, David, 2005, “Democracy and Education Spending in Africa,” American
Journal of Political Science, vol. 49, no. 2, pp. 343-348.
Vegas, Emiliana and Ilana Umansky, 2005, What Can We Learn From Education
Reforms in Latin America? Washington, D.C.: World Bank.
World Bank, 1993, World Development Report: Investing in Health, Oxford University
Press: World Bank.
World Bank, 2004, World Development Report: Making Services Work for Poor People,
Oxford University Press: World Bank.
42
Introduction
Governments in most developing countries are poor providers of social services
ranging from education to health. Poverty itself has been blamed for this, to
the extent that the poor have a low demand for high quality social services or
that poor country governments lack capacity to provide them. If poverty itself
explains service delivery failures, financial assistance to developing countries is
an immediate and obvious response. The case for aid is strengthened if service
delivery improvements, such as those that improve education and health, also
drive growth (see, for example, Weil’s 2005 estimates of the large explanatory
effect of health status on cross-country differences in income). If better service
delivery leads to growth, then aid to improve service delivery may trigger sus-
tainable reforms in which the need for external assistance eventually falls.
This picture is overly optimistic, however. Citizen demand for quality edu-
cation or health care, even among the poor, is not low; and, except in the very
poorest and smallest countries, even poor countries have shown themselves able
to build capacity in areas of importance to their governments. Instead, then, it
appears that service delivery dysfunction is rooted in weak political incentives
to provide quality social services to citizens. Under these circumstances, finan-
cial transfers are insufficient for improved service delivery.
Although most donor activity continues to focus on such transfers, an awa-
reness of the political obstacles to improved service delivery are evident in im-
portant, though scattered donor efforts to address service delivery inadequacy
by enhancing user participation, publicizing information about financing, or
assembling and disseminating citizen report cards. These innovative strategies
are all implicitly based on improving political incentives to provide services.
The expansion of these efforts is hobbled, however, by our lack of a precise
understanding of why politicians tolerate poor service provision by government
in the first place.1
The difficulty is best illustrated with a puzzle. Elections are a basic form of
citizen participation in government, in principle endowing citizens with signi-
ficant leverage over policy makers, since they allow citizens to expel poor per-
formers from office at low cost to themselves. However, social services in poor
countries with elected governments are little better and sometimes significantly
worse than in poor countries with non-elected governments. Table 1 demon-
strates the depth of the puzzle. It compares service delivery across poor demo-
cracies and non-democracies. In all cases, the differences between democracies
Disclaimer: The findings, interpretations and conclusions expressed in this paper are entirely those
of the author and do not necessarily represent the views of the World Bank, its Executive Directors,
or the countries they represent.
1
Mansuri and Rao (2004) criticize community-based development efforts in part precisely
because their efforts to encourage popular participation run into political obstacles.
43
Primary school completion rate (of age appropriate children) (s.d.= 26.8) 70.7 (40) 80.7 (26)
Child mortality (deaths of children under 5, per 1000) (s.d.= 68.4) 110.1 (54) 85.3 (36)
Acute respiratory illness, 1995 – 2000, (s.d.=12.0) 14.5 (36) 21.4 (31)
Of those, how many treated?, 1995 – 2000, (s.d.=19.3) 50.2 (37) 43.9 (31)
Diarrhea, cut off average, 1995 – 2000, (s.d.=7.4) 17.3 (44) 19.2 (35)
Of those, how many oral rehydration salts? 1995-2000, (s.d. = 13.4) 30.3 (45) 32.5 (36)
% with DPT immunization, 12-23? Year=1998 (s.d.=22.1) 71.1 (54) 78.2 (42)
Total government expenditures/GDP, 1998, (s.d. = .09) .27 (21) .24 (18)
Private sector development
Days in customs 6.8 (15) 8.15 (16)
Paved roads/total roads, 1998 44.8 (11) 38.9 (14)
Corruption (0 – 6, least corrupt = 6), 1997 2.7 (25) 2.9 (34)
Bureaucratic quality (0 – 6, 6 = highest quality), 2000 2.3 (28) 2.4 (30)
Rule of law (0 – 6, 6 = highest quality), 2000 3.7 (28) 2.9 (30)
Note: Poor countries are those poorer than the mean country for the years from which the data
comes. Corruption, Bureaucratic Quality and Rule of Law from Political Risk Services, International
Country Risk Guide. All other indicators from World Development Indicators, The World Bank.
The comparison in Table 1 also demonstrates that the puzzle of democratic per-
formance goes beyond service delivery. Along a variety of dimensions regarded
as important both to sound public sector administration and to private sector
development, democracies and non-democracies exhibit nearly the same per-
formance, with non-democracies sometimes exceeding democratic outcomes.
The explanation for this puzzle explored below, focused on political market
imperfections that dissuade politicians from pursuing service delivery impro-
vements, also can explain these broader performance characteristics of demo-
cracies.
and their own welfare. With regard to the first, past research has demonstrated
that political credibility is a crucial determinant of government policy choices
(Keefer and Vlaicu, 2005). In democracies, political competitors use credible
promises to citizens to influence their voting decisions. However, they have the
strongest incentives to make promises to citizens who believe them.
When politicians can make credible promises to only a fraction of the ci-
tizenry, they focus on targeted goods that benefit these citizens narrowly and
less on governments programs that broadly improve the welfare of all citizens.
The latter deliver benefits to citizens who do not believe their promises, and
therefore offer no political payoffs to the politicians. Under these circumstan-
ces, government jobs and infrastructure become the currency of politics: they
are easily targeted to narrow groups. Moreover, political restraints on corruption
ease, since the costs of corruption are easily shifted to the broad segments of the
population to which political competitors cannot make credible promises. Ser-
vice delivery improvements have substantial public good attributes, since the
steps that politicians must take to improve service delivery tend to require im-
provements that benefit all service recipients and are difficult to target. Social
services should therefore be under-provided in countries in which politicians
cannot make credible promises to most voters.
Information is also key to government accountability, however, and public
goods are heterogeneous in the extent to which citizens can easily track po-
litical responsibility for better or worse provision of them. As Majumdar, et
al. (2004) have shown, the greater are the information asymmetries between
citizens and politicians associated with a particular public good, the more likely
are politicians to subordinate that public good to others in which their actions
are more easily monitored by citizens.
Service delivery is, in particular, rife with information asymmetries. It is dif-
ficult for citizens to identify how political actions have affected their health or
the educational progress of their children; it is easy, on the other hand, for them
to assign political responsibility for the appearance of electricity to or roads near
their homes. If citizens only observe their own private experience with service
providers, they may not be able to determine even whether it was good or
bad, much less whether to attribute bad or good performance to the actions of
the service provider, to the bureaucratic apparatus standing behind the service
provider or to the decisions of politicians. The effects of this information asym-
metry are magnified because the benefits of service provision depend heavily
on private characteristics of the service recipients – how healthy they are or
how academically astute, for example. As a consequence, even when politicians
develop the capacity to make credible promises to most citizens, they may still
have limited incentives to provide public goods that are subject to severe infor-
mation asymmetries (see Keefer and Khemani, 2005, for a review).
The fact that “contractual” problems of information and credibility limit the
ability of citizens to hold government officials accountable for public good pro-
vision in democracies does not fully explain why poor democracies perform
about as well as non-democracies. One might expect that these contractual
problems would be at least as severe in non-democracies, where citizens are
apparently even less able to sanction non-performing officials. In fact, as the
brief discussion at the end of this chapter indicates, although leaders of non-
45
democracies need not appeal to voters, their survival is enhanced to the extent
that they can make credible promises to citizens, since broad citizen support is
a deterrent to potential challengers. Those non-democracies that solve this pro-
blem of credible commitment may therefore perform better than those demo-
cracies that do not. The ability to overcome contractual obstacles to politician
accountability to citizens therefore explains not only why some democracies
perform better than others, but also why some democracies perform worse than
some non-democracies.
Evidence reported below affirms earlier research about the importance of
credibility for public good provision: when politicians can make credible pro-
mises only to limited constituencies, corruption is higher, bureaucratic quality
and the rule of law are lower, and the political emphasis on non-public goods,
such as jobs in the government, is higher. However, service delivery indicators,
particularly those associated with health, appear to be less influenced by these
measures of the breadth of the citizenry to whom politicians can make credible
promises. This is true in both democracies and non-democracies. Information
asymmetries are the most likely explanation for this, and therefore loom large as
additional barriers to service delivery. That is, overcoming political obstacles to
service delivery reform requires that interventions allow citizens both “to see”
that the country’s leaders have improved their welfare through their service
delivery efforts, and “to believe” the promises of political competitors that they
will undertake actions to improve broad social welfare.
This message contrasts sharply with the main thrust of development stra-
tegies, particularly in struggling countries. For example, one conclusion of the
development community’s work on fragile states is to focus first on service de-
livery in post-conflict or other troubled states (World Bank, 2002; Chauvet and
Collier, 2004). The evidence here suggests that, as a political matter and across
all countries, service delivery is the most challenging area of policy to improve.
Not only does it require that politicians have incentives to reach broad seg-
ments of the population, but also that they have the administrative apparatus
to deliver services and a population sufficiently informed to give them credit
for their efforts.
2
Rent-seeking, or corruption, are greater when political decisions have less of an influence over
electoral outcomes. When political competitors are not credible to most voters, they can do little
to use policy to sway those voters in the event of adverse popularity or other shocks. The electoral
price of corruption is therefore lower (since with a greater probability than if they were credible,
they are expelled from office independent of their performance – see Persson and Tabellini, 2000,
Chapter 8).
46
3
This leads to a counter-intuitive result. The typical conclusion of models of political competition
with credible parties and ideologically motivated voters is that the smaller the fraction of swing
voters, the greater the focus on targeted goods and the less on public goods (see Persson and Tabel-
lini, 2000). However, in the absence of ideologically motivated voters, when all voters are essentially
swing but in which only individual candidates are able to swing them, it is actually targeted goods
provision that increases.
4
Competitive elections are determined by values of the legislative and executive indices of
competitive elections from the Database of Political Institutions, Beck et al. (2001). When these
are both equal to the maximum, seven, indicating that multiple parties can and do compete in the
election and no party gets more than 75 percent of the vote, elections are competitive.
5
Spending on education can be slotted either to public goods (the quality of education) or tar-
geted goods (school buildings, patronage positions for teachers). The political variables in Table 2
capture the extent to which that spending, controlled for in the education regressions, is spent “pu-
blicly”; to the extent that it is not, we expect school enrollment and completion to fall as families
react to lower quality by withdrawing their children.
48
N.B. t-statistics in parentheses. Every specification controls for total population, land area, ppp-
adjusted income/capita, percent population young, and percent rural. The two education regres-
sions both control for total government education spending/GDP. The gross secondary school
enrollment regression controls for gross primary school enrollment. Regressions are time-series,
cross section, OLS. Robust standard errors are reported, clustered by country so that multiple
country-regimes from the same country are not treated as independent observations. Election year
results are unchanged when the units of observation are democratic episodes (as in Keefer 2005b)
rather than country years. Only country-years with competitive elections are considered (where
the legislative and executive indices of competitive elections from Beck, et al. 2001 are at their
maximum values of seven). All regressions include a constant (not reported).
6
Effects on targeted goods are also more ambiguous, but this is more likely due to measurement
error, since although it is most likely that high public investment and government wage spending
are targeted, they need not be and could, on the contrary, represent government efforts to provide
public goods.
7
The relative insensitivity of service delivery to political variables may also be due to demand-
side explanations. For example, immunization campaigns are easily observed, but they are not
particularly popular. Where demand is weak, so are the political payoffs, so credible politicians
expend efforts on other public policies. Nevertheless, it should still be the case that a government
motivated to provide the public good of immunization would also invest in the public good of
information about the benefits and necessity of immunization.
50
who will defend the regime from challenges. The reliance on personal ties, in
non-democracies as in democracies, diminish political incentives to improve
bureaucratic quality or to curb corruption, simply because personal relation-
ships with the leadership must play an important role in personnel decisions,
displacing merit and performance. These distortions potentially diminish over
time, to the extent that leaders who remain in power for a longer period of
time build up a personal reputation with a larger fraction of the citizenry, much
as political parties in democracies can become more credible or programmatic
with the passage of time (Clague, et al., 1996 also make this argument). Finally,
unelected leaders can build up an institutionalized ruling party to extend an
umbrella of credibility over a broad group of citizens, even those who lack a
personal connection to the leaders. The broader is the group of citizens who
believe the leadership, the more efficient it is for leaders to improve public good
provision (see Haber, forthcoming and Keefer, 2006).
As in the case of elected leaders, the core empirical question is how to mea-
sure the credibility of non-elected leaders. Party characteristics again provide
one solution. A key characteristic of institutionalized ruling parties is that they
do not depend on the leader in power. Such parties exhibit some degree of
intra-party checks and balances, relatively institutionalized transitions from one
leadership group to the next, and promotion procedures inside the party that
are transparent and non-arbitrary.8 In fact, those autocracies known for high
levels of service delivery, particularly communist countries such as the former
Soviet Union or Cuba, also exhibit large and relatively institutionalized ruling
parties.
The extent to which a ruling party is institutionalized can be captured, first,
by looking at the years an individual leader has been in office. The effect of this
on institutionalization is not clear, a priori. The longer the leaders’ tenure, the
greater the opportunity the leader has had to build up an independent reputa-
tion and an institutionalized party. However, it could also be that the long-lived
leader has also had a greater opportunity to shape the ruling party around his
personal leadership. The results below support the former interpretation. Data
on the tenure of the executive is available from the Database of Political Institu-
tions (Beck, et al., 2001).
Second, we can ask whether ruling party has a programmatic identity – to the
extent that it does, it can make credible commitments to a broader fraction of
the population. The regressions below use a dummy variable, coded as one if
the ruling party is categorized as right, left or center in the Database of Political
Institutions (Beck, et al., 2001). Third, we can look at the age of the leader’s
party and ask whether it is older than the leader’s tenure. Specifically, the reg-
ressions below ask whether a leader’s party has existed at least 10 years longer
than the leader herself has been in power. To the extent that it has, controlling
for the leader’s own personalized influence, we can assume that the ruling party
is more institutionalized and the capacity of the regime to provide public goods
is greater.
The last two indicators are highly correlated. Focusing only on countries with
8
The Chinese Communist Party has these characteristics; most autocracies do not (see Whiting,
2001 and Yang, 2004 for ways in which the CCP has institutionalized).
52
no elections, 77 percent of parties at least ten years older than the leader’s
tenure are also programmatic; only 11 percent of non-programmatic parties are
also old. Two specifications are therefore employed. Both include the tenure of
the executive, while one controls for whether the ruling party is programmatic
and the other controls for its age.
The regressions in Table 2 investigate only countries where there were com-
petitive elections and no party won more than 75 percent of the votes or se-
ats. The regressions in Table 3, in contrast, examine all of those countries in
which there were either no elections or, if there were, no opposition party won
any seats at all.9 The first set of regressions reports the estimated effects of the
variables capturing leader tenure and whether the ruling party was program-
matic, using the same specifications as in Table 2. The second set of regressions
does the same, substituting the ruling party age dummy for the programmatic
dummy. The state ownership of newspapers is omitted because sample sizes are
too small.
Overall, the results suggest that policy is in fact sensitive to the institutiona-
lization of political parties in non-democracies and to the tenure of leaders, but
as with the democracy regressions, the effects are least pronounced for service
delivery and most pronounced for those related to the quality of the public ad-
ministration. At least one of the credibility variables is a statistically significant
determinant of corruption or bureaucratic quality in each of the two sets of reg-
ressions in Table 3. In the case of the three service delivery variables, however,
only the years in office of the executive is ever significant, and then only with
respect to primary school completion rates. Ruler and ruling party characteris-
tics are not strong with regard to targeted goods: both programmatic and long-
lived parties are associated with lower spending on public sector employment,
though only the programmatic variable is significant; long-lived parties are also
associated with higher, not lower, public investment.
One reason for the ambiguous results is that nearly all (93 percent) of the
observations classified as programmatic happen to be left-wing governments.
Although those governments do not do much to improve public good pro-
vision, they do appear to redistribute income: total government spending,
taking into account the controls used in the specifications of Table 3, is a full
12 percent of GDP greater in the countries with programmatic parties than in
countries without them. This likely affects the public investment regressions, as
well. In order to implement their program and to retain their credibility, these
governments in all likelihood engage in significant redistribution (about which
we have little information). Information is therefore a likely explanation of the
paradox: citizens can more easily observe and give credit to government for
redistribution (and large-scale public investment projects), but not for service
delivery improvements.
9
That is, only countries for which the legislative and executive indices of electoral competition
are both less than six.
53
N.B. t-statistics in parentheses. Specifications and other estimation details are the same as in the
corresponding Table 2 regressions, substituting the respective political variables. See Table 2 notes.
Only country-years with no competitive elections are considered (where the legislative and exe-
cutive indices of competitive elections from Beck, et al., 2001 are below six).
In troubled or fragile states, the same goal must be pursued, of increasing the
credibility of and information about political promises to as broad a fraction of
the population as possible. However, authentic and serious capacity constraints
and the intrinsic social divisions that often characterize such countries may
make traditional donor strategies of public good provision or improved service
delivery impractical vehicles for accomplishing these objectives. The response
of donors to these difficulties has often been to support non-governmental pro-
vision of services in the short and medium run. This does not build crucial
political credibility, however. In order to accomplish this, donors should also
pursue a parallel track of supporting government efforts to make and fulfill pro-
mises for which it does have the capacity. These promises may take the form
of unconditional handouts, such as food baskets or cash grants, available equally
to all citizens. In a perfect world, such handouts are an inferior development
strategy and one that donors universally abjure. However, as a basis for sustaina-
ble political development in a country with weak institutions and non-credible
politicians, they may be optimal: promises about such handouts are easily com-
municated; the fulfillment of those promises is easily observed.
These are not easy lessons. Transfers are costly and the effort to oversee go-
vernment implementation is expensive. Moreover, well-functioning political
systems have at least two broadly credible political competitors. Donor support
of government credibility necessarily disfavors the development of credible po-
litical competitors outside of government. Donors can address this problem by
operating both through government and through large NGOs with significant
grassroots presence. In this case, NGOs themselves can become potential politi-
cal entities able to make credible promises to broad segments of the population.
This strategy in turn raises its own issues, not least of which are those related to
sovereignty. Nevertheless, any attack on the fundamental obstacles to develop-
ment requires that these issues be confronted in a forthright manner.
References
Beck, Thorsten, George Clarke, Alberto Groff, Philip Keefer and Patrick Walsh, 2001,
“New Tools in Comparative Political Economy: The Database of Political Institutions”, World
Bank Economic Review, vol. 15, no.1, pp. 165-176.
Besley, Timothy and Robin Burgess, 2002, “The Political Economy of Government Respon-
siveness: Theory and Evidence from India”, Quarterly Journal of Economics, vol. 117, no. 4,
pp. 1415-1451.
Chauvet, Lisa and Paul Collier, 2004, “Development Effectiveness in Fragile States: Spillo-
vers and Turnarounds,” Centre for the Study of African Economies, Department of Economics,
Oxford University, Mimeo (January).
Clague, Christopher, Philip Keefer, Stephen Knack and Mancur Olson, 1996, “Property and
Contract Rights under Democracy and Dictatorship”, The Journal of Economic Growth, vol. 1,
no. 2, pp. 243-276.
Djankov, Simeon, Caralee McLiesh, Tatiana Nenova and Andrei Shleifer, 2003, “Who owns
the media?” Journal of Law and Economics, vol. 46, no. 2, pp. 241-382.
Haber, Stephen, forthcoming, “Authoritarian Government”, in Barry Weingast and Do-
nald Wittman (eds.), The Oxford Handbook of Political Economy, Oxford: Oxford University
Press.
55
Keefer, Philip, 2006, “Governance and Economic Growth in China and India”, in L. Alan
Winters and Shahid Yusuf (eds.), Dancing with Giants: China, India and the Global Economy,
Washington: The World Bank.
Keefer, Philip, 2005a, “Clientelism, Credibility and the Policy Choices of Young Democra-
cies”, Development Research Group: The World Bank, Mimeo.
Keefer, Philip, 2005b, Programmatic Parties: Where do they come from and do they mat-
ter? Paper presented at the 2005 meetings of the International Society for New Institutional
Economics, Barcelona, Spain.
Keefer, Philip and Razvan Vlaicu, 2005, Democracy, Credibility and Clientelism, Policy
Research Working Paper no. 3472, The World Bank.
Keefer, Philip and Stuti Khemani, 2005, “Democracy, Public Expenditures, and the Poor:
Understanding Political Incentives for Providing Public Services”, The World Bank Research
Observer, vol. 20, no. 1, pp. 1-28.
Majumdar, Sumon, Anandi Mani and Sharun Mukand, 2004, “Politics, Information and the
Urban Bias”, Journal of Development Economics, vol. 75, no. 1, pp. 137-165.
Mansuri, Ghazala and Vijayendra Rao, 2004, Community-based (and driven) development:
a critical review, World Bank Policy Research Working Paper no. 3209.
Persson, Thorsten and Guido Tabellini, 2000, Political Economics: Explaining Public
Policy, Cambridge, MA: The MIT Press
Strömberg, David, 2004, “Radio’s impact on public spending”, Quarterly Journal of Econo-
mics, vol. 119, no. 1, pp. 189-221.
Weil, David N., 2005, Accounting for the Effect of Health on Economic Growth, Brown
University Economic Working Paper No. 2005-07.
Whiting, Susan H., 2001, Power and Wealth in Rural China: The Political Economy of
Institutional Change, New York: Cambridge University Press.
World Bank, 2002, “World Bank Group Work in Low Income Countries Under Stress:
A Task Force Report”, World Bank Fragile States: the LICUS initiative, September.
Yang, Dali L., 2004, Remaking the Chinese Leviathan: Market Transition and the Poli-
tics of Governance in China, Stanford: Stanford University Press.
56
Stuti Khemani*
Introduction
Strategies to promote “good governance” are centre-stage in current develop-
ment policy discussions. One element that has been emphasized by several in-
ternational aid agencies is to use pre-identified and, as far as possible, objective,
governance indicators to determine aid flows. The underlying arguments for
this approach consist both of fulfilling fiduciary responsibilities to donors, and
to promote competition among countries to improve governance indicators
in order to access more aid. In parallel, there is increasing discussion around
how international agencies might strengthen the hands of citizens in develo-
ping countries to hold their own governments accountable. For some time now
the aid community has engaged in large-scale consultation with civil society
in developing countries, and moved away from the more traditional model of
dealing only with political representatives and bureaucrats. This paper makes a
new proposal of a structured intervention along these lines, based on political
economy research on determinants of government accountability, with the po-
tential of improving poor citizens’ capacity to demand and receive basic servi-
ces for development.
The proposal lies in an area of growing interest—to use information as a
tool to empower citizens in developing countries to hold their public agents
accountable. Civil society organizations have recently experimented with grass-
roots “information campaigns”, organizing public meetings around how public
resources are allocated and what services are delivered, to bear collective pres-
sure upon local service providers and compel them to improve performance
(Jenkins and Goetz, 1999; Goetz and Jenkins, 2001; Paul, 2002; Olken, 2005;
Banerjee, et al., 2006; Bjorkman and Svensson, 2006). Governments have re-
cently pursued legislation on “right to information” or “freedom of information”
to increase transparency and enable their citizens to probe allocation of public
resources and achievement of results (Banisar, 2004).
This paper reviews the literature on information campaigns and the role of
mass media in influencing public policy and argues that there are significant
gaps in knowledge and experience in this area, and that much work remains in
exploring “information” as a tool to strengthen political accountability for deve-
lopment. It argues that local information campaigns whose immediate objective
is to generate local participation cannot have sustainable or large-scale impact
on public services unless they change incentives of politicians who have ulti-
* I am grateful to Shanta Devarajan, Karla Hoff, Philip Keefer, Santosh Mathew, Ingrid Widlund, and
other EGDI workshop participants for helpful comments. The findings, interpretations, and conclu-
sions expressed in this chapter are entirely those of the author, and do not necessarily represent the
views of the World Bank, its Executive Directors, or the countries they represent.
57
mate authority over the management of public employees and public budgets.
What kind of information to citizens can change political incentives to serve the
poor? This is the question addressed in the paper.
The answer that is proposed is information on development outcomes, that
public policies are designed to address, that is statistically representative at the
level of the lowest electoral district, or similarly disaggregated politically rele-
vant jurisdiction in a country. The paper argues that if such information is made
available in a credible manner to citizens on a regular basis, so they can compare
performance in one electoral district to another in achieving development indi-
cators, and monitor improvements (or lack thereof) over time within a district,
it can potentially lead to voter coordination in evaluating governments on this
basis, and thereby promote yardstick competition between districts to improve
development performance.
The next section discusses why information might be a binding constraint
on improving accountability of public providers and political representatives
for basic services. Section 3 describes emerging policy experiments to address
this potential constraint—information campaigns to improve public services for
the poor by mobilizing local collective action. Section 4 proposes a new type
of information campaign—designed to make basic human development outco-
mes more salient in political competition—and thereby strengthen incentives
of political representatives to find solutions to development problems. Section
5 concludes.
sibility for key policy outcomes (Ferejohn and Kuklinski, 1990). These studies
show that voters adopt simple voting criteria based on very limited information
about politics and public policies. Fiorina (1990) emphasizes that the informa-
tion that people use when they vote comes from the ordinary performance of
social and economic roles and is therefore “free”. Fiorina and Shepsle (1990)
and Chappell and Keech (1990) argue that citizens can employ voting rules
requiring very little information and still motivate politicians to pursue policies
in their interest. Ferejohn (1990: 8-9) captures this process as follows: “find a
way to get the electorate to commit itself to act as though it is a simple principal
with a one-dimensional set of rewards. In this way, incumbents will be preven-
ted from taking advantage of the conflicting interests in the electorate.” This
suggests that information can play a coordinating role, focusing public attention
on particular policies for greater political accountability.
Keefer and Khemani (2005) argue that the nature of information likely to be
“freely” available in developing countries is responsible for a “political market
failure” in that it creates for undue public attention to inefficient, and someti-
mes ineffective, policies of targeted transfers in the form of public-sector jobs
and price subsidies, shifting effort and resources away from public policies for
basic development outcomes. Their analysis points to two kinds of information-
related political market failures. One, although people know whether their own
child died or not, and whether their neighborhood health worker did anything
to help them, they are unlikely to hold their remotely located politician ac-
countable for this private experience. They typically do not have information
on aggregate development consequences that is more likely to be linked to the
actions of their political representatives—they don’t know for example, how
many children in their politician’s electoral district do not survive beyond the
age of five, and whether the politician is taking action in terms of greater im-
munization efforts or public health campaigns. Professional journalists are not
trained to be able to generate the kind of data needed to calculate development
indicators at the level of disaggregated political jurisdictions which citizens
could use to evaluate their representative.
Two, even if they can guess that others in their neighborhood are suffering
similar tragedies, people might be apathetic to using actual development out-
comes as an indicator of politician performance, focusing instead on simple
actions they can directly observe, such as announcement of a price subsidy or
provision of jobs, because they do not expect politicians to be able to improve
these difficult outcomes. They don’t have access to or the capacity to understand
“expert” studies that argue that politicians could impact outcomes like the rate
of child mortality and incidence of disease through simple policy levers. If a so-
ciety is stuck in this vicious cycle of low performance and low expectations for
broad development outcomes, with efforts instead expended upon scrambling
for private benefits from public resources, incumbent politicians have stronger
incentives to provide targeted benefits rather than broad public goods, and op-
position politicians have trouble mounting a credible challenge on the platform
of broad development issues.1
1
Banerjee and Duflo (2006) describe such general public apathy to improving broad social ser-
vices in developing countries.
60
ned that they were not getting these scholarships, whilst teachers complained
that parents inappropriately enroll under-age children, that can’t and don’t at-
tend school, just to lay claim to the scholarships. The second issue that attracted
attention was a new government mid-day meal program. Actual learning levels
attracted the least attention, and the facilitators had a difficult time steering the
conversation away from scholarships and school meals to the broader issue of
learning. That is, the field experience lends support to the story of Keefer and
Khemani (2005) that citizens often mobilize to demand private transfers from
public policies, but less so for broad public goods such as improvements in ser-
vice quality for everybody.
The average attendance in these meetings consisted of about 108 villagers,
which seems a large gathering, with village total population (all ages) ranging
from five hundred to five thousand. These meetings were followed-up with
small group meetings with education committee members who were provided
with pamphlets about their roles and responsibilities in education service deli-
very. The hypothesis behind these interventions was that once key community
members were informed about local agency, they would participate more acti-
vely through it to improve services, and citizens at large would thence become
informed and aware of local agency.
Follow-up surveys were undertaken in the same 280 villages in March 2006,
3-6 months after the information campaigns were implemented, and we are
currently in the process of analyzing impact. The most surprising fact emerging
is that the campaigns did not lead to any improvements in citizens’ lack of
knowledge of local agency, specifically of the formal education committees. We
also find no effect on school performance and learning, but this is less surprising
given the narrow time frame in which we attempt to measure impact, and our
field experience which suggested that the odds were heavily stacked against the
interventions with existing public apathy to learning. But at the very least, if
community participation is to have any hope of influencing service delivery, we
expected a significant number of citizens to become aware of the existence of
local education committees and its potential role in improving service quality.
In sharp contrast to this experience with education services in India, a similar
experiment with citizen participation in health services in Uganda resulted in
substantial improvements in service provision in village health clinics, reduc-
tions in under-five child mortality, and weight-gains of infants (Björkman and
Svensson, 2006). Why these experiences are so different requires more time to
answer convincingly—these results are only just emerging. However, we explo-
re some preliminary hypotheses below for why experiences might be sensitive
to local context.
All of the approaches for “information campaigns” described above are based
on the assumption that information alters the incentives of frontline service
providers and local bureaucrats by mobilizing citizens for collective action.
However, a large literature on local participatory institutions suggests that suc-
cess is context-specific – some areas with pre-existing institutions of civic enga-
gement might succeed, whilst others with entrenched social inequalities might
fail (Baland and Platteau, 1999; Mansuri and Rao, 2003). Poor people might
indeed be well aware that teachers, doctors and local officials are not doing their
job, but feel powerless to change anything because teachers and doctors are
63
elite members of the community, or people with political influence. Drèze and
Gazdar (1996) recount how a village school in Uttar Pradesh can be non-func-
tional for as long as ten years due to teacher absenteeism and shirking, without
any collective protest being organized. Developing countries are characterized
by historical institutions of social inequality and social polarization whose im-
pact is persistent (Banerjee and Iyer, 2005), which likely makes local collective
action for improvements in public goods for everybody difficult to sustain.
Furthermore, Keefer and Khemani (2004, 2005) argue that public service
providers have weak incentives to improve performance quality because their
jobs are protected by political agents – politicians have stronger incentives to
provide secure public-sector jobs as teachers, health workers, and local bureau-
crats, than to pressure these job-holders to improve service delivery. If informa-
tion campaigns are designed to galvanize local collective action to improve local
services, they might be successful in temporarily improving outcomes around
the campaigns, but the political analysis suggests that their impact will die out
once “politics-as-usual” kicks in unless they also change broader political incen-
tives. From this perspective, it would be worthwhile to study successful expe-
riences like the one in Uganda analyzed by Bjorkman and Svensson (2006) over
time to see if the impact is sustained.
the policy platforms on which elections are contested. Finally, we can evaluate
whether making such data and information available eventually improves public
policies and the development indicators.2 The assumption behind why such a
minimalist strategy might work and is worth testing is that local media markets
and civil society organizations would pick-up and publicize these indicators on
their own because they are dramatic demonstrations of development, or lack
thereof, at politically relevant levels of aggregation; and because by dissociating
the production of the data with its dissemination it might be easier to establish
the credibility and political independence of the data producers.
At an intensive interventionist level, this campaign might partner with
local NGOs and civil society organizations, with local credibility in genuine-
ly working towards improving development outcomes, to share the informa-
tion with citizens through participatory methods like the “report card” cam-
paigns discussed in The previons section. The difference with the “report card”
campaigns that have happened thus far is the nature of the information being
shared—combining public education about public policies with measurable
and regularly available indicators of their success or failure at the level of politi-
cally-relevant jurisdictions.
At an intermediate level of intervention, brief advertisements might be pre-
pared for district-level radio and/or TV broadcasts announcing the district’s de-
velopment outcomes, relative to the country or regional average, and the official
view of the major public policies being pursued to address them, ending with an
advocacy message that citizens should raise questions with their political repre-
sentatives on how these outcomes can be improved. These advertisements need
to be designed to capture the attention of a large number of people, so they
are likely to be brief, but can alert listeners to newspaper advertisements for
more details. That is, the radio and TV spots could be followed by a newspaper
advertisement which provides the same information in writing, with some more
detail. This campaign could be intensely pursued both around elections and in
the middle of a political term in office. After an initial investment in direc-
tly publicizing these district-level development indicators, and official view of
public policies designed to address them, the development agency engaged in
this effort can revert to the minimalist agenda of ensuring public availability.
The driving assumption behind the idea proposed here is that the informa-
tion constraint that reduces government accountability for broad development
outcomes cannot be usefully viewed as simple “lack of” information, and hence
be addressed by providing more and more information, because of serious limits
on people’s cognitive abilities. Recent work on the psychological underpinnings
of social communication indicates that getting information to have the desired
impact on actual outcomes is a particularly difficult mechanism design problem
(Lupia, 2003). The problem is as follows–suppose there is a target audience
that lacks sufficient information to accomplish certain tasks; what kind of infor-
2
In keeping with cutting-edge evaluation methodology, the strategy to evaluate impact can be
one of randomized experimentation. Data on development performance could be collected for a
selected sample of jurisdictions in a small geography, and the information would be revealed in the
first and experimental stage for only a randomly chosen sub-set of jurisdictions constituting the
“treatment” group. The group whose information would be collected but not revealed would be the
“control” group.
66
mation and dissemination strategy is necessary and sufficient to give the target
audience the requisite skill to accomplish those tasks? Lupia (2003) provides
a framework for thinking about the content and nature of information disse-
mination–that it should be designed to maximize the possibility that the target
audience digests and retains the information, and actually uses it when making
a choice for which it is relevant.
If the target audience is poor citizens in developing countries, and the task
is for them to hold their governments accountable for basic development out-
comes, the new idea proposed here is based on the assumption that providing
information on the final outcomes, which have dramatic value, to large numbers
of citizens can shift public focus to these outcomes, and make them politically
salient in the sense that voters can coordinate on them as the basis for evaluating
their governments. Such coordination is likely to be further facilitated if these
indicators are developed at the level of electoral districts or local governments,
because they can then be used to compare performance across political jurisdic-
tions, thus generating yardstick competition among political representatives.
Voter coordination and yardstick competition among politicians on the basis
of development outcomes shifts the burden of designing policies and allocating
scarce public resources effectively to accomplish them, onto political represen-
tatives with access to “experts”. This is similar to ideas in the literature on cor-
porate governance where the principal-agent problem is almost identical–there
are multiple principals, “shareholders”, with attendant problems of collective
action and lack of expertise, and agents, “managers”, whose actions and their
impact on outcomes, “profits and dividends” are difficult to observe and un-
derstand. Prat (2005) draws upon this literature to show that if the agent is an
“expert” and principals are amateurs, the expert agent’s interest will be more
aligned with the principals’ interest when the latter is informed about the con-
sequences of the agent’s actions rather than the actions itself.
The above is an idea, not a formal argument based on systematic evidence
on what people know of development outcomes and public policies, nor on
systematic evidence on how poor people evaluate governments, because of lack
of such evidence from developing countries. The proposal here is for a policy-
relevant research program that collects such evidence, designs a campaign along
the lines suggested above, experimentally implements the campaign in a few ju-
risdictions, and evaluates impact in terms of changes in politics, public policies,
and finally, the development outcomes.
Conclusion
This paper has argued that there is a role for information campaigns to shift
the platforms of political competition in developing countries away from inef-
ficiently targeted programs such as public-sector jobs and subsidies and on to
broad public policies that effectively (and efficiently) promote basic develop-
ment outcomes. It has presented a proposal for a specific kind of campaign
that requires investment in collecting and publicizing data on development
outcomes—such as child literacy and mortality, and incidence of preventable
diseases—representative at the level of politically-relevant jurisdictions, such
as the lowest level of government or an individual electoral district. The broad
67
References
Adsera, A., C. Boix, and M. Payne, 2003, “Are you being served? Political account-
ability and the quality of government”, Journal of Law, Economics and Organization, vol.
19, no. 2, pp. 445-490
Baland, Jean Marie and Jean-Philippe Platteau, 1999, ”The Ambiguous Impact of
Inequality on Local Resource Management”, World Development, vol. 27, no. 5, pp. 773-
788
Banerjee, A., R. Banerji, E. Duflo, R. Glennerster and S. Khemani, 2006, Can informa-
tion campaigns spark local participation and improve outcomes?, Policy Research Working
Paper No. 3967, Development Research Group, The World Bank: Washington DC
Banerjee, A. and E. Duflo, 2006, “Addressing Absence”, Journal of Economic Perspecti-
ves, vol. 20, no. 1, pp. 117-132
Banerjee, A. and L. Iyer, 2005, “Historical Institutions and Economic Performance:
The Legacy of Colonial Land Tenure Systems in India”, American Economic Review, vol.
95, no. 4, pp. 1190-1213
Banisar, D, 2004, Freedom of Information and Access to Government Record Laws
Around the World, available at: www.freedominfo.org.
Barro, R., 1973, “The Control of Politicians: An economic model”, Public Choice, vol.
14, pp. 19-42.
Besley, T. and R. Burgess, 2002, “The Political Economy of Government Responsive-
ness: Theory and evidence from India”, Quarterly Journal of Economics, vol. 117, no. 4,
pp. 1415-1452.
Björkman, M. and J. Svensson, 2006, Power to the People: Evidence from a Randomized
Experiment of a Community Based Monitoring Project in Uganda, Mimeo. IIES: Stockholm
University
Brunetti, A. and B. Weder, 2003, “A Free Press is Bad News for Corruption”, Journal of
Public Economics, vol. 87, no. 7-8, pp. 1801-1824.
68
Chappell, H., and W. Keech, 1990, “Citizen Information, Rationality, and the Politics
of Macroeconomic Policy,” in John Ferejohn and James Kuklinski (eds.), Information and
Democratic Processes. Urbana and Chicago, Ill.: University of Illinois Press.
Dreze, J. and H. Gazdar, 1996, “Uttar Pradesh: The Burden of Inertia”, in J. Dreze and
A. Sen (eds.), Indian Development: Selected Regional Perspectives, Oxford and New Delhi:
Oxford University Press.
Dreze, J. and A. Sen, 1989. Hunger and Public Action, Oxford: Clarendon Press
Ferejohn, J., 1986, “Incumbent Performance and Electoral Control”, Public Choice,
vol. 50, no. 1-3, pp. 5-25.
Ferejohn, John, 1990, “Information and the Electoral Process”, in John Ferejohn and
James Kuklinski (eds.), Information and Democratic Processes, Urbana and Chicago Ill.:
University of Illinois Press.
Ferejohn, John, and James Kuklinski, 1990, Information and Democratic Processes. Ur-
bana and Chicago, Ill.: University of Illinois Press.
Fiorina, M., 1990, “Information and Rationality in Elections,” in John Ferejohn and
James Kuklinski (eds.), Information and Democratic Processes, Urbana and Chicago, Ill.:
University of Illinois Press.
Fiorina, M. and K. Shepsle, 1990, “A Positive Theory of Negative Voting”, in John
Ferejohn and James Kuklinski (eds.), Information and Democratic Processes, Urbana and
Chicago, Ill: University of Illinois Press.
Goetz, Anne Marie and Rob Jenkins, 2001, “Hybrid Forms of Accountability: Citi-
zen Engagement in Institutions of Public-Sector Oversight in India”, Public Management
Review, vol. 3, no. 3, pp. 363-684
Iyengar, S., 1990, “Shortcuts to Political Knowledge: The Role of Selective Attention
and Accessibility,” in John Ferejohn and James Kuklinski (eds.), Information and Demo-
cratic Processes, Urbana and Chicago, Ill.: University of Illinois Press.
Jenkins R. and A.M. Goetz, 1999, “Accounts and Accountability: Theoretical Impli-
cations of the Right-to-Information Movement in India”, Third World Quarterly, vol. 20,
no. 3, pp. 603-622
Keefer, P. and S. Khemani, 2005, “Democracy, Public Expenditures, and the Poor,”
World Bank Research Observer, vol. 20, no. 1, pp. 1-27.
Keefer, P. and S. Khemani, 2004, “Why do the Poor Receive Poor Services?” Economic
and Political Weekly, vol. 39, no. 9, pp. 935-943.
Krishna, A., 2003, Poor People’s Participation in Democracy at the Local level: Informa-
tion and Education matter more than Wealth and Social Status, working paper no. SAN03-
04, Terry Sanford Institute of Public Policy: Duke University
Lupia, Arthur, 2003, “Necessary conditions for improving civic competence”, Mimeo,
University of Michigan, available at http://www-personal.umich.edu/~lupia/necessary.
pdf
Mani, A. and S. Mukand, 2006, “Democracy, Visibility and Public Good Provision”.
Journal of Development Economics, vol. 83, no. 2, pp. 506-529.
Mansuri, G. and V. Rao, 2003, “Community Based (and Driven) Development: A
Critical Review”, World Bank Research Observer, vol. 19, no. 1, pp. 1-33.
Olken, B, 2005, Monitoring Corruption: Evidence from a field experiment in Indonesia,
NBER Working Paper No. 11753, National Bureau of Economic Research: Cambridge,
MA
Paul, Samuel, 2002, Holding the State to Account: Citizen Monitoring in Action, Banga-
lore: Books for Change.
Prat, A., 2005, “The Wrong Kind of Transparency,” American Economic Review, vol. 95,
no. 3, pp. 862-877.
Reinikka, R. and J. Svensson, 2005, “Fighting Corruption to Improve Schooling:
Evidence from a Newspaper Campaign in Uganda”, Journal of the European Economic
Association, vol. 3, no. 2-3, pp. 259-267
69
Anuradha Joshi
Introduction
Public services in poor countries continue to fail poor people. Recent tracking of
progress on the Millennium Development Goals shows that several countries in
Africa and Asia are still behind in providing basic services of education, health,
water and sanitation to poor people, despite the focused attention paid by the
international community to the issue (ESCAP, 2005). There is evidence that
although in some countries increased spending on public services is necessary,
in many others, money is not the issue.
The problem is an old one. How can we ensure that democratically elected
governments use the resources they raise to provide appropriate basic services,
especially for the poor? In other words, how can those who exercise public aut-
hority be made accountable to the people who have granted them the power
to act on their behalf? Recently, two sets of literatures have been converging on
similar answers. One, the literature on service delivery reforms, underpinned by
New Public Management (NPM), has focused on a separation of policy-making
and implementation functions, establishing formal accountability mechanisms
such as performance contracts and the creation of customer orientation in ser-
vice delivery agencies through tools such as citizen charters (Hood, 1991; Man-
ning, 2001). Similarly, the central message of the recent World Development
Report on “Making Services Work for Poor People” is that service delivery can
be significantly improved by strengthening accountability mechanisms between
three sets of actors: policy-makers, providers and citizens (World Bank, 2003).
Second, the literature on new ways of deepening democracy through direct
citizen participation expects that the creation of formalized participatory spaces
will democratize and rationalize the state and offer marginalized groups a say in
policy-making. Direct participation is viewed as a way of filling the ‘democratic
deficit’ that beleaguers traditional representative democracy (Gaventa, 2006;
Heller, 2001). Some authors see participation in such ‘invited spaces’ not only
as a possible avenue for articulation of preferred public goods but as a means
of civic education, the creation of engaged citizens (Fung and Wright, 2003).
The answers to improved service delivery offered by these literatures empha-
1
The ideas in this paper critiquing the WDR were developed in collaboration with Ann Marie
Goetz and Mick Moore at IDS. The focus on the impact of reforms on collective action presented
in this paper is the product of a collaborative effort with Peter Houtzager (IDS) and Adrian Gurza
Lavalle (CEBRAP, Brazil) and part of a larger research project comparing urban basic services in
three large metropolises in the developing world supported by the DFID-funded Centre for the
Future State at the Institute of Development Studies. I am grateful for the comments and sugges-
tions received from participants in the seminar on “The politics of service delivery in democracies”
hosted by the EGDI in Stockholm, 27-28 April 2006.
71
size two features: a) formal systems of direct accountability and b) direct (often
individual) citizen participation in policy-making and implementation.
This paper examines two sets of issues that arise from these prescriptions.
The first set relates to some of the problems of relying on formal, direct ac-
countability mechanisms for service providers—the problem of ‘watching the
watchdogs’, the transactions costs of direct action, the danger of being misled
by arrangements that contain little effective accountability and the fact that
in poor countries, much service delivery is accomplished through informal in-
stitutions in which accountability may be difficult or obscure. There might be
other, perhaps informal arrangements through which accountability might be
effectively established in contexts where formal direct mechanisms are difficult
to establish or enforce (Masud, 2002). We also emphasize the need for placing
some trust in public service providers, particularly at the frontline, rather than
subjecting them to more and more accountability.
The second set of issues relates to the assertion that it is individual citizen
action of some form – through choice or voice mechanisms in particular – that
will increase accountability and responsiveness of providers. There is of course
an underlying assumption that institutional changes that increase voice or choi-
ce (e.g. decentralization, formalized participation or pluralization) will automa-
tically lead citizens to engage with providers, make demands and elicit greater
responsiveness. In fact, there is little evidence to suggest that individual citizens
are effective agents in holding either public or private providers to account, and
securing greater responsiveness. Rather, collective action, particularly for poor
people, seems essential if direct accountability is to work. The emphasis then
has to be on understanding the conditions under which collective action influ-
ences service delivery in terms of policy-making and implementation. Further,
we also need to understand the impact that recent reforms of the state (decen-
tralization, participation and pluralization) have had on the ability of citizens
to act collectively. Both these issues have received relatively little attention in
the literature.
We argue in this paper that responsiveness and accountability in service deli-
very, at least to poorer sectors of the population, depend on the one hand on the
ability of users/beneficiaries to engage in collective action and on the other on
the ability of the state to negotiate consensuses between competing demands.
Collective action by users contributes to more responsive governance when
the state is able to lead a process of consensus construction. Collective action
without negotiation can lead to either demand overload or policy fragmenta-
tion, as public officials respond to multiple particularistic demands, or can lead
to new inequalities because some groups of people have greater capacity to
engage in collective action than others.
The key questions then are: under what conditions are poor people likely to
engage collectively and demand better services and accountability in the con-
text of service delivery reforms outlined above? And what kinds of reforms are
needed within the public sector to increase the capacity of public officials to en-
gage with citizens directly and create incentives for increasing responsiveness?
The paper is organized in the following manner. Sections two, three and four
deal with the issue of formal direct accountability. In section two, we outline
some of the key propositions of those that emphasize direct formal accountabi-
72
lity as exemplified in the WDR 2004 and point to some key distinctions that we
need to make regarding accountability as a concept. In section three we point
to some problems with overemphasis on direct accountability, and section four
highlights some useful alternatives. Sections five and six relate to the issue of
direct citizen participation. The policy prescriptions of decentralization, plu-
ralization and institutionalized participation and their expected outcomes are
briefly outlined in section five. In section six, we discuss why individual direct
action might be problematic and why a focus on collective action is important
in a) understanding the extent to which direct action can improve services and
b) understanding the actual impact of prevailing service delivery reforms. Sec-
tion seven concludes.
Accountability
A recent and powerful analysis in the World Development Report 2004, “Ma-
king Service Work for Poor People”, identifies the fundamental problem of
service delivery as one of accountability failures. The central argument is that
accountability relationships between the key stakeholders in service delivery—
citizens, policy-makers and service providers—are not transparent, formalized
or effective. This, the report argues, leads to a whole host of problems that pla-
gue services—inadequate spending on basic public services, the skewing of the
services that are provided towards the rich, inadequate coverage particularly for
the poorest, lack of resources at the point of delivery, corruption, absenteeism
of staff and poor quality (World Bank, 2003).
The problem is evident in the three sets of relationships between three sets
of actors. First, mechanisms to ensure that governments adopt policies that
meet the needs of their citizens (including the poorest ones) are weak. Politici-
ans simply do not have incentives to improve service delivery when political in-
stitutions enable them to win and retain power by appealing to religion, caste or
ethnicity rather than developmental performance. Citizens have few channels
to make themselves heard other than through the ballot box. The lack of infor-
mation on the performance of politicians, social fragmentation along religious
and ethnic lines, and a lack of credibility among the population regarding poli-
tical promises, creates a climate in which there are few incentives for politicians
to focus on improving services (Keefer and Khemani, 2003). Policy-makers in
turn do not have effective ways of holding service providers accountable. This
can be a particularly acute problem when providers are part of the state itself
(as is common in many countries where services are directly provided by the
state) and there is little transparency or clarity about expectations and respon-
sibilities. Finally, in most polities there are few mechanisms for citizens to hold
service providers directly accountable for service delivery.
2
There is a large literature on accountability. For a recent overview see Goetz and Jenkins, 2004.
74
the end, the developmental logic for reforms has to match a political logic for
reforms, without which reforms are unviable. In fact, cases of successful policy
delivery—education in Kerala and Madhya Pradesh in India, or health policies
in Brazil—are clearly rooted in the initiatives of mass-based political parties.
Detailed examination of some of these initiatives reveals that ‘scheming for the
poor’, as the title of Ascher’s book on reforms in Latin America suggests, is an
important part of the story (Ascher, 1984). In other words, these reforms often
come about through the ‘politics of stealth’ rather than through the insulation
of policy-makers, the separation of policy-makers from providers or the amplifi-
cation of citizen voice. Understanding how service delivery works must include
an appreciation of the processes through which political bargaining occurs and
the role of politicians and political parties within them.
Informal institutions
The WDR framework for service delivery emphasises formal accountability
mechanisms including citizen compacts and provider contracts. Such a focus on
formal accountability mechanisms could lead to a relative neglect of informal
institutions and their potential influence on service delivery. Informal institu-
tions, often based on clan, kin, religion or ethnicity, can have a considerable im-
pact on service outcomes (Leonard, 1977). At their best, informal institutions
within the state are exemplified by a sense of ‘public service’ that can permeate
organizations to produce committed, honest public officials. At worst, informal
institutions can sustain networks of corruption and rent seeking. Boundaries
between informal institutions within and outside the state are not strict—they
overlap and often operate across the public-private divide.
Informal institutions are often perceived negatively, as constraining the pro-
per operation of formal political systems and limiting accountability. They are
seen to prevent users of public services from mobilizing broadly on the basis
of shared interests in improved service provision. There is some basis of truth
in such negative views. Indeed, informal institutions can be inequitable—inter-
nally, when discrimination against some members of the group (women, elders,
lower castes etc.) is accepted and justified, as well as externally, when there is
discrimination against outsiders.
Yet research shows that such negative beliefs are not universally true. Re-
search shows that the introduction of local level elected formal institutions in
India (through the passing of the Panchayati Raj Act) strengthened traditional
institutions which in turn contributed to improving the performance of formal
institutions (Ananthpur, 2004). The Citizen Police Liaison Committee (CPLC)
in Karachi is another case in point (Masud, 2002). Since 1989, the CPLC has
been directly involved in supporting core policing activities in the city. The
collaboration between the police and the CPLC violates our ideas of the sepa-
ration of public and private interests in a core area of state activity. However,
a strong web of informal relationships between the forty wealthy businessmen
76
that comprise it and society at large that revolve around trust, reputation, sha-
red interests and careful institutional design has enabled it to establish credibili-
ty and become accountable despite the lack of formal contractual arrangements.
If service delivery improvements are to be realized in difficult contexts such
as these, reformers will have to consider how these informal institutions and
systems of insider privilege can be reformed to work with formal accountability
systems rather than against them.
Collective action
Rather, collective action seems essential for the poor if direct accountability is
to work. Theories of collective action and social movements as well as empi-
rical work on good governance and the rights revolution suggest that whereas
individuals have little power to influence public action, collective actors can
(McAdam, et al., 2001; Tendler, 1997; McCann, 1994). At the level of policy
there is experience suggesting that engagement by collective actors representing
the poor has in fact made public services more responsive to their needs. The
recent classic cases are the role of the health movement in Brazil in securing
universal public health care (Duarte, This volume) and the role of social move-
ments in the securing of the right to work and the right to information in India
(MacAuslan, forthcoming; Pande, forthcoming). Furthermore, the existence of
such collective actors creates constituencies for the services in question that
make it more likely the reforms will be sustained over time (Skocpol, 1992). At
the level of implementation, collective actors can increase the credibility and
uptake of reforms as well as pressurize providers to implement reforms.
There is a large literature on collective action that outlines some of the
circumstances that enable (or constrain) collective action, an area that is bey-
ond the scope of this paper. Here, we argue for the need to focus attention on a
narrower but important question—what happens to collective action by service
users/beneficiaries when we move from large-scale public delivery to small-
scale (decentralised or fragmented) privately-mediated delivery (through sub-
contracting, outright privatisation, provision through charitable organizations)?
And, how does such a move affect the state’s ability to negotiate agreements/
consensus around policies? The question is important because polity centred
analyses suggest that the way state institutions are organized will influence who
engages in collective action and around what types of issues. Pluralization, de-
centralization and institutionalized participation impact on the incentives faced
by potential collective actors. Insofar as the literature deals with the question,
there is an assumption suggesting a positive impact of these changes on collec-
tive action (Robinson, 2007).
Yet there are reasons to be cautious about such optimism. The literature
on collective action and social movements shows that direct public provisio-
ning can facilitate the construction of alliances between groups of poor citi-
zens and reformist professionals within the state involved with service delivery
(Fox, 1992; Joshi, 1999; Tendler, 1997). These alliances are often forged out of
direct experience with service delivery. Location of providers within the state
can increase their institutional leverage over policy issues (Watson, 1996). An-
other potential causal factor is ‘targetness’ – that is, it is easier for people to
mobilise against a large single target whose actions directly affect many people
(McAdam, et al., 2001; Houtzager, 2003). This is more likely to be a govern-
ment agency than either a single monopolist or a diverse set of non-state pro-
viders. A large single provider creates a base for collective action: an interest
80
State responsiveness
The reforms – pluralisation, decentralization and citizen participation – pro-
posed to improve services not only assume citizen capacity and willingness to
engage, but also assume provider capacity to respond to citizen demands. Po-
licy-makers and public officials are expected to take on new roles from ‘direct
providers’ to ‘enablers’ through regulation, engaging with citizens, adjudicating
competing claims and negotiating policy changes. These capacities are unlikely
to emerge automatically and require significant reforms within the public sec-
tor including recruitment, training and organizational changes – reforms that
are aimed at changing the incentives and organizational culture within agencies.
And there is no reason why public agencies that performed poorly on direct
provision are likely to do any better at carrying out these new, and more chal-
lenging tasks.
In fact, there are few studies examining the changes in public officials’ ‘work
culture’ as a result of these reforms (for an exception see Crook and Ayee,
2006). A large-scale study that examined public service reforms in Asia and
Africa concluded that reforms largely failed primarily due to the inability of pu-
blic officials to take on the new tasks of regulation and citizen engagement and
81
the lack of citizen participation in or support of the reforms (Batley and Larbi,
2004). There needs to be greater attention paid to the roles and work culture
of frontline workers at the ‘cutting edge’ of the state/provider-citizen interface
(Tendler, 1997). Their actions (or inaction) to a great extent determine how
services are experienced by citizens (one can see this in the impact of teacher
absenteeism on education). Without understanding how reforms impact their
work culture and building capacity to deal with their new roles, it is unlikely
that accountability mechanisms alone will improve performance.
Conclusions
Recent thinking about service delivery reforms seems to have coalesced around
two features – direct formal accountability and direct citizen action. There are
good reasons why we should pay attention to these features of service provisio-
ning systems – they have the potential to improve services through increased
accountability. They also appeal to broad constituencies by offering simple ways
of thinking about a wide diversity of issues in a wide diversity of environme-
nts.
Yet such simple frameworks also obscure the ways in which service delivery
and accountability may work in the complex political and social environments
of developing countries. Service delivery reforms will be more sustainable to
the extent that they are a product of political negotiation between states and
citizens. We have argued in this paper that we need to pay greater attention
through empirical research to collective actors – the way in which reforms im-
pact their ability to engage as well as the extent to which their involvement
is likely to lead to better service outcomes. Understanding of these issues will
enable us to judge whether and under what conditions the high expectations
of direct citizen action are justified. We also suggested that new capacities are
needed within the state and provider organizations that will have to be consci-
ously built before the benefits of accountability-driven reforms can be reaped.
In terms of reforms, what this means is that greater caution is needed before
advocating particular solutions (e.g. privatization) in specific contexts. We need
some empirical testing of the recommendations implicit in the WDR 2004 to
understand the conditions under which particular combinations of solutions
are likely to be successful. And we need to understand not only the impact of
reforms on service delivery improvements, but also their impacts in the long
term on collective actors representing the poor as well as their impact on the
capacity and organizational culture of public officials.
References
Ananthpur, Kripa, 2004, Rivalry or Synergy? Formal and Informal Local Governance in
Rural India, Working Paper No. 226, Brighton: IDS.
Ascher, William, 1984, Scheming for the Poor: The Politics of Redistribution in Latin
America, Cambridge: Harvard University Press.
82
Batley, Richard and Larbi, G, 2004, “Changing Approaches to Public Sector Manage-
ment”, in R. Batley and G. Larbi (eds.), The Changing Role of Government: the Reform of
Public Services in Developing Countries, Basingstoke: Palgrave.
Batley, Richard, 2004, “The Politics of Service Delivery Reforms”, in Development and
Change, vol. 35, no. 1, pp 31-56.
Bloom, Gerald and Hilary Standing, 2001, Pluralism and Marketisation in the Health
Sector: Meeting Health Needs in Contexts of Social Change in Low and Middle Income
Countries, IDS Working Paper 136, Brighton: IDS.
Caseley, Jonathan, 2003, Blocked Drains and Open Minds: Multiple Accountability Re-
lationships and Improved Service Delivery Performance in an Indian City, IDS Working
Paper No. 211, Brighton: IDS.
Crook, R. and Ayee. J., 2006, “Urban service partnerships, ‘street-level bureaucrats’
and environmental sanitation in Kumasi and Accra, Ghana: coping with organisational
change in the public bureaucracy”, Development Policy Review, vol. 24, no. 1, pp. 51-73.
DiIulio, J.D., 1994, “Principled Agents: the Cultural Bases of Behaviour in a Federal
Government Bureaucracy,” in Journal of Public Administration Research and Theory, vol.
4, no. 3, pp. 277-318.
DiMaggio, P. and W. Powell., 1983, “The Iron Cage Revisited: Institutional Isomorp-
hism and Collective Rationality in Organizational Fields”, in American Sociological Re-
view, vol. 48, no. 2, pp. 147-160.
ESCAP, 2005, A Future Within Reach: Reshaping Institutions in a Region of Disparities
to Meet the Millennium Development Goals in Asia and the Pacific, Regional Report no. 2,
UN Economic and Social Commission for Asia and the Pacific, UNDP: New York
Espinoza, Vicente, 2000, The Medical Association and the Teacher’s Association: Public
Action in the Establishment of the Relationship between Civil Society and the State in Chile
in the 1990s, Santiago: Institute for Advanced Studies.
Fox, Jonathan, 1992, The Politics of Food in Mexico, Ithaca: Cornell University Press.
Fung, A. and E.O. Wright, 2003, Deepening Democracy: Institutional Innovation in Em-
powered Participatory Governance, London: Verso.
Gaventa, J., 2006, Triumph, Deficit or Contestation?: Deepening the ‘Deepening Demo-
cracy’ Debate, IDS Working Paper no. 264, Brighton: IDS.
GHK, 2005, Tackling Social Exclusion in Health and Education in Asia, Draft mimeo,
Delhi: Unicef.
Goetz, Anne Marie and John Gaventa, 2001, Bringing Citizen Voice and Client Focus
into Service Delivery, IDS Working Paper no. 138, Brighton: IDS.
Goetz, Anne Marie and Rob Jenkins, 2004, Reinventing Accountability: Making Demo-
cracy Work for the Poor, London: Palgrave/Macmillian.
Goetz, Anne Marie, Anuradha Joshi and Mick Moore, 2004, Diversity Accountabilities
and Service Delivery, mimeo, Brighton: Institute of Development Studies.
Grindle, Merilee, 1997, “Divergent Cultures? When Public Organizations Perform
Well in Developing Countries”, in World Development, vol. 25, no. 4, pp. 481-495.
Grindle, Merilee, 2004, Despite the Odds: The Contentious Politics of Education Re-
forms, Princeton: Princeton University Press.
Gurza Lavalle, Adrian, Peter P. Houtzager, and Arnab Acharya, 2005, “Beyond Com-
parative Anecdotalism: How Civil and Political Organizations Shape Participation in São
Paulo, Brazil”, in World Development, vol. 33, no. 6, pp. 951-964.
Heller, P., 2001, “Moving the State: The Politics of Democratic Decentralization in
Kerala, South Africa and Porto Allegre”, in Politics and Society, vol. 29, no. 1, pp. 131-
163.
Hood, C., 1991, “A Public Management for All Seasons,” in Public Administration, vol.
61, no. 1, pp. 3-19.
Houtzager, Peter, 2003, “Introduction: From Polycentrism to the Polity,” in Peter P.
83
Houtzager and Mick Moore (eds.), Changing Paths: The New Politics of Inclusion, Michi-
gan: University of Michigan Press.
Hunt, Alan, 1993, Explorations in Law and Society: Toward a Constitutive Theory of
Law, New York: Routledge.
Joshi, Anuradha and Mick Moore, 2000, “The Mobilizing Potential of Anti-Poverty
Programmes: Concepts and Cases”, in Journal of Development Studies, vol. 37, no. 1, pp.
25-36.
Joshi, Anuradha, 1999, Progressive bureaucracy: An oxymoron? The case of Joint Forest
Management in India, Rural Development Forestry Network Paper No. 24a, London:
ODI.
Joshi, Anuradha, “Do Rights Work? Law, Activism and the Employment Guarantee
Scheme”, in M. Moore, A. Joshi, and Sujata Patel (eds.), Employment Guaranteed? Politics
and Mobilization in Maharashtra, forthcoming.
Keefer, Philip and Stuti Khemani, 2003, The Political Economy of Social Service Provi-
sion, Background Paper for the WDR 2004, Washington D.C.: World Bank.
Leonard, David, 1977, Reaching the Peasant Farmer: Organization Theory and Practice
in Kenya, Chicago: University of Chicago Press.
MacAuslan, I., The Politics of Pro-Poor Policy Change in India: The National Rural Em-
ployment Guarantee Act, draft, forthcoming, Sussex: IDS.
Malick, Badal, 2006, “Citizen Voice and Client Power: How to Build Effective and
Sustainable Accountability in Urban Services”, mimeo, Water and Sanitation Program-
me, World Bank-UNDP: New Delhi.
Manning, N., 2001, “The Legacy of the New Public Management in Developing Coun-
tries”, in International Review of Administrative Sciences, vol. 67, no. 2, pp. 297-312.
Masud, Omar, 2002, Co-producing Citizen Security: The Citizen Police Liaison Commit-
tee in Karachi, IDS Working Paper No. 172, Brighton: IDS.
Matin, I., 2004, Targeted Development Programmes for the Extreme Poor, Experiences
from BRAC Experiments, CPRC Working Paper No. 20, Manchester: CPRC.
McAdam, Doug, Sidney Tarrow and Charles Tilly, 2001, Dynamics of Contention,
Cambridge: Cambridge University Press.
McCann, Michael W., 1994, Rights at Work: Pay Equity Reform and the Politics of Legal
Mobilization, Chicago: University of Chicago Press.
O’Donnell, Guillermo, 1999, “Horizontal Accountability in New Democracies”, in
Andreas Schedler, Larry Diamond and Marc F. Lattner (eds.), The Self-Restraining State:
Power and Accountability in New Democracies, London: Lynne Reinner.
Pande, S., Societal Accountability and the PDS Campaign, draft, forthcoming, Sussex:
IDS.
PRAJA, 2000, Citizen Satisfaction Survey 2000, executive summary, PRAJA: Mum-
bai.
Reich, M., 1995, “The Politics of Health Sector Reform in Developing Countries:
Three Cases of Pharmaceutical Policy,” in Peter Berman (ed.), Health Sector Reform in
Developing Countries, Cambridge, Ma.: Harvard University Press.
Robinson, M., 2007, “Does Decentralization Improve Equity and Efficiency in Public
Service Delivery Provision,” in IDS Bulletin, vol. 38, no. 1, pp. 7-17.
Schedler, Andreas, 1999, “Conceptualising Accountability”, in Andreas Schedler,
Larry Diamond and Marc F. Plattner, The Self-Restraining State: Power and Accountability
in New Democracies, Lynne Reinner: London.
Skocpol, Theda, 1992, Protecting Soldiers and Mothers: The Political Origins of Social
Policy in the United States, Cambridge: Harvard University Press.
Standing, Hilary, 2004, Understanding the ‘demand side’ in service delivery, Health
Systems Resource Centre, Issues Paper, London: DFID.
84
Tarrow, Sidney, 1998, Power in Movement: Social Movements and Contentious Politics,
Cambridge: Cambridge University Press.
Tendler, Judith, 1997, Good Government in the Tropics, Baltimore: The John Hopkins
University Press.
Watson, Gabrielle, 1996, Good Sewers Cheap? Agency-Customer Interactions in Low
Cost Urban Sanitation in Brazil, Washington, DC: The World Bank.
Wilson, J.Q., 1989, Bureaucracy: What Government Agencies Do and Why they Do It,
New York: Basic Books.
World Bank, 2003, World Development Report 2004: Making Services Work for Poor
People, Washington, DC: World Bank.
85
James Manor
A current dilemma
Development practitioners and analysts know that it is difficult to ensure that
services and other development initiatives actually reach people at the local le-
vel in less developed countries. Many of them also know that special difficulties
often arise when development programmes pass downward through the space1
between the intermediate level (the ‘district’ level in Commonwealth countri-
es, the ‘departement’ level in Francophone countries, etc.) and the local (village,
town or city) level. In other words, serious difficulties often arise as programmes
traverse what we might call ‘the last mile’ before reaching the local level.
This paper argues that this problem is more severe than most people think
and also suggests 1) how the gap between intermediate and local levels might
be bridged, and 2) how certain approaches can assist in making development
initiatives more likely to succeed within that space.
International development agencies, governments and enlightened civil so-
ciety organisations in less developed countries which seek to improve service
delivery, to promote development, and to reduce poverty are flying semi-blind.
Their understanding and effectiveness are impaired by the division of labour
among social scientists. We know much about events and conditions at the local
level – thanks mainly to anthropologists. And thanks to analysts from several
disciplines, we know much about events and conditions between the national
level and the intermediate level. But we know far too little about what happens
in the space between the intermediate and the local level.
This is a serious problem, because it is in that space that service delivery,
governing and politics, as ordinary (not least, poor) people experience them,
mainly occur. It is there that state and society mainly interact. And – to lend real
urgency to the problem – events within that space often determine whether
efforts to deliver services and reduce poverty succeed or misfire.
When we consider events in the space between local and intermediate levels,
we often know that something happened (for example, that an initiative to
deliver services failed), but not why it happened. We therefore cannot see how
things might be changed for the better – even though there appear to be ways
of encouraging constellations of actors and of altering processes so as to facili-
tate service delivery and development.
In addition, there exists an abundance of studies of the supply of services
1
The use of the word ‘space’ is different from e.g. ‘interface’. ‘Interface’ implies greater proxim-
ity and connection between intermediate and local levels than we usually find in the real world.
The word ‘space’ is preferable precisely because it does not imply these things – the gap or ‘space’
between these levels is often bridged either poorly or not at all.
86
and of governance initiatives, while the demand side has received less attention.
This is, in this context, highly unsatisfactory, since we have clear evidence from
some less developed countries that people at the local level increasingly reach
upward into the space between local and intermediate levels to make demands
and get things done (see for example, Karanth, et al., 2004). So this space is
increasingly important, as is the demand side of governance.
This paper is especially concerned with ‘politics’ (the interplay of actors, in-
terests and ideas in the pursuit of power). All sorts of actors who have little or
no connection to governments operate in the space between local and interme-
diate levels. They include landowners, labourers, pastoralists, traditional leaders,
religious leaders, leaders of castes in South Asia or clans elsewhere, money len-
ders, petty traders, criminal elements, locally organised self-help groups, citizens-
based organisations, and others. Alongside them, we often (but not always) find
government actors such as elected members of low-level councils, low-level
bureaucrats, line ministry employees and the police. And we often find other
actors who are not officials, but who seek to engage with government and play
mediating roles. They include small-time lawyers and contractors, members of
‘user committees’ (water users associations, parent-teacher associations, etc.),
political party activists, and activists working for civil society organisations that
reach down from above towards the local level – plus political entrepreneurs
or ‘fixers’ who live locally and reach up into the space between e.g. village and
district, to contact government actors and access government programmes.
It is the interplay of these various types of actors that constitutes ‘politics’
within the space between local and intermediate levels – some of which entails
interactions with government, and some of which does not. We know far too
little about this. But recent studies have indicated that certain approaches are
likely to prove promising in promoting service delivery and other development
initiatives.
3
These studies also stress the necessity for key figures in development programmes to learn by
doing and by listening to people at the grassroots, to proceed in an open-minded manner, and to
adjust their approaches often on the basis of lessons learned.
4
This was an important conclusion of the two largest and most objective investigations of civil
society mounted thus far: the studies produced by the Third Sector project and related exercises
at Johns Hopkins University, and the Ford Foundation’s ‘Civil Society and Governance’ project,
coordinated out of the Institute of Development Studies, University of Sussex.
88
boost the morale of civil society activists who deserve encouragement. But this
tendency still undermines the analytical value of much of this material.
More objective studies raise questions about the capacity and promise of
enlightened civil society organisations to bridge the divide between interme-
diate and local levels in less developed countries. Two problems are especially
important here.
First, enlightened civil society organisations often lack sufficient ‘reach’. Their
resources are limited, so they cannot engage with large numbers of localities.
Thirteen years ago, this writer asked a prominent civil society leader in Bang-
ladesh what proportion of the population of that country was reached by all
civil society organisations there. His reply was “at best, about 20 per cent”,5 and
recent enquiries indicate that the figure has increased very little since then. That
was in a country where civil society organisations are immensely formidable.
The figures for most other less developed countries would be less impressive.
To reinforce this point, consider another piece of evidence. In almost every
less developed country, development-oriented civil society organisations at the
national level claim to have networks that extend all the way down to the local
level across wide areas. Such claims are not surprising, since they help to attract
funding from international development agencies. But these claims are usually
inaccurate. In 2002, this writer asked civil society specialists at two very dif-
ferent organisations – the World Bank and the Ford Foundation – to provide a
list of countries in which these claims were true. Despite their very different
perspectives, both provided the same list, and it is very short: Brazil, India,
Bangladesh and the Philippines. Inadequate ‘reach’ is clearly a serious problem.
(For more detail, see Manor, 2003.)
It can be eased somewhat (though not removed) if governments and donors
develop congenial ties to civil society organisations – usually by drawing them
into efforts to promote service delivery. But this raises a further problem. Many
governments manipulate such arrangements to disarm and co-opt civil society
organisations, and to foster invidious divisions between those civic organisations
that are contracted into such partnerships and those that are not. Even when
governments do not set out to achieve these things, they often happen anyway.
All of this raises serious doubts about civil society as a bridge between inter-
mediate and local levels. To say this is not to suggest that donors and governme-
nts should cease their attempts to encourage enlightened civil society organisa-
tions to reach the local level. This is a splendid idea. But they should be realistic
– and sceptical – about how fully those organisations can perform that task.
Political Parties
Democracy, or at least competitive politics, has spread to a great many less
developed countries in recent years. So we might expect political parties to
contribute much to the bridging of the gap between the intermediate and local
levels. That is supposedly one of their key tasks. But this has happened to a
surprisingly limited degree – far too little to integrate higher levels with the
localities in most places.
5
Interview with Qazi Farouque Ahmed of PROSHIKA (a centre for human development) in
Dhaka, 19 February 1993.
89
Governments
Government institutions and programmes represent a third set of agencies that
might reach down to the local level. They are more promising than the first
two – because they usually have comprehensive ‘reach’ and penetrate more ef-
fectively. But governments usually bring with them major problems – not least
corrupt, complacent, sclerotic bureaucracies. How might they be changed so
that they penetrate in a less coercive, more open and responsive manner – so
that they enable the resources available at the local level to contribute to the
political and policy processes?
Several recent studies have shown that many leading politicians and techno-
crats are acutely fearful that reforms which make governments more open will
threaten their control over and the predictability of development processes. But
some of those studies also indicate that it is possible to persuade senior leaders
that such changes are non-threatening and offer abundant benefits to them and
to their governments (Kulipossa and Manor, 2007).
Readers will also be aware of work on changes in incentives for low-level
bureaucrats and line ministry employees, so that they become more responsive
to local-level preferences and thus draw local energies into the development
process. Some analyses and numerous governments have concentrated on disin-
centives. But there is also great promise in creating new, positive incentives
(Tendler, 1997).
Democratic decentralisation, which will be further discussed in the next sec-
tion, when it gives elected representatives significant leverage over government
employees (in ways that help the latter to perform more effectively) can also
help here. But if this is to occur, it is essential that substantial powers and re-
sources be devolved onto elected bodies at lower levels – and most of the many
governments that have experimented with decentralisation have provided too
little of both. Senior politicians need to be persuaded that by parting with po-
6
It indicates that India’s Bharatiya Janata Party – one of those with genuine organisational strength
in several senses – largely lacks such a penetrative capacity.
90
wers and resources in this way, they and their governments will become more
influential, popular and legitimate – that this is no zero-sum game. This will
enable them to recognise that such generosity is in their political interests.
Finally, we often find that special agencies or institutions are created to add-
ress urgent needs which by-pass mainstream government institutions that are
thought to be too ineffective to perform adequately. Social funds are one ex-
ample. So are many single-sector programmes and the ‘user committees’ that
are created by them. Recent studies indicate that it is crucial to integrate these
parallel agencies and bodies with mainstream government institutions at a relati-
vely early stage. If they remain unintegrated, they will undermine the legitimacy
of mainstream institutions. This is immensely damaging because it is a matter of
urgency that government legitimacy be enhanced, and that governments should
be persuaded to make greater responsiveness a core characteristic of their ma-
instream institutions. Also, if such parallel institutions sap the legitimacy of go-
vernments, politicians who make key decisions within those governments will
never develop an adequate sense of ownership of them.
is far less need or room for ‘fixers’ in such regions. But most Indian parties lack
the organisational capacity to penetrate effectively. So in most of India, ‘fixers’
have ample room in which to operate.
Like government agencies – but unlike civil society organisations and (in
most regions) political parties – ‘fixers’ are ubiquitous. They are therefore al-
most certainly a more promising force than either civil society organisations or
parties when we look for actors that may forge links between the local and in-
termediate levels. (They are not as effective at doing this as are demand-driven
programmes, discussed below, but they still have great potential.)
Politicians are increasingly recognising that the vast army of ‘fixers’ offers
them a major new political resource that can help to make development pro-
grammes – and not least, service delivery – more effective and responsive to
people at the local level. They can help to inform local residents about (i) the
services that are available to them and (ii) how to gain access to those services.
They can also – like elected members of councils at lower levels (see below)
– explain why certain services are beneficial using a language that local residents
can understand, so that the uptake on vital services increases.
One final comment needs emphasis here. ‘Fixers’ are, above all, political ani-
mals. They do not fit neatly into any development programme’s technocratic
blueprint. They are the untidy, somewhat unpredictable progeny of open ‘poli-
tics’. Their considerable, if largely unacknowledged importance in India’s deve-
lopment process reminds us of the importance – at the local level, as well as at
the apex of the political system – of ‘politics’. Like ‘fixers’, ‘politics’ needs more
of our attention – and international development agencies need to be tolerant
of its untidiness and unpredictability.
tion, and because they overload subordinates with many other tasks so that they
have no time to consult databases.7
So this approach to finding out what governments need to know does not
work. There is, however, another way that can work. (And it is this that makes
the evidence from India especially relevant to other less developed countries,
because in many of those countries such databases are far less reliable than
in India.) Development programmes which are demand-driven – which enable
people at the local level to influence action by governments – can provide this
and still more crucial information.
Consider one such programme that involves ‘service delivery’ – the Educa-
tion Guarantee Scheme in the Indian state of Madhya Pradesh. It is a useful
region to examine because it is decidedly under-developed – its low human
development indicators make it comparable to many less developed countries
where people face serious deprivation.
In 1997, two brilliant civil servants presented the state’s Chief Minister –
who was himself quite brilliant as Indian politicians go – with a one-page memo
proposing what they expected to be a relatively modest but useful new deve-
lopment programme. He immediately approved it, also expecting it to be rather
modest.8 It gave any village that did not have a school within it or quite nearby
the right to demand a school and a para-professional teacher drawn from the
locality to provide primary education at the government’s expense. Funds to
pay the new teachers would be channelled through the elected village council,
to which the teachers would thus be accountable.
By 2003, over 26,000 villages had demanded and received such schools, and
1.23 million pupils were studying in them. This was no ‘modest’ programme.
Examination results from the new schools were comparable to those in con-
ventional government schools – largely because the accountability of the new
teachers to village councils dramatically reduced teacher absenteeism, which
plagues conventional schools.
Madhya Pradesh contains 52,000 villages. Thus, just over half of those vil-
lages received new schools as a result of this programme. But the key thing
to note here is that at the outset, the Chief Minister and his two aides – all of
whom were perceptive – did not know that half of the state’s villages lacked
schools. They only discovered this because the Education Guarantee Scheme
was a demand-driven programme. It was the residents of the most deprived vil-
lages of the state whose demands informed them of this. The lesson from this
case – which is reinforced by evidence from the set of studies of ‘low income
countries under stress’ mentioned above – is clear. Demand-driven programmes
can tell governments what they need to know in order to govern effectively and
responsively.
They can also – to return to our main theme – enable the gap between
intermediate and local levels to be bridged. But just as the discussion on local-
7
Interview with N.C. Saxena, former head of both the national administrative academy and the
Ministry for Rural Development, New Delhi, 8 September 2005.
8
This discussion is based on interviews with one of the two civil servants, Amita Sharma, Bhopal,
4 December 2003; and with the then Chief Minister, Digvijay Singh, New Delhi, 18 September
2005.
93
level political ‘fixers’ indicated, the bridging of that gap is not entirely about
top-down actions and programmes. ‘Fixers’ operate from the bottom up as they
try to access resources from top-down programmes. And demand-driven pro-
grammes are by definition powerfully influenced by bottom-up pressures, even
though they also entail the top-down flow of goods, services and funds from
governments. (The same can be said of generously empowered systems of de-
mocratic decentralisation – one of which played a key role in this episode.)
In other words, the problem of the gap between intermediate and local levels
is best addressed by combining top-down and bottom-up processes. And that
can only be done successfully if significant powers are provided to those opera-
ting from the bottom up. They have long been substantially without power.
use intimidating needles for ante- and post-natal care. Intermediate-level health
sector managers in India and Zambia have told this writer that the uptake on
vital services has increased as a result – so that lives have been saved.
It also makes sense to give elected councillors significant influence over sing-
le-sector ‘user committees’ or ‘stakeholder committees’. This is true because
the selection of members of such committees is often influenced by bureaucrats
in the relevant line ministries – so that they are far more hesitant to express
local preferences and discontents than are democratically elected members of
decentralised councils (Manor, 2004).
Concluding remarks
One further point is worth reiterating. Each of the processes that facilitates
substantial bottom-up influence – the activities of ‘fixers’, democratic decen-
tralisation and demand driven programmes – entails the opening up of ‘politics’
within the space between intermediate and local levels. It also entails extending
‘politics’ – the interplay of ideas and interests in the pursuit of power – below
the intermediate level, so that it is no longer thoroughly dominated by actors
at higher levels. In other words, far from insulating development programmes
from unwelcome intrusions by ‘politics’, the argument here favours the enlar-
gement of the role that ‘politics’ is permitted to play. Despite its untidiness and
unpredictability, ‘politics’ is worth encouraging – in ways that direct it into con-
structive channels, as these types of bottom-up influence tend strongly to do.
Finally, some readers may ask why politicians at higher levels would decide
to open things up in this way. The answer – which has been provided by many
imaginative politicians in less developed countries – is that it serves their inte-
rest to do this. When they open the system up, they themselves, their parties
and their governments gain greater legitimacy and popularity among people at
the local level. So there are good ‘political’ reasons for them to promote more
bottom-up involvement in ‘politics’.
References
Karanth, G.K., V. Ramaswamy, Ruedi Baumgartner and Ruedi Hogger, 2004, “Emerg-
ing Rural Leadership and Sustainable Management of Natural Resources: Evidence from
Two South Indian Villages” in R. Baumgartner and R. Hogger (eds.), In Search of Sustain-
able Livelihood System: Managing Resources and Change, New Delhi, Thousand Oaks and
London: Sage Publications.
Krishna, Anirudh, 2002, Active Social Capital: Tracing the Roots of Development and
Democracy, New York: Columbia University Press.
Kulipossa, F.P. and James Manor, 2007, “Decentralized District Planning and Finance
in One Mozambican Province”, in J. Manor (ed.), Aid that Works: Successful Development
in Fragile States, Washington: World Bank.
Manor, James, 2000, “Small-Time Political Fixers in India’s State Politics: ‘Towel over
Armpit’,” Asian Survey, vol. 40, no. 5, pp. 816-835.
Manor, James, 2003, Civil Society and Poverty Reduction in Less Developed Countries: A
Guide for Development Practitioners, Stockholm: Sida.
95
Omar Azfar
Introduction
According to the theory of democratic local government, democracy and de-
volution are expected to improve service delivery through better preference
matching and better accountability. Local officials are supposed to care about
re-election, and hence find out what citizens want and provide them with these
goods and services. While it makes interesting theoretical predictions, much of
the literature on decentralization and preference matching – e.g. Hayek (1945)
and Oates (1972 and 1999) – is not empirical. These classical works provide litt-
le evidence on whether preference matching actually happens and the evidence
provided is primarily on the United States and not on developing countries.
This chapter presents results from research on Uganda and the Philippines
conducted at the IRIS Center of the University of Maryland. In this research
we collected data from households on their preferences and data from public
officials on what they thought people wanted, and on actual resource alloca-
tions. The results from both countries showed only a weak match between ci-
tizens’ preferences and what was being provided. Also officials at lower levels
(Filipino municipalities and Ugandan sub-counties) were more likely to know
the population’s preferences than higher level (Filipino provinces and Ugandan
districts) local officials
The question arises as to what to do about this poor knowledge of preferences.
One reform that has been tried recently is participatory planning: communities
are directly involved in prioritization, budgeting and monitoring. These reforms
which originated in Porto Alegre, Brazil, have now been adopted across Brazil
and are in the process of being adopted in many other developing countries
– Kerala, India is a notable example. The second part of the paper will review
this and other recent experiences.
Improving the quality of information flows is most likely to be effective if the
poor are empowered to respond to the information and provide clear incentives
to public officials to improve their performance. Elections would be one such
mechanism, but this is not a question that we could directly address in this pa-
per. This paper does not examine the role of elected public officials in improving
preference matching by linking the depth of democracy in the two countries
to the quality of preference matching. The Philippines is more democratic than
Uganda with more genuinely contested multi-party elections at the local level,
but a simple comparison of the accuracy of preference matching would not
persuasively answer the question of whether democracy improves preference
matching because it would only be a comparison of two data points.
A more in-depth study of how preferences were communicated through elected of-
ficials may have been able to shed some light on the role of elections and democracy
97
in communicating preferences but we did not collect the information we would have
needed to answer that question. Thus while it is possible that local elections do lead to
improvements in preference matching, we cannot tell if they do.
The Philippines
In 1986 the Filipino Revolution led by Corazon Aquino overthrew the infa-
mous government of Ferdinand Marcos and established a vibrant democracy.
Promises of decentralization were an important part of the revolutionary rhe-
toric and the new government delivered the promised devolution of authority
with the Local Governments Act of 1991. The devolution of authority in the
Philippines was serious. The Local Government Code enacted in 1991 and im-
plemented in 1992-93 significantly increased the responsibilities and resources
of sub-national governments: 77 provinces, 72 cities, 1526 municipalities and
over 40 000 barangays1. In addition, the Local Government Code mandated the
creation of local democracies at each level, with regular elections being held for
both executives and legislative bodies. 2
Expenditures devolved to sub-national governments covered a wide range
of government activities, the most prominent of which was health, which ac-
counted for more than 53 per cent of the total 7.23 billion pesos (~USD 100
million) of devolved expenditures. Sixty per cent of health personnel and a
large percentage of the personnel of the social welfare, agriculture and budget
ministries, were devolved to lower-level governments. However, with the ex-
ception of social welfare the percentage of central ministry funds reallocated to
lower-level governments was often significantly smaller than the percentage of
devolved personnel.
To defray the cost of devolved expenditures, Section 284 of the Local Go-
vernment Code provided for 40 per cent of central government revenues col-
lected three years before to be transferred back to sub-national governments
as internal revenue allotments (IRAs). Provinces and cities received 23 per
cent each of the total transfer, municipalities 24 per cent and barangays 20 per
cent.
The combination of the devolution of real expenditure authority, substantial
transfers to local governments so they can implement many of their mandates,
and the creation of a vibrant local democracy makes it fair to call the decentra-
lization of the Philippines in the early 1990s a serious devolution.
Nine different sets of respondents were surveyed in the Philippines during
the summer of 2000: households, provincial heath officials, provincial admi-
nistrative officials, provincial education officials (Department of Education,
Culture and Sports, DECS), municipal heath officials, municipal administrative
officials, municipal education officials, health clinic officials, and elementary
school principals.
1
These numbers change as new units are created or old ones combined (Miller, 1997). A baran-
gay is the lowest level of government corresponding to “a neighbourhood”.
2
The interested reader can read Azfar, et al. (2001) for a more detailed descriptive account of
decentralization in the Philippines and Uganda.
98
Uganda
President Museveni’s NRM government has been devolving power by stages to
local governments for several years. Local “Resistance Councils” were created in
1988 by the NRM government. In 1993 the government decentralized many
of its functions to the district level. This action was reinforced by the Local
Government Act of 1997.
The Local Government Act gives both legislative and executive authority
to local governments. Legislative authority is limited by the requirements that
laws and by-laws enacted by local governments be consistent with the consti-
tution, national laws and – in the case of lower-level governments – district
ordinances. There are five levels of local government: village, parish, sub-county,
county and district. Of these, the 45 districts are allocated the most responsibi-
lities and the 800 sub-counties are next in importance. The average population
of a district is about half a million, and that of a sub-county, 28 thousand, but
the range of variation is large.
Many responsibilities within health and education were devolved to local go-
vernments. Part 2 of Schedule 2 of the Local Government Act devolves educa-
tion and health responsibilities to the district level, and recommends the further
devolution of nursery and primary education to lower levels of government.
The two most important sources of revenue are the graduated tax – which
is similar to an income tax – and grants from the central governments. Local
governments have some discretion in setting tax rates. Additionally, district go-
vernments are supposed to distribute 30 per cent of revenues raised in the
district to lower levels of government according to a formula based on child
mortality, the number of school-age children, population and area. The central
government gives three main kinds of grants to local governments: uncondi-
tional grants, conditional grants and equalization grants – the latter directed to
those localities lagging behind in some kind of public service provision. In sum-
mary the Local Government Act appears to give local governments significant
and sometimes overlapping authority over the supply of several public services.
In section 3 we describe an empirical attempt to measure de facto autonomy in
both the Philippines and Uganda by analyzing the responses of public officials
to questions about their authority.
Surveys of eight different groups were undertaken in Uganda: households,
district health officials, district education officials, sub-county health officials,
sub-county education officials/chiefs, health facilities, and primary school prin-
cipals. The surveys were conducted in spring 2000. Data were collected from
75 sub-counties, chosen randomly from 10 quasi-randomly selected districts.3 In
3
Some districts were taken out of the data pool before the sample was randomly selected, be-
cause they could not be surveyed for safety reasons. Thus our sample might be biased towards bet-
ter run districts with more exacting political disciplines. Since most of our results are negative, it is
possible they would be even stronger if we had data on a random sample of districts.
99
Adjustability
The first question to ask in the study of preference matching is whether public
officials have the authority to adjust the supply of public services to demand.
The Local Government Act of the Philippines does provide a fair amount of
formal authority to local government, but it is important to assess whether this
is borne out in fact. In order to evaluate the extent of adjustability in different
government offices, the survey asked public officials two kinds of questions.
The first one inquires about the extent to which officials are able to respond to
people’s suggestions:
The second set of questions evaluates the authority of officials to hire or fire
staff, to decide on the salaries of personnel, to reallocate funds among services,
and to offer different services. An aggregate index is formed from these ques-
tions to represent the authority of local officials to make a decision without the
approval of a higher government agency.4
There is an element of arbitrariness in constructing any such index from
ordinal answers, but this seemed like the least controversial way to do so. The
index is constructed so that it takes values between 0 (no adjustability) and 1
(easy adjustability).
We find that provincial and municipal officials have similar flexibility to re-
spond to local demand (0.83 vs. 0.79, Table 1). We conducted a similar exercise
in Uganda. We found adjustability was 0.56 for the sub-county health office,
0.59 for the sub-county education/chiefs office, 0.68 for the district health of-
fice and 0.63 for the district education office (Table 1).
In summary, the survey results indicate that local governments at both the
district and sub-county level in Uganda do in fact have some autonomy in the
delivery of public services. Local governments in the Philippines appear to have
even more autonomy. The next section begins examination of whether there
is reason to expect that this autonomy will improve preference matching in
Uganda and the Philippines.
4
Adjustability Index = (0x Cannot adjust + 1x Can adjust with great difficulties + 2x Can adjust
with some difficulties + 3x Can adjust easily) / 3.
100
Table 1. Adjustability: “How easily can you respond to suggestions of local people.
Percentage of respondents.”
Uganda
Sub-County Health 12 24 47 17 56
Sub-County Education 5 26 52 16 59
District Health 0 10 75 15 68 20
District Education 0 22 67 11 63 18
The question is asked in this way to make it easier for respondents to un-
derstand the question and be able to answer it. In principle, better data could
be collected if respondents were asked to rank several different publicly pro-
vided goods, but this is liable to confuse respondents as we have discovered in
subsequent attempts to do so. The answer, which measures deficient demand,
also conflates demand and supply. Adequate public provision of a public good,
which saturates demand, might lead respondents to pick some other good as
their first choice. Our interpretation of the results is, however, based on the
101
assumption that this conflation of supply and demand affects household and
official responses in a similar fashion.
Most of the households (35 per cent) rated investment for infrastructure
(classified as building bridges, roads and canals) as their top preference, followed
by new jobs, education, aid to the poor, health and water services. Municipal
public officials also emphasize infrastructure as the first priority of households
(49 per cent) followed by health, new jobs, and aid to the poor. They give a low
priority to education (5 per cent) as compared to households (10 per cent).
There are also large differences in households’ demands across provinces, and
across municipalities within provinces, for those services demanded by more
than 5 per cent of the population – education, health, roads, new jobs, aid to the
poor and water. (The procedure used for detecting such differences is described
in the technical Appendix.)5
Given that there is apparently a wide dispersion in demands for publicly pro-
vided goods, and apparently a fair amount of discretion on resource allocation
by local officials, if local officials actually knew about people’s demands, then
local governments might be able to better provide public services that people
actually want. The survey also asked public officials about how they learn about
households’ demands. All public officials cited some efforts to find out local
citizens’ demands as is shown in Table 3. Many of them (about 25 per cent)
choose to consult with community leaders and local civic organizations. Quite
a few also conduct surveys or on-field inquiries (around 15 per cent).
Therefore we next examine the awareness of public officials about the de-
mands of people living in their jurisdictions. We found that in 10 out of 20
provinces households’ first priority is correctly anticipated by most municipal
officials. Provincial officials, on the other hand, are able to cite the good deman-
ded by most households only in three cases (they would incidentally do better
by simply stating “Roads” in all cases6).
We tested for a match between households’ demands and public officials’
knowledge of these demands by regressing public officials’ statements about
what households want on what households say they want (The formal procedu-
re is described in the Appendix.) Municipal officials demonstrate a marginally
significant knowledge of citizens’ demands (t= 1.86, Table 2, Figure 1). Howe-
ver, provincial officers demonstrate no knowledge at all of citizens’ demands
– the coefficient is actually negative (Azfar, et al., 2001).
In summary our analysis of the structure of demand shows that information
about demands of the households is conveyed to municipal officials with a mo-
dest success, whereas provincial officials show no evidence of any knowledge
about local demands. Given that municipal officers are much “closer” to the
people – there are approximately 1500 municipalities with an average popula-
tion of around 50 000 each as opposed to approximately 70 provinces with an
average population of around 1 million – we can say we find some weak eviden-
5
Trivial levels of demand make differences in demands conceptually less meaningful and statisti-
cally more difficult to detect.
6
I am reminded of a similarly dismal performance predicting Oscars in 1998 when I would have
done better by simply guessing “Titanic” on all categories, including the ones for which the movie
was not even nominated!
102
ce that being closer to the people does help in knowledge of their demands.
For future replications of studies of this nature, we would recommend that
a more sophisticated question be administered to public officials, asking their
opinions about the percentage of people who would want public resources al-
located to each type of good. This would provide much more detailed informa-
tion to match citizens’ demands with officials’ knowledge of these demands. It
may also make sense to interview some central government officials to compare
their knowledge with that of local officials.
In the last year, on which activity did this sub-county government spend most of its local tax re-
venues?
No evidence was found of any correlation between household demands and either public officials’
knowledge of these demands or resource allocation decisions at the district level (The procedure
is described in the Technical Appendix). Both regressions of 100 observations (=10 goods*10 di-
stricts) have coefficients of essentially 0. Figure 1 shows the lack of any match between households’
demands and district officials’ decisions at the district level.
This exercise was then repeated at the sub-county level, where we found evi-
dence of some congruence between household and public official preferences.
The coefficient of 0.19 is significant (Table 2). A literal interpretation of the
coefficient would be “sub-county officials reflect 20 per cent of the population’s
preferences.” However, the coefficient should not be taken literally, because me-
asurement error in the X variable may be dragging down the coefficient (since
we did not know the measurement error, and could not accurately estimate
it, we did not try to correct for this bias). Sub-county officials, unlike district
officials, do seem to reflect the demand of the citizenry to some extent.
In Uganda, we were also able to conduct an analysis of whether actual resour-
ce allocation was adjusted in response to preferences. Similar but weaker results
were found for actual resource allocation: there is a positive but insignificant
relationship (t=1.4 in a robust regression) between household demand and sub-
county-level resource allocation. One possible explanation for the weakness of
this relationship may be that if resources were actually allocated to a public
good in the previous year, the marginal demand for the good this year would
fall. Figure 1 shows the partial correlation between demands of households, the
knowledge of these demands by sub-county officials, and actual resource alloca-
tion decisions at the sub-county level.
In general there is some evidence of preference matching at the lower
levels of local government in the Philippines and Uganda (municipality or sub-
county), but no evidence of preference matching at the higher level of local
government (province or district). This adds an important nuance to the argu-
ment for decentralization. There is thus some evidence that decentralization to
levels that are really close to the people might result in some improvement in
preference matching.
Sub-county/Municipal Officials
Uganda Philippines
Household Preferences 0.188** 0.150*
(2.47) (1.86)
Table 3. What have you done to find out about local preferences?
Philippines
Municipal Health Municipal
Municipal DECS
Administrator
Survey 6 10 6
Consultation with community leaders/organizations 27 14 31
On-field inquiry 10 8 2
Meeting/consultation with staff 10 12 3
Other 61 62 75
Nothing 0 0 0
Municipal
Were these suggestions... Municipal Health Municipal DECS
Administrator
Incorporated into municipal/city
67 70.83 54.58
development plans
Actually implemented 65.83 65.42 56.67
N 80 80 80
Uganda
District Health District
Sub-County Health Sub-County Edu- Office Education
Office cation Office Office
Nothing 9 14 5 0
Survey 18 10 68 35
Community Meeting 68 72 68 94
Interviewed Staff 17 30 36 52
Other 30 20 63 41
Were these suggestions
incorporated into..
.9875
e( public official prefrences | X)
-.3859
-.3442
e( household preferences| X ) .5843
.972973 .685778
-.256757 -.314222
meetings, and providing them with consultants, was implemented using a ran-
domized protocol. Some districts were randomly selected for treatment and
others as controls. Within the treatment districts, some settlements were se-
lected as treatments and others as controls. Hence the study will be able to
examine the impact of a program to support participation on outcomes. This is
a multi year study and results should be available by 2010.
A second mechanism may simply be to survey citizens about preferences
and communicate the results to local officials. Compared with participatory
budgeting, this has the drawback that information does not flow back and forth
with ease. For instance in a participatory budgeting process citizens could com-
municate their preferences, public officials could respond with what services
are feasible in the given budget, and the meeting could come to some agre-
ement about how to proceed. This would not be feasible in a survey-based
process – even if several rounds of surveys were conducted, the process would
be ungainly. On the other hand the survey-based process has the advantage that
preferences could be collected from a representative sample rather than the eli-
tes – or busy bodies – that are likely to come to and dominate a town meeting.
Compromises could be designed like explicitly inviting randomly selected
people to town meetings, or even offering food or refreshments which may
provide stronger incentives for poorer people to attend. Alternatively, surveys
could be combined with participation, with information collected using surveys
before and after participatory meetings. Such a reform may combine the bene-
fits of two-way communication and the reflection of representative preferences
in public decision making.
Technical appendix
A.1.2 Variance of preferences and preference matching in the Philippines
To formally test for differences in demand for education, we created a dummy
variable for whether the household responded that the money should be spent
on primary education. We subtracted the average number of households that
responded “primary education” from this variable to create a variable of mean 0
and regressed this variable on district dummies.
X i =demand for (say) education in orvince i , measured by the percentage of respondents
who picked education.
xi Xi X i
Regress without intercept
xi b1 D1 .........b10 D10
The F statistic for joint significance of the dummies, tests for the equality of
demands across districts. The estimated statistic of 2.01 is almost significant at
1 per cent. We repeated this exercise for all other services – health, roads, new
jobs, aid to the poor, water – demanded by more than 5 per cent of the popula-
tion and found significant differences in demands across provinces.7
7
Trivial levels of demand make differences in demands conceptually less meaningful and statisti-
cally more difficult to detect
109
References
Azfar, Omar, Satu Kahkonen and Patrick Meagher, 2001, Conditions for effective decen-
tralized governance: A synthesis of research findings, mimeo IRIS: University of Maryland
College Park.
Azfar, Omar, and Jeffrey Livingston, 2001, Federalist disciplines or local capture: An
empirical analysis of decentralization in Uganda, mimeo IRIS: University of Maryland,
College park.
110
Oates, Wallace, 1972, Fiscal Federalism, New York: Harcourt Brace Jovanovich.
Oates, Wallace, 1999, “An Essay on Fiscal Federalism”, Journal of Economic Literature,
vol. 37, no. 3, pp. 1120-1149.
Oates, Wallace and Robert Schwab, 1988, “Economic Competition Among Jurisdic-
tions: Efficiency Enhancing or Distortion Inducing?”, Journal of Public Economics, vol. 35,
no. 3, pp. 333-354.
Olken, Ben, 2004, “Monitoring Corruption: Evidence from a Field Experiment in
Indonesia”, November 2005, NBER WP11753.
Pande, Rohini, 2003, “Can Mandated Political Representation Provide Disadvantaged
Minorities Policy Influence? Theory and Evidence from India,” American Economic Re-
view, vol. 93, no. 4, pp.1132-1151.
Rao, Vijayendra and Ana Maria Ibanez, 2003, The social impact of social funds in Ja-
maica - a mixed-methods analysis of participation, targeting, and collective action in commu-
nity-driven development, Policy Research Working Paper Series 2970, Washington D.C.:
World Bank.
Tiebout, C., 1956, “A Pure Theory of Local Expenditures”, Journal of Political Eco-
nomy, vol. 64, no. 5, pp. 416-424.
World Bank, 2002, Social Capital, Local Capacity, and Government: Findings from
the Second Indonesian Local Level Institutions Study: Overview Report, Washington D.C.:
World Bank
112
Introduction
This paper discusses a case of successful reform in the educational sector – the
Fundef initiative implemented by the Cardoso government (1995–2002) in
1996. This reform can be considered successful because not only has it produ-
ced tangible results – in particular an impressive expansion of pupils’ enroll-
ment as well as the deepening of decentralization in the sector – but it has also
passed the test of political sustainability. In fact, the current Workers Party (Par-
tido dos Trabalhadores) administration of Luis Inacio da Silva, “Lula”, (2003– )
has maintained the scheme, and has even recently presented a constitutional
amendment to upgrade it, drawing on the same basic approach and modus
operandi (Melo, 2007). Fundef’s success is all the more remarkable considering
the hostile fiscal environment in which it has been implemented. Proposed in
1996, it was implemented amidst international crises (the Russian, Asian and
“Tequila” crises) – all of which impacted Brazil’s fiscal situation and created a
highly unstable fiscal situation thenceforth.
Several general lessons can be drawn from this experience that are of broad
interest. This case is particularly interesting because successful comprehensive
educational reforms are non-anticipated events. Social service reforms are a sub-
set of reforms that in the development literature are usually termed second-ge-
neration reforms, mostly because of their institutional focus and emphasis on
institutional reform (Navia and Velasco, 2003; Graham and Naím, 1999). They
share the problems associated with the latter. A non-exhaustive list of the dif-
ficulties encountered includes the following1. First, reforms are public goods.
These face the classic collective action problem of having to face the resistance
of small groups despite the collective benefits that they might generate. This
occurs both at the level of initiation and implementation of reforms.2 Putting in
place efficiency-enhancing measures encounters the opposition of many vested
interests that can produce a stalemate.
Second, these reforms require the cooperation of many actors to be effective-
ly implemented. These include mayors, unions, teachers and service providers,
among others. Severe coordination and incentive-compatibility problems have
to be resolved.3 Third, unlike fiscal and economic reforms – and even initiatives
1
For extended discussions of this issue and updated case studies, see Kauffman and Nelson
(2004a) and Grindle (2004).
2
When good policies turn bad, this collective action problem generates a bad equilibrium trap
(Grindle, 2007 forthcoming).
3
This problem is compounded at the micro-organizational level by the severe contracting prob-
lems (as defined by agency theory) affecting the work of teachers. What actually happens at the
classroom level is hard to observe, thereby creating moral hazard problems, and making change
difficult to promote (Navarro, 2005).
113
in the area of social security – there are no clear models to follow (pay-as-you-
go, etc). A variety of alternatives have been proposed to implement general gui-
ding principles such as decentralization and the introduction of performance-
based incentives (Nelson, 2000, 2005). Finally, the costs of non-reform are low
and diffuse. Those more affected by reforms – students and their families – bear
the cost of non-reform. Despite ubiquitous references to educational crisis, no
sense of urgency exists.4
However, despite all these factors that make reform initiatives unlikely, they
do happen (Grindle, 2004). Despite the odds, many reform initiatives have
been attempted in Latin America in recent years, though with varying degrees
of success (Kauffman and Nelson, 2004b). However, failure is not pervasive:
rather, it is analysts that have been unable to detect successes (McCourt and
Bebbington, 2007). The conditions that favor the initiation of reforms are dif-
ferent from those that lead to effective implementation. In this paper I will
concentrate on the latter.
In this paper I will attempt to demonstrate that programs for educational re-
form that include elements of expansion of coverage may convert the zero sum
game situation of efficiency-enhancing measures into positive sum games. In
addition, I argue that in decentralized contexts, such as that characterizing Bra-
zil and increasingly also other Latin American countries, sub-national political
actors (mayors, governors and associations of municipal/provincial education
secretaries) may represent important pro-reform actors (the almost exclusive
focus in much of the literature on teachers unions has led to an underestima-
tion of these other actors). These actors – particularly mayors – supported the
program because their municipalities stood to receive additional resources and
because they could claim credit for the benefits accruing to teachers and fami-
lies involved – a major consequence of the initiative.
This paper is organized with an introduction and three further sections. The
second section presents background information on Fundef, its creation and
evolution, and its accomplishments and results. Section three discusses Fundef
with reference to the framework set out in the introductory section. The final
section concludes.
4
The existing literatures on policy reform and policy successes also explain success as the result
of reform champions and of hard-to-specify factors such as political leadership, political will or
political commitments. I am skeptical of these explanations. A useful alternative concept is policy
entrepreneurship. Allegations of a lack of political will can be a blame-shifting strategy used by
multilateral institutions for programs that fail for other reasons (McCourt and Bebbington, 2007).
114
of freeing up existing funds for pro-poor programs by shifting funds from other
programs, and administrative reforms associated with changes in incentives to
enhance the quality of services and extend their coverage.
As indicated, the Fundef mechanism represents a highly successful initiative
of the Cardoso government that led to the deepening of decentralization of
education in Brazil. Based on state-level data and controlling for other supply-
and demand-side determinants of school enrollment, Mello and Hoppe (2006)
found that the increase in Fundef transfers to finance sub-national service deli-
very had a positive effect on actual enrollment rates, rather than just an expan-
sion of the school network. They also found that decentralization encouraged
enrollment: the lower the proportion of students enrolled in the state, rather
than the municipal, school network, the faster the increase in enrollment. Ex-
tending the empirical analysis to the municipal level confirms the results of the
state-level regressions. The initiative also helped improve working conditions
and salaries of teachers, particularly in the most remote areas (an average of
almost 12% in one year, but in some cases salaries doubled or trebled). The
proportion of lay teachers in municipal systems has been reduced significantly
and the increase in coverage has also been noteworthy.
Education expenditures considering all levels of government in Brazil in-
creased in the period 1995–2000 from 4.2% to 5.6% of GDP. The net enroll-
ment rate at the primary level increased from 89% to 96% in the period 1996–
2001 (World Bank, 2002). Municipal governments accounted for 34% of public
primary education enrollment in 1996, but for 54% of enrollment in 2001.
At the same time, the relative weights of the different levels of government
have changed, with municipal governments now accounting for nearly 38% of
expenditures, compared to 27% of expenditures in 1995. Municipalities as a
group spent nearly R$ 24 billion on education in the year 2000, nearly twice
what they were spending, in real terms, in the year 1995 (World Bank, 2002).
Antecedents
In Brazil, primary and secondary education is not a functional competence of
any specific level of government. According to the Constitution of 1988, pri-
mary education was to be provided by the municipalities with the financial and
technical assistance of the Federal Government and of the state (Article 30 of
the Constitution). Constitutional Amendment 14 contains articles calling for
priorities to be given by each level of government but does not mandate a very
clear specialization of competence.5 However, it introduced an incentive sche-
me that punished municipalities that did not specialize in primary education.
Education financing in Brazil involves a complex distribution of resources
that reflects the overlapping distribution of competencies in the area (Castro,
1998). The Constitution of 1988 contained provisions for hard-wiring resour-
ces for education. It determined an increase in the resources set aside by the
Federal Government to 18% and established a level of 25% of net tax revenues
5
According to Article 211 of the Constitution, the municipal governments would “give priority”
( “atuarão prioritariamente”) to pre-school education and primary education (and, according to the
Education Guidelines Law, especially grades 1 to 4); the states would prioritize primary (similarly,
especially grades 4 to 8) and secondary education; and the Federal Government was to prioritize
higher education.
115
6
It should be noted that unlike countries such as Argentina and Mexico, primary education was
already fairly decentralized. It would be more appropriate to describe the process as the deepening
of decentralization in an already decentralized institutional arrangement.
116
The use of low per capita spending meant in practice that the transfers that the
Federal Government had to make to Fundef declined in the period 1998–2003
(see Table 1 and 2). Between 1998 and 2002, nominal GDP grew by 46.0%,
Fundef’s revenue went up by 76.5% and pupil enrollment rose 5.3%. However,
the minimum per capita value increased only 42.1%. This factor, combined
with the fact that Fundef’s revenues were systematically underestimated, led to
a very low level of execution of the budgeted Federal Government transfers to
Fundef (an average of 67.8% for the period 1998–2003; see Table 2). The states’
and municipalities’ share rose correspondingly.
The law that created Fundef was ambiguous as to how the minimum per ca-
pita spending should be calculated. The government argued that because Fun-
def was a state-level fund, the national per capita rate was to be calculated as
an average of all state funds (Ministry of Education, 2003). This point became
the subject of a huge controversy involving mostly the opposition parties, the
representative institutions of the educational sector, and the Public Prosecutors
Office. If a national minimum was calculated on the basis of the expected total
revenue from earmarked Fundef sources (divided by the total number of enroll-
ments), the national minimum would be set at a much higher level and the Fe-
deral Government contributions would consequently be vastly superior. Indeed,
according to this latter criterion the national per capita level would be set at R$
418.7 (instead of R$ 315 in 1998), and the Federal Government’s share would
rise by 423%. The number of states benefiting from the scheme would climb
from 7 to 17. For the entire period 1998–2002, the Federal Government con-
tributions represented a mere 15.6% of the estimated total amount it should
have contributed if the national minimum parameter had been used (Ministry
of Education, 2003: 8).
The rationale for the Federal Government was to control and secure sub-
national spending in a context of rapid decentralization and consequently high
uncertainty over outcomes. The unification of funding in a single account was
key to securing federal control over sub-national spending. Consistent with
their preferences for fiscal expansion at the local level, sub-national govern-
ments have pressured the government to raise the national universal per capita
118
level. However, fiscal needs have led the national government not to readjust
it. The Federal Government’s desire is to control sub-national priorities and
spending while keeping its own preferred federal fiscal targets as the ultimate
adjustment variable.
Impact of Fundef
What has been the impact of Fundef on the quality of education? The increase
in enrollment rates has been well documented but not the increase in quality, if
any. A few indicators have been presented in the introduction but they will be
discussed here more systematically. Efficiency and quality gains in the case of
Fundef seem to be unanticipated effects of the politics of expansion. There is
indirect evidence that quality may have improved as well. First there has been a
reduction over time in drop-out rates (from 35.9 to 27.7% during 1999–2002)
and average class size (from 36 to 33.9 students during 1999–2003), as well
as repetition rates and grade-age gaps (Mello and Hoppe, 2006). Other factors
may be contributing to these outcomes, but Fundef seems to be playing a role.
Levels of remuneration and qualification play a key role in the quality of
service provision. Teacher pay increased by 38% between December 1997 and
June 2001. In the northeastern region, pay rose by 70% on average between
1997 and 2000. In addition, since 2002, compensation for non-certified teachers
can no longer be financed through the 60% share of revenue earmarked for
teacher compensation, leading to a shift in demand for teachers with better
qualifications (Mello and Hoppe, 2006). As a consequence, the percentage of
lay teachers in total teaching staff countrywide fell from 12% to 5% between
1997 and 2000. In the same period, some 150 000 new teachers were hired,
with the total number of teachers climbing to 750 000 (Souza, 2004).
7
Increases in municipal resources have come about mainly as a result of a redistribution of re-
sources from states to municipalities. The Fundef mechanism leads to redistribution only within a
state rather than from rich states to less affluent ones. According to the World Bank (2002), state
governments in the Northeast and the state of Rio de Janeiro account for a large proportion of the
redistributive transfers. A relatively small number of state governments account for a bulk of the
redistribution of resources, even though there is some redistribution from rich to poor municipali-
ties.
119
showed some support for Fundef. However, the unions were dominated by left-
wing movements and the parties opposed to the Cardoso government. The lar-
gest union, the Confederação Nacional dos Trabalhadores em Educação (CNTE),
was a member of the CUT – the peak trade union association linked to the PT.
A confederation of state level unions that had states and municipalities as
employers, CNTE was fragmented at the national level. CNTE had made raising
teachers’ salaries one of its banners and pressed for the setting up of a nationally
defined minimum. In fact, CNTE (together with UNDIME and CONSED, see
below) had pressed for an agreement with the leftist government of Itamar
Franco for this purpose. In this agreement, signed in 1994, a national minimum
salary for teachers was set at R$ 300. The government, however, opposed the
proposal on political and technical grounds.8 First, it was thought that it would
lead to the creation of a much stronger trade union movement in the educatio-
nal sector, which was dominated by the PT. Although many top advisers inside
the government supported the proposal, it was dropped (ibid). Second, many
advisers argued this proposal was not consistent with the Brazilian federal struc-
ture. Municipalities showed vast economic, fiscal and cost of living differences.
Third, many municipalities paid teachers for a specific number of classroom
hours, making it impossible to adopt the national minimum.
The other stakeholders involved in the discussion of Fundef consisted of pu-
blic sector interests. Two institutions were the main actors in the discussion of
the initiative: the National Association of Municipal Secretaries of Education
(UNDIME) and the National Council of State Secretaries of Education (CON-
SED). Although they had links (though much weaker) to the opposition parties,
they appeared as non-partisan entities. The most active entity was UNDIME,
which shared many interests with CNTE (CNTE, 1999; UNDIME, 1999).
UNDIME was set up in 1986 during the democratization process and soon
became a forum for the articulation of demands in the educational sector during
the Constituent Assembly and for the discussion of the New Law on Educatio-
nal Guidelines of 1996. It was an umbrella institution for the interests of the
forces that criticized educational policy under the military regime. It should be
noted that many of UNDIME’s demands were incorporated into the proposal
for Fundef, including the proposals to ‘municipalize’ fundamental education
and to find more resources for teachers’ pay and training. A former President
of UNDIME in 1994–1995 – Maria Helena Castro – was appointed to several
positions within the Ministry of Education (the positions of Executive Secre-
tary of the Ministry of Education and President of the INEP, the Ministry’s
think tank). CNTE and UNDIME were strong critics of the level set for the
per capita minimum and took the government to court for alleged breach of
the Constitution (for not implementing the constitutional amendment it had
itself proposed). CNTE and UNDIME campaigned for the inclusion in Fundef
of illiterate adults and disadvantaged pupils. The Ministry of Education opposed
this latter proposal because it would mean that “many people could potentially
be enrolled and we could control attendance as we do in normal schools”9. The
main opposition came from the Finance Ministry. This created a disincentive to
8
Maria Helena Castro, interview with the author, July 2004 (various dates).
9
Ibid.
120
municipalities to enroll these two types of students, because they would not be
able to count them as enrollments.
Although it praised the initiative, UNDIME argued that Fundef was a me-
chanism for “redistributing resources that were already available at the state
level” and that by setting the per capita rate at such a low level, the Federal
Government was able to avoid spending more resources on primary education
(UNDIME, 1999). During the approval of Fundef, the government vetoed th-
ree provisions, which subsequently attracted a lot of criticism from both UN-
DIME and CNTE. These included provisions that allowed young people and
adults in the program; that forbade the Federal Government to use its quota of
the employer’s tax for education (salário educação) as part of its equalization
transfer to Fundef; and that called for the automatic transfer of the states’ quota
to municipalities rather than to Fundef. These vetoes aimed at guaranteeing
that the Federal Government could reduce its contributions to the fund and
retain control of Fundef and of municipalities’ behavior. It is interesting to note
that these associations ended up playing not the role of critics and opponents
of Fundef, but rather that of its strongest supporters. This was so in virtue of
the alleged non-enforcement or lack of implementation of all of the provisions.
In terms of the predictions in the literature on second-generation reforms, this
outcome is a paradox. Rather than opposing the reform initiative, the opposi-
tion forces pressed for a more effective implementation of the reform.
More explicit opposition to the reform can be found in the case of the go-
vernors of states where decentralization was already quite substantial. Castro10,
who participated actively in the negotiations, argued that:
we only managed to go ahead with Fundef because we ignored the governors and
started a political negotiation directly with the mayors and federal deputies…The big
problems were Rio Grande do Sul and Rio de Janeiro. When their governors found out
how much they would be losing to the municipalities, they panicked. It was amazing on
the day the amendment was finally put to the vote, it was very difficult to coordinate
because the governors were pressing very hard, because they hadn’t done their calcula-
tions right.
During the negotiation of the amendment and of Law 9424, governors from
the ‘loser’ states pressed for a compensation scheme. This came in the form of
a program for the refurbishment and enlargement of secondary schools. This
program in the Plan Avança Brasil was the ‘political payment’ for approval of
the Plan.11
Much of the change was accomplished at a low cost to the Federal Executive.
It managed to design an incentive structure that caused a revolution in the in-
centives that mayors and governors faced in the education sector. It actually re-
sonated with society’s and government bureaucratic elites’ concern about edu-
cation and at the same time helped ensure that the Federal Government kept
control over the process. It was also consistent with the Federal Government’s
concern with fiscal issues. Fiscal constraints led the government not to increase
10
Maria Helena Castro, interview with the author, July 2004 (various dates).
11
Ibid.
121
the per capita level but only to adjust it to the inflation rate. In fact, the go-
vernment actually reduced the share its equalization funds contributed to total
revenue over time.
allows them to clearly identify the authorship of proposals. They can only see
the final outcomes, and these were tangible. After all, pay rises for teachers are
municipal executive orders.
Fourth, even when resources are shrinking major reforms can be accomplis-
hed. The fact that the Federal Government’s contribution declined over time
was non-anticipated. But Fundef’s success even when resources were diminis-
hing is a remarkable example of the positive effects of a good incentive struc-
ture. Fifth, although Brazil does not have a national teachers’ union, unions did
play a crucial role at the state level. Instead of the predicted resistance to re-
forms, unions actively participated in implementation, calling for the proposal
to be effectively implemented as ‘planned’.
A final comment is that the lack of a sense of urgency may find a functional
equivalent when the issue of educational underachievement acquires the extra-
ordinary visibility it did in Brazil. Public discussion of educational problems was
indeed much enhanced because of the program.
Concluding remarks
The case of Fundef illustrates how reform design can help to overcome resis-
tance to reform. It is usually assumed that second-generation reforms produce
only ‘losers’ and few or no clear ‘winners’. The benefits of reform are usually
assumed to be public goods. In the case of Fundef, it is clear that the reform
benefited a specific constituency – the teachers and smaller municipalities. The-
refore, teachers’ unions and educational sector interests turned out to be not
veto points but pro-reform constituencies. Although most of these interests
were ideologically opposed to the Cardoso government, they were unable to
oppose the reform proposal as such. The case of Fundef also shows that a very
important class of losers – the state governors – could not hinder the reforms.
They were partially compensated by the Federal Government. The fact that the
proposal was designed by a team of domestic experts and was not a transplant
from abroad also reflects the importance of the process of professionalization
that occurred in the education sector. In addition, the case of Fundef also il-
lustrates how the Federal Government could put the reform into practice in a
context of fiscal austerity. Increased rates and improved collection of the taxes
from which Fundef was tapped caused the compensation funds from the Fede-
ral Government to gradually decline, but with more resources being channeled
to primary education.
References
Castro, Jorge Abrahão, 1998, O Fundo de Manutenção e Desenvolvimento do Ensino e
Valorização do Magistério (Fundef) e seu Impacto no Financiamento do Ensino Fundamen-
tal. Discussion Paper no. 604. Brazil: IPEA.
CNTE, 1999, Confederação Nacional dos Trabalhadores em Educação, National Plan of
Education, seminar on Fundef, Brazil, unpublished manuscript.
Costa, Bruno L., 2004, Desigualdade Educacional e Coordenação Intergovernamental:
O Caso do Fundef (Fundo de Desenvolvimento do Ensino Fundamental e Valorização do
Magistério) em Minas Gerais, paper presented at the CLAD Congress, Madrid, Novem-
ber.
123
Cláudio Duarte
1
Brazil is a federal republic composed of a Federal Government, 27 State governments and 5.565
local authorities
2
Until the introduction of the Unified Health System in the beginning of the 1990s, only salaried
workers had the right to medical assistance at federal government units. The poorest sectors of the
population and the unemployed were treated by federal government hospitals or by non-profit
organisations.
125
1985, a coalition called the New Republic. At the Ministry of Health, the team
was made up of technical staff with experience in managing health policies. The
Ministry of Health initiated the implementation of programmes to decentralise
resources and responsibilities to the State Health Secretariats with a view to
promoting the extension of coverage and the implementation of preventive
action.
The new Constitution included the main points endorsed by the 8th Na-
tional Health Conference. This was a significant and unprecedented chapter in
the history of health. It represented the beginning of a new era, in which the
process of institutionalisation and operationalisation of the SUS was the most
important feature.
3
At each level of government, the SUS is managed by the health manager: the Ministry of
Health, at federal level, the State Health Secretariat, in the states, and the Municipal Health Secre-
tariats, in the municipalities.
126
regulates and inspects health insurance plans4. FIOCRUZ is the main agency
for research and staff training. FUNASA takes part in the execution of health
policies, especially in poor rural areas, and coordinates national health policies
for indigenous peoples5.
From the mid-1990s onwards, the roles of providing medical assistance and
other kinds of action of a preventive nature, up to then restricted to federal and
state levels of government, were decentralised to the Municipal Health Secre-
tariats. Today, the Ministry of Health is no longer responsible for medical care
except in special cases, such as the National Cancer Institute, a federal organ set
up in the 1940s and responsible for coordinating national cancer care policies
as a national reference point. The management of care for hospital patients and
outpatients is thus carried out principally by the State and Municipal Health
Secretariats. Generally speaking, the State Secretariats are responsible for pro-
viding more specialised hospital services or those of hospitals that are regional
in scope. The Municipal Health Secretariats manage the basics and hospital
care of medium- to low complexity. Towards the end of the 1990s, large-scale
municipalities6 and state capitals progressively took on more responsibility for
the management of hospital medical care involving more specialised services.
These municipalities are considered to be full managers of the SUS and are
therefore mandated to draw up health plans and programmes with comprehen-
sive budgets, covering everything from the funding of medical care to preven-
tive programmes, health promotion, health and epidemiological surveillance
and control of communicable diseases. Small-scale municipalities have, in parti-
cular, benefited from the decentralisation of health services, as they have come
to receive monthly allocations of federal resources to fund basic health services,
as well as laboratory and hospital treatment of low complexity. In the past four
years, the State Secretariats have drawn up programmes for providing services
in collaboration with municipalities where the responsibilities in terms of medi-
cal care for their inhabitants is decided. For diseases of national significance,
such as AIDS, malaria, dengue fever, tuberculosis, and leprosy, the Ministry of
Health still has important functions, especially in the coordination and funding
of national plans to control and eradicate such diseases.
In the past ten years the Ministry of Health has been trying to change the
model of medical care based on hospitals and outpatient centres with medical
specialisations, which dominated the country until the 1980s. For this reason, it
has been extending and bolstering its basic health services, which are delivered
directly by the Municipal Health Secretariats. The main initiatives to this effect
were the introduction of the Community Health Agents Programme in 1991,
the Family Health Programme in 1994, and the Basic Care Minimum in 1997.
The Community Health Agents Programme (PACS) emerged as a local ex-
periment in the State of Ceará in the 1980s. Health agents operated during
the rainy season to combat infant mortality. This experiment was turned into a
4
In Brazil, around 40 million people have private health insurance plans.
5
Around 350 000 indigenous people live in Brazil, most of them living in the Amazon region and
in the mid-west.
6
231 municipalities have more than 100 000 inhabitants, 33 more than 500 000 inhabitants, and
4037 less than 20 000.
127
national programme by the Ministry of Health in 1991. From then on, munici-
palities received financial and technical incentives to introduce the programme
in their governments. PACS, which is run by the Municipal Health Secretari-
ats, hires individuals from the community where they are based, who, through
specific training carry out preventive actions, health education and detection of
risk factors. The agents are also a link between the community and the primary
care units, thereby facilitating access and promoting better relationships with
the health services. Each community health agent is responsible for around 170
families through house calls, using a register of the beneficiary households. The
activities of the programme have contributed to a growth in vaccination figures
and have encouraged more mothers to breastfeed, thereby backing up efforts to
reduce infant mortality and to improve women’s reproductive health (Macinko,
et al. 2006).
In 1994 the Family Health Programme (PSF - Programa Saúde e Família) was
introduced. The PSF was adopted as a strategy to complement PACS activities
by including doctors, nurses and technicians in the teams. Each PSF team is
responsible for 1000 families full-time, carrying out clinical consultations, edu-
cating, performing outpatient procedures and treating minor emergencies. The
programme rapidly spread across the country, especially after the year 2000,
also in large-scale municipalities and metropolitan centres (Magalhaes and Sen-
na, 2006). In 2006, Brazil had 26700 PSF teams and 85 million people were
covered by PSF throughout the country.
Table 1. Aim and evolution of the number of implemented family health teams – Bra-
zil, 1994–2006
30,0
25,0
20,0
15,0
10,0
5,0
0,0
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
Year
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
AIM 0,3 0,7 0,8 1,6 4,0 5,0 10,5 17,0 20,0 21,0 23,0 26,0 26,0
CARRIED
OUT 0,3 0,7 0,8 1,6 3,1 4,3 8,6 13,2 16,7 19,1 21,2 24,6 26,7
Training centres for the PSF teams involving universities and nursing colleges
have also been set up. One of the main hindrances for expanding the program-
me is the lack of professionals, especially doctors, with suitable experience in
primary care. To address this problem, medical and nursing colleges are being
encouraged to change their syllabuses and fields of practice for students, with
the introduction of teaching in the communities, attended by the PSF.
The main impact of the Family Health Programme has been the reduction
of hospital admissions for diseases that are treatable with primary care, greater
control of chronic diseases such as hypertension and diabetes, improvement in
reproductive health and increased coverage for pre-natal consultations, as well
as a more widespread use of family planning techniques (Cufino, et al. 2000;
Paiva, et al. 2006).
The Ministry of Health, in a partnership with the World Bank, has sought
to back up the expansion and consolidation of the PSF in large urban centres.
One programme, known as the Programme for Expansion and Consolidation
of Family Health, PROESF, allocated in its three phases funding of USD 550
million (50% Federal Budget and 50% World Bank). With the expansion of the
Family Health Programme in urban centres, it is hoped that a change will occur
in the model of medical care, with primary care and health promotion gaining
power. Some of the aims are to promote greater rationalisation of spending on
hospital and emergency care, improve access to actions to control and prevent
chronic diseases through improved clinical supervision and the implementation
of community-based actions, thereby encouraging changes in the habits and
lifestyles of individuals (especially the poorest members of the community)
(Health Ministry, Brazil, 2005).
Another way of encouraging the expansion and improvement of primary care
was the introduction of the Minimum Primary Care Transfer (PAB). Launched
in 1998 by the federal government, PAB allocates resources to the municipali-
ties on a per capita basis and transfers them directly to municipal health funds7.
These resources are earmarked for the funding of improvements in infrastruc-
ture, purchase of equipment, payment of salaries and the cost of drugs and other
consumer goods. At first, the value of the PAB stood at 10 Reais per inhabitant
per year, but was readjusted in 2005 to 13 Reais per inhabitant per year.8
The introduction of the Minimum Primary Care Transfer has improved the
distribution of resources for basic care for the benefit of the poorest and small-
est municipalities. The programme also led to better regional distribution of
SUS resources, for the benefit of the northern and north-eastern regions of the
country. Furthermore, it brought about a rise in the number of professionals
working with primary care, caused an expansion of the outpatients network in
small and medium-sized municipalities with a tenfold increase in the number
of medical clinics and a twelve-fold rise in the number of Family Health Pro-
gramme teams. Another highly significant result was the expansion of action to
control and prevent diseases. There was a 54% rise in the control of diabetes,
7
Federal Law No. 8142/92 set up Health Funds at the three levels of government. It is through
these funds that all SUS resources are transferred. The funds are public accounts subject to inspec-
tion by Congress and the Health Councils.
8
10 Reais are equivalent to USD 4.9 (April 2007).
129
40% in the control of cervical cancer, 49% in all-round children’s health care,
46% in vaccination for pneumococcus, 54% for influenza, and 43% for hepatitis
B (Health Ministry, Brazil, 2002).
In spite of increased spending on basic care, most SUS resources are still
earmarked for hospital and outpatient care of high complexity. The country has
5739 hospitals: 2111 in the public sector, 3481 in the private sector, and 147
teaching hospitals. More than 11 million people are admitted into hospitals
each year at a cost of more than 6 billion Reais (Health Ministry, Brazil, 2006).
Services are contracted from private hospitals, whether they operate for profit
or not, by State and Municipal Health Secretariats. The main form of payment
for hospital services is by means of admissions. More recently, some Municipal
Secretariats have been introducing management contracts with key hospitals
where treatment targets and quality control parameters have been introduced
as well.
In the last ten years, the federal government, with the backing of the Inter-
national Bank for Reconstruction and Development, and the InterAmerican
Development Bank, has invested in improving the quality of hospital care, by
means of the REFORSUS project. REFORSUS resources have been allocated to
repairing and building public sector hospitals and laboratories, training hospital
managers, purchasing equipment and setting up networks for the treatment of
trauma and the management of high risk care. The REFORSUS project came
to an end in the beginning of 2001, having invested around USD 650 million.
Despite improvements in the structure of hospital care networks, there are still
a number of challenges to be overcome. The quality of hospital management is
one point that affects the performance of hospitals, mainly public ones. Some
hospitals are introducing projects to pay more respect to patients’ wishes, es-
pecially in public sector maternity hospitals, which are a point of reference in
high risk management. The number of projects, however, is somewhat limited.
Difficulties are still encountered in terms of access to specialised care and there
are long waiting-lists for transplants and non-urgent surgery. The links between
hospitals and PSF units are still somewhat weak. The process of integrated coor-
dination and management of these two levels of care is still under development,
especially in the case of the country’s large-scale municipalities and metropo-
litan regions.
Despite the growth in spending, Brazil still invests less in health than other
countries in the region. The distribution and effectiveness of spending has been
criticised by specialists for the past 20 years (Uga, Piola, Porto, et al. 2003).
Spending on primary care is still not enough to make it the main gateway into
the health system, controlling access to more specialised care. Distortions in the
pattern of spending on admissions to hospitals still exist, such as in the area of
psychiatry.
collaboration with prosecutors are also worthy of note. Furthermore, the ex-
istence of a network of Council members, which meets every two years, has
allowed a greater exchange of experiences and greater political maturity on the
part of representatives of the general public. It is estimated that there are more
than 80 000 council members in the whole country, operating in more than
5500 Municipal Health Councils. This puts significant pressure on governments
and mobilises a large number of people to defend effective and efficient public
health policy.
As its first measure, the Health Secretary’s team adopted an emergency plan
aiming to ensure that patients were attended as soon as possible and that the
health care network was provided with basic supplies. The administration’s plan
of action focused on the following points:
9Increasing the Municipal Secretariat’s resources through the certification of the
municipality in the Ministry of Health;
9Introduction of the Family Health Programme;
9Setting up a distribution centre for the supply of drugs;
9Introduction of a policy for integrated women’s health care to reduce infant mor-
tality;
9Reopening of the municipal maternity hospital and an emergency clinic in one of
the communities;
9Appointment of a coordinator responsible for women’s health;
9Restructuring of the Municipal Health Council and convening a municipal health
conference;
9Installation of an information and epidemiological surveillance unit;
9Elaboration of a project to refurbish the local state hospital.
The four main lines of action were: a) the introduction of the Family Health
Programme as a way of reorganising the whole system of preventive medicine
and health promotion; b) certification of the municipality in the decentralised
management of the SUS with greater funding from the Ministry of Health;
c) physical restructuring of the care network; and d) social mobilisation and
strengthening of partnerships with the Municipal Health Council and with civil
society organisations, especially grass-roots and women’s movements.
The process of municipalisation of services was already complete by the end
of the first year of the new administration and gave rise to a four-fold increase
in the quantity of resources allocated for health. The City Hall managed to form
strong links with the Ministry of Health and with the state government, thereby
making it possible to refurbish the local hospital by the end of the second year
of the administration. The municipal maternity hospital was reopened and a
133
programme for giving women greater choice in the birth process and encou-
raging natural childbirth was introduced (Rede Saude, 1999). The provision of
drugs was planned with a view to expanding the network of services in the first
two years with standardised medication introduced for mental health, hyper-
tension, diabetes, and women’s and children’s health.
The greatest challenge, however, was the restructuring of outpatients care
with a change from the traditional pattern of curative treatment to a model
with greater emphasis on prevention and health promotion. This approach was
chosen in an effort to universalise basic health care with the introduction of the
Family Health Programme. The team was fully aware of the potential that the
programme should have in order to change the existing model of care, and the
strategy for its introduction was planned to be guided by public participation.
The implementation plan was discussed at the Municipal Health Council.
At first there was some scepticism on the part of the Council as to whether
the initiative would be successful. Such a programme had never been tried in
the municipality. There were few doctors living in the city and too few health
units to accommodate the planned teams. The programme required direct ac-
cess to emergency services, the allocation of ambulances in the poorest neigh-
bourhoods and treatment carried out by specialist doctors. The administration
was not unaware of these issues. The implementation plan was drawn up along
with solutions to these needs. This allowed the Council to participate progres-
sively in the process as the teams set to work.
One coordinating group was responsible for implementation. The team was
made up of a doctor with experience in planning and five technical experts in
epidemiology, women’s health, children’s health, service programme-building,
and financial administration. All the units were set up in accordance with the
following routine steps: discussion in the Health Council of the district served;
presenting the organisation of the programme to the community9; start of work
with introductions to the professionals involved; registering all individuals co-
vered by the programme; evaluation of the functioning of the programme in a
meeting with the community after thirteen months.
This process helped to get the local people involved in the programme. As
this was a new scheme, there was some suspicion on the part of the general pu-
blic regarding the way the programme was to function. Many complained that
medical care did not involve the presence of specialists, such as paediatricians.
With time, when twelve teams had already been set up and the programme had
been running for six months, the inhabitants of Cabo had already fully accepted
the initiative.
The main features that contributed to this were the following: ease of access
to units for patients; the ability of the team to solve the main health problems;
the availability of a full-time service; good relationships between the teams and
the community; and frequent meetings between the programme coordinators
and the community to discuss the introduction of the new scheme.
The setting up of the programme in the first two years of the new administra-
9
One of the greatest difficulties encountered when setting up the PSF was the lack of suitable
physical accommodation. Most of the communities were very poor, without suitable buildings for
the purposes of the programme.
134
tion was the main strategy for restructuring the local health system as a whole.
This is because, at the end of this period, when 28 teams had been established
and the maternity hospital and emergency clinic reopened, there was a high
level of implementation for specific health care programmes, such as those for
mental health, all-round women’s health care, children’s health care, control of
hypertension and diabetes, and control of tuberculosis and leprosy.
A number of precautions were taken in the course of this process. Adequate
installation of the units, the selection of a professional staff genuinely commit-
ted to the strategy and with a suitable profile for working in the community,
supervision and continuous support for the work of the teams, training of pro-
fessional staff with an emphasis on strengthening the clinical skills of doctors,
accountability to the community from the teams. Thus, the way in which the
Family Health Programme was introduced may have contributed to the high
level of sustainability of the initiative. Nine years after the first team was set up,
the number of teams has grown to 32, which attend to 72 per cent of the total
population of the municipality.
The Family Health Programme is the main way of introducing primary care
into the health system in Brazil and as such it was adopted as an experiment in
Cabo. Its characteristic feature is its capacity to interact with and mobilise the
general public to act proactively in defence of their health. Its community base,
translated into permanent action on the part of health officers, stimulates the
participation of the community and allows the sharing of experience of reorga-
nising health practices around the everyday lives of patients in their real social
and cultural context.
The participatory nature of the process has probably been a key factor in
ensuring the sustainable functioning of the programme in the past few years
and may have contributed to the impact made on the state of health of the
population of Cabo de Santo Agostinho.
References
Brazil, Ministério da Saúde (Health Ministry), 2002, Custo e Avaliação de Impacto da
Implantação da Parte Fixa do Piso de Atenção Básica – PAB, Report of Research, Execu-
tive Summary, Health Ministry, vol. 1, no. 5, June, Mimeo, available at www.dtr2002.
saude.gov.br/caadab, cited on 12.04.2006.
Brazil, Ministério da Saúde (Health Ministry), 2005, Monitoramente dos Indicadores
de Componente I do Proesf, Secretaria de Atencão à Saúde, Departamento de Atenção
Básica, powerpoint presentation available with the author.
Brazil, Ministério da Saúde (Health Ministry), 2006, DataSUS, available at www.
dataSUS.gov.br, cited on 05.04.2006.
Cufino, Svitone E., R. Garfield, MI Vasconcelos and V. Araujo Craveiro, 2000, “Pri-
mary health care lessons from the northeast of Brazil: the Agentes de Saude Program,” in
Rev Panam Salud Publica, vol. 7, no. 5, pp. 293-302.
Escorel, Sarah, 1998, Reviravolta na Saúde: origem e articulação do movimento sanitá-
rio, Rio de Janeiro: Editora Fiocruz.
Macinko, James, Frederico C. Guanais and Maria de Fátima Marinho de Souza, 2006,
“Evaluation of the impact of the Family Health Program on infant mortality in Brazil,
1990-2002, Journal of Epidemiological Community Health, 2006: 60, 13-19.
135
Magalhaes, Rosana and Mônica de Castro Maia, Senna, 2006, “Local implementation
of the Family Health Program in Brazil,” in Cad. Saúde Pública, vol. 22, no. 12, pp. 2549-
2559.
Paiva, Daniela Cristina Profitti de, Ana Aparecida Sanches Bersusa and Maria Mer-
cedes L. Escuder, 2006, “Healthcare assessment for patients with diabetes and/or hyper-
tension under the Family Health Program in Francisco Morato, São Paulo, Brazil,” in Cad.
Saúde Pública, vol. 22, no. 2.
Rede Nacional Feminista de Diretos Feministas de Saúde e Diretos Reprodutivos,
1999, Cairo-Brasil. 5 anos de experiências relevantes em saúde e direitos reprodutivos e
sexuais, Brasil.
Uga, Maria Alícia, Sérgio Francisco Piola, Sílvia Marta Porto and Solon Malgahaes
Vianna, 2003, “Decentralization and resource allocation in the Brazilian National Health
System (Sistema Único de Saúde - SUS),” in Ciênc. saúde coletiva, vol. 8, no. 2, pp. 417-
437.
136
Persisting inequalities1
Brazil has a public health system that was set up in 1930, with the creation of
the Ministry of Health, and grew in strength in the 1960s with the creation of
the National Institute for Social Security (INPS), which was substituted for the
National Institute for Medical Assistance and Social Security (Inamps) in 1977.
However, from the end of the 1970s different studies began to be presented by
researchers in the field of public health, managers of the public health system,
users associations of these services and international agencies pointing to sig-
nificant distortions in the distributive profile of the resources allocated in the
area of health. On the one hand, during the mid-1980s, it was already a recog-
nized fact that 40 percent of public resources for health went towards financing
complex medical procedures, whilst basic services were not a priority. On the
other, only those workers who were in formal employment, and thus had all
the requisite working papers, had the right to the services provided by Inamps,
whilst those who worked in the informal sector, and thus did not pay direct
taxes, were not entitled to these services. Furthermore, there was a significant
inequality in the distribution of resources between the different regions of Bra-
zil. For example, the Southeast region, the richest and most developed area in
Brazil, had 43.79 of the population but received 59.28percent of the resources
in 1986 (Souza, 2003).
With the aim of changing this situation, the Unified Health System (SUS), a
public health system with universal and unconditional coverage, was set up in
Brazil towards the end of the 1980s. The SUS constitutes an innovative model
for the nationwide provision of public health services and facilities, incorpora-
ting new forms of management aiming at the decentralization and democrati-
zation of the Brazilian health policy. In its original conception, it was envisioned
that the SUS would integrate the branches of preventative and curative medi-
cine, as well as public and private services, in a single national system.
1
This paper presents partial results of the research “Transformations in the distribution of public
health services in the municipality of São Paulo”, carried out by the Centre of Metropolitan Studies/
Brazilian Centre of Analysis and Planning, Cebrap, with the support of the Foundation for Research
Support of the State of São Paulo and the Citizenship Development Research Centre/Institute of
Development Studies at the University of Sussex.
137
It is worth noting that the implementation of this system during the 1990s
went against the tendency of the reform of the Social Welfare State in the
1980s, which came in the wake of increasing fiscal and demographic pressures.
The Welfare State reforms agenda defended only the implementation of assis-
tance programmes deemed essential – such as basic education, primary health
care, distribution of basic food baskets – designated for poverty-stricken mem-
bers of the population (Carneiro Junior, 2000). In this context, the organization
of the SUS required a great deal of effort, on the one hand in breaking away
from the model upon which the Brazilian health system had been structured
– centralized, privately-operated and access to which required participation in
the formal work market. On the other, it involved confronting the restrictions
set up by an international and national political and economic context which
did not favour the institution of universal social policies.
As argued by Cornwall and Shankland (2007), the impetus that led to the
creation of the Brazilian SUS grew out of a conjunction of elements. First, there
was the democratization of political and societal institutions in the post-dicta-
torship period and a strong political commitment from social movements and
left-wing parties to the provision of publicly-funded services to all Brazilians.
Second, we can point out the successful mobilisation by the movimento pela re-
forma sanitária (movement for health reform) that was supported by Christian
grassroots communities, intellectuals, students and public health physicians and
workers and gathered momentum and influence over the course of the 1980s.
Third, there were innovative institutional experiments, which will be discussed
through the article, which provided the inspiration for mechanisms for popular
involvement, accountability and decentralization within the SUS architecture.
These factors were decisive to sustain a compact between state and citizens
which could ensure the political sustainability of the SUS.
In order to advance the SUS, a variety of programmes and management stra-
tegies were adopted during the 1990s. The results of studies which aimed at
analyzing the impact of these initiatives at the turn of the new millennium
show that whilst the mechanisms put into action by the SUS have significantly
increased access to health services, the distributive profile of this access has not
changed greatly. Marques and Arretche (2004), analyzing the distribution of
public health services throughout Brazil, observed that:
Coelho and Pedroso (2002) through their analysis of the distribution of public
services in São Paulo, in 2001, point out that:
These findings follow in the same vein as the analysis presented by the World
Development Report 2004, “Making Services Work for the Poor”, that highligh-
ted how service provision is skewed in favour of the better off . Furthermore,
the report illustrates that it is difficult to change such a distributive tendency,
even though there is, as has been briefly set out above and will be discussed in
greater detail in the next section, a deliberate effort by agents, both from within
and without the government, to change this situation.
Whilst the situation is a long way from being ideal, it is nevertheless possible
to identify a reduction in the inequalities in the inter-regional distribution of
public health resources if a comparison is made between the pre- and post-SUS
situation.
In the same vein, in a recent work Coelho and Silva (2005) followed the distri-
bution of public health services in the municipality of São Paulo between 2001
and 2005. They came up with results that confirmed the already recognized
fact that the highest levels of consumption were concentrated in the richest
and most educated areas, which also had the best epidemiological indicators.
139
Nevertheless, they also pointed out a new fact, that the consumption of services
was increasing at a greater rate in the poorest sub-municipalities, as well as the
fact that there has been a reduction in the disparity between the consumption
of services across the sub-municipalities with the best and worst indices of in-
come, education and health.
In the next section, data is presented which shows the distribution of public
health services in the Municipality of São Paulo and the changes that have been
taking place in this distribution. In the third section we aim to identify the
mechanisms that have contributed towards improving the access of the poorest
groups and regions to health services. Finally, we conclude with a brief note
about the capacity of survival of a system like the SUS which is committed to
providing complete and universal health coverage.
Map: CEM/Cebrap.
2
The MHDI is constructed for each sub-municipality from the following variables: the per capita
household income, the household head’s average number of schooling years , the illiteracy rate of
those aged 15 years and over (data provided by the Demographic Census of the IBGE) and the
population’s life expectancy (provided by the Fundação Seade/SP).
140
The percentages of SUS users were calculated for each sub-municipality3 and
thereafter the consumption rates for primary appointments4 and for hospital
admissions in the 31 sub-municipalities.
It is important to clarify that the SUS-user is a citizen without a private
health insurance, who uses the public health system. The share of such users
is important for the analysis of differences in health consumption, over time
and space. According to Neri and Soares (2002), in Brazil, among the poorest
10percent of the population, around 2.8 percent have some kind of private
health plan, a figure that reaches 74 percent for the wealthiest 10percent. In
São Paulo’s case 54 percent of total population use exclusively SUS services.
The differences in distribution measured here are between poor people living in
different areas of the city, rather than between poor and non-poor as such.
Table 2 presents the distribution of SUS users in the Municipality of São Pau-
lo (54percent of the total population) according to the MHDI and the Health
Index5. The table shows that sub-municipalities with the worst socio-economic
and health indicators have the highest concentration of SUS users6.
3
In conducting our analysis, we have used an estimate of the population of the city of São Paulo
per sub-municipality, based on the growth rate for 1991–2000, according to data provided by the
IBGE Demographic Census of 2000. Next, we applied to this population an estimated proxy of
the population using the SUS in each sub-municipality. The SUS user population estimate was
calculated by the CEInfo by statistic inference based on data extracted from the Pesquisa de Con-
dições de Vida (Research on Living Conditions) of 1998 (Fundação Seade/SP) and from the 2000
Demographic Census (IBGE).
4
Since there is no information to allow identification of the beneficiary of a given appointment,
we have assumed a plausible premise that this kind of service tends to be produced in a decentral-
ized fashion and consumed locally.
5
The Health Index is generated by the Municipal Department of Health of São Paulo, varying
from 0 to 1, such that lower figures are associated with the worst epidemiological cases and vice-
versa. The indicators used to calculate the Index are as follows: the infant-mortality coefficient,
premature mortality due to chronic non-infectious diseases, the coefficient for tuberculosis and the
coefficient for death from external causes. Pearson’s correlation between the Health Index and the
sub-municipalities HDI: 0.842** (significant with 99 percent confidence).
6
The proportion of SUS users is negatively correlated with the sub-municipalities HDI. Pearson’s
correlation: 0.967**.
141
The areas with the best socio-economic and epidemiological indicators have
on the other hand been providing more SUS-funded primary appointments.
The number of primary appointments/year per SUS user in São Paulo rose
between 2001 and 2005 from 1.32 to 1.76, a value close to that recommended
by the Ministry of Health, of two appointments per person per year. However,
the average for more deprived sub-municipalities was 1.50, which significantly
contrasts with 2.12 appointments per year in the wealthier sub-municipalities.
In the case of hospital admissions there was also a sharp rise in the number of
SUS inpatients, which hit 847 admissions for every 10,000 SUS users in 2005,
up from 714 in 2001. In this case, once again we verify that consumption was
concentrated in centrally-located sub-municipalities, which presented the hig-
hest admission ratio, with 1,073 admissions for every 10,000 SUS users in 2005
which, again, significantly contrasts with the admission rate of 648 admissions
in the poorest sub-municipalities 7.
The finding, however, that the consumption of health services is concen-
trated in territories with higher Municipal Human Development and Health
indices, should not necessarily be construed as the capturing of the SUS by the
wealthy who, in fact, rely primarily on private services. Rather, what the data
suggest is that access to health services differs depending on whether the poor
live in Jardim Ângela (the area with São Paulo’s lowest MHDI) or in Pinheiros
(highest MHDI). In this case, the physical network of public services, which
is concentrated in older and more central regions, is in itself one of the main
factors accounting for the gaping intra-municipal differences in the distribution
of health services. In other words, an important consumption determinant is
linked to components intrinsic to the offer of services, which remain concentra-
ted in the more privileged areas.
To better understand the distributive tendency, it is more important to con-
sider the variation of the consumption of these services over the period (2001–
2005), rather than the levels of consumption per se. Below, Chart 1 shows that
in the case of primary appointments there was a significant increase in the offer
of these services in the sub-municipalities in the first and second quartiles (the
poorest) of MHDI. It should also be noted that the number of appointments
offered across the four quartiles is becoming closer8.
7
In the case of hospital admissions it is possible to identify the address of those admitted, thereby
confirming that the procedures were consumed by the residents of a particular sub-municipality.
Pearson’s correlation coefficient between admission rate and sub-municipalities’ MHDI: 0,616**
(significant with 99 percent confidence). A detailed description of the distribution of hospital ad-
missions according to the quartiles is presented in Chart 2 in the Appendix.
8
The standard deviation in the level of primary appointments across the sub-municipalities,
classified by quartile, decreased from 0.88 in 2001 to 0.29 in 2005, which points to an increasing
homogenization in the distribution.
142
3,00
Number of Appointments
2,50
2,00
1,50
1,00
0,50
0,00
2001 2003 2004 2005
Year
Source: DataSUS – Ministry of Health, Brazil, 2001, 2003, 2004, 2005. Graph: CEM/Cebrap.
Available at http://w3.datasus.gov.br.
Chart 2 shows, in the same way, that the number of hospital admissions grew above
all in the first and third quartiles and the disparity between the levels of admission in
the sub-municipalities in the fourth quartile (the best located) and those of the first (the
worst positioned) decreased.
1.200
1.000
800
Ratio
600
400
200
0
2001 2003 2004 2005
Year
Source: DataSUS – Ministry of Health, 2001, 2003, 2004, 2005. Graph: CEM/Cebrap. Available at
http://w3.datasus.gov.br.
143
As already mentioned, this data illustrates that, whilst there was a significant
expansion in the offer of health services between 2001 and 2005, the distribu-
tive profile remained inequitable: the highest levels of use are to be found in
the richest areas with the best epidemiological indicators in the municipality
of São Paulo. The good news to be highlighted is that there is some evidence,
feeble though it may be, that this pattern might change. A higher increase in
the consumption in the poorest sub-municipalities and a narrowing of the con-
sumption gap across sub-municipalities with the highest and lowest MHDIs
may result in the reversal of the current distributive trend. If this trend rever-
sion continues, we might witness, in the medium term, the emergence of a more
equitable distribution pattern of public health services between locations. Thus,
it might be assumed that a more equitable distribution of financial resources
and of the sub-municipalities’ services will eventually lead to a reduction in the
geographic inequalities hindering access to the public health system.
In the next section, the results described above will be discussed with respect
to the health policies implemented in the 1990s and the first years of the new
millennium with the aim of identifying factors that contributed to improving
the access of the poorest to health services.
From the middle of the 1990s, the Ministry of Health has been strengthe-
ning the system of automatic transfers, (…) [which] prioritizes attention
9
Such percentages were specified in the Constitutional Amendment and they should gradually
increase year by year to reach the figure of 12 percent for the States and 15 percent for the Munici-
palities.
10
Per capita public spending on health has fluctuated around USD100. According to Vianna
(2003) in 1996 spending reached USD151.
144
on health, investing the mayor with the role of manager of the Universal
Health System (SUS) and establishes plans of action for health in accor-
dance with the local reality (Ministry of Health, 2003:13).
The author also emphasizes that such a role represented by the federal gover-
nment is even more important in countries marked by relevant inter-regional
disparities, as is the case in Brazil.
145
To sum up, from the point of view of this analysis, it is important to highlight
the relevance of Federal automatic transfers for the financing of the basic health
programmes; of the greater autonomy given to the municipalities that partici-
pated in the modality of “full management”; the intra-municipal decentraliza-
tion and of social participation. All of these factors combined may have led to
a process, which, as was seen in the previous section, has contributed towards
promoting greater equality in the public health system. As part of the research
programme it will be necessary in future work to indicate in a more systematic
way what the effective contribution was of each of these processes – the form
of financing, decentralization and social participation – upon the identified im-
provements in distribution.
References
Almeida, Cláudia Maria, 1996, “Os modelos de reforma sanitária dos anos 80: uma
análise crítica”, Saúde e Sociedade, vol. 5, no. 1, pp. 3-57.
Almeida, Cláudia Maria, 1999, “Reforma do Estado e reforma de sistemas de saúde:
experiências internacionais e tendências de mudanças”, Ciência & Saúde Coletiva, vol. 4,
no. 2, pp. 263-289.
Almeida, Cláudia Maria, 2000, “Saúde e eqüidade nas reformas contemporâneas”,
Saúde em Debate, vol. 24, no. 54, pp. 6-21.
Arretche, Marta, 1996, “Mitos da descentralização: mais democracia e eficiência nas
políticas públicas?”, Revista Brasileira de Ciências Sociais, vol. 11, no. 31, pp. 44-66.
Arretche, Marta, 2004, “Federalismo e Políticas Sociais no Brasil: problemas de coor-
denação e autonomia”, Revista São Paulo em Perspectiva, vol. 18, no. 2, pp. 17-26.
Arreche, Marta and Eduardo Marques, 2002, “Municipalização da saúde no Brasil:
diferenças regionais, poder do voto e estratégias de governo”, Ciência & Saúde Coletiva,
vol. 7, no. 3, pp. 455-480.
Berlinguer, Giovanni, 1999, “Eqüidade, seletividade e assistência à saúde”, Lua Nova,
no. 47, pp. 59-71.
Brasil, 1988, Constituição da República Federativa do Brasil, Brazil, DF: Senado.
Carneiro Junior, Nivaldo, Maria Celina Andrae, Miriam Mazziero, R. C. Pinheiro, N.
D. Souza, S. M. Silva and Marco Akerman, 2000, “A construção da equidade no acesso
às políticas públicas: uma experiência de articulação intersetorial e interinstitucional na
área central do município de São Paulo”, Revista de Administração Pública, vol. 34, no.
6, pp. 95-104.
Carneiro Junior, Nivaldo, 2000, O setor público não-estatal: as organizações sociais como
possibilidades e limites na gestão pública da saúde, PhD Thesis – Medicine Scool, Univer-
sity of São Paulo.
CEM (Centro de Estudos da Mettrópole), 2002, “Espaço, Política e Políticas na Me-
trópole”. Novos Estudos–Cebrap, no. 64, pp. 121-176.
Coelho, Vera, 1998, “Interesses e instituições na política de saúde”, Revista Brasileira
de Ciências Sociais, vol. 13, no. 37, pp. 115-128.
Coelho, Vera and Marcel Pedroso, 2002, “Distribuição de Serviços Públicos de Saúde
no Município de São Paulo”, Novos Estudos–Cebrap, no. 64, pp. 141-154.
Coelho, Vera, 2004a, “Brazil’s Health Councils: The Challenge of Building Participa-
tory Political Institutions”, IDS Bulletin New Democratic Spaces?, vol. 35, no.2, pp. 33-
39.
Coelho, Vera, 2004b, “Conselhos de saúde enquanto instituições políticas: o que está
faltando?” in Vera Coelho and Marcos Nobre (eds.), Participação e Deliberação: teoria
democrática e experiências institucionais no Brasil contemporâneo, São Paulo: 34 Letras.
Coelho, Vera, Barbara Pozzoni and Mariana Cifuentes, 2005, “Participation and public
policies in Brazil”, in J. Gastil and P. Levine (eds.), The Deliberative Democracy Handbook,
San Francisco: Jossey Bass.
Coelho, Vera and Nílian Silva, 2005, “A distribuição dos serviços públicos de saúde na
cidade de São Paulo está se tornando mais eqüitativa?”, available at http://www.demo-
craciaparticipativa.org/.
Coelho, Vera, 2006, “Brazilian Health Councils: including the excluded?”, in Vera
Coelho and Andrea Cornwall (eds.), Spaces for Change? The Politics of Participation in
New Democratic Arenas, London: Zed Books.
Cornwall, Andrea and Alex Shankland, 2007, “Engaging Citizens: Lessons from Build-
ing Brazil’s National Health System”, Social Science and Medicine, forthcoming.
Cunill, Nuria, 2005, Sobre a intersetorialidade, Working Paper.
IPEA (Instituto de Pesquisa Econômica Aplicada), 1998, Visor Ipea, Ano III; no. 7,
November.
148
Levcovitz, Eduardo, L. D. Lima and Cristiani Machado, 2001, “Política de saúde nos
anos 90: relações intergovernamentais e o papel das Normas Operacionais Básicas”,
Ciência & Saúde Coletiva, vol. 6, no. 2, pp. 269-291.
Marques, Eduardo and Marta Arretche, 2004, “Condicionantes Locais da Descentrali-
zação das Políticas de Saúde”, Cadernos do CRH, vol. 39, pp. 55-81.
Mendes, Rosilda and Juan Fernandez, 2004, “Práticas Intersetoriais para a qualidade
de vida na cidade”, in Roberto Garibe and Paulo Capucci (eds.), Gestão Local nos Territó-
rios da Cidade, São Paulo: Mídia Alternativa Comunicação e Editora.
Misoczky, Maria Ceci, 1995, “A agenda para reformas do Banco Mundial e a política
de saúde: algumas notas para reflexão”, Saúde em Debate, no. 47, pp. 4-7.
Neri, M. C. and W. L. Soares, 2002, “Desigualdade social e saúde no Brasil”, Cadernos
de Saúde Pública, vol. 18, pp. 67-76.
Souza, Renilson, 2003, “Redução das desigualdades regionais na alocação de recursos
federais para a saúde”, Ciência & Saúde Coletiva, vol. 8, no. 2, pp. 449-460.
Vianna, S. M., 2003, “Racionalidade Econômica e Qualidade de Vida”, in P. Golden-
berg (ed.), O Clássico e o Novo, Rio de Janeiro: Fiocruz.
World Bank, 2003, World Development Report 2004: Making Services Work for the Poor,
New York: Oxford University Press.
Appendix
Table 1. SUS Primary Appointments – Municipality of São Paulo 2001–2005.
Relative
Number of appointments Number of appointments
Sub-municipalities’ MHDI Increase
2001 2005 2001–2005
Quartiles MHDI Total Per SUS user Total Per SUS user %
1
Percentage of the population using the SUS in the sub-municipalities.
2
A detailed description of the distribution of primary appointments according to the quartiles is
presented in Table 1 in the Appendix.
149
Peter Spink
Introduction
Over forty years ago, community activist Saul Alinsky when criticizing the
USA’s internal “War on Poverty” program commented that: “poverty means not
only lacking money but also lacking power. An economically stable Negro in
Mississippi is poor. When one lives in a society where poverty and power bars
you from equal protection, equal equity in the courts and equal participation
in the economic and social life of your society, then you are poor” (Alinsky,
1965). Key to Alinsky’s argument is the need to go beyond the restricted vision of
poverty seen as a reference to those without the necessary monetary resources
to support themselves and adopt a more expanded approach to poverty, that
considers not only the economic means of sustaining a viable livelihood, but the
many different social aspects of this livelihood, including gender and ethnicity,
freedom of organization and the presence and the access to those services that
are judged as basic to everyday life. Concern is with the many different, subtle
and not so subtle ways in which poverty is produced and reproduced. Increa-
singly, this more expanded view (see for example Sen, 1992, 1999; Friedmann,
1992), is being expressed through the language of rights and, within this langu-
age, the social notion of dignity is emerging as a pivotal expression of voice and
collective respect.
Thus, whilst it is important to point out, as many recent international deve-
lopment reports have done, that basic services are failing poor people world-
wide either through their absence or their skewed provision, it would be more
accurate to also go on to say that in addition, because of their absence or skew-
ness, these services are actively producing poverty. A municipal health center
that opens between 9.00 am and 5.00 pm and has no outreach mechanism for
the surrounding rural area, or a hospital that ignores the presence of traditional
midwives in local communities is not just failing to respond to conditions, but
also actively producing them. One of the first rules that a working women’s
community collective laid down for their independent day care center in urban
Santiago, Chile, was that it would never close until the last mother had arrived
home from work; thus countering the common practice in the public crèches
of a fixed closing time that would leave working mothers frantically running
through the streets in order to be “on time”. In a small municipality in interior
of São Paulo State, the local government created a night care center specifically
to support single mothers who work in the chicken processing industry where
part of the work – preparing the birds for transportation from farm to factory
– is done at night, when the birds are calmer. Before, the mothers had to depend
151
to 99 per cent and infant mortality from 15 per thousand to 250 per thousand;
where 80 per cent of the population are considered urban, yet the meaning of
urban may vary from some four thousand very small municipalities through to
twenty-five large metropolitan regions in which 25 percent of the country’s po-
pulation occupy only 2 per cent of the territory. Current estimates place some
60 per cent of the economically active population in informal work relations
and twenty-five per cent of young people between 17 and 25 in the major met-
ropolitan regions are both out of work and without access to education.
1
The program’s data base, publications and many videos can be accessed freely through its web-
site (http://inovando.fgvsp.br).
153
cultural movement in dense urban areas aimed at increasing the social inclusion
of young people and joint work by church, local government and NGOs with
vulnerable groups such as street children, child prostitutes and victims of family
violence.
To begin with, the discussions and debates showed how far apart are the
many different groups that supposedly make up Brazil’s social action arena.
Dialogue is restricted, as could be expected in a country with high degrees
of implicit – and explicit – socio-economic stratification. Activists are used to
talking to activists and NGOs with NGOs. Academics were criticized for being
too academic, activists were criticized for being too political, local government
technical staff wanted to know how to solve problems and community leaders
complained of government inactivity. As the discussions went on in, we learned
that part of the problem lay in this very lack of connection. Important practices
and bits of the jigsaw puzzle were being lost through the lack of dialogue. Gra-
dually discussion began to take place and people began to listen to each other
and, as a result, lessons began to emerge.
Early lessons
The first of these lessons was that nothing is better than people’s own organiza-
tions for voicing their concerns. This was an early observation of Alinsky in his
Chicago project during the 1940s and 1950s and continues valid in today’s local
Brazil. Respecting peoples’ own organizations – no matter how odd they may
appear to be, how disparate may be the interests and identities they support,
how informal or how loosely organized – is the key to recognizing their voices
and to hearing what has to be said. This is not participation, nor empowerment,
but presence and respect. It is being prepared to listen and is about working to-
gether. Most people are aware of the cultural explosion that takes place in Brazil
over the weekend before the beginning of lent (the catholic period of partial
fasting and preparation for Easter). Yet when they see the carnival processions
and its many different rhythms that change from one part of the country to
the other, they are less aware of the myriads of community associations and
clubs that are behind the floats, the bands and the costumes; be these highly
orchestrated as in the big set piece competitions of Rio de Janeiro, São Paulo
and elsewhere, or the loosely coupled linkages between ideas and traditions that
in Recife will put 1,500,000 people dancing the local frevo behind 29 bands,
forming the biggest carnival block in the world : the galo de madrugada (the
sunrise cockerel).
The second lesson was the importance of recognizing that communities in
poverty circumstances are not passive. On the contrary people are actively en-
gaged in trying to survive and in doing so need to rely on each other because
there is frequently no other option. In doing so, many such communities have
developed notions of what is public that are broader than those used by local
governments who are not necessarily seen as having the public interest at heart.
At the best, these different “publics” will overlap to some extent, at worst they
may disconnect. In both cases, they are evidence of competence and capacity.
Some riverside communities in the Amazon Basin have purchased their own
154
boats in order to guarantee effective low cost “public” transport and it is not
uncommon for one person to make the long trip to the urban center of the mu-
nicipality taking with them the payment cards for the social benefits of all com-
munity members plus a shopping list of needed goods. Obviously it is necessary
to ask why the municipal government cannot be as equally creative and build
different forms of service provision, but the key point from the example is that
in the absence of one sphere of public action, the public will create another.
Many communities in shanty towns support their own radio stations in which
everybody takes their turn to play music and have their say, guaranteeing voice
on community issues. Sometimes these are nothing more than loudspeakers
mounted on telegraph poles yet they are often under threat by the broadcasting
authorities.
Thirdly, a constant theme in the discussions was a very keen sense of the
place, the lived in territorial, material and social horizon of the day to day, as the
starting point for activities. The place may have hierarchy and difference, but it
is where action is possible. People would say in discussion, the place is where
we are, it looks like us, it is where things can start. When these three lessons
come together, the acceptance of people’s own organizations, the recognition
of competence and of the importance of place, the results can be remarkable
and happen in relatively short periods of time. The widely known participatory
budget in Porto Alegre in the South of Brazil is one such case in point; key to
its success was the municipal government’s acceptance of the competence of
closely packed networks of neighborhood associations and its disposition to
discuss individual proposals for specific urban investments that had to do with
very local community priorities, on a block to block basis, ignoring the integra-
ted priorities of planners.
Fourthly, none of the successful experiences that we have followed have star-
ted out life as development projects. That is with objectives, goals, indicators
and precise time frames. None of them started life as policies. At the most
they have been local public challenges that have resulted in local public practi-
ces, that have raised questions and gathered other resources and made alliances
along the way. Participatory budgeting is a case in point. They may have become
policies later on, but that was not their initial intention. They may have started
life in the community and been adopted by local government, or have started
out through the actions of those public servants who are in direct contact with
the day to day of communities in poverty settings. Usually it is a mix of both
and within the Latin American context, this blending together in the local set-
ting of public policy and social action, of public officials, activists and civic
leaders working together to solve problem is being increasingly referred to as
the space of “public action”.
For example, ten years ago we listened to the mayor of Tupandi, a poor ru-
ral municipality (pop. 3,000) in the south of the country tell the story of the
municipal agricultural development fund. The municipality was in decline, the
local administration had very little income given the widespread informality
of the rural economy and the youngsters were no longer showing interest in
farming, preferring instead to move to the larger cities in search of unskilled
factory work. Part of the reason was diagnosed as the difficulty that the small
farmers had in marketing their produce (chicken, pigs, milk and fruit) on the
155
2
For further discussion of this issue, see Olsson & Wohlgemuth, 2003
157
health and the early years of life; Growing up and Developing in Sobral bring
together educational, cultural, moral, ethical and artistic aspects of children
and adolescents and the third Sobral Child Citizens , focuses both the rights of
children and adolescents as well as the professional development. The results
have had considerable impact.
However when these different arenas fall prey to private interests or even to
the good natured and well intentioned interests of those seeking to speak on
behalf of those in poverty situations, these new forms of articulation and joint
working can become the reverse. Those that are best positioned to take advan-
tage of the information and opportunities provided do so and those that can’t,
don’t. Sometimes it can be so simple as a choice of time and place that makes
attendance impossible for women community leaders who have to be indoors
by informal curfew times imposed by urban violence or even the lack of safe
urban transport. Some peoples organizations, such as the rural landless workers
movement (MST) (Branford and Rocha, 2002) have preferred to keep their
distance from such spaces, adopting a policy of conscious disconnection and in-
dependence, but in the urban environment disconnection is less easy, especially
when the distribution of a variety of tied federal and state funds are required to
be approved by a number of multi-sector committees.
References
Alinsky, Saul, 1965, “The war on poverty – political pornography”, The Journal of
Social Issues, vol. 11, no. 1, pp. 41-47.
Bauman, Zygmunt, 1998, Work, consumerism and the new poor, London: Open Uni-
versity Press.
Branford, Sue and Jan Rocha, 2002, Cutting the wire: the story of the landless movement
in Brazil, London: Latin American Bureau.
Camarotti, Ilka and Peter Spink, 2000, Estratégias Locais para redução da pobreza:
construindo a cidadania, São Paulo: Getulio Vargas Foundation.
Castel, Robert, 1995, Les metamorphoses de la question sociale, Paris: Libraire Arthème
Fayard.
Frederickson, H. George, 1990, “Public administration and Social Equity”, Public Ad-
ministration Review, vol. 50, no. 2, pp. 228-237.
Friedmann, John, 1992, Empowerment: the politics of alternative development, Malden
MA: Blackwell.
Geremek, Bronislaw, 1994, Poverty: a history, Oxford: Blackwell.
Isin, Engin F. and Patricia K. Wood, 1999, Citizenship and Identity, London: Sage.
Jacobi, Pedro and José Antonio Pinho (eds.), 2006, Inovação no campo da gestão públi-
ca local, Rio de Janeiro: Editora da Fundação Getulio Vargas.
Lindblom, Charles, 1959, “The science of muddling through”, Public Administration
Review, vol. 19, no. 2, pp. 79-88.
Marshall, Thomas, H., 1992, Citizenship and Social Class, London: Pluto Perspec-
tives
Olsson, Jan and Lennart Wohlgemuth, 2003, Dialogue in Pursuit of Development,
EGDI Studies no. 2, Stockholm, Ministry for Foreign Affairs
Rorty, Richard, 1999, Philosophy and Social Hope, New York: Penguin Books.
Sabatier, Paul and Hank Jenkins-Smith, 1993, Policy Change and Learning: a advocacy
coalition approach, Boulder: Westview Press.
Sen, Amartya, 1992, Inequality reexamined, New York and Cambridge Massachusetts:
Russell Sage Foundation.
Sen, Amartya, 1999, Development as Freedom, Oxford: Oxford University Press.
Shiva, Vandana, 1993, Monocultures of the Mind, London: Zed Books.
Spink, Peter, 2000, “The rights approach to public administration: experiences from
Brazil”, Revista de Administração de Empresas, vol. 40, no. 3, pp. 45 – 65.
Wilson, William Julius, 1989, The ghetto underclass: social science perspectives, New-
bury Park CA.: Sage Publications.
162
Omar Azfar is Associate Professor of Economics at the John Jay College of Criminal
Justice and at the Graduate Center of the City University of New York. He conducts
research on corruption, local government and other aspects of governance and economic
development.
Vera Schattan P. Coelho is a political scientist. She is researcher and project coordinator
at the Brazilian Center of Analysis and Planning (CEBRAP) in São Paolo, Brazil. Her
interests centre of new forms of citizen participation, deliberation and consultation to
improve social policies and democracy.
Anuradha Joshi is a research fellow in the Governance Team at the Institute for Deve-
lopment Studies, IDS, Sussex University. Her work focuses on institutional analysis and
public policy and she convenes the research programme on collective action and service
delivery as part of the Development Research Centre for the Future State at IDS.
Philip Keefer is Lead Economist of the Development Research Group, The World Bank.
His work has focused on the impact of political and social institutions on development
and currently concentrates on how the dynamics of political competition influence the
policy choices of governments.
Stuti Khemani is an Economist in the Development Research Group, The World Bank.
Her research focuses on understanding the political economy of public policy choices, its
impact on fiscal and service delivery performance, and its implications for institutional
reform to promote better outcomes.
Anirudh Krishna is Assistant Professor of Public Policy Studies and Political Science at
Duke University. His research investigates how poor communities and individuals in
developing countries cope with the constraints that result in poverty and powerlessness,
and how poor community groups interact with states and markets.
James Manor is the Emeka Anyaoku Professor of Commonwealth Studies at the Insti-
tute of Commonwealth Studies, University of London, and the V.K.R.V. Rao Professor
at the Institute for Social and Economic Change, Bangalore, India. He specialises in
political and state-society relations in less developed countries, especially India.
Marcus André Melo is Professor of Political Science and Director of the Center for Pu-
blic Policy at the Federal University of Pernambuco, Recife, Brazil. His current research
interests include independent audit institutions in Latin America and Sub-Saharan Afri-
ca, checks and balances at the sub-national level and social policy in Brazil.
marcus.melo@uol.com.br
the Smithsonian Institution. Her primary research interest is the politics of economic
reforms, and the interactions between market-oriented reforms and democratization in
middle and low-income countries.
Nílian Silva has a bachelor’s degree in Social Sciences and is currently doing her master’s
degree in Political Science at Universidade de São Paulo, Brazil. Her interests centre on
new forms of citizen participation to improve social policies and democracy and to over-
come political, social and economical inequalities.
Peter Spink is Professor of Public Administration and Chair of the Center for Public
Administration and Government, São Paulo School of Business Administration, Getulio
Vargas Foundation. Current research interests: innovation in government, public admi-
nistration reform, poverty reduction, local development.
164
Appendix:
Introduction
In his opening remarks, EGDI’s Head Secretary Torgny Holmgren summarised
the motivation of the workshop: bringing together insights and arguments on
the politics of service delivery in democracies for scrutiny and discussion in or-
der to help sharpen the Ministry’s own understanding of the issues involved.
The main findings of the World Development Report 2004, Making Services
Work for Poor People, served as a backdrop for the workshop discussions. Shan-
tayanan Devarajan highlighted how the provision of publicly financed basic
services, like primary health and education, is systematically skewed in favour
of the richer citizens in most developing countries. The quality and quantity
of service provision can in addition be unrelated to existing supply. How and
to whom basic services are provided is a deeply political matter, much beyond
factors like available resources and programme design. The WDR 2004 there-
fore suggests a framework of accountability relations, between policy-makers,
providers, and poor people, for analysing and identifying solutions to provision
failures.
When either policy-makers/politicians or providers are not held accountable
for their actions and decisions, service provision suffers. The most problematic
relationship is the one between poor people and politicians. Even countries
with a long democratic record and predominantly poor voters fail in providing
services to its poorer citizens. Why? Asymmetric information and problems of
attribution seem to be important explanations. Clientelism, or the practice of
favouring political supporters at the expense of non-supporters, is one manifes-
tation of the problem.
delivery also in terms of its impact on citizenship. Manor analyses the space
between the local and intermediate levels of the state, which is operated by in-
dependent political entrepreneurs, so called ‘fixers’, thereby filling a void where
neither political parties nor civil society organisations manage to reach. Larbi
has studied formal and informal non-state service providers in Nigeria and their
often-uneasy relationship with the state. These providers have a vital function
in a context of state withdrawal or failure, but are not always pro-poor in na-
ture.
Joshi observed some common themes and questions in this connection.
1. An emphasis on what is working on the ground, rather than on pre-set mo-
dels. But is it possible to scale-up good practices without the use of models?
2. A focus on state vs. non-state actors. What factors strengthen or constrain
them, respectively? Alternative providers are not needed in states with high
capacity. Is there then reason to support non-state providers rather than sup-
porting the state itself?
3. A broad view of state capacity, which includes the capacity to negotiate bet-
ween competing claims and regulations.
Views on the state and its capacity to provide basic services, especially to the
poor, are linked to views and observations of politicians and what drives them.
Politicians and their incentives therefore naturally re-emerged in the discussion.
It was, for example, stated that the objective utility of a government programme
is never enough to make it work, its political utility must come across to those
in power. Views clearly differed on what to realistically expect from elected
politicians but, as someone noted, we do not know why political mobilisation
and vote canvassing differ so much between democracies.
One participant remarked in this connection that NGOs – in contrast to po-
liticians – tend to be characterised in purely positive terms in the debate. With
reference to Guatemala, he argued that the increased reliance on NGOs for e.g.
basic service delivery has led to the fragmentation of ideas and programmes.
Civil society therefore needs to have better connections to political processes.
State capacity, in turn, needs to be strengthened. Someone noted that the role
of national parliaments was absent in the discussion. ‘They lack power’ was the
response in short. The fact that most parliaments, for whatever reason, have
little leverage makes them a lesser concern for researchers in politics.
Many agreed on the importance to build upon what works on the ground.
To understand why a particular programme or policy is successful, one must go
closely into the specifics: what was it about this programme and the circums-
tances around it? In response to Joshi’s question, Spink argued that local inno-
vations of the kind he has studied are amenable for scaling up. The trick is to
match government agendas and priorities with local efforts and organisations,
which technocrats may be unfamiliar with, or even dislike. This is something
different than making local organisations implement government policy. A case
was also made for the key role of government staff and institutions. One parti-
cipant noted, for example, that three persons are said to be crucial for a policy
intervention to work in India: the petty contractor, the local policeman and the
lowest functionary in the revenue department.
169
On Swedish policy
Remarks on the challenges facing development assistance donors had been
made throughout the workshop, but took centre stage during the fifth and final
session, ‘what can governments and donors do?’, an open discussion moderated
by Shantayanan Devarajan.
Choosing whether to work through governments or through non-govern-
ment actors can be a dilemma in this context. Torgny Holmgren of the Foreign
Ministry’s Department of Development Policy described, with particular refe-
rence to service provision, how Swedish ODA is channelled to central govern-
ments mainly in the form of budget support and, in nations with weak govern-
ment capacity, to civil society organisations. However, as Devarajan questioned,
if popularly elected politicians do not act in the interests of poor people, why
deal with politicians at all? On the other hand, how can democratic proces-
ses and institutions be strengthened if donors keep circumventing government
institutions in favour of NGOs and other civil society organisations? In more
general terms, why are donors so concerned with democracy at all, given the
intrinsic problems many democratic regimes have in being accountable to their
poorer citizens?
These questions framed parts of the debate. Helena Bjuremalm of the Di-
vision of Democratic Governance at the Swedish International Development
Cooperation Agency (Sida) explained that democracy is not a condition for
Swedish aid, although it is one of eight central components in Sweden’s policy
for global development. Sida’s work aims rather at supporting existing pro-
cesses in a democratic direction. Democracy is, moreover, regarded to have a
value in its own right, on top of potential gains for e.g. service delivery. There is
admittedly a tendency to expect countries to become democratic more rapidly
than is realistic. Many current regimes can today be described as ‘semi-authori-
tarian’. Elections are held, but in practice other processes determine who will
govern. However, Sweden cannot deny a country its support for democratic
development on the grounds of being at ‘too early a stage’ in the process. It is
in such cases key to be very cautious in the choice of partners and processes to
support.
Dorrit Alopaeus-Ståhl, of the Swedish Foreign Ministry’s Department of
Multilateral Development Cooperation, remarked that half of Swedish ODA
is channeled through multilateral organizations, in which the partner countries
are also members. Thus, Sweden faces a dilemma in its ambitions to promote
democratic governance as a condition for the organizations’ activities in non-
democratic countries. The government representative from Guatemala, Hugo
Beteta, noted from his perspective of a ‘partner country’ that donors will fail
to advance the cause of the poor as long as they do not coordinate their activi-
ties better. Currently, donors all work through different channels and therefore
cannot push for policy reform. From a different standpoint, one participant sug-
gested that donors should instead be guided by their comparative advantages.
What in the field of development does e.g. Sweden do better than others? One
result of this focus on coordination is that some things are not being done.
170
stacle in some contexts. This would be a way for donors to make a constructive
contribution without taking political sides. The reasoning was that providing
such resources would also be a way of ensuring long-term sustainable outcomes.
What works in one context will not necessarily work in another. The best ideas
emerge in the country itself and strong results generally require a long process
of trial and error. Donors were therefore advised to encourage the study of poli-
cies that have survived changes in government. However, as someone remarked,
policies with such survival capacity need not be the most constructive ones for
poverty reduction. In India, for example, subsidised electricity is prioritised over
basic health services.
Another message to donors, aside from questions on choices and approaches,
was simply to improve the survival of an institutional memory in partner
countries. There can be a fairly frequent turnover of staff in some places and if
no one remembers what has been said before (there are examples of this), it be-
comes difficult to support long-term reform processes. In addition, the impera-
tive to make use of funds according to original schedules often takes precedence
over considerations of impact and efficiency.
Closing
The rich and energetic discussion of this final session did not result in any una-
nimous recommendations on how donors can best contribute to strengthened
democratic processes and improved service delivery or how to generally un-
derstand and deal with these problems. This was, however, not to be expected
in view of both the complexities involved and the varied perspectives of the
participants. Still, everyone did agree on the importance of understanding the
political equations behind social and economic outcomes, in democracies and
elsewhere.
The workshop was closed by Shantayanan Devarajan and Ingrid Widlund,
who thanked all participants for making the event such an energising and infor-
mative exercise. The debates will continue.
172
Basic services in health, education and sanitation are failing poor people world-
wide, even in democratic systems with free media and institutional mechanisms
for ousting leaders who govern badly. Why, and what can be done about it?
This volume brings together a collection of short papers that address issues
and practical experiences related to these questions. The contributing authors
represent different disciplines, approaches and opinions but share a sensitivity
to the political dimension of service delivery provision.