Richard M. Bird, Francois Vaillancourt-Fiscal Decentralization in Developing Countries (Trade and Development) (1999) PDF

Fiscal Decentralization in Developing Countries
Recent years have seen a worldwide trend toward fiscal decentralization. In

particular, many developing countries are turning to various forms of fiscal
decentralization as an escape from inefficient and ineffective governance,
macroeconomic instability, and inadequate growth.
Fiscal Decentralization in Developing Countries, edited by Professors Bird
and Vaillancourt and featuring important, original, and up-to-date research
from leading scholars, assesses the progress, problems, and potentials of
fiscal decentralization in a variety of developing countries around the world.
With rich and varied case-study material from countries as diverse as India,
China, Colombia, Bosnia-Herzegovina and South Africa, this volume com-
plements neatly the recent collection Fiscal Aspects of Evolving Federations,
edited by David Wildasin and also published by Cambridge, which pre-
sented theoretical advances in the area of research.
Fiscal Decentralization in Developing Countries is the latest volume in the dis-
tinguished Cambridge series Tr a d e a n d Development.
Richard M. Bird is Professor of Economics, University of Toronto and a

consultant to the World Bank. He has previously taught at Harvard
University and has worked at the Fiscal Affairs Department of the
International Monetary Fund.
François Vaillancourt is Professor of Economics, Université de

Montréal and a consultant to the World Bank. He has previously been a
Visiting Fellow at both the Federalism Research Centre, Australian National
University and the Institut d’Etudes Européennes, Brussels, as well as a
Visiting Lecturer at the Shastri Institute.
TRADE AND DEVELOPMENT
A series of books on international economic relations and economic issues in
development
Academic editor
Ron Duncan, National Centre for Development Studies,
The Australian National University
Advisory editors
Ross Garnaut, The Australian National University
Reuven Glick, Federal Reserve Bank of San Francisco
Enzo R. Grilli, The World Bank
Mario B. Lamberte, Philippine Institute for Development Studies
Executive editor
Maree Tait, National Centre for Development Studies,
The Australian National University
Other titles in the series

Helen Hughes (ed.), Achieving Industrialization in East Asia
Yun-Wing Sung, The China–Hong Kong Connection: The Key to China’s Open Door
Policy
Kym Anderson (ed.), New Silk Roads: East Asia and World Textile Markets
Rod Tyers and Kym Anderson, Disarray in World Food Markets: A Quantitative
Assessment
Enzo R. Grilli, The European Community and Developing Countries
Peter Warr (ed.), The Thai Economy in Transition
Ross Garnaut, Enzo Grilli, and James Riedel (eds.), Sustaining Export-Oriented
Developments: Ideas from East Asia
Donald O. Mitchell, Merlinda D. Ingco, and Ronald C. Duncan, The World Food
Outlook
David C. Cole and Betty F. Slade, Building a Modern Financial System: The
Indonesian Experience
Ross Garnaut, Guo Shutian, and Ma Guonan (eds.), The Third Revolution in the
Chinese Countryside
David Robertson (ed.), East Asian Trade After the Uruguay Round
Yiping Huang, Agricultural Reform in China: Getting Institutions Right
Christopher Manning, Indonesian Labour in Transition
Fiscal Decentralization in
Developing Countries
edited by
RICHARD M. BIRD
and
FRANÇOIS VAILLANCOURT
CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo, Delhi
Cambridge University Press

The Edinburgh Building, Cambridge CB2 8RU, UK
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9780521641432
© Cambridge University Press 1998
This publication is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 1998

This digitally printed version 2008
A catalogue record for this publication is available from the British Library
Library of Congress Cataloguing in Publication data
Fiscal decentralization in developing countries / edited by Richard M. Bird, François

Vaillancourt.
p. cm. – (Trade and development)
ISBN 0 521 64143 8 (hardbound)
1. Fiscal policy – Developing countries. 2. Intergovernmental fiscal relations –
Developing countries. 3. Decentralization in government – Developing countries. I. Bird,
Richard Miller, 1938–. II. Vaillancourt, François. III. Series: Trade and development
(Cambridge, England)
HJ1620.F538 1999
336.1´8´091724 – dc21 98-20491 CIP
ISBN 978-0-521-64143-2 hardback

ISBN 978-0-521-10158-5 paperback
Contents
List of tables page viii

List of contributors xi
Preface xiii
1 Fiscal decentralization in developing countries: an overview 1

Richard M. Bird and François Vaillancourt
2 China: evaluating the impact of intergovernmental fiscal reform 49
Roy W. Bahl
3 India: intergovernmental fiscal relations in a planned economy 78
M. Govinda Rao
4 Indonesia and Pakistan: fiscal decentralization – an elusive goal? 115
Anwar Shah
5 Morocco and Tunisia: financing local governments – the impact
on infrastructure finance 152
François Vaillancourt
6 Colombia: the central role of the central government in fiscal
decentralization 172
Richard M. Bird and Ariel Fiszbein
7 Argentina: fiscal federalism and decentralization 206
Ernesto Rezk
8 South Africa: an intergovernmental fiscal system in transition 239
Junaid K. Ahmad
9 Bosnia-Herzegovina: fiscal federalism – the Dayton challenge 271
William Fox and Christine Wallich
Index 301
vii
Tables
1.1 Fiscal decentralization: the case-study countries page 19

1.2 Fiscal decentralization: other countries 20
1.3 A taxonomy of fiscal decentralization 23
2.1 China: distribution of budgetary revenues, 1992 and 1995 50
2.2 Tax revenue performance, 1985–95 58
2.3 Tax collection and expenditure of central and local
governments, 1980–95 60
2.4 Fiscal and economic concentration in rich and poor
provinces 67
2.5 Regressions of per capita local government revenues and
expenditures on selected variables 67
2.6 Tax effort indices and rankings, 1990 and 1995 70
2.7 Requirements for effective decentralization 73
3.1 India: revenue receipts of the central and state governments 83
3.2 Share of state governments in total expenditures 84
3.3 Trends in vertical fiscal imbalance 90
3.4 Vertical fiscal imbalance in selected countries 91
3.5 Revenues and expenditures of the states, 1993–94 93
3.6 Coefficients of variation in per capita state government
expenditures 94
3.7 Interstate redistribution through regional, loan, and credit
policies 98
3.8 Current transfers from the center to the states 99
3.9 Criteria for tax devolution: Tenth Finance Commission,
1990–95 101
3.10 Formula for distributing state Plan assistance 103
3.11 Per capita federal fiscal transfers and Plan outlay in the states
during the Seventh Plan 104
3.12 Equalizing effect of transfers 106
3.13 Effect of federal transfers: Gini coefficients of fiscal variables 107
3.14 Regression results 108
viii
List of tables ix
4.1 A comparative impressonistic review of selected aspects of

the fiscal system in Indonesia and Pakistan 117
4.2 Indonesia: fiscal transfers to subnational governments,
1995–96 123
4.3 Pakistan: legislative responsibility and actual provision of
services by different levels of government 130
4.4 Pakistan: federal transfers to provinces, 1994–95 137
4.5 Pakistan: constitutionally mandated revenue-sharing
program, 1997–98 138
4.6 Pakistan: vertical imbalances, 1994–5 140
4.7 Principles and best practices in grant design 150
5.1 Morocco and Tunisia: sociodemographic characteristics 153
5.2 Morocco and Tunisia: subnational governments 154
5.3 Morocco: parameters of urban tax and supplementary urban
tax 157
5.4 Morocco: sources of funds, local governments, 1991 and 1993 163
5.5 Morocco: sources of funds by type of local government, 1993 163
5.6 Tunisia: sources of funds, local governments, 1992 and 1993 163
5.7 Tunisia: sources of funds by type of local government, 1993 164
5.8 Morocco and Tunisia: main characteristics of local
government investment finance, 1995–96 167
5.9 Tunisia: sources of funds for investments by local
governments, 1995 167
5.10 Morocco and Tunisia: sources of investment financing,
five-year periods 168
6.1 Colombia: structure of the public sector 176
6.2 The system of intergovernmental transfers 180
6.3 Earmarking and mandates in specific municipalities 188
6.4 Macroeconomic effects of increased transfers 193
7.1 Argentina: tax assignment 211
7.2 Tax-sharing systems, 1935–95 214
7.3 Distribution of tax revenue by source and level of
government, 1995 215
7.4 Vertical imbalance, 1995 216
7.5 Total spending by level of government, 1991 217
7.6 Total spending by level of government, 1993 218
7.7 Unconditional and conditional transfers to provinces and the
city of Buenos Aires, 1983–95 219
7.8 Vertical fiscal imbalance coefficients, 1983–95 222
7.9 Provincial social and economic indicators 226
7.10 Per capita provincial revenue and spending, 1994 228
x List of tables
7.11 Provincial public employment and spending on personnel,

1995 230
7.12 Secondary participation by provinces, selected years 231
8.1 South Africa: shares of revenue, expenditure, and transfers by
tier of government, 1990–91 240
8.2 Revenue structure of BLAs in Central Wits: 1991–92 budget 243
8.3 Central Wits: contributions to net income, 1991–93 243
8.4 Shares of property taxes in total revenue: actual and required 244
8.5 Jurisdictional changes 248
9.1 Bosnia-Herzegovina: tax structure, November 1995 284
9.2 Recommended assignment of public functions 292
9.3 Summary of suggested tax assignments 294
9.4 Tax revenue assignments for the Federation, 1997 295
9.5 Tax revenue distribution for Bosnia, 1997 295
Contributors
Junaid K. Ahmad, World Bank

Roy W. Bahl, Georgia State University
Richard M. Bird, University of Toronto
Ariel Fiszbein, World Bank
William Fox, University of Tennessee
M. Govinda Rao, Australian National University
Ernesto Rezk, Universidad Nacional de Córdoba
Anwar Shah, World Bank
François Vaillancourt, Université de Montréal
Christine Wallich, World Bank
xi
Preface
Interest in fiscal decentralization has grown greatly all over the world in
recent years. The public finances of many developed economies have to
varying degrees become more decentralized as one way of attempting to
accommodate the fiscal realities of the “post-welfare state” era.
Throughout eastern and central Europe new systems of local and inter-
governmental finance are being established as part of the evolution away
from the old central planning system. Finally, an increasing number of
developing countries are turning to various forms of fiscal decentraliza-
tion as one possible way of escaping from the traps of ineffective and inef-
ficient governance, macroeconomic instability, and inadequate economic
growth in which so many of them have become mired in more recent
years. Discussion of various aspects and issues of fiscal decentralization
is thus in the air more or less everywhere these days. Economic theorists
are theorizing about fiscal decentralization, applied economists are
attempting to measure its potential effects in various dimensions, and
policy economists are busily flying around the world dispensing advice
about it.
In many developing countries, moreover, fiscal decentralization is not
only in the air but also, to varying degrees, already on the ground. The
studies in this book describe and analyze some of the many varieties of
fiscal decentralization found throughout the developing world. Although
what has happened in the ten very different countries covered here can in
no sense be considered a representative sample, these case studies should
prove of considerable interest to a wide variety of readers both in devel-
oping countries and in the increasingly wide circles of those throughout
the world, whether in the private or public sectors, on whose activities and
interests the manifold and changing dimensions of intergovernmental
fiscal relations described here will impinge, whether they realize it yet or
not.
The contributions to this volume were originally presented at the
xiii
xiv Preface
PARADI/ICTS* Conference held in Montreal, Canada on September

19–20, 1996. They were subsequently revised and edited for inclusion in
this volume, together with an initial overview chapter prepared by the
editors. We thank André Martens, director of PARADI, who first suggested
that this conference be organized. We are also most grateful not only to the
various authors represented here, but also to the discussants and com-
mentators whose inputs are reflected in the volume: Leonard Dudley,
Nicolas Marceau, Louis Massicotte, David Sewell, France St-Hilaire,
William Watson, and Stanley Winer. In addition, we are extremely grateful
for the financial support provided by the International Development
Research Centre (IDRC), the PARADI Program, and the Social Sciences
and Humanities Research Council of Canada. Finally, we acknowledge the
indispensable administrative support provided by Marie-Christine
Thirion in Montreal and the substantial assistance provided by Duanjie
Chen in Toronto. Coordinating nine papers on ten countries by ten authors
currently living in five countries has not been the easiest task we have ever
undertaken, but it has proved surprisingly pleasant, and we think that the
results are worth it.
Richard M. Bird
François Vaillancourt
* The Programme d’analyses et de recherches économiques appliquées au développement

international (PARADI) is funded by the Canadian International Development Agency
(CIDA). The institutions affiliated with the PARADI program are the Centre de recherche
et développement en économique (CRDE) of the Université de Montréal and the Centre de
recherche en économie et finance appliquées (CREFA) of the Université Laval. The
International Centre for Tax Studies (ICTS) is located at the Faculty of Management of the
University of Toronto.
1
Fiscal decentralization in developing countries:

an overview
RICHARD M. BIRD AND FRANÇOIS VAILLANCOURT
Recent years have seen worldwide interest in fiscal decentralization.

Developed countries are reshaping their intergovernmental fiscal struc-
ture to be more in tune with the realities of the “post-welfare state”
(Bennett, 1990; Wildasin, 1997a).1 The countries in transition in eastern and
central Europe are busily setting up new systems of local and intergovern-
mental finance (Bird, Ebel, and Wallich, 1995). Many developing countries
are also turning to various forms of fiscal decentralization as one possible
way of escaping from the traps of ineffective and inefficient governance,
macroeconomic instability, and inadequate economic growth in which so
many of them have become mired in recent years.2 Each country does what
it does for its own peculiar reasons, but when so many countries in so
many different circumstances do somewhat similar things, there is likely
to be more at work than meets the local eye. The principal purpose of this
book is to take stock of the progress, problems, and potentials of fiscal
decentralization in developing countries by bringing together a set of
studies from a variety of countries around the world.
Decentralization in developing countries sometimes seems to be viewed
as either a panacea or a plague – either a cure for all the ills of such coun-
tries or an addition to their already heavy burdens. Some argue for decen-
tralization on grounds of improved economic efficiency, some on grounds
of cost efficiency, some in terms of improved accountability, and some in
terms of increased resource mobilization.3 On the other hand, others argue
that none of these virtuous outcomes is likely to be achieved in countries
in which citizen preferences are unlikely to be reflected in budget out-
comes and the institutional capacity of existing subnational (state and
local) governments is close to nil.4 From this perspective, decentralization
seems likely to result in increased costs, lessened efficiency in service deliv-
ery, and probably greater inequity and macroeconomic instability
(Prud’homme, 1995).
In general terms, it is not difficult to defend and elaborate either side of
this controversy. With respect to efficiency, for example, the standard
1
2 Richard M. Bird and François Vaillancourt
economic view is that the existence of different tax-spending packages in

different jurisdictions, coupled with individual mobility, is sufficient to
ensure that there will be efficiency-producing interjurisdictional competi-
tion in service provision (Tiebout, 1956). Similarly, empirical evidence
from a number of countries supports the proposition that locally con-
trolled services are likely to be provided at lower costs than centrally pro-
vided ones (Campbell, Peterson, and Brakarz, 1991). On the other hand,
reaping these benefits appears to require the prior existence of such rare
conditions in developing countries as significant local administrative
capacity and locally responsive and responsible officials with substantial
discretionary financial control (Bahl and Linn, 1994).
Interesting as such generalities may be, they largely miss the mark. The
essence of decentralization is that it does not occur in general but rather in
a particular country – in a country with its own history and traditions and
its own specific institutional, political, and economic context. Moreover, as
the various case studies in this book demonstrate, decentralization has
taken many different forms in different countries at different times, and
even exactly the same variety of decentralization may have very different
effects under different conditions – compare, for example, the decentrali-
zation in Morocco with that in Tunisia (Vaillancourt, this volume), two
countries similar in language, religion, and colonial administrative legacy.
As the ten case studies that constitute the bulk of this book illustrate, eco-
nomic theorists are theorizing about fiscal decentralization, applied econo-
mists are attempting to measure its effects in various dimensions, and
policy economists are busily flying around the world dispensing advice
about it. But just what is meant by fiscal decentralization? What advice
does the academic literature suggest should be given? And how does this
advice relate to what is actually taking place in the real world? In this first
chapter we attempt both to answer these questions to provide some
general background to the case studies that follow and also to draw some
general conclusions relevant to these questions from those case studies.
In the next section, we discuss what fiscal decentralization is and why it
is a matter of policy concern. We then provide a brief quantitative over-
view of the extent and nature of fiscal decentralization in the countries
covered in this volume and, for comparison, in a few others as well. But
numbers alone cannot depict adequately the complex institutional reality
that constitutes fiscal decentralization in any country. The third section of
this chapter therefore sketches briefly some of the important patterns of
fiscal decentralization to be found in the world, referring to some of the
key points developed in more detail in the country analyses that constitute
the bulk of this book. Finally, in the concluding section of the chapter, we
Fiscal decentralization in developing countries 3
draw some general lessons from the very diverse experiences recounted
here with respect to both the substance and the process of fiscal decentrali-
zation in developing countries.
Fiscal decentralization: what and why?

Four questions are addressed in this section. First, how do we define “fiscal
decentralization” and assess its success? Second, what macroeconomic
questions are associated with decentralization, and, in particular, how
should subnational borrowing be managed? Third, how do local mana-
gerial capacity and local taxation impact on decentralization? Fourth, what
do theory and experience suggest are necessary and sufficient conditions
for “successful” fiscal decentralization?
Definition and assessment

Three varieties of fiscal decentralization may be distinguished, corre-
sponding to the degree of independent decision-making exercised at the
local level.5 First, deconcentration means the dispersion of responsibilities
within a central government to regional branch offices or local adminis-
trative units.6 Second, delegation refers to a situation in which local govern-
ments act as agents for the central government, executing certain functions
on its behalf. Third, devolution refers to a situation in which not only imple-
mentation but also the authority to decide what is done is in the hands of
local governments.7
How one assesses decentralization clearly depends in part upon
whether what has occurred is best characterized as deconcentration,
delegation, or devolution. It also depends upon whether one views decen-
tralization from the top down or from the bottom up (Bird, 1980). The
approach to fiscal decentralization from the bottom up generally stresses
political values – improved governance in the sense of local responsive-
ness and political participation, for example – as well as allocative effi-
ciency in terms of improving welfare (as in the decentralization theorem of
Oates, 1972). The political literature is replete with passages praising the
virtues of decentralization. Not only will it produce more efficient and
equitable service delivery through making better use of local knowledge,
but it will also, so we are sometimes told, lead to greater participation and
democracy and hence result in more popular support for government, and
presumably in improved political stability. When to these good qualities
are added such further ascribed virtues as increased resource mobilization
and reduced strain on central finances, greater accountability, and more
responsive and responsible government in general, it is not surprising that

some have seen decentralization to be intrinsically valuable.
Whatever the precise outcomes of adopting a decentralized (in the sense
of devolved) system of decision-making, such outcomes are presumed,
from this perspective, to be satisfactory simply because the process itself is
intrinsically desirable. Local people may make wrong decisions from the
perspective of the central government or of an outside observer, but if they
make them, the decisions must, by definition, be assumed to be right for
them. From this perspective, then, decentralization is intrinsically good
because it institutionalizes the participation of those affected by local deci-
sions. The results of a good process must themselves be good.
But matters may not seem so clear if one looks at the process from the
top down, rather than from the bottom up. From the top down (the central
government) the rationale for decentralization may be, for example, to
make the life of the central government easier by shifting deficits (or at
least some of the political pressures resulting from deficits) downward.8 Or
it may be a desire on the part of the central government to achieve its
allocative goals more efficiently by delegating or decentralizing authority
to local governments, as appears to have been the case to some extent in
Colombia (Bird and Fiszbein, this volume) and Indonesia (Shah, this
volume). The goal of the central government may even be to increase the
level of national welfare, as often assumed in theoretical discussion (for
example, Boadway, Roberts, and Shah, 1993). Whatever the rationale, this
top-down approach suggests that the main criterion for evaluating fiscal
decentralization should be how well it serves the presumed national
policy objectives.
An initial problem in analyzing fiscal decentralization in any specific
setting is thus to determine whether a “good” fiscal decentralization is one
which better achieves the goals of the central government (or improves
national welfare as a whole, if one prefers) or one which frees local govern-
ments most from central dictates (or, if one prefers, improves local welfare
most). Decentralization may have many virtues: it may, for instance,
improve accessibility, local responsibility, and the effectiveness of govern-
ment. But it is not likely to yield, for instance, precisely the same expendi-
ture pattern the central government would choose to implement except in
the extremely unlikely case that the goals of central and local government
precisely coincide. In a geographically heterogeneous society, this is
simply not possible. Conflicts between central and local governments as to
what should be done are inevitable, even if each government tries faith-
fully to serve the interests of its (different mix of) constituents.
The choice of perspective is thus essential in approaching issues of fiscal
decentralization. The bottom-up perspective may be particularly

appropriate for countries like India (Rao, this volume), South Africa
(Ahmad, this volume) or Bosnia-Herzegovina (Fox and Wallich, this
volume) in which heterogeneity among different territorial units on
various dimensions is high, and to a considerable extent reflects political
decisions intended to make the national state at least potentially viable. In
most of the other countries covered in this volume, however, as in devel-
oping countries more generally (Bird, 1993), the top-down perspective
seems more likely to be appropriate. In China, for example, Bahl (this
volume) suggests that the recent reforms of taxation and intergovern-
mental finance were intended (1) to reassert macroeconomic control and
(2) to secure adequate resources for the central government to achieve such
objectives as developing important interregional infrastructure.9
Macroeconomic aspects of decentralization

The stringent conditions for successful decentralization have recently been
emphasized with respect to developing countries.10 In particular, it has
been argued that not only may decentralization fail to improve local
service delivery, it may even risk national destabilization. This risk is great-
est when revenues are decentralized without adequate steps being taken
to ensure both that local revenue mobilization is maintained and that local
authorities are capable of carrying out the corresponding expenditure
responsibilities. Argentina in the 1980s is a commonly cited example,11 but
others are not hard to find in the transitional economies of eastern and
central Europe (Bird, Ebel, and Wallich, 1995). As just noted, similar fears
appear to have played an important role in the recent Chinese fiscal
reforms (Bahl, this volume).
International experience indeed suggests that if countries decentralize
more expenditure responsibilities than revenue resources, either service
levels will likely fall or else local governments will press – successfully, it
is usually assumed – for either more transfers, or more loans, or both. One
of the clearest, and most analyzed, cases exhibiting this phenomenon is the
Russian Federation (Wallich, 1994). On the other hand, if more revenues
than expenditures are decentralized, it is often argued that local revenue
mobilization may decline and again macroeconomic imbalances may
emerge.12 Countries such as Colombia, Argentina, and Brazil are fre-
quently cited as bad examples in this respect.13 Even if both sides of the
budget are decentralized in a balanced fashion, it is often feared that local
governments may not have adequate administrative or technical capacity
to carry out their new functions in a satisfactory fashion.14 Such problems
may give rise to particular concern in developing countries where local

governments are charged with important social and economic infra-
structure investments (Bird, 1994a) – an aspect stressed in this volume in
the chapter on Morocco and Tunisia.
In view of the apparent widespread concern about the destabilizing
effects of fiscal decentralization, it may surprise some readers that some of
the case studies in this book – see especially those on Colombia and South
Africa – lend little support to the more dire predictions of macroeconomic
disaster ensuing from fiscal decentralization. Nonetheless, such concerns
continue to rank high in many countries, and care must clearly be taken to
avoid unwanted outcomes in this respect. The key to unlocking this
problem, we shall argue, is to ensure that decentralization is undertaken
in such a fashion as to increase rather than decrease accountability.
Concern for macro imbalance lies behind the common recommendation
that strict limits be imposed on the borrowing ability of subnational
governments.15 Some fear that, unchecked, subnational governments, par-
ticularly those highly dependent on national transfers, may increase
current expenditures well above their capacity to finance them out of
current revenues and then close the gap through borrowing. Others argue
that since macroeconomic stabilization is properly a national government
task, it is important that the national government have full control over all
the instruments of policy it needs to carry out this task properly, including
borrowing – and particularly borrowing abroad.
Such arguments (or variants of them) have been made in many coun-
tries, and the result has often been the imposition of a variety of restraints
on provincial and local borrowing, for example, limiting such borrowing
to financing capital expenditures, limiting debt service to a maximum per-
centage of current revenues, or requiring prior approval of central govern-
ment for borrowing (Bird, Ebel, and Wallich, 1995).
In fact, however, a properly designed local finance system would not
appear to require any specific controls on debt beyond those imposed by
a well-functioning private capital market – something which may not, of
course, be considered to exist yet in many developing countries. With
respect to foreign borrowing by subnational governments, however, there
may be a special problem owing to the apparent assumption by many
lenders that all “public” debt is (implicitly) guaranteed by the central
government. A possible (partial) solution to this problem might be through
“semi-privatizing” subnational borrowing as much as possible, for
example, through “revenue bonds” which are guaranteed explicitly and
solely by specific (related) revenue sources. This approach would have the
additional virtue of increasing one of the potential advantages of
decentralizing public borrowing in the first place – namely, risk diversifica-

tion – by increasing the (so to speak) portfolio of public debt on offer,
assuming (as seems not implausible) that the revenue streams attributable
to different components of the public sector are not highly correlated.16
Imposing debt limits to prevent local governments from making fiscal
mistakes may produce more perverse results than the public insurance of
savings deposits so often condemned by economists. Like deposit insur-
ance, debt limits and similar controls may be perverse precisely because
they prevent market discipline from being applied (see Ahmad, this
volume). Potential lenders to governments, unlike ordinary citizens choos-
ing a bank in which to place their savings, can reasonably be expected to
be capable and motivated with respect to finding out what risks they are
running with their money. From this perspective, much of the concern
about irresponsible local governments getting themselves into trouble
seems like another instance of inappropriate and misconceived paternal-
ism – the “father knows best” attitude so common with central govern-
ments facing the uncomfortable prospect of losing control as a result of
decentralization. In life, children seldom learn to save unless they suffer
the consequences of not having done so. And local governments are as
unlikely to be well managed if they are saved from the possibility of
making mistakes by the imposition of arbitrary limits as they are if they
know they will always be bailed out by the central government.
If a national government wants to avoid macro problems arising from
subnational debt, it can do so by not subsidizing such borrowing and by
letting subnational governments that borrow too much go bankrupt. This
is exactly what was done in Morocco, where the government changed the
subsidy scheme for local governments from one of budget-balancing
grants, in which both capital and interest payments on loans increase
transfer receipts, to a formula-based equalization transfer which takes no
account of borrowing (Vaillancourt, this volume). In addition, lenders
were explicitly told not to count on financial bail-outs.17 The possible prob-
lems arising from misguided foreign borrowing, however, may require
more careful national attention, for instance, requiring explicit prior
approval from the central government before any such loans may be con-
tracted.
Unless subnational governments are able, so to speak, to “save them-
selves” from fiscal crises by drawing on their taxing powers, however,
their only options in practice in many countries may be bankruptcy or bail-
out. In the end, the only way to reduce the moral hazard implicit in this
situation may be by imposing strict limits on subnational borrowing. What
needs to be emphasized, however, is that the root of the problem lies in the
very limited taxing powers available to subnational governments that are

expected and required to carry out a much wider range of functions than
they can finance on their own without extensive reliance on central
support (either direct through transfers or indirect through bail-outs).18
Unless local “ownership” of the tax base is extended considerably beyond
the narrow limits existing in most developing countries (see later discus-
sion) it may not be desirable to loosen borrowing rules.
The difficulty of envisioning, let alone carrying out, bankruptcy in the
public sector provides good reason to require that there be fairly stringent
conditions on all subnational borrowing – not for macro reasons, however,
but in order to ensure local government accountability. Along the same
lines, all local borrowing should be reported immediately and in a trans-
parent fashion so that no one can shift hidden debts onto the next
administration, and so that both local voters and the national government
can have a better handle on what is going on. Moreover, since the only case
for local borrowing – and it is a good one – is to finance capital investment,
no borrowing should be permitted for other purposes, no matter how
worthy.19
Finally, one matter that sometimes gives rise to concern appears to be
largely a non-problem, namely, the ability of local governments to borrow
on the basis of the increased cash flow as a result of transfers (or, for that
matter, royalties). As long as the borrowing is not subsidized, why is this
a problem? The portion of these transfers that is not specifically earmarked
constitutes “own revenues” of local governments, and if some agency is
willing to lend money based on this security it should be free both to do so
and to bear the consequences if the loan goes bad. Of course, transfers
make good security only if they are predictable, which has by no means
always been the case in developing countries (Bird, 1990). Generally, trans-
fers are more likely to be used for this purpose when they are enshrined in
law (Tunisia) or, even better, in the constitution (Colombia, Argentina,
South Africa), than when, as in Morocco, they are made by ministerial
directive or, as in China, effectively negotiated on an ad hoc basis.
Local capacity and taxation

An essential ingredient in improving the life of the poor in many countries,
both immediately and in terms of enhanced productivity in the long run,
is the improvement of basic infrastructure – roads, water, sewerage, and
electricity (World Bank, 1994). A number of countries have used inter-
governmental transfers to guide and shape local investments in these
areas, as emphasized, for example, in the chapter on Morocco and Tunisia.
Similarly, in Indonesia, specific grants are provided for provincial and dis-
trict road improvement (Shah and Qureshi, 1994; Shah this volume). The
program is designed to provide minimum standards of road service across
the nation and to facilitate the development of an internal common market.
The grant allocation formula is related positively to indicators of poor
roads and low motor vehicle registrations (proxies for road expenditure
needs). As in the case of other Indonesian transfers, local discretion in the
use of this grant is restricted, which may limit its effectiveness. The use of
the grant is confined to the repair and upgrading of existing roads: new
roads have to be financed from other sources. Projects for the repair of
roads have to be approved by districts and then forwarded to the central
government.
Some have been concerned that local governments subject to less
detailed guidance and control than in this case may not have the capacity
to handle such critical functions. Colombia, for example, has a much less
“guided” system, which nonetheless appears to have been moderately
successful in directing infrastructure investment to the poor (World Bank,
1996c). Under the so-called “coparticipation” system, local communities
provide labor and local materials, and municipal governments contribute
a portion of the cost. This fund not only fosters community involvement
in identifying needs and choosing projects but also promotes community
participation in the execution, operation, and maintenance of the works.
Municipalities have to prepare projects which are then appraised by the
fund against technical and environmental criteria. The other important
requirement is that the beneficiaries should be low-income rural families.
Projects may be carried out by any of a number of types of contractor
(private firms, non-government organizations, state agencies, or uni-
versities), who compete to supply the works and services.
Although partial and preliminary, the evidence so far concerning local
capacity to carry out such functions in Colombia is surprisingly encour-
aging. A recent study of a sample of sixteen municipalities found numer-
ous beneficial results of decentralization in terms of the enhancement of
local capacity in the areas of labor, capital, and technology (World Bank,
1995a). Colombian municipalities are, for example, increasing the skills of
local bureaucracies through such means as competitive hiring, sharing the
services of professionals among municipalities, training municipal
employees, and rotating personnel through different departments in the
same municipality. Capacity in terms of capital has also been increased.
One municipality has totally privatized road maintenance; another has put
private developers in charge of the construction of urban roads.20
Computers have been introduced to monitor water and sanitation services
in other localities. Municipalities have started to share certain equipment.

They have also improved their technological capability in terms of inter-
nal organization, planning, and monitoring to ensure better management
of municipal projects.
Underlying these improvements is a more basic change: Colombian
municipalities have been moving to a “demand-driven” (bottom-up)
approach to public services as opposed to the previous “supply-driven”
(top-down) one. Increasingly, reflecting both the new liveliness of local
politics and (with substantial variations from area to area) more extensive
community participation, people are getting what they want, rather than
what someone in the capital thinks they should want. In practice, empha-
sis has been put on roads, education, and water projects: these are the
needs people perceive, and these are the needs that the newly empowered
and responsive local governments are attempting to satisfy. Opinion
surveys suggest that the resulting sectoral allocation of resources is con-
sistent with community preferences, with most respondents indicating
that they trust the local government more than the national government to
deliver goods and services (World Bank, 1995a).
All this is most encouraging for believers in the potential allocative and
democratic virtues of decentralization. As in such well-known Asian cases
as the Orangi project in Karachi, Pakistan (Bird, 1995), such popular
participation both reveals strong preferences for the project being built and
tends to keep costs down. Depending on the precise nature of the project,
such community involvement may also enhance substantially the effective-
ness of “targeting” in terms of poverty alleviation. Participants in such com-
munal work projects are in effect “self-selected,” being poor and willing
enough to volunteer their major asset, their labor, without remuneration.
A recent review of experience with the social investment funds set up on
roughly similar lines in a number of Latin American countries (Glaessner
et al., 1994) concluded that such funds have proved to be generally effec-
tive because
(1) they have been demand-driven, thus requiring a high degree of local
involvement,
(2) their operations have been transparent and hence accountable,
(3) they have been carefully targeted to low-income groups, and
(4) they have been relatively autonomous in their operation, usually being
run by private-sector managers and freed from much official red tape.
Most important is the direct involvement of the beneficiary groups in
both the management of the fund, and in the selection, operation, and
financing of projects. In particular, it appears to be critical to require cost-
sharing from even the poorest communities – usually their contribution

takes the form of land and labor – on the well-founded grounds that people
take more interest in what they have to pay for and are hence more likely
to be interested in ensuring that they get value for their contributions. On
the whole, Latin American experience clearly shows that it is possible to
deliver local public services relatively effectively and efficiently even in the
face of adverse macroeconomic circumstances and, in some cases, very
poorly functioning public administrations. Where there is a will, the way
seems to have been shown.
As for local taxes, the “correct” revenue assignment in a multi-level
government structure is by no means clear in principle, and is generally
controversial in practice. The fundamental problems are twofold. First, the
central government can inherently collect most taxes more efficiently than
can local governments. Second, the potential tax bases that can be reached
by the latter vary widely from region to region. The first of these problems
gives rise to vertical imbalance; the second produces horizontal imbal-
ance.21
Two basic principles of assigning revenues to subnational governments
may be suggested. First, “own-source” revenues should ideally be suffi-
cient to enable at least the richest subnational governments to finance from
their own resources all locally provided services primarily benefitting local
residents. Second, to the greatest extent possible, subnational revenues
should be collected from local residents only, preferably in relation to the
perceived benefits they receive from local services. More specifically,
among the characteristics that might be sought in an “ideal” subnational
tax are the following:
(1) the tax base should be relatively immobile, to allow local authorities
some leeway in varying rates without losing most of their tax base;
(2) the tax yield should be adequate to meet local needs and sufficiently
buoyant over time (that is, it should expand at least as fast as expendi-
tures);
(3) the tax yield should be relatively stable and predictable over time;
(4) it should not be possible to export much, if any, of the tax burden to
non-residents;
(5) the tax base should be visible, to ensure accountability;
(6) the tax should be perceived to be reasonably fair by taxpayers;
(7) the tax should be relatively easy to administer efficiently and effec-
tively.
Not everyone would agree that all these characteristics are necessarily
or equally desirable – for example, is it unequivocably good that local
governments should be insulated from either the tax base consequences of

their tax rate choices or from inflation? Moreover, it is by no means simple
to draw from the array of (subjective) information any clear conclusions
about different possible local taxes that may be included within this frame-
work. But some candidates come to mind, as discussed to varying extents
in the country studies included in this volume, notably property taxes,
taxes on automobiles, and, perhaps most importantly if local governments
are to have major expenditure responsibilities (for example, for education),
some form of surcharge on national taxes. Most of the desired (and desir-
able) aims of decentralized revenue policy may be achieved solely by
allowing variation of the rates of such surcharges, perhaps subject to a con-
straint on minimum rates to restrict competition for tax base. In addition,
since “tax-exporting” breaks the critical link between local spending deci-
sions and the taxes borne by local residents, care should also be taken to
prevent provinces from exporting their tax burdens – for example, by lim-
iting access to the taxation of business.
Conditions for success

Experience in a variety of settings suggests that two conditions seem par-
ticularly important for successful decentralization, whether success is
defined in terms of macro balance or micro efficiency. First, the local deci-
sion process must be democratic in the sense that the costs and benefits of
decisions are transparent and that all those affected have an opportunity
to influence the decision. Given the inevitable imperfection of democratic
institutions, and the ability of the rich and powerful to come out on top in
most systems, this is obviously a counsel of perfection. Nevertheless, the
implication is clearly that decentralization means something quite differ-
ent in countries such as China, Indonesia, and others in this volume that
are notably not democratic in this sense than in others such as Argentina
and South Africa that, on the whole, are. In the terms used earlier, in the
absence of meaningful local democracy, only the “top-down” delegation
approach seems to make sense.
Second, and more amenable to policy design, the costs of local decisions
must be fully borne by those who make the decisions. That is, there should
be no significant “tax-exporting,” and no funding at the margin from trans-
fers from other levels of government.22 This means that local governments
need to control the rates (and perhaps bases) of at least some taxes. When
these rather strict conditions are satisfied, devolution is sensible, whether
viewed instrumentally or intrinsically. When they are not, it may not be.
Even when one or more of the conditions set out above does not hold,
the delegation of implementation responsibilities to local bodies may still

make instrumental sense provided that the incentives facing local decision-
makers are properly structured, that is, structured to produce the results
desired by the central government (in its capacity as representing the
population as a whole). In the absence of the right incentive structure, the
effects of either delegation or devolution on the efficiency and equity of
resource allocation may be much less beneficial than often alleged. What
is required for decentralization to produce efficient outcomes is what has
been called a hard budget constraint with respect to devolved functions in
order to ensure accountability, combined with incentive-compatible rules
(prices, monitoring) with respect to delegated functions (Bird, 1993).
Accountability is a complex concept, with many dimensions. Political
accountability requires political leaders at all levels to be responsive and
responsible to their constituents, and those constituents to be fully
informed about the consequences of their (and their leaders’) decisions.
Administrative accountability requires a clear legal framework with
respect to who is responsible for what, what financial reports are to be
made in what form, to whom, and when, and so on. Economic accountabil-
ity requires that local residents are responsible for paying for local services,
which in turn requires that local authorities can set some tax rates. The
critical point in this respect is accountability at the margin. It is perfectly
possible (in principle) for a local government to be 90 percent dependent
on central transfers and still be fully accountable – to its citizens and/or
the central government, depending on circumstances.23 For this reason, the
best form of intergovernmental transfer is one the amount of which is fixed
in advance and will not be altered as a result of any (in-period) action by
the recipient (Ahmad and Thomas, 1997). Such a lump-sum transfer by
formula implies that, at the margin, local actions to raise or lower local rev-
enues or expenditures will directly affect outcomes – which is what is
needed to ensure accountability.24
If decentralization is to work, those charged with providing local infra-
structure and services must be accountable both to those who pay for them
and to those who benefit from them. Unfortunately, enforcing accountabil-
ity at the local level is not easy. It requires not only clear incentives from
above but also the provision of adequate information to local constituents
as well as the opportunity for them to exercise some real influence or
control over the service delivery system. “Informal” organizations must be
structured like this almost by definition or they cannot exist. But it can be
a challenge in the political and social circumstances of many developing
countries to introduce a similar degree of responsiveness into formal
governmental organizations.
Just as accountability is the key to improved public sector per-

formance, information is the key to accountability. The systematic collec-
tion, analysis, and reporting of information that can be used to verify
compliance with goals and to assist future decisions is a critical element
in any decentralization program. Such information is essential both to
informed public participation through the political process and to the
monitoring of local activity by central agencies responsible for super-
vising and (usually) partially financing such activity. Unless local
“publics” are made aware of what is done, how well it is done, how
much it cost, and who paid for it, no local constituency for effective
government can be created. Unless central agencies monitor and evalu-
ate local performance, there can be no assurance that functions of
national importance will be adequately performed once they have been
decentralized.
An important accompaniment to any top-down decentralization
program is thus an improvement in national evaluation capacity.
Decentralization and evaluation (for example, using cost-benefit analysis)
are not substitutes; they are complements. Another essential element of the
“hard budget constraint” system needed to induce efficient local decisions
is adequate central enforcement capacity in the shape of credible informa-
tion-gathering and evaluation. The “carrot” of central financial support of
local efforts must be accompanied by the “stick” of withdrawn support if
performance is inadequate, which of course requires both some standard
of adequacy and some way of knowing whether performance is satisfac-
tory.
Various mechanisms for building such evaluative capacity into a decen-
tralization program are conceivable. One might be to build “sunset” pro-
visions into the program, that is, to provide that (say) the newly prominent
role given to local institutions in the water supply area will be subject to
renewal in a number of years, provided they pass some kind of inde-
pendent evaluation of their performance. As noted in Bird and Fiszbein
(this volume), such a sunset clause was incorporated in Colombia’s 1991
constitutional reform. Another approach, as in India (Rao, this volume),
may be to establish a Finance Commission to re-examine intergovern-
mental fiscal relations periodically and adjust them if necessary. South
Africa has adopted a somewhat similar approach with its Financial and
Fiscal Commission (Ahmad, this volume). A quite different approach
might be to use the likely need for some centrally supported access to
capital markets to finance local infrastructure not only as a screening
device to reject obviously flawed projects but also as an evaluation system
to build up “ratings” of local capacity.25
A quantitative overview of fiscal decentralization

The way in which any country organizes its public sector invariably
reflects its history, its geography, its political balance, its policy objectives,
and other characteristics that vary sharply from country to country.
Nonetheless, since all countries, except perhaps the very smallest, have
more than one level of government, they necessarily all have some kind of
intergovernmental fiscal system. Four big questions must be answered
with respect to intergovernmental finance in any country:
(1) Who does what? – the question of expenditure assignment.
(2) Who levies what taxes? – the question of revenue assignment.
(3) How is the (virtually inevitable) imbalance between the revenues and
expenditures of subnational governments that results from the
answers to the first two questions to be resolved? – the question of ver-
tical imbalance.
(4) To what extent should fiscal institutions attempt to adjust for the differ-
ences in needs and capacities among different governmental units at
the same level of government? – the question of horizontal imbalance,
or equalization.
Ideally, these questions should be approached in the specific circum-
stances of each country in a manner that is consistent with achieving the
relevant policy objectives – not only the normal public finance trio of effi-
ciency (allocation), equity (distribution) and stabilization but also eco-
nomic growth, as well as such nebulous (but politically resonant) goals as
“regional balance.” In many instances, of course, there will be conflicts not
only between these objectives but also between local and central percep-
tions of the weights to be attached to them. Moreover, like all public poli-
cies, intergovernmental fiscal policies must be developed taking into
account both political constraints (for example, the strength of different
regions and groups in political decisions) and economic constraints (for
example, the stage of development of financial markets) facing policy-
makers. Finally, all policy changes proposed must of course start from the
given set of initial conditions: every country has a history, and the current
state of its fiscal institutions in large part reflects the product of an accre-
tionary process of policy change over time.
Any country’s intergovernmental finance system inevitably reflects in
many ways the complex reality of the country. Clearly, countries differ on
many dimensions with respect to intergovernmental finance: how many
local governments are there? What are their relative sizes in terms of
population and economic activity? How different are they in terms of per
capita income? Natural resource wealth? What are the historical origins of
the move to fiscal decentralization: bottom-up or top-down, peaceful or
violent? What is the geography of the country: compact or dispersed? How
homogeneous is it: in terms of language, in its ethnic groups, and in its uni-
fying cultural myths? To what extent do state boundaries coincide with
heterogeneity in any of these dimensions? Are regional interests explicitly
represented in the central political structure? How?
Moreover, large countries tend to have more complex and formal
systems of intergovernmental finance, often explicitly “federal” in
nature.26 Developing countries tend to face different problems, and
possibilities, than developed countries in this as in other areas, in part
arising from political instability, in part reflecting their economic status,
and in part from lesser technical capacity. And countries in the process of
transition from a central-planning to a more “market-oriented” environ-
ment have still different problems – for example, with respect to privatiza-
tion and the allocation of assets among governments.27
Given this enormous variety, the optimal (not to mention feasible) solu-
tions to intergovernmental fiscal problems will be quite different from
country to country, depending upon where they are starting and what they
are trying to do.28 Nonetheless, although comparative analysis is most
unlikely to yield a clear prescription as to what should be done at one par-
ticular time in one particular country, it may be helpful in understanding
just how and why certain institutional structures work (or do not work) in
particular circumstances.29 While subject to misinterpretation in the
absence of complete knowledge of all relevant aspects of the institutions
and settings being compared, reference to experience abroad is often the
only guide available and, despite its obvious limitations, such experience
may provide useful lessons in assessing the potential strengths and weak-
nesses of decentralization in any country.
In particular, the reasons for success or failure, so far as they can be dis-
cerned, may help focus attention on some key factors in the process and at
best may highlight a feature that appears to require more emphasis if success
is to be achieved. At the very least, a comparative approach may help correct
the belief, sometimes held by otherwise well-informed people, to the effect
that there must be a simple solution in existence somewhere that could
replace the seemingly unending complexity and negotiation characterizing
intergovernmental financial arrangements in their own country.
Some international comparisons

How important is subnational spending? Bahl and Linn (1992) found that
between 6 percent and 50 percent of total government spending, with an
average of 23 percent, was accounted for by local governments in the

twenty-one developing countries for which they had data. For ten devel-
oped countries, OECD (1991) found the comparable range to be from 12
percent to 53 percent, with an average of 26 percent. On the other hand,
using a different source, and sample, Bird (1995) found an average local
expenditure share of 22 percent for 18 developed countries and only 9
percent for 16 developing countries.30 Comparative data on the impor-
tance of local (and other subnational) spending are surprisingly hard to
find, and to interpret. Nonetheless, it is probably safe to say that, on the
whole, the degree of fiscal decentralization in terms of spending tends to
be greater in richer than in poorer countries (Wasylenko, 1987), but that
there is considerable variation within each of these groups, however
defined.
An alternative, though clearly less satisfactory, way of indicating the
relative importance of subnational governments is in terms of the propor-
tion of central government expenditure that goes to subnational govern-
ments in the form of transfers. From this perspective, there appears to be
no necessary relation between the level of a country’s development, the
size of its central government, and the importance of subnational fiscal
transfers. For example, Bird (1995) found that on average local govern-
ments in the industrial countries in his sample financed only 62 percent of
their expenditures from their own resources, compared to 88 percent for
the developing countries. As usual, of course, there is substantial varia-
tion within each group. In Chile and Malaysia, for example, local govern-
ments financed more than 60 percent of their expenditures from own
revenues in 1990, while Argentina, India, and Pakistan had lower levels
of financial autonomy (38 percent to 50 percent), and in Indonesia, the
proportion was only 21 percent in 1989 (UNDP, 1993). In most countries,
fiscal decentralization in the form of increasing the importance of local
expenditures is in practice likely to be accomplished only by simultane-
ously increasing the importance of national fiscal transfers. The level and
design of such transfers is therefore invariably a central aspect of any
decentralization process.
Two further broad characteristics of decentralization around the world
emerge from reviewing the scarce comparative data available. First,
however much local governments spend, and whatever they spend it on,
the revenues under their direct control are almost invariably less than
their expenditures (Bird, 1995). The only exceptions are a few countries
– rich and poor – in which local governments are without any impor-
tance as spenders, and a very few rich countries (mostly in Scandinavia)
in which local governments have substantial access to large and elastic
tax bases usually by levying surcharges on national income taxes (Bird
and Slack, 1991). Countries clearly have considerable discretion in

arranging the structure of their public sectors regardless of their level of
income.
A similar conclusion emerges from consideration of the second
characteristic common to all countries: not all subnational governments
are equal. Even in quite small and homogeneous countries, there are big
cities and small towns, heavily urbanized areas and rural municipalities.
The resulting unevenness in access to local public resources can be
marked: in 1993, per capita revenues in low-income provinces of Vietnam
were only 9 percent of those in high-income provinces. However, owing to
central government transfers, expenditures in the poorest provinces of
Vietnam were 59 percent of those in the richest provinces (Bird, Litvack,
and Rao, 1995). Similarly, in Indonesia, Timor (one of the poorest
provinces) has a per capita own-source revenue equivalent to 4 percent of
Jakarta’s. Again, however, owing to transfers from the central government,
Timor’s per capita expenditures are 40 percent of those in Jakarta (Shah
and Qureshi, 1994). Transfers are especially important in determining the
pattern and level of expenditures in the poorer regions of countries. In
Argentina, for instance, although the per capita GDP in low-income
provinces is only 39 percent of that in high-income provinces, per capita
public spending in the low-income provinces is 130 percent of that in the
wealthier provinces (Rezk, this volume).
In the remainder of this section, we present some key comparative
information both for the countries included in this volume and for a few
other countries that seem relevant for various reasons, namely, the United
States, Germany, Russia, Brazil, Vietnam, Australia, Switzerland, and
Canada. Russia, like China, is large and in “transition.” Vietnam is a
smaller representative of the same group (though not nearly as small as
Bosnia-Herzegovina). Brazil, like India (and Pakistan), is a large develop-
ing country and also a formal federation. The United States, of course, is
the largest federal developed country. Germany, Australia, Canada, and
Switzerland are smaller (and more equalization-minded) examples of the
same group – more comparable in size to South Africa or Argentina, the
other formal federations included in this volume.
As shown in tables 1.1 and 1.2, there are, of course, wide variations in
many respects among this set of countries. The data shown in these tables,
even for the same country, are often from different sources or for different
years, and in some instances are estimated in whole or in part. Many of the
specific data for individual countries may no doubt be questioned.31
Nonetheless, the broad comparative patterns shown in these tables seem
unlikely to be much changed by most plausible corrections.
Table 1.1. Fiscal decentralization: the case-study countries
Bosnia- South
Argentina Herzegovinaa China Colombia India Indonesia Morocco Pakistan Africab Tunisia
Demographic
(1) Population (millions) 34.0 3.1 1191.0 36.0 914.0 190.0 26.0 126.0 40.0 9.0
(2) Density (no./sq.km.) 12.0 n.a..0 124.0 32.0 278.0 10.0 59.0 159.0 33.0 54.0
(3) Population growth (%) 1.2 n.a..0 1.2 1.9 1.8 1.6 2.0 2.9 2.2 1.9
(4) Urban (%) 88.0 n.a..0 29.0 72.0 27.0 34.0 48.0 34.0 50.0 57.0
Economic
(5) GNP per capita (US$) 8 110.0 728.0 530.0 1 670.0 320.0 880.0 1140.0 430.0 3 040.0 1 790.0
(6) GNP per capita growth (%) 2.0 n.a..0 7.8 2.4 2.9 6.0 1.2 1.3 ⫺1.3 2.1
(7) Inflation (%) 317.2 n.a..0 8.4 25.6 9.7 8.9 5.0 8.8 14.3 6.3
(8) Income of the lowest 20% (%) n.a..0 n.a..0 6.2 3.6 8.5 8.7 6.6 8.4 3.3 5.9
Public sector
(9) Number of regions 23.0 10.0 30.0 33.0 25.0 27.0 65.0 4.0 9.0 23.0
(10) Size of government (%) 27.0 n.a..0 23.0 32.0 25.0 19.0 24.0 24.0 40.0 32.0
(11) Subnational as % of total exp. 43.0 57.0 60.0 40.0 58.0 20.0 15.0 33.0 52.0 8.0
(12) Subnational as % of total rev. 40.0 57.0 60.0 21.0 41.0 7.0 9.0 9.0 14.0 5.0
(13) Subnational rev. as % of 82.0 100.0 106.0 55.0 49.0 33.0 61.0 12.0 7.0 63.0
(13) subnational exp.
Notes and sources:

a
Data cover only the “Federation” (Bosniac-Croat section) and fiscal data refer to planned 1997 budget.
b
Fiscal data are for pre-1994 period.
(1) Population in millions, mid-1994. World Bank, 1996d, table 1.
(2) Number of persons per square kilometre, mid-1994. Calculated from data in World Bank, 1996d, table 1.
(3) Average annual rate of population growth, 1990–94. World Bank, 1996d, table 4.
(4) Urban population as percentage of total population, 1994. World Bank, 1996d, table 9.
(5) GNP per capita in US dollars, 1994. World Bank, 1996d, table 1.
(6) Average annual growth of GNP per capita, 1985–94. World Bank, 1996d, table 1.
(7) Average annual GDP deflator, 1984–94. World Bank, 1996d, table 2.
(8) Percentage share of income received by the lowest 20 percent of households/population, various years. World Bank, 1996d, table 5.
(9) Number of major regional units (some special territories may be excluded). Various sources.
(10) Total government expenditures as percentages of GNP, estimated from data on central government expenditures as percentage of GNP,
1994. World Bank, 1996d, table 14; IMF, 1993; UNDP, 1993, tables 4.2 and 4.3, and other information from a variety of country sources that are
not directly comparable.
(11)–(13) Estimated from data from same sources as (10).
Table 1.2. Fiscal decentralization: other countries
Australia Brazil Canada Germany Russia Switzerland USA Vietnam
Demographic
(1) Population (millions) 18.0 159.0 29 82.0 148.0 7.0 261.0 72.0
(2) Density (no./sq.km.) 2.0 19.0 3 228.0 9.0 170.0 28.0 22.0
(3) Population growth (%) 1.1 1.7 1.3 0.6 0.0 1.0 1.0 2.1
(4) Urban (%) 85.0 77.0 77 86.0 73.0 61.0 76.0 21.0
Economic
(5) GNP per capita (US$) 18 000.0 2 970.0 19 510 25 580.0 2 650.0 37 930.0 25 880.0 200.0
(6) GNP per capita growth (%) 1.2 ⫺0.4 0.3 n.a..0 ⫺4.1 0.5 1.3 n.a.0
(7) Inflation (%) 4.1 900.3 3.1 n.a..0 124.3 3.7 3.3 102.6
(8) Income of the lowest 20% (%) 4.4 2.1 5.7 7.0 3.7 5.2 4.7 7.8
Public sector
(9) Number of regions 6.0 26.0 10 16.0 89.0 26.0 50.0 53.0
(10) Size of government (%) 40.0 35.0 49 52.0 45.0 32.0 38.0 25.0
(11) Subnational as % of total exp. 43.0 34.0 56 41.0 48.0 52.0 44.0 33.0
(12) Subnational as % of total rev. 28.0 23.0 50 35.0 66.0 51.0 42.0 58.0
(13) Subnational rev as % sub. exp. 55.0 57.0 76 75.0 100.0 41.0 75.0 177.0
Notes and sources:

(1) Population in millions, mid-1994. World Bank, 1996d, table 1.
(2) Number of persons per square kilometre, mid-1994. Calculated from data in World Bank, 1996d, table 1.
(3) Average annual rate of population growth, 1990–94. World Bank, 1996d, table 4.
(4) Urban population as percentage of total population, 1994. World Bank, 1996d, table 9.
(5) GNP per capita in US dollars, 1994. World Bank, 1996d, table 1.
(6) Average annual growth of GNP per capita, 1985–94. World Bank, 1996d, table 1.
(7) Average annual GDP deflator, 1984–94. World Bank, 1996d, table 2.
(8) Percentage share of income received by the lowest 20 percent of households/population, various years. World Bank, 1996d, table 5.
(9) Number of major regional units (some special territories may be excluded). Various sources.
(10–13) Estimated from data on central government expenditures as percentage of GNP, 1994. World Bank, 1996d, table 14; IMF, 1993; UNDP,
1993, tables 4.2 and 4.3, and other information from a variety of country sources that are not directly comparable.
Broadly and with some overlapping of categories, five types of countries

are included in these quantitative comparisons:
(1) developed-country federations, two large (the United States and
Germany) and three small (Canada, Australia and Switzerland);
(2) developing-country federations, both large (India, Pakistan and Brazil)
and small (Argentina, South Africa, and Bosnia-Herzegovina);
(3) unitary developing countries, both large (Indonesia and China) and
small (Colombia, Morocco, and Tunisia);
(4) transitional countries, one large and “federal” (Russia), one small and
federal (Bosnia-Herzegovina), one large and unitary (China), and one
small and unitary (Vietnam); and
(5) “new” countries (South Africa and Bosnia-Herzegovina).
The tables depict both some basic demographic and income character-
istics of the selected countries and some important features of their inter-
governmental fiscal systems. An important characteristic which is not
captured in these numbers is the degree of “homogeneity” of the country
in both ethnic and economic terms. For example, Russia is clearly consid-
erably less homogeneous than China in most relevant respects.32
Experience suggests that countries with a small number of constituent
units and with substantial ethnic and other differences between the units
(such as Canada), may work quite differently from countries with a larger
number of less heterogeneous units (such as the United States), but this
question cannot be discussed in detail here.
To paraphrase Shakespeare, some countries may be “born” federations
in the sense that a federal structure from the beginning played a real role
in the political and fiscal structure (India), some may “become” federations
in order to reconcile regional differences and preserve the nation (South
Africa), and some may have federation “thrust upon them” in a desperate
attempt to keep them alive (Bosnia-Herzegovina). Matters do not always
stay the same, however. Australia, for example, was born a federation
because, given its history as a set of widely separated and separate
colonies, there was no other viable model at the time. As time went on, and
distances and distinguishing factors shrank, the center became much
stronger. In contrast, as Rezk (this volume) notes, the opposite may take
place in another basically “non-natural” federation, Argentina. Germany
had federation thrust upon it after the Second World War in a deliberate
effort to weaken the central state, but the recent addition of the eastern
Länder, with their very different history, has probably strengthened the
political basis for federation. On the other hand, “natural” federations
such as multilingual Switzerland, Canada, and India are composed of
entities that were distinct at the foundation of the country and have largely
remained distinct.33
An interesting point that emerges in tables 1.1 and 1.2 is that the transi-
tional countries stand out as those in which the basic flow of resources
is from below to above. Because most revenues are collected at the local
level, in effect transfers appear to flow upward rather than downward
as in every other country. To a considerable extent this is simply an illu-
sion, particularly in China and Vietnam. Since the central government in
these countries in principle controls all revenue, by allowing different
localities to retain different proportions of collections, it is in effect
behaving as central governments in other countries that first collect the
revenues and then distribute some part of them to lower levels of
government.34 Many of the recent intergovernmental fiscal problems in
Russia reflect the inappropriate combination of this sort of “bottom-up”
revenue collection system with a federal structure which has made it
increasingly difficult for the central government to control local govern-
ments. As the chapter on Bosnia-Herzegovina discusses, this problem is
likely to prove even more critical in that country. Moreover, it was at
least in part to avoid similar problems in the face of growing local de facto
economic independence that China’s recent reform centralized the
collection of the major taxes, thus creating the need to establish, for the
first time in that country, a formal system of general intergovernmental
fiscal transfers.
Patterns of fiscal decentralization

As the brief comparisons in the previous section suggest, as wide a range
of institutional structures and relations is found within nominally federal
as within nominally unitary countries (Norregaard, 1995). Indeed, the
difference between a tight federation such as Malaysia or Germany and
most unitary countries is probably less than that between such federations
and looser federations such as India and Canada, in which state govern-
ments have more power to act independently with respect to expenditure
and taxing patterns. The constitutional label matters less than the reality of
how intergovernmental relations work in practice. Broadly, however, we
suggest that two “models” of intergovernmental fiscal relations can be dis-
cerned in the literature. In addition, three broad patterns of intergovern-
mental fiscal relations may be discerned in the world, as shown in the
matrix set out in table 1.3. The entries in the cells in this table are intended
as illustrations only: they include both the countries listed in the earlier
tables and a few others.
Table 1.3. A taxonomy of fiscal decentralization
Fiscal federalism Federal finance
Developed countries France, Japan USA, Canada

Developing countries Indonesia, Colombia, Morocco, India, Brazil, Argentina, Pakistan,
Tunisia South Africa
Transitional countries China, Vietnam Russia, Bosnia-Herzegovina
Fiscal federalism and federal finance

The dominant economic literature on “fiscal federalism” sets out a norma-
tive model under which, in effect, the central government, acting as the
benevolent interpreter of the will of the people, is given guidance as to how
to structure the institutional rules of the intergovernmental system in
order to ensure that local government agents act as the central government
(and presumably the people as a whole) would wish. Even if the central
government in many countries is not all that benevolent, these rules may
still provide useful guidelines with respect to intergovernmental fiscal
relations (Bird, 1993).
In this approach, local governments may for most purposes be consid-
ered to be agents of the central government – or, in some federal countries,
of state governments – rather than independent actors.35 France and
Britain illustrate this pattern among developed countries (Bennett, 1990),
while Indonesia, Morocco and Tunisia offer examples among the develop-
ing countries covered in this volume, as does Vietnam among the transi-
tional countries (World Bank, 1996a). Among the transitional countries,
China seems intermediate between Russia and Vietnam in this respect,
although clearly closer to the latter. From this perspective, the appropriate
framework for assessing decentralization is, in the terms used earlier,
clearly “top-down” and takes the form of deconcentration or at most
delegation, and the appropriate analytical framework is that of agency
theory.36
A good example of where the fiscal federalism model clearly applies is
with respect to intergovernmental fiscal transfers in Indonesia. Indonesia
has had the same government since 1965. Authority is highly concentrated
in the central government, which confirms the appointments of the
governors of the twenty-seven provinces and which has officials dispersed
throughout the country. On the other hand, the government has consis-
tently emphasized both economic growth and regional development,
including a marked degree of centrally directed decentralization.
Although financial autonomy – the share of local expenditures financed by

local revenues – varies widely across provinces (from 0.08 to 0.70), on
average local governments currently raise less than one-quarter of their
own resources (Shah and Qureshi, 1994), with their major revenue source
being a local property tax that is actually collected by the central govern-
ment.37 Funds to provinces or localities are intended to promote regional
autonomy and improve local infrastructure, but they may only be spent
subject both to the central government’s general guidelines and to the
requirement that they be spent in specific sectors. Among the stated ratio-
nales for transfers in Indonesia are two that require explicit conditionality
to be effective: (1) to alleviate inefficiencies that arise from interjurisdic-
tional spillovers and (2) the equal provision of minimum standards in ser-
vices.
As one might expect in view of these objectives, most intergovernmental
fiscal transfers in Indonesia allocate funds by formula, but require the
funds to be spent on specified objectives. Grants for health, for example,
are determined on a per capita basis. These grants include allowances for
medicine, standardized expenditures for health centers, estimated expen-
ditures on rehabilitation and renovation of health centers, and the costs of
doctors and paramedics. Some components of the grant take the form of
transfers in kind (for example, medicine). There is a small matching
element in health grants in the sense that local governments must finance
the cost of acquisition of land for health centers. The central government
is directly involved in the provision of health through the appointment of
medical personnel, the provision of medicine and other supplies, and the
construction of health centers.
In contrast to the fiscal federalism model illustrated by the Indonesian
case, what has been called a “federal finance” model seems more appropri-
ate for some countries, particularly those in which there are important geo-
graphic or ethnic differences.38 In such countries, a certain degree of local
autonomy often emerges in practice as a means of achieving and retaining
political stability even if the constitutional structure is formally unitary. In
the traditional world of fiscal federalism (or multi-level finance), as set out
by Oates (1972), in principle everything – boundaries, assignments, the
level and nature of transfers, etc. – is subject to revision by the central
government in its search for efficiency and, perhaps, equity. In the federal
finance model, however, jurisdictional boundaries and the assignment of
functions and finances must generally be considered as fixed at some
earlier (constitutional) stage and not open to further discussion in normal
circumstances. More importantly, where a federal structure is politically
required to hold a country together – contrast Canada to Australia or the
United States (at least in this century) or Switzerland to Germany (Bird,

1986a) – the central concern in designing intergovernmental fiscal relations
is invariably the extent to which they contribute (or not) to political stabil-
ity (Breton, 1996).39
In the fiscal federalism model, the central government’s policy prefer-
ences are clearly dominant. In the federal finance model, however, rather
than simply assuming consensus on, for example, the degree of fiscal and
regulatory harmonization to be attained, or even for that matter the degree
to which an internal common market should be sought, such matters
should ideally be determined jointly by both levels of government in some
appropriate political (constitutional) forum.40 In addition, both levels of
government may properly pursue their own distributive policies (Tresch,
1981), again with no presumption of central dominance. In such a federal
setting, intergovernmental transfers are appropriately often both equal-
izing and unconditional, as, for example, Shah (1994) and Ahmad (1997)
emphasize.
The federal finance setting is in principle one of bargaining between
principals (who are not necessarily equal) – what one Canadian author
(Simeon, 1972) has called, in a useful analogy to the world of international
relations, “federal–provincial diplomacy.” The appropriate analytical
framework is not that of principal–agent relations but rather one of nego-
tiation among equals – in the words of Wheare’s (1969: 10) classical politi-
cal treatment of federalism among federal and state governments that are
“. . . each, within a sphere, co-ordinate and independent.” In some coun-
tries – for example, Canada and Switzerland – this may not be a bad
description of federal–state relations (Dafflon, 1977, 1991; Bird, 1986a).
With caution, a similar framework can clearly be applied to India, as
shown by Rao (this volume) and a similar pattern may be emerging in
Argentina, which has long been a “formal” federation but has only
recently begun to act in what is called here a “federal finance” way (World
Bank, 1996b; Rezk, this volume) as well as, perhaps, South Africa (Ahmad,
this volume). In the extreme case of Bosnia-Herzegovina, the model is in a
sense the reverse, with the subnational governments being the principals
and the central government their agent with respect to relations to the
outside (financial) world (Fox and Wallich, this volume).
In a fiscal federalism setting, however – which seems to be the best way
to characterize cases as diverse as the Colombian, Moroccan, and Chinese
ones considered here – the appropriate analytical framework is clearly a
principal–agent model in which the principal (the central government)
may unilaterally alter both local government revenue and expenditure
responsibilities and intergovernmental fiscal arrangements in its attempt
to overcome the familiar agency problems of information asymmetry and

differing objectives between principal and agent – as indeed China has just
done. In this setting, the case for both equalization transfers and for
unconditional transfers is by no means as clear as is commonly assumed.41
The accepted rationale for equalization transfers in the fiscal federalism lit-
erature is essentially to ensure that comparable bundles of public services
may be provided throughout the country at comparable tax rates
(Boadway and Flatters, 1982). Such arguments may be appealing to many,
but there is clearly no agreement on either the desirability or effects of
equalizing general intergovernmental transfers – as illustrated by the
absence of such transfers in the United States.42
The two “models” of fiscal decentralization sketched above may apply
in developed, developing, and transitional countries. We shall discuss only
the developing countries here. As noted above, with the exception of a few
large countries (such as India and Brazil), the appropriate normative
framework for institutional design in most such countries is almost cer-
tainly the traditional fiscal federalism approach (Bird, 1993). In a few other
countries – for example, Pakistan, Argentina and South Africa – the theory
might be “federal finance” but the practice to date has been essentially
“fiscal federalism.” As the World Bank (1995b) emphasizes, however, the
observed reality of intergovernmental fiscal relations in most developing
countries may perhaps better be characterized as “overcontrolled,”
“undercontrolled,” or “perversely controlled.”
Local governments are frequently overcontrolled in the sense that they
are subject to so many conflicting and unattainable rules that they are, as
it were, prevented from behaving efficiently even if they wanted to do so.
Similarly, localities may be subject to incentives that lead them to behave
perversely. Moreover, the rules may change from place to place and time
to time, depending upon the shifting winds of political favor. Finally,
central capacity to monitor behavior may be so limited that local govern-
ments are in effect subject to no effective controls at all – an approach
which seems equally flawed from the point of view of attaining the objec-
tives of the central government efficiently and effectively. All in all, the
picture in many developing countries is one of ad hoc negotiation between
governments and regions, of considerable instability in intergovernmental
relations, and of substantial difficulty in knowing exactly what is going on.
The situation may be somewhat different in developing transitional
countries such as China and Vietnam. As Bird, Ebel, and Wallich (1995)
emphasize, the task facing would-be designers of intergovernmental fiscal
institutions in such countries is difficult for at least three additional
reasons. First, government is almost invariably being downsized in such
countries. Not only is it difficult to restructure incentives constructively

when budgets are being cut, but it is virtually inevitable that establishing
effective local governments will come very low on the priority list of hard-
pressed central governments. Second, a critical part of the “transition”
process is to develop not just an effective private market economy but also
an effective public sector, to work with and guide such an economy. Again,
it is hard enough to create a public sector in this sense at one level of
government; to do so at more than one level is doubly difficult. Where
political liberalization is part of the transitional process, this effort may
have to be made in any case in order to prevent the development of separ-
atist movements, particularly in the more prosperous parts of the
country.43 But when the political system remains centrally controlled, as in
China, such concerns are again likely to fall to the end of the queue. Finally,
both downsizing and disentanglement require the clear separation of
“business” from “government” – something which has definitely not
occurred in China to any significant extent to date.
Matters may also be quite different in essentially “new” countries. The
studies of South Africa and Bosnia-Herzegovina in this volume are
particularly interesting because they illustrate clearly the essentially polit-
ical rather than economic nature of fiscal decentralization. South Africa is
a country which, so to speak, had to become federal in structure – and
perhaps will soon become federal in reality – as a necessary condition for
its peaceful creation. In contrast, Bosnia-Herzegovina is a country which
has, so to speak, been created in a federal form in the perhaps forlorn hope
that this structure may enable it to exist at all.
These examples illustrate one extreme of a broader question that arises
in many other countries, namely, the difficulty or even impossibility of
assessing the economic aspects of fiscal decentralization in abstraction
from the political goals that may be the primary rationale for decentral-
ization. In many countries, particularly those encompassing territorially
based ethnic minorities, decentralization is inevitably as much about
issues of political power as it is about economic efficiency or equity. Even
when there are no ethnic differences, but some regions are much richer
than others or there is a marked regional divergence in growth patterns,
political strains may make the potential role of intergovernmental finance
as a means of maintaining national harmony critical.44 When the center
loses its grip, to them that have shall more be given, whether what they
have is oil, a credible separation threat, or a strategically important border
area.
A recent analysis of intergovernmental finance in the Russian
Federation, for example, found that transfers have for the most part gone
to those regions least satisfied with the central government – and, indeed,
argued that this bias is probably one important reason why Russia has
(with the minor exception of Chechnya) remained united while
Yugoslavia, Czechoslovakia, and the Soviet Union all collapsed (Treisman,
forthcoming). Students of federal finance in other countries would be sur-
prised to see any other result (Bird, 1986a).45 An important question for
China is whether the political tensions created by the unbalanced nature
of its growth process will require it either to rein back decentralization (and
hence perhaps growth)46 or to mute growth to some extent by redirecting
more of the proceeds to the less favored regions on political grounds – that
is, to move closer to a “federal finance” model by employing regional fiscal
transfers as one means of maintaining political balance among disparate
regions. It is quite unclear, however, whether it would do so in a formal,
transparent grant system47 or, as in the case of Russia (Treisman, forth-
coming), by continuing the discretionary way in which intergovernmental
relations have traditionally been resolved.48 While most dramatically
posed in Bosnia-Herzegovina, this question is also important in countries
as diverse as Pakistan and South Africa.
The role and design of transfers

No matter how revenues and expenditures are reassigned a problem of
vertical imbalance will almost certainly remain, or so worldwide experi-
ence suggests. Even if the tax base of the richest province enabled it to
balance “own” revenues and expenditures, imbalances would remain for
all the rest. Moreover, history and international experience both suggest
strongly that the differential elasticity of expenditures and revenues
assigned to different government levels would in any case soon lead to a
re-emergence of a vertical imbalance problem even for the richest
province. One way or another, structural “gaps” will emerge and will have
to be dealt with.
In principle, there are at least four ways in which any revenue–expendi-
ture gap might be closed.
(1) Revenues could be increased at the provincial level. But this is unlikely
to be either feasible to any great extent or (if the system has been prop-
erly set up in the first place) desirable.
(2) Provincial expenditures could be reduced. While always popular (with
central governments), and perhaps sometimes necessary, this
approach is also unlikely to be advisable if the system has been prop-
erly set up at the outset.
(3) Expenditure functions could be transferred up to the level with more

revenue (or revenue-raising power) or down to the level with more
expenditures – though this is again both unnecessary and unwise if the
basic structure of the system is correct.
(4) Finally, some centrally collected revenues could be transferred to pro-
vincial governments. In the end, in every country, it is this alternative
that almost always prevails.
Intergovernmental fiscal transfers may of course have many objectives
in addition to simply closing the fiscal gap. Transfers may, for example, be
designed to equalize revenue effort, or expenditure levels, or outcomes in
terms of services provided. Such equalization may be desired for purposes
of income redistribution, or to ensure that for the same revenue effort, citi-
zens obtain the same expenditures (or outcomes) regardless of where they
live, or to provide at least minimum standards of key public services to
everyone, or to provide everyone with an equal opportunity to access
public services. Transfers may also be intended to achieve objectives more
directly related to growth and efficiency in resource allocation such as
encouraging local governments to build critical public works or to provide
more services that provide “spillover” benefits to residents of other areas.
They may also be one way in which to recognize that certain areas in a
sense may have some “right” to certain incomes, for example, because the
natural resources that give rise to those incomes are physically located
there. Finally, as emphasized earlier, transfers may have more explicitly
political objectives, such as making it possible for even the poorest areas to
sustain a certain level of public sector activity or increasing the acceptabil-
ity of other policies that may affect certain regions adversely.
The extent of equalization in federal countries varies widely from
country to country, and appears to reflect differences in national history
(preferences) as much or more than any theoretical consideration (Ahmad,
1997). Moreover, there appears to be little, if any, relation between the exis-
tence of formal equalization systems and the extent of regional equaliza-
tion actually effected through the fiscal system (Bird, 1986a). Each country
must find its own solution, in accordance with its own imperatives. What
any country does with respect to regional “rebalancing,” and how it does
it, should be decided in light of the real objectives of policy, not in terms of
someone else’s theory or practice.
Nonetheless, international experience suggests that even general
purpose transfers can probably be expected to deviate in practice from the
rigid lines of an equalization formula, in response to the overriding need
to maintain political stability in the face of rapid economic change. This
need may manifest itself in two ways, both of which may reduce the
amount of equalization attainable through a formal transfer system. First,
to keep the richer regions on side, it may be necessary to direct some trans-
fers to them even if they do not “deserve” them in equity terms.49 Second,
not all the equally poor may be considered to be equally deserving in
political terms, so some may do better than others. In Indonesia, for
instance, the frontier province of Irian Jaya receives more than three times
the average provincial per capita transfer. Frontier regions also receive
special attention from central governments in countries such as Argentina
and Chile. Such developments may be deplored by those who view equity
as the only appropriate objective of fiscal policy. This focus is not only
naive but unduly limited in any real-world country. The design of inter-
governmental transfers is always and everywhere an exercise not solely in
normative economics but also in political economy.50
Transfers thus constitute the heart of subnational finance. In themselves,
transfers are neither good nor bad: what matters are their effects on such
policy outcomes as allocative efficiency, distributional equity, and macro-
economic stability. There are only three basic ways to determine how much
is to be distributed through intergovernmental fiscal transfers:
(1) as a fixed proportion of central government revenues;
(2) on an ad hoc basis, that is, in the same way as any other budgetary
expenditure;
(3) on a “formula-driven” basis, that is, as a proportion of specific local
expenditures to be reimbursed by the central government or in relation
to some general characteristics of the recipient jurisdictions.
Variants of all three methods are found around the world.
Examples of transfer systems that determine transfers as a proportion of
national current revenues may be found in the Philippines, where 20
percent of national internal revenue collections are distributed among local
governments (on the basis of population and land area), and in Colombia,
where approximately 25 percent of (non-earmarked) national current rev-
enues are distributed to the departmental (intermediate-level) governments
(in part in equal portions and in part on the basis of population). Similar
systems operate with respect to most major taxes in some developed coun-
tries: for example in Austria, where local governments receive about 12
percent of income and value-added taxes; and in Japan, where local govern-
ments receive 32 percent of income and alcohol taxes. In both these exam-
ples, the resulting total is distributed in accordance with a formula taking
into account such factors as population and community size. Large federal
countries such as India, Brazil, and Nigeria also tend to use such systems.
Many other countries (including most of the transitional countries) have

“tax-sharing” systems that distribute a fixed share of certain national taxes
– for example, the income tax or the value-added tax – among local govern-
ments. Although many of these systems of “sharing” particular central
taxes attempt to allocate all or part of the total thus determined in accor-
dance with the origin (or “derivation”) of the tax revenues being shared,
others (such as those in Germany, Hungary, and Morocco) allocate the total
set by the shared tax amount in accordance with a formula that attempts
to take into account both needs and capacity. On the whole, despite its
popularity, sharing specific national taxes is less desirable than sharing all
national taxes because it leads central governments over time to tend to
increase those taxes which they do not have to share.
An alternative system is used in Canada with respect to the largest
federal transfer to the provinces (the Canada Health and Social Transfer),
which was initially set in per capita terms to be equal in amount to certain
(matching) transfers it replaced, and has since been increased as a function
of a three-year moving average of nominal GDP growth. Something
similar has recently been suggested for Argentina (World Bank, 1996b),
although it should perhaps be noted that under increasing budgetary pres-
sure, the Canadian federal government has, over time, both weakened the
link to GDP growth (the adjustment factor is now GDP growth less 3
percent) and imposed a “cap” on the absolute amount of transfers going
to the richest provinces. Such measures may relieve federal finances, but
they obviously vitiate to some extent the stability of revenue flow accruing
to the provinces (Smart and Bird, 1997). Another approach might be to
have a “horizontal equalization” transfer, as in Germany and Denmark,
under which, in effect, rich local governments directly transfer resources
to poor localities without directly affecting central revenues (Lotz, 1997).
Among developing countries, only Chile appears to have such a system in
operation (World Bank, 1993), although the new Moroccan law creating
sixteen regional entities provides for an equalization scheme funded in
part by contributions from the richer regions.
Once the total to be distributed in transfers to provinces is determined,
how should it be distributed? Most experts agree that a formula incorpo-
rating both local “needs” for expenditures and local “capacities” to finance
such expenditures out of their own resources should be used for this
purpose. Such a formula, if properly designed, has the additional virtues
of stimulating local governments to make at least an average level of
“effort” to finance the expenditures they carry out and ensuring that they
remain accountable to local residents at the margin.
If horizontal fiscal balance (another name for equalization) is interpreted
in the same gap-filling sense as vertical fiscal balance, an equalization

transfer might seem to imply that sufficient transfers are needed to equal-
ize revenues (including transfers) and the actual expenditures of each local
government. Such “fiscal dentistry” makes no sense, however. Closing all
gaps between actual outlays and actual own-source revenues for all local
governments ignores differences in local preferences for public and private
goods and thus vitiates the basic economic rationale for local government
in the first place. Moreover, this approach to equalization ignores local
differences in needs, in costs, and in own revenue-raising capacity. Finally,
equalizing actual outlays clearly discourages both local revenue-raising
effort and local expenditure restraint, since under this system those with
the highest expenditures and the lowest taxes would get the largest trans-
fers.
For these reasons, in all countries with formal systems of equalization
transfers, the aim is either to equalize the capacity of local governments to
provide a certain level of public services or the actual performance of this
level of service by local governments.51 The performance criterion, which
adjusts the transfer received in accordance with the need for the aided
service (and which may also allow for cost differentials) is in principle
more attractive to central governments – or those concerned primarily
with the provision of certain services such as education or social assistance
– since the level of service to be funded is determined centrally and the
transfer can be made conditional on the provision of that level of service.
Unfortunately, this approach may suffer from the same disincentive effect
on the revenue side as equalizing actual outlays, for unless adjustment is
made for differential fiscal capacity, the government which tries least again
gains most.
In principle, then, any sound design for general fiscal transfers to
provinces should pivot on some notion of revenue capacity. At one extreme,
the aim might be to provide each local government with sufficient funds
(own-source revenues plus transfers) to deliver a (centrally) prede-
termined level of services. Because such capacity-based transfers are
generally based on measures of potential revenue-raising capacity and not
on actual revenues, no disincentive to fiscal effort is created by this
approach. Differentials in the cost of providing services may or may not be
taken into account. Ensuring that the recipient governments will in fact use
the funds they receive as the central government might wish, requires that
receipt be conditioned in some way on performance (and compliance be
monitored in some way).
A number of developing countries use some variety of formula intended
both to equalize public expenditures in localities with differing needs and
capacities, and to stimulate local fiscal efforts, although severe data prob-
lems often constrain the parameters employed in such formulas.52 Brazil
and India, for example, allocate some proportion of transfers in accordance
with per capita income levels in the different states, but few other devel-
oping countries do so owing to data difficulties. Argentina had some
experience with a transfer formula along these lines with the 1973
coparticipation law, which distributed 65 percent in accordance with
population, 10 percent in accordance with the inverse of population
density, and 25 percent in accordance with an index of the “developmental
gap,” which in turn was based on measures of the quality of housing, the
number of vehicles per inhabitant, and the level of education.
Sometimes, cruder adjustments are made simply by reserving a larger
share of the transfer for parts of the country considered to be especially
poor or needy, for example, mountainous regions (as in Switzerland) or
remote regions (in the systems of municipal transfers of some Canadian
provinces), or those with large concentrations of particularly poor groups
(India). Perhaps the most common formula elements found in developing
countries are population and land area, perhaps with some adjustment for
some of the factors just mentioned (for example, remoteness from central
markets) or for the size of the municipality (as in some proposed systems
in Colombia – Bird, 1984).
Transfers intended to finance particular types of service, for example,
road maintenance or education, are often linked to particular measures of
need – or existing capacity – such as length of roads or number of students.
At one extreme, this approach leads to the sort of “norms” found in
Vietnam and a number of other transitional countries (such as Hungary),
and may give rise to problems, including allocating funds on the basis of
installed capacity, which may reflect past political decisions, rather than
need. More careful determination of expenditure needs may have some
role with respect to conditional grants – for example, for basic education53
– but it is not likely to prove useful with respect to grants intended to
finance general local expenditures. Experience in countries such as
Australia and Canada suggests that a high level of reliable disaggregated
data is required before the detailed “norm” approach makes sense. In the
absence of such data, simpler approaches, based on, for example, popula-
tion and a simple “categorization” of localities (by size, by type, or perhaps
by region), seem more likely as a rule to prove useful in measuring general
expenditure needs.
Despite the strong theoretical reasons for doing so, however, few devel-
oping countries appear to include explicit measures of fiscal capacity in
their formulas. Examples of where this is done are India and Nigeria,
where some measure of tax effort is included in the basic distributional

formula to states, and Korea, which assumes that a standard tax rate is
applied by cities and lowers the transfer if the actual rate is lower (Ahmad,
1997). Colombia also has such an element in one of its transfer programs.
Chile goes further and, as noted above, actually “taxes” richer localities to
some extent by reducing their transfers and raising those granted to poorer
localities. Capacity measures of various sorts are more common in trans-
fer formulas in developed countries. In Spain, for example, 25 percent of
local transfers are allocated in accordance with local tax collections (and 70
percent on population). Denmark and Sweden, like Canada and Australia,
explicitly calculate local transfers on the assumption that the average
“national” local tax rate is applied. The result is that those localities that
levy above average local taxes are not penalized by having their grant
reduced while those that levy below average taxes are not rewarded by
having their grant increased. Of course, this approach makes sense only if
local governments have the ability to vary local tax rates. The absence of
significant local autonomy with respect to local taxes combined with data
difficulties probably explains the relatively few examples of transfer pro-
grams incorporating explicit capacity measures in developing countries.
Often, as with the Planning Commission grants in India (Rao, this
volume), transfers vary by type of project and are essentially means of
implementing national planning goals.
Compliance with conditions imposed on transfers can be monitored in
part through requirements for uniform and timely local financial reporting
and through periodic national inspections and audits of local facilities,
although no developing country currently seems to do much on either of
these lines. At the very least, the central government should make every
reasonable effort to improve local financial reporting – for example,
making the provision of such reports a condition for receiving grants – as
well as attempting to improve its information base on what is happening
with respect to the provision of local public services. Unfortunately, few
developing countries appear as yet to have progressed very far along these
lines.
What have we learned about fiscal decentralization?

Two general lessons emerge from the brief consideration of intergovern-
mental arrangements around the world in this chapter and the more
detailed case studies that follow. The first lesson is that since every country
both is unique and in some sense constitutes an organic unity, the signifi-
cance of any particular component of its federal finance system – for
example, the assignment of taxes or the design of intergovernmental trans-

fers – may be understood only in the context of the system as a whole. One
cannot pick an institution from a specific setting, plant it in the alien soil of
another environment, and expect to obtain the same results. Policy
recommendations in the area of intergovernmental finance must be firmly
rooted in understanding the rationale of the existing intergovernmental
system and its capacity for change if they are to be acceptable and, if
accepted, successfully implemented.
What is feasible and desirable in any particular setting depends very
much upon what that setting is, and why it is that way. In the world of
intergovernmental finance, one size does not fit all: simple general pro-
nouncements – for example, that unconditional transfers are better (or
worse) than conditional ones or that income taxes should always be
assigned to central governments – are worse than useless as a guide to
policy: they may be positively dangerous. Thus good policy recommenda-
tions with respect to matters of intergovernmental finance must, in a sense,
be tailor-made for each country.
The second lesson suggested by experience is that in the end what may
matter more than the precise nature of the technical solutions found in the
different countries is the process through which such solutions are reached.
In federal countries, for instance, the continued viability of the component
units – the states – is generally considered an important and explicit objec-
tive of policy. The form that regional policy takes often provides a good
indication of the extent to which the federal system reflects underlying
sociopolitical realities. The greater the value placed, for political or social
reasons, on regional survival, the more emphasis is likely to be placed on
relatively unfettered regional tax powers supplemented by equalization
(as in Switzerland) or large unconditional regional transfers (as in Canada
and India). Such factors, together with the institutional structure of deci-
sion-making, play a larger role in all federations than efficiency concerns
in determining the assignment of taxing powers, the degree of vertical and
horizontal fiscal coordination, and the size and nature of intergovern-
mental fiscal transfers.
Fiscal arrangements invariably constitute perhaps the most important
component of the intergovernmental system in any country. Changes in
such arrangements reflect, and may sometimes also induce, changes in
that system, and the design of such changes is of course a proper subject
for economic analysis. In the end, however, any governmental system is
invariably a political creation with primarily political ends. The fiscal
system existing in any country must therefore be understood and assessed
within a political as well as an economic framework. What matters most in
intergovernmental finance is thus who determines the rules of the game

and how those rules are changed.
Analysis of technical issues is interesting and important, but if such
analysis is to play a meaningful role in the essentially political process of
intergovernmental bargaining, the institutional framework must be one
that accommodates the analysis in the process of achieving sufficient con-
sensus for decision-making purposes in a society which is complex and
divided. Those who would improve the economic outcomes of the inter-
governmental finance system in any country would be well advised to
develop and publicize accurate and relevant information in order to influ-
ence the outcome of the inevitably political processes that shape that
system.
Lessons for policy in any one country may be drawn only with caution
from the welter of confusing details and often contradictory evidence that
constitutes the experience of other countries that have utilized somewhat
similar policies in inevitably rather different circumstances. Nonetheless,
several possibly useful guidelines for decentralization policy seem to
emerge, with varying degrees of support, from the international experi-
ences reviewed in this book. Some obvious potential conflicts between
these guidelines may perhaps be averted by clever policy design and
implementation. But one big conflict cannot be dodged: the greater the
weight that the central government places upon specific policy goals, such
as the equitable delivery of essential services to specific groups of poor citi-
zens, the less the weight that can be given to the autonomy of local govern-
ments in the sense of leaving them free to spend as they see fit. As noted
earlier in this chapter, decentralization, properly carried out, may have
many virtues; but doing exactly what the central government wants is not
likely to be one of them.
Would-be decentralizers should not be in too much of a hurry. Building
viable, efficient, and equitable local governance structures is the work of
decades, not years. Rome was not built in a day; nor is a village water
supply agency or a functioning regional health center, or an efficient
municipal council. Any central government that is serious about decentral-
izing needs not only the will and the resources to do so but also, and criti-
cally, a clear and sustained strategy and an adequate (central) institutional
structure to support the decentralization effort.
Whenever possible, new institutions should be built on existing founda-
tions. Decentralization will work better if it relates more closely to existing
community structures and organizations. “Small” may not always be
beautiful, but it is more likely to reflect and yield what local people really
want. How, and how far to follow this principle in large and diverse urban
areas is one of the central problems facing any decentralization policy; it is

difficult, but not impossible (Bird and Jenkins, 1993).
As emphasized above, what local people want may not be what the
central government wants. The political aspect of decentralization in the
sense of better reflecting local preferences is the key to achieving its effi-
ciency advantages, but to the extent local policy choices are driven by local
demand they are less likely to reflect central policy preferences.
Decentralization policy must therefore set out clearly in which, if any, areas
local differences are not to be permitted to develop – for example, in pro-
viding services to the poor – and then to develop institutions that will
achieve the desired results.
Local autonomy should not include the right to spend “other people’s
money” freely. To ensure accountability at the margin (which is where it
matters) in principle, local spending decisions should relate to locally
raised (and locally borne, not exported) revenues. However, since under
almost any conceivable assignment of expenditure and revenue
responsibilities some localities will inevitably be dependent upon fiscal
transfers – and under some assignments, all will be – the appropriate
design and monitoring of transfers is invariably a critical element in
effective decentralization policy. What design is best depends upon
whether the central government views local governments as its agents in
implementing national policy (in which case transfers should, for
example, be conditioned on the delivery of basic health and education
services) or as independent actors who can do what they (local citizens)
want.
Money should follow functions, not precede them. Perhaps the worst of
all decentralization policies is to dump a lot of money on an ill-prepared
(and often ill-conceived) local government structure without first consid-
ering carefully:
• the appropriate expenditure tasks for local government;
• what interest the central government has in how well, or how ill, these
tasks are carried out;
• if it has such an interest, how it can best be achieved in the new system
of intergovernmental finance.
Good transfer design can assist in implementing a well-thought-out
policy in this regard, but it cannot replace it.
Many local governments may need help to do their jobs properly.
Local governments can and should carry out a variety of local public ser-
vices (such as infrastructure development) with little central control or
direction. If they are to do so effectively, however, careful attention has
to be paid to three critical and too often neglected aspects of decentral-

ization:
• developing a viable staffing policy – including the often troublesome
problem of incorporating officials from other levels of government
without unduly inflating wage bills or unduly lowering service quality;
• implementing a sound information system for accounting and financial
reporting – to ensure accountability to local residents and to enable the
central government to monitor and evaluate the decentralization
process;
• providing adequate technical support – both for the previous two points
and for such matters as project development and contracting and pro-
curement.54
The focus should be on local services, not local governments. Local
governments need not – indeed, often should not – themselves produce
most local services. Instead, particularly in smaller localities, private and
other providers should be used as extensively as possible to provide local
public services efficiently. The path to the decentralized twenty-first
century for most developing countries seems unlikely to be via the
twentieth-century “big government” model, but via more eclectic and
imaginative modes of service provision.55 To make such “decentralized”
decentralization work, however, will require very close attention by the
centre to issues of technical assistance and support. Such matters are not
minor “add-ons” to the big decisions on, for example, tax and expenditure
assignment. Developing an adequate institutional framework for decen-
tralization is essential for success no matter what the big decisions may be.
Above all, intergovernmental finance should not be unnecessarily com-
plicated. In all countries, fiscal relations between governments are
inevitably complex and usually unsatisfactory in some respects to all
parties concerned. If what the central government wants to do is accom-
plish a specific goal such as delivering specific services to specific (poor)
households, it should, whenever possible, do so without further compli-
cating intergovernmental finance. Many of the complications, and com-
plaints, characterizing intergovernmental fiscal issues in most countries
result from overloading the system with tasks for which it is ill-equipped,
such as targeted poverty alleviation. Whenever feasible, direct transfers to
the poor are better than indirect transfers to localities – even poor local-
ities – that are intended primarily to help poor households. In short, if the
provision of a nationwide basic uniform level of health or educational ser-
vices is an important objective of national policy, the national government
should either provide such services itself or directly (such as through
some variant of vouchers) transfer the needed resources to the target

population.
When local governments are expected to play a major role in delivering
social services, they must inevitably depend in large part upon central
fiscal transfers for the funds needed to do so. Two quite different
approaches may be taken to the design of such transfers. On one hand, to
the extent that their primary objective is to ensure that all regions of the
country that are expected to provide such services at acceptable minimum
standards have adequate resources to do so, simple “lump-sum” transfers,
with no conditions other than the usual requirements for financial audit-
ing, are called for (Shah, 1994). In this “federalist” approach, it is essen-
tially assumed that the fact that the funds flow to locally responsible
political bodies will ensure sufficient accountability and that it is neither
necessary nor desirable for the central government to attempt to interfere
with, or influence, local expenditure choices. On the other hand, if the
central government is, in effect, using local governments as agents in exe-
cuting national policies – for example, to provide primary education at a
specified level throughout the country – then it would seem to make sense
to make the transfer conditional upon the funds actually being spent on
education or on the attainment of some level of educational performance
(Bird, 1993).
If key services are to be provided through decentralized governments,
careful attention has to be paid to:
• getting the prices facing service providers right (for example, by means
of a well-designed system of matching grants);
• setting up an information and inspection system sufficient to ensure that
the desired services are in fact delivered to the target groups;
• devising some system (such as a nationally organized “fail-safe” provi-
sion) for dealing with the non-compliant without punishing the inno-
cent.
None of these tasks is easy. None is costless. And the results are not likely
to be very satisfactory. The moral is simply that if it is really important to
the central government that something should be done in a particular way,
it should, if at all possible, do it itself.
Decentralization is often, but not always, a good policy, but it must be
done correctly. Decentralization is not a panacea; it cannot and will not solve
all problems. Indeed, it may sometimes create new ones (or at least make
them more obvious), and, inappropriately implemented, it may sometimes
make matters worse (for example, for the poor). However, decentralization
has many virtues – not least letting people take more control over their
own lives, better satisfying local preferences, and, often, lowering the costs
of service provision. Careful planning, carefully monitored and supported
implementation, and, above all, careful and continuous efforts both to
involve local people in the process and to make them face the opportunity
costs of their decisions are essential to success, which even in the best condi-
tions is unlikely to come quickly or easily. As several of the countries dis-
cussed in this volume have demonstrated, however, these obstacles can be
overcome and at least some of the virtues of decentralization realized
without undue sacrifice of either efficiency or equity – indeed, on the con-
trary, with improvements in both dimensions. A good decentralization
policy is not easy to design, and is even harder to implement. But it can be
done, and it is worth doing.
Notes
1 For other recent collections of papers on this topic, see Prud’homme (1991) and
Owens and Panella (1991).
2 See Bahl and Linn (1992), Shah (1994), Ahmad (1997), and CEPAL/GTZ (1996)
for numerous examples.
3 For empirical examples from a number of countries, see Campbell, Peterson,
and Brakarz (1991).
4 As Bahl and Linn (1994: 5) say,
the situation in a developing country that could provide maximum gains
from a more decentralized local government structure would include: (1)
enough skilled labor, access to materials, and capital plant to expand
public service delivery when desired; (2) an efficient tax administration;
(3) a taxing power able to capture significant portions of community
income incrementally; (4) an income-elastic demand for public services;
(5) popularly elected local officials; and (6) some local discretion in
shaping the budget and setting the tax rate.
They might have added: (7) a tradition of local decision-making.
5 Of course, fiscal decentralization is only one aspect, and not necessarily the
most meaningful one, of decentralization, and measuring the degree of even
fiscal decentralization, however defined, is itself a complex and often treacher-
ous exercise (Bird, 1986b), but neither of these points can be further discussed
here.
6 Deconcentration may take the form of “field administration,” under which
local administrative units (or branches of central ministries) have responsibil-
ity for service delivery but the staff remains under the direction and control of
the center, or “local administration,” under which local units have more
responsibility for policy and program implementation but are still under the
technical supervision of central ministries. For further discussion of this and
other forms of decentralization, see World Bank (1996a).
7 With delegation, local governments basically have the responsibility for the
function, but remain accountable to the central government with respect to how
well they perform it. With devolution, local governments are fundamentally
accountable to local residents.
8 Wallich (1994) suggests that this motive has been important in Russia, for
example.
9 See also Qiang (1995) who explicitly cites these two rationales for reform.
10 In particular, Prud’homme (1995) at times seems to suggest that all versions of
decentralization are necessarily defective. In reality, viewed from any reason-
able perspective there may be “good” or “bad” decentralization in any country:
it depends upon what is done, and how. See McLure (1995) and Sewell (1996)
for persuasive critiques of Prud’homme’s extreme position. A more balanced,
though still basically negative, approach may be found in Tanzi (1996).
11 See World Bank (1990); for a more skeptical view of the extent to which decen-
tralization per se was the destabilizing villain even in the Argentine case, see
World Bank (1996b).
12 In countries such as China, Russia, and Vietnam in which most central govern-
ment revenues have traditionally been collected at the local level, there is
perhaps more danger of discouraging collection effort by equalizing transfers,
which from the perspective of local governments amount to implicit pro-
gressive taxes (Yu, 1996). Ma (1996), for example, argues that the constantly
changing nature of revenue-sharing arrangements in China – in effect, the lack
of pre-commitment – has discouraged local collection efforts.
13 But see Bird and Fiszbein (this volume) on Colombia and World Bank (1996b)
on Argentina for counter-arguments.
14 As Lewis (1966) said long ago, however, giving more responsibility to local
governments can often utilize previously unused entrepreneurial and mana-
gerial resources and hence strengthen rather than weaken the overall level of
public administration in a country. As noted in the chapters on Colombia (see
also World Bank, 1995a) and South Africa in this volume, there appears to be
some merit in this argument.
15 For a balanced assessment of this issue, see Ter-Minassian (1996). A useful theo-
retical framework for considering the interaction of local accountability and
access to capital markets is presented in Wildasin (1997b).
16 For an argument along these lines, see Brean and Bird (1986).
17 Given the impossibility of bankruptcy proceedings against local governments,
reducing this moral hazard will have to depend upon the institution of a cred-
ible review/control system for debt work-outs. See the chapter on South Africa
(Ahmad, this volume) for further discussion of how to make the “no bail-out”
option credible; also World Bank (1996b) for a similar proposal for Argentina.
18 This is the nub of the analysis in Eichengreen and Von Hagen (1995): there is an
inverse correlation between local taxing powers and central controls over local
borrowing.
19 Some arrangements may have to be made for smoothing out cash flows, but this
should not be a concern in most muncipalities which are, and will remain, trans-
fer-dependent, that is, will largely get their funds in monthly cheques from the
central government.
20 Similar innovative local methods of dealing with roads in Indonesia are
described in Bird (1995).
21 For a more precise discussion of these concepts see the chapters in this volume
on India and Argentina.
22 Although this point cannot be further discussed here (but see the chapter on
South Africa), there should also be no “bail-outs” of improvident local author-
ities (World Bank, 1996b).
23 No one has ever argued, for instance, that the province of Prince Edward Island
in Canada, which is 80 percent dependent on federal transfers, is less account-
able to its citizens than the province of Ontario, which is less than 20 percent
transfer-dependent.
24 Two exceptions to this rule are when transfers are explicitly intended either to
ensure the provision of specific services at specific levels (that is, the transfers
are essentially payments to provinces acting as agents) or to induce provinces
to provide more of certain services than they would otherwise do.
25 Note, however, that this approach may not be entirely compatible with the idea
of using capital markets as an independent source of “fiscal discipline” or
accountability.
26 See Bird and Chen (1996) for the argument that in most such countries the
system may more properly be described as one of “federal finance” than “fiscal
federalism.”
27 These problems are discussed extensively with respect to the former socialist
countries of Europe in Bird, Ebel, and Wallich (1995).
28 Compare, for example, the very different intergovernmental systems in neigh-
boring countries that are similar in many respects – for example, Canada and
the United States, Switzerland and Germany, India and Pakistan, or Argentina
and Brazil (all of which countries have formally federal structures).
29 This argument is developed in Bird (1986a, 1994b).
30 These figures exclude the important regional-level governments in some
federal countries.
31 Those who have not looked into the question seldom believe how difficult it is
to obtain such figures in most countries. Take the case of Brazil, for example.
Table 1.2 presents data based largely on the IMF’s Government Finance Statistics.
In contrast, although Afonso and Lobo (1996) indicate that the size of the public
sector in 1995 is roughly comparable (35 percent) to that shown in table 1.2, their
breakdown between the central and subnational sectors is completely different,
with the latter accounting for about 44 percent of the total with respect to both
revenues and expenditures. Without further detailed examination of the under-
lying data used by Afonso and Lobo (1996), it is not possible to reconcile these
figures for Brazil with those in table 1.2. Similar stories could be told about a
number of the other countries included in these tables.
32 Contrast, for example, Zhu (1995) with Litvack (1994).
33 The case of Canada is particularly interesting in the sense that it is both a
“natural” (linguistic) federation in the case of Quebec and a “non-natural”
federation like Australia with respect to most of the rest of the country, which
is perhaps one reason Canadian federation is uneasy these days.
34 The mechanism used for the interregional allocation in both (pre-1994) China
and Vietnam is (1) to allow the provinces that collect the revenues a (variable)
share of what they collect and (2) to provide, in addition, a variety of explicit
central government grants for specific purposes (see the chapter on China in
this volume and World Bank, 1996a).
35 This is broadly the normative perspective taken in Bird (1993).
36 Bird (1993) argues that this is the appropriate framework in most developing
countries. It is also that employed by Ma (1996) to analyze the Chinese case.
37 For further discussion of the Indonesian case, see Shah (this volume), Bagchi
(1995), and Qureshi (1997).
38 Such a model is sketched in Bird (1994b) and Bird and Chen (1996).
39 Both models may apply within one country at different levels. In Canada, for
example, Bird and Chen (1996) suggest that the federal finance model charac-
terizes federal–provincial fiscal relations and the fiscal federalism model char-
acterizes provincial–municipal fiscal relations.
40 Contrast Ahmad (1997: 4), which simply assumes that a unified market is a sine
qua non of any federal system, with Bird (1989), which argues that Canada is in
some respects less unified than the European Union – and probably correctly
so.
41 See, for example, the various studies in Ahmad (1997).
42 On this, see the recent debate between Oakland (1994) and Ladd and Yinger
(1994).
43 In most countries of eastern and central Europe, the fall of communist regimes
brought with it the rejection of all central-planning institutions, including the
previously centralized control of local budgets. If such countries were to remain
functioning wholes, they had to develop more autonomous and effective sub-
national governments (Bird, Ebel, and Wallich, 1995). All have found this task
difficult in the face of all the other changes taking place, and some have found
it impossible – the former Yugoslavia, the former Czechoslovakia, and the
former Soviet Union – and perhaps in future also Bosnia-Herzegovina?
44 Consider, for instance, the recent discussion about adopting a more federal
structure in Italy in response to the secessionist movement (“Padania”) in the
north; or, again, the extensive and expanding form of federalism being imple-
mented in Spain, particularly in Catalonia and the Basque Country (Vizcaya);
while both Spanish cases have ethnic components, Catalan separatism has a
clear economic motive as well.
45 For a theoretical rationale along related lines, see Breton (1996: 9).
46 For divergent views on the direction of the relationship between decentraliza-
tion and growth in China, see Zhang and Zou (1996) and Qiang (1995).
47 As has been urged by many authors, for example, Lou (1997).
48 See, for example, Qiang (1995).
49 Bird (1993) calls this the “social security” argument, by analogy with the rela-
tively greater success achieved in many countries in delivering income to the
poor through a general transfer system such as social security as opposed to a
more targeted scheme of poverty assistance.
50 See, for example, the discussion in Bird (1986a) of the constant revision, largely
on political grounds, of equalization schemes, even in such strongly federal
countries as Canada and Switzerland.
51 The original formulation of many of these arguments was done by Musgrave
(1961). For a recent full treatment of most of these design issues, see Ahmad
(1997).
52 Most of what follows is based in part on Shah (1994) and Ahmad (1997).
53 A strong case might perhaps be made for tying equalization transfers to the pro-
vision of particularly important public services, for example, education and
perhaps basic health care, as is argued in the chapter on Colombia in this
volume.
54 Note that this does not mean this support must necessarily, or even desirably,
come from the central government.
55 For examples, see Roth (1987) and Jenkins and Sisk (1993).
References
Afonso, José Roberto Rodrigues, and Lobo, Thereza (1996), “Fiscal decentralization
and participation in delayed democratic experiences,” paper for Tinker Forum
on the Role of the State in Latin America and the Caribbean, Cancún, Mexico
(October).
Ahmad, Ehtisham, ed. (1997), Financing Decentralized Expenditures: An International
Comparison of Grants, Cheltenham: Edward Elgar.
Ahmad, Ehtisham, and Thomas, Ravi (1997), “Types of transfers – a general frame-
work,” in Ahmad (1997).
Ahmad, Ehtisham, Qiang, Gao, and Tanzi, Vito, eds. (1995), Reforming China’s
Public Finances, Washington, DC: International Monetary Fund.
Bagchi, Amaresh (1995). “Intergovernmental fiscal relations: the cases of India and
Indonesia,” in Jayanta Roy, ed., Macroeconomic Management and Fiscal
Decentralization, Washington, DC: World Bank.
Bahl, Roy, and Linn, Johannes (1992), Urban Public Finance in Developing Countries,
New York: Oxford University Press.
(1994), “Fiscal decentralization and intergovernmental transfers in less devel-
oped countries,” Publius: The Journal of Federalism, 24: 1–19.
Bennett, Robert J., ed. (1990), Decentralisation, Local Governments and Markets,
Oxford: Clarendon Press.
Bird, Richard M. (1980), Central–Local Fiscal Relations and the Provision of Urban
Public Services, Canberra: Centre for Research on Federal Financial Relations,
Australian National University.
(1984), Intergovernmental Finance in Colombia, Cambridge, MA: Harvard Law
School International Tax Program.
(1986a), Federal Finance in Comparative Perspective, Toronto: Canadian Tax
Foundation.
(1986b), “On measuring fiscal centralization and fiscal balance in federal states,”
Environment and Planning C: Government and Policy, 4: 389–404.
(1989), “Tax harmonization in federations and common markets,” in Manfred
Neumann and Karl W. Roskamp, eds., Public Finance and Performance of
Enterprises, Detroit: Wayne State University Press.
(1990), “Intergovernmental finance and local taxation in developing countries:
some basic considerations for reformers,” Public Administration and
Development, 10: 277–88.
(1993), “Threading the fiscal labyrinth: some issues in fiscal decentralization,”
National Tax Journal, 46: 207–27.
(1994a), “Decentralizing infrastructure: for good or for ill?” Policy Research
Working Paper 1258, World Bank, Washington, DC (February).
(1994b), “A comparative perspective on federal finance,” in K. G. Banting, D. M.
Brown, and T. J. Courchene, eds., The Future of Fiscal Federalism, Kingston, Ont.:
Queen’s University School of Public Policy.
(1995), Financing Local Services: Patterns, Problems, and Possibilities, Toronto:
Centre for Urban and Community Studies, University of Toronto.
Bird, Richard M. and Chen, Duanjie (1996), “Federal finance and fiscal federalism:
the two worlds of Canadian public finance,” Discussion Paper No. 6,
International Centre for Tax Studies, University of Toronto (July).
(forthcoming) “Intergovernmental finance in China in international per-
spective,” in Donald J. S. Brean, ed., Taxation in Modern China.
Bird, Richard M. and Gendron, Pierre-Pascal (1998), “Dual VATs and cross-border
trade: two problems, one solution?,” International Tax and Public Finance, 5:
429–42.
Bird, Richard M. and Jenkins, Jerry (1993), “Expanding consent in the finance and
delivery of urban services,” in Jenkins and Sisk (1993).
Bird, Richard M. and Slack, Enid (1991), “Financing local governments in OECD
countries: the role of local taxes and user charges,” in Owens and Panella (1991).
Bird, Richard M., Ebel, Robert, and Wallich, Christine, eds. (1995), Decentralization
of the Socialist State: Intergovernmental Finance in Transition Economies,
Washington, DC: World Bank.
Bird, Richard M., Litvack, Jennie, and Rao, M. Govinda, (1995) “Intergovernmental
fiscal relations and poverty alleviation in Vietnam,” Policy Research Working
Paper 1430, World Bank, Washington, DC (March).
Boadway, Robin W. and Flatters, Frank (1982), “Efficiency and equalization pay-
ments in a federal system of government: a synthesis and extension of recent
results,” Canadian Journal of Economics, 15: 613–33.
Boadway, Robin W., Roberts, Sandra, and Shah, Anwar (1993), “The reform of fiscal
systems in developing countries: a federalism perspective,” paper for
Conference on Fiscal Reform and Structural Change, International
Development Research Centre, New Delhi (August).
Brean, Donald J. S. and Bird, Richard M. (1986), “Fiscal risk of state-owned enter-
prise,” in Bernard F. Herber, ed., Public Finance and Public Debt, Detroit: Wayne
State University Press.
Breton, Albert (1996), Competitive Governments: An Economic Theory of Politics and
Public Finance, Cambridge: Cambridge University Press.
Bruce, Neil (1995), “A fiscal federalism analysis of debt policies by sovereign
regional governments,” Canadian Journal of Economics, 28: S195–S206.
Campbell, Tim, Peterson, George, and Brakarz, J. (1991), Decentralization to Local
Government in LAC, Washington DC: World Bank.
CEPAL/GTZ (1996), Descentralización Fiscal en América Latina: Balance, Principales,
Desafíos, Proyecto Regional de Descentralización Fiscal CEPAL/GTZ,
Santiago, Chile.
Dafflon, Bernard (1977), Federal Finance in Theory and Practice with Special Reference
to Switzerland, Bern: Verlag Paul Haupt.
(1991), “Assigning taxes in a federal context: the experience of Switzerland,”
paper for seminar on Fiscal Federalism in Economies in Transition, OECD,
Paris (April).
Eichengreen, Barry, and von Hagen, Jurgen (1995), “Fiscal policy and monetary
union: federalism, fiscal restrictions and the no-bailout rule,” Working Paper
C95–056, Center for International and Development Economics, University of
California, Berkeley (September).
Glaessner, Philip, et al. (1994), “Poverty alleviation and social development funds:
the Latin American experience,” Discussion Paper 261, World Bank,
Washington, DC.
IMF (International Monetary Fund) 1993, Government Finance Statistics 1993,
Washington DC.
Jenkins, Jerry, and Sisk, David E., eds. (1993), Development by Consent: The Voluntary
Supply of Public Goods and Services, San Francisco: ICS Press.
Ladd, Helen, and Yinger, John (1994), “The case for equalizing aid,” National Tax
Journal, 47: 211–24.
Lewis, W. Arthur (1966), Development Planning, London: George Allen & Unwin.
Litvack, Jennie I. (1994), “Regional demands and fiscal federalism,” in Wallich
(1994).
Lotz, Jorgen (1997), “Denmark and other Scandinavian countries: equalization and
grants,” in Ahmad (1997).
Lou, Jiwei, (1997), “Constraints in reforming the transfer system in China,” in
Ahmad (1997).
Ma, Jun (1996), Intergovernmental Relations and Macroeconomic Management in China,
New York: St. Martin’s Press.
McLure, Charles E., Jr. (1993a), “The Brazilian tax assignment problems: ends,
means, and constraints,” in A Reforma Fiscal no Brasil, São Paulo: Fundação
Instituto de Pesquisas Econõmicas.
(1993b), “A North American view of vertical fiscal imbalance and the assignment
of taxing powers,” in D. J. Collins, ed., Vertical Fiscal Imbalance and the Allocation
of Taxing Powers, Sydney: Australian Tax Research Foundation.
(1995), “Comment on Prud’homme,” World Bank Research Observer, 10: 221–26.
(1997) “Topics in the theory of revenue assignment: gaps, traps,and nuances,” in
Mario I. Blejer and Teresa Ter-Minassian, eds., Macroeconomics and Public
Finance, London: Routledge.
Musgrave, Richard A. (1961), “Approaches to a fiscal theory of political federal-
ism,” in Public Finances: Needs, Sources, and Utilization (National Bureau of
Economic Research Special Conference Series), Princeton: Princeton
University Press.
(1983), “Who should tax, where, and what?” in Charles E. McLure, Jr., ed., Tax
Assignment in Federal Countries, Canberra: Australian National University
Press.
Norregaard, John (1995), “Intergovernmental fiscal relations,” in P. Shome, ed., Tax
Policy Handbook, Washington, DC: International Monetary Fund.
Oakland, William (1994), “Fiscal equalization: an empty box?,” National Tax Journal,
47: 199–209.
Oates, Wallace E. (1972), Fiscal Federalism, New York: Harcourt, Brace, Jovanovich.
OECD (Organisation for Economic Co-operation and Development) (1991), “The
role of intermediate and local levels of government: the experience of selected
OECD countries,” background document for Seminar on Fiscal Federalism in
Economies in Transition, Paris.
Owens, J. and Panella G., eds. (1991), Local Government: An International Perspective,
Amsterdam: North-Holland.
Prud’homme, Remy, ed. (1991), Public Finance with several Levels of Government, The
Hague/Königstern: Foundation of Public Finance Journal.
(1995) “The dangers of decentralization,” World Bank Research Observer, 10:
201–20.
Qiang, Gao (1995), “Problems in Chinese intragovernmental fiscal relations, tax-
sharing system, and future reform,” in Ahmad, Qiang, and Tanzi (1995).
Qureshi, Zia (1997), “Fiscal transfers in Indonesia,” in Ahmad (1997).
Roth, Gabriel (1987), The Private Provision of Public Services in Developing Countries,
Sewell, David (1996), “The dangers of decentralization according to Prud’homme:
some further aspects,” World Bank Research Observer, 11: 143–50.
Shah, Anwar (1994), The Reform of Intergovernmental Fiscal Relations in Developing
Countries and Emerging Market Economies (Policy and Research Series No. 23),
Shah, Anwar, and Qureshi, Zia (1994), “Intergovernment fiscal relations in
Indonesia,” Discussion Paper No. 239, World Bank, Washington, DC.
Simeon, Richard (1972), Federal–Provincial Diplomacy, Toronto: University of
Toronto Press.
Smart, Michael, and Bird, Richard (1997), “Federal fiscal arrangements in Canada:
an analysis of incentives,” in National Tax Association, Proceedings of the
Annual Tax Conference, Washington, DC.
Tanzi, Vito (1996), “Fiscal federalism and decentralization: a review of some effi-
ciency and macroeconomic aspects,” in Michael Bruno and Boris Pleskovic,
eds., Annual World Bank Conference on Development Economics 1995,
Ter-Minassian, Teresa (1996), “Borrowing by subnational governments: issues and
selected international experiences,” IMF Papers on Policy Analysis and
Assessment PPAA/96/4, International Monetary Fund, Washington, DC
(April).
Tiebout, Charles M. (1956), “A pure theory of local government expenditure,”
Journal of Political Economy, 64: 416–24.
Treisman, Daniel (forthcoming) After the Deluge: The Politics of Regional Crisis in
Post-Soviet Russia, Ann Arbor, MI: University of Michigan Press.
Tresch, Richard (1981), Public Finance, Plano, TX: Business Publications.
UNDP (United Nations Development Program) (1993), Human Development Report
1993, New York: Oxford University Press.
Wallich, Christine, (ed.) (1994), Russia and the Challenge of Fiscal Federalism,
Wasylenko, Michael (1987), “Fiscal decentralization and economic development,”
Public Budgeting and Finance, 7: 57–71
Wheare, Kenneth (1969), Federal Government, 4th edn., Oxford: Clarendon Press.
Wildasin, David, ed. (1997a), Fiscal Aspects of Evolving Federations, Cambridge:
Cambridge University Press.
(1997b) “Externalities and bailouts: hard and soft budget constraints in inter-
governmental fiscal relations,” paper presented to IIPF Congress, Kyoto.
World Bank (1990), Argentina: Provincial Government Finances, Washington, DC.
(1993), Subnational Finance in Chile, Washington, DC.
(1994), World Development Report 1994, New York: Oxford University Press.
(1995a), Colombia: Local Government Capacity – Beyond Technical Assistance,
Washington, DC.
(1995b), Better Urban Services, Washington, DC.
(1996a), “Vietnam: fiscal decentralization and the delivery of rural services,”
Report No. 15745-VN, Washington, DC (October).
(1996b), “Argentina: Provincial Finances Study,” Report No. 15487-AR,
Washington, DC (July).
(1996c), “Colombia: reforming the decentralization law,” Report No. 15298-CO,
Washington, DC.
(1996d), World Development Report 1996, New York: Oxford University Press.
Yu, Xiaoping (1996), “Intergovernmental transfer system in China,” paper for
workshop on International Experience on Intergovernmental Fiscal Transfer,
Hanoi, Vietnam (March).
Zhang, Tao, and Zou, Heng-fu (1996), “Fiscal decentralization, public spending,
and economic growth in China,” Policy Research Department, World Bank.
Zhu, M. Zhenjun (1995), “China’s financial policies for minority nationalities and
poor areas,” in Ahmad, Qiang, and Tanzi (1995).
2
China: evaluating the impact of

intergovernmental fiscal reform
ROY W. BAHL
In China’s fiscal system, the distinction blurs as among tax policy reform,
tax administration reform, and intergovernmental fiscal reform. So linked
are they that changes in any one of the three legs of the public financing
structure will automatically affect the other two:
• all tax rates and bases are centrally determined, so structural changes
have direct impacts on local as well as central government revenues;
• revenues from these centrally determined taxes are shared between the
central and local governments, partly on a derivation basis with the
retention rates varying by tax, and partly in the form of an ad hoc grant;
• the tax administration system is decentralized enough that local govern-
ments have derived some fiscal autonomy from a creative implementa-
tion of the tax assessment and collection system.
The comprehensive fiscal reform in China in 1994 altered the structure
of the major taxes, changed the responsibilities for tax administration, and
modified the revenue-sharing arrangements. It also curtailed the ability of
local governments to use “back-door” approaches to revenue mobiliza-
tion. As a result, the entire intergovernmental system was redefined. This
chapter is an attempt to evaluate the impact of these changes.1
In the next sections, the Chinese intergovernmental fiscal system before
and after the 1994 reform is described and the underlying revenue con-
cerns that drove the reform are considered. Most of the remainder of the
chapter is an evaluation of the impacts of the 1994 reform on the inter-
governmental fiscal system.2 A concluding section presents a summary
comparison of Chinese fiscal decentralization with the world practice.
Revenue structure
The Chinese tax structure is highly centralized, with the rates and bases of
all taxes in the system set by the central government. However, the enter-
prise income tax and sales taxes together account for about 90 percent of
49
50 Roy W. Bahl
Table 2.1. China: distribution of budgetary

revenues, 1992 and 1995 (percentages)
1992 1995
Industrial and commercial taxesa 68.1 76.0

Tariffs 6.5 4.8
SOE income tax 19.0 12.6
Collective income tax 2.9 2.0
Other 3.6 4.6
Note: a Primarily sales taxes.

Source: Statistical Yearbook of China, 1996.
total budgetary revenue (table 2.1). The other thirteen taxes in the system
give considerably less than 10 percent of total revenues, with the balance
coming from miscellaneous non-tax revenues.3 With the enterprise income
tax in decline, and the individual income tax very narrowly based, China
has moved to heavier reliance on sales taxes. By 1992, indirect taxes
accounted for more than 70 percent of total collections.
Enterprise income tax

Before the 1994 reform, income tax on enterprises comprised a family of
taxes. There were separate tax treatments for large state-owned enter-
prises, small state-owned companies, collectives, individual household
enterprises, joint ventures, and foreign enterprises. There were five differ-
ent rate structures, and differences in the definition of the tax bases for the
several surtaxes applied. Taxable profit was defined as the difference
between gross sales and allowable costs, but there were some notable
departures from the conventional practice; for example, repayment of loan
principal was an allowable deduction, the depreciation deduction was an
actual cash expense, and bonuses and some fringe benefit payments were
not allowable labor cost deductions.
A combination of this complex tax structure, the uneven quality of
record-keeping by enterprises, and the manually operated collection,
assessment, and audit system made this an especially difficult tax to
administer. To complicate matters further, most state-owned enterprises
did not pay according to the legal rate schedule, but according to a con-
tractual arrangement with the local government.4 This “contract,” agreed
to by the local government administration and the enterprise (but not by
the central government), usually required the enterprise to pay a quota
amount of tax or to guarantee a quota level of taxable profit. In effect, con-
China 51
tracting was a step backward to the time when enterprises were not taxed
but paid their profits to the government sector.
Not only did these practices affect the efficiency with which the
economy operated, but also they had built-in biases against some
provinces and in favor of others, depending on the structure of their
economy. Those provinces with profitable enterprises of course did better,
and those with a greater concentration of state-owned enterprises were
favored because the tax system did not fully reach the collective and
private enterprise sectors. Moreover, provinces with more profitable, state-
owned enterprises were in a position to use preferential tax treatment in
effect to transfer enterprise profits from (shared) general taxes to local
extrabudgetary accounts.
The 1994 reform dramatically changed the enterprise income tax and the
intergovernmental fiscal system. The surtaxes were eliminated, all domes-
tic enterprises were brought under the same regime regardless of owner-
ship, and there was a reduction in the top rate from 55 percent to 33
percent. Tax-contracting between enterprises and government was pro-
hibited, and the system of deductions was restructured to move closer to
international practice. Central–local sharing of enterprise income tax rev-
enues was dropped in favor of assignment of all enterprise income tax rev-
enues to provincial governments.5 The new system was meant to be
revenue-neutral in total, but it is not revenue-neutral for every province
individually.
Indirect taxes
Before the 1994 reform, there were three sales taxes in China: product tax,
business tax and value-added tax (VAT).6 The product tax was levied on
the total sales value of specified manufactured and imported goods. The
base of the tax was the price paid by the buyer, with rates ranging from 3
percent to 60 percent. Over the last few years, the VAT has gradually
replaced the product tax as the standard levy on manufactured and
imported goods. The pre-reform VAT had twelve different rates. The base
of the Chinese VAT is similar to international practice in that exports are
zero-rated, and exemptions include agriculture and most services. But it is
different in that exemptions include the construction, transport,
communications, wholesale, and retail sectors. Credit is not allowed for tax
paid on capital inputs. The method of calculating the base is a major depar-
ture from international practice. Taxes paid on intermediate goods were
credited, but, until recently, according to presumptive methods rather than
invoices. The service sector was taxed through a turnover tax referred to
52 Roy W. Bahl
as the business tax. It was levied on gross receipts, with rates ranging from
3 to 10 percent.
The sales tax was overdue for reform. It was terribly complicated and
costly to administer. Earlier the sales tax had been used as a device to
regulate profits and prices, and by the late 1980s, there were over 250
different sales tax rates. But, with most prices decontrolled, this was no
longer necessary and created significant distortions in the economy
(World Bank, 1990). Evasion seemed to be a problem, and the revenue
income elasticity was less than unity. Finally, the structure and
administration of the VAT departed too much from good practice and did
not give China the revenue productivity and self-policing advantages that
it has given other countries.
The structure of the pre-reform indirect tax system also raises issues of
intergovernmental fiscal relations, because some provinces fared better
than others under the taxing and revenue-sharing rules that governed the
indirect taxes. First, the indirect taxes were part of the sharing pool, so a
larger sales tax base meant greater revenues under China’s derivation-
based sharing system. The VAT was a counter-equalizing form of
revenue-sharing. Second, some of the structural features of the indirect
taxes had implications for the relative fiscal capacity of provinces: the
failure to credit capital expenses both raised more revenues from capital-
intensive enterprises and discouraged capital investment; export credits
were paid by the central government thereby advantaging exporting
provinces; state-owned enterprises were taxed more heavily than collec-
tives or private enterprises; and the complicated rate structure made the
revenue yield a function of provincial economic structure rather than total
output.
The 1994 reform program substantially restructured the sales tax
system. The VAT was expanded to absorb most sectors covered under the
product tax, and the latter will eventually be phased out. The wholesale
and retail sectors are taxed under VAT rather than under the business tax.
A two-rate VAT system was adopted (17 percent and 13 percent), and pre-
sumptive crediting was eliminated in favor of an invoice method. Because
of the potential revenue cost, there was no proposal to permit a credit for
taxes paid on capital goods.
From the point of view of intergovernmental relations, the 1994 reform
was a major change. The VAT was made a central government levy, not
subject to sharing, though a transitional sharing arrangement was put in
place. Administrative arrangements were also changed, and beginning in
1994, the VAT was fully administered by the central government without
local involvement.
China 53
Tax administration
The 1994 reform established separate national and local tax administration
services. This was a major national policy change with far-reaching
implications. Economists have long written about the folly of separating
tax structure reform from tax administration considerations (Bird, 1989),
but in China it is also true that intergovernmental reform cannot be
divorced from tax administration reform. Reform of the revenue-sharing
and taxation systems in 1994 could not have achieved its objectives unless
the administrative system was also changed.
The responsibility of the National Tax Service (NTS) is to collect central
and shared taxes. The NTS has about 500,000 employees and operates
down to the county level. Prior to the 1994 reform, two-thirds of these
employees operated at the lowest level of government, with relatively little
central supervision. Consequently, they developed close ties with the local
finance offices, and with the locally owned enterprises (IMF, 1994). This led
to a kind of collusion that was thought to weaken the position of the central
government in revenue-sharing (World Bank, 1990).
Under the reformed system, a separate local tax service operates in each
province with responsibility for collection of “local taxes,” including the
enterprise income tax on locally owned enterprises, collectives, and
private businesses.7 The local tax service is now part of the local finance
departments, and its operations are funded from local resources. The
central tax service is responsible for collecting the VAT and all other central
or shared taxes. The NTS still operates on a regional basis, but is inde-
pendent from the local tax service.
This reform is an attempt to solve a longstanding problem of divided
loyalties among the local officers in the former State Tax Bureau. There no
longer will be a question about the objectives of the tax administration
office as regards the choice between full taxation under the law and
maximization of retained local revenue. The NTS now has every incentive
to collect fully the indirect taxes due, since these are the mainstay of the
central government revenue structure. The local tax service must decide
how it will treat local enterprises under the income tax, but if it chooses to
grant preferential treatment it must absorb whatever revenue costs occur.
Revenue-sharing
China’s revenue-sharing system is primarily a division of national sales
and profit taxes among the central, provincial, and local governments.
Whereas in most countries taxes are collected by the central government
54 Roy W. Bahl
and then allocated to the subnational governments, in China they have

been collected by the local governments and “shared-up” to the higher
levels. The amount of shared tax revenue that finally shows up in the local
government budget depends on the centrally determined tax base and tax
rates, the tax administration, the assignment of revenues to the local
government, and the sharing formulas. These formulas, in turn, have two
elements: the proportion of revenues from any given tax that is to be
shared with the province, and the method by which this amount is distrib-
uted amongst local governments within the province. In addition, there
are earmarked grants which are given to provincial governments by the
center. To understand the revenue-sharing system in China, and to under-
stand the implications of the 1994 reform, one must understand all of these
dimensions. It is a very complicated matter.
In the following subsections, the three major issues surrounding
revenue sharing are taken up: the actual distribution formulas for shared
taxes, the distribution of earmarked grants, and the pattern of extrabud-
getary revenues of local governments. The last of these is properly part of
the discussion of intergovernmental fiscal relations in China, because
extrabudgetary funds reflect the success of Chinese local governments in
mobilizing resources outside the sharing pool.
Shared taxes
The revenue-sharing formulas have been regularly adjusted since the fiscal
reform began in 1983. The 1994 reform can be best understood as another
step in this transition. The formula has changed over time and was not
always strictly applied, but the government has held to a basic philosophy:
that local governments should retain enough revenue to cover a “basic”
level of services and should turn the remainder over to the center. The
debate and negotiation usually centered around what should be the
“approved level” of expenditure and by how much the approved level
should grow each year.8
The revenue-sharing formula in operation from 1985–88 provided for
three categories of provinces: those which retain a fixed percentage of what
they collect (fifteen provinces), those which keep all they collect and
receive a subsidy, and those which retain all they collect and pay a fixed
subsidy to the center (Guangdong Province).
There was dissatisfaction with this system of tax-sharing. The subna-
tional governments felt that their shares were too low, and both sides felt
that the system was too arbitrary. The ability of the central government to
use fiscal policy as an instrument of macroeconomic policy and regional
China 55
equalization was limited, local government incentives for increased

revenue mobilization were weak, and local government incentives to
move revenues to the extrabudgetary sector were strong. Some ad hoc
adjustments were made in 1987 to address the incentives issue (an incre-
mental sharing rate on revenues above a target level was introduced in
Jiangsu, Zhejiang, Hebei, Beijing, and Tianjin) but this did not quiet the
calls for a complete reform.
Beginning in 1988, the system was changed to an even more negotiated
approach. The idea was still to allow the local governments to retain an
amount of revenue that would enable them to cover a basic level of expen-
diture needs. The base year chosen to define the expenditure amount was
1987. Local governments retained the base amount plus a share of any
increase in revenues, as negotiated in advance. This agreement, originally
planned to be in force for three years, led to six different arrangements for
tax-sharing.9
The fundamental problem, however, remained. The system was nego-
tiated and not transparent. This led to criticisms that it was unfair. The local
governments continued to have an incentive to channel tax money to their
own treasuries, outside the formal revenue-sharing system. And, though
neither the central nor the local governments felt that the division of the
revenues was correct, the flow of resources to the central government and
the overall level of revenue mobilization declined.
The 1994 reform also directly addressed the revenue-sharing system. The
most important element of the revenue-sharing reform was the designation
of the value-added tax as a central tax which would be shared 75/25
between the center and the provinces.10 The enterprise income tax (other
than that collected from centrally owned enterprises) and the individual
income tax would be assigned fully to the local governments. Local govern-
ments would also have responsibility for income tax collection. This would
seem to be a first step toward an assignment system – though as noted above,
the local governments have no authority to set the rate and base of the tax.
Earmarked grants
In addition to shared taxes, local governments receive earmarked grants
from the central government. The purposes of this subsidy include
appropriations for capital construction projects, price subsidies for urban
grain consumption, social relief funds, and special subsidies for health and
education of the poor, of ethnic minorities, and in border provinces.11 There
is no set formula to determine the amount of earmarked grants for distri-
bution in any given year, and the distribution among the provinces
56 Roy W. Bahl
appears to be ad hoc rather than formula-based. Earmarked grants have

grown significantly, and were greater in magnitude than tax-sharing trans-
fers to the central government in 1992.12 In 1990, earmarked grants were
equivalent in amount to 14.4 percent of local government budgetary
expenditures, slightly less than in 1985.
Some would argue that a primary purpose of earmarked grants is equali-
zation. In fact, China has no equalization grant program. Simple correla-
tions show that per capita earmarked grants are distributed in significantly
larger amounts to higher income provinces.13 These are project rather than
entitlement grants, and it is not unusual for more developed regions to
absorb more of such grants because of their greater capability at project
preparation and their greater ability to “buy in” on a matching basis.
Extrabudgetary revenue
The 1980s and early 1990s saw a rapid growth in revenue outside the bud-
getary accounts. Local governments in particular had found a way to
siphon revenues away from the normal budgetary accounts – and the
sharing pool – to special purpose spending. The declining ratio of taxes to
GNP over the past decade reflects this transfer of funds.
There are two types of extrabudgetary revenue in China. The first is the
“fiscal extrabudgetary funds” of the government. These are earmarked for
capital purposes and include a set of taxes and charges that are controlled
by the local government finance department. The most important is the
public utility surcharge – a 10 percent tax on the utility bills of consumers
– and the urban construction and maintenance tax. There are also some
minor taxes and charges in this category, including the surcharge on the
agricultural tax, revenues received from public housing and public prop-
erty, and some institutional income that accrues to the various city enter-
prises. The latter include such items as fees and charges from hospitals,
road maintenance charges, advertisement fees, etc. The World Bank (1994:
31) estimates that fiscal extrabudgetary revenues grew from 2.6 percent of
GDP in 1978 to 4.2 percent in 1993, and now are equivalent to about one-
fourth of total budgetary revenues.
The other type of extrabudgetary revenue is the retained earnings and
depreciation funds of locally owned enterprises. Extrabudgetary funds of
this type expanded very rapidly in the 1980s, and in 1993 were equivalent
in amount to about 12 percent of GDP. In principle, these funds should not
be classified as part of the government budget because they are not
resources over which the local governments have direct control. Nor is
their rapid growth necessarily a cause for alarm. It could reflect increased
China 57
profitability, increased enterprise savings, and government policy to allow

enterprises more flexibility in managing their resources (World Bank, 1993:
17–19). However, the flagging level of enterprise income tax collections
suggests that the increase in extrabudgetary funds was at least partly a
result of the contracting system between local governments and enter-
prises that led to a channeling of resources from the budgetary to the extra-
budgetary side. Some of these revenues were used for social purposes and
to support the economic development goals of the local government
administration, and their growth was heavily influenced by local taxing
and contracting practices.14
Extrabudgetary funds are by their very nature sensitive to the business
cycle. In 1992, total extrabudgetary revenues of local governments were
equivalent to about 89 percent of local government budgetary collections.
This share increased after 1986 when enterprise contracting began to grow
significantly. The responsiveness of extrabudgetary revenues to GNP
would appear to be greater than the responsiveness of budgetary revenues
– there is a positive and significant relationship between the ratio of local
government extrabudgetary revenues to budgetary collections, and to
GNP, over the 1983–92 period.15
Objective of the reform

The 1994 reform of the fiscal system was prompted by the slow growth in
central government revenues. The primary contributing factors were a
long-term decline in tax revenue collections as a share of GNP and a
reduced share of total revenue going to the central government. The 1994
reform was an attempt both to increase the revenue income elasticity of the
system and to recentralize. It also reflected an attempt by the center to
regain control over fiscal policy.
The declining tax ratio

The revenue trend over the period 1985–95 is described in table 2.2. The
national tax ratio (collections as a share of GNP) had fallen to less than half
of its level of about 23 percent at the time the enterprise income tax was
introduced.16 By 1995, government revenue mobilization was about 11
percent of GNP, a low share by international standards.17 A declining
government share of GNP in China is not a problem per se because it
reflects the narrowing of the scope of government responsibility.18
However, this decline was greater than was planned and fiscal deficits
arose.19 Between 1985 and 1993, the revenue income elasticity of the
58 Roy W. Bahl
Table 2.2. Tax revenue performance, 1985–95
Enterprise income
Total tax revenue Total tax revenue as tax revenues as a
(billions of yuan) a percentage of GNP percentage of GNPa
1985 208.5 23.18 8.23

1986 213.3 20.90 7.20
1987 218.3 18.26 5.91
1988 244.2 16.36 4.87
1989 279.1 16.49 4.51
1990 290.0 15.59 4.27
1991 306.5 14.14 3.72
1992 335.7 12.59 2.93
1993 430.5 12.45 2.11
1994 512.7 11.01 1.52
1995 603.8 10.54 1.53
Note: a Includes direct remittances by enterprises.

Chinese tax system was only 0.60, that is, the rate of revenue growth was
a little more than half the rate of growth in GNP.20 Such a slow growth in
revenue will almost certainly create budgetary balance problems and
restrain expenditure growth. Government policy is to increase the income
elasticity of the revenue system, and reduce the size of the deficit (Zhongli,
1996).
The decline in the revenue share of GNP is largely due to the declining
budgetary contributions of the enterprise income tax. On the one hand,
this is neither unexpected nor undesirable. The enterprise reform in
China was meant to reduce payments to the government and to give
enterprises more control over their resources (World Bank, 1994). On the
other hand, this industrial policy alone would not seem to have dictated
so great a reduction in enterprise tax payments. There are several
explanations for the large reduction: the offloading of some enterprises
from the government sector, the granting of tax incentives by local
governments, and enterprise losses. Another reason for the weak
revenue performance of the enterprise income tax is the failure to expand
the tax base adequately to include the private and collective enterprise
sector. In the early 1990s, about 60 percent of GNP was accounted for by
the non-state sector, but almost 80 percent of tax revenues were derived
from the state-owned enterprises (IMF, 1994: 28). Whatever the reason, at
a 1.5 percent share of GNP, the enterprise income tax has become much
less a force in economic policy than it was at over 8 percent of GNP in its
early years (see table 2.2).
China 59
Division of revenues
The revenue–expenditure balance between central and local governments
changed markedly in the 1980s and early 1990s. At the time of the income
tax reform in 1983, the local government sector was spending an amount
equivalent to 74 percent of what it collected and turning a net 26 percent
over to the central government for national purposes. By 1992, the local
government sector was spending all of its collections and receiving an
additional subsidy from the center equivalent in amount to about 2 percent
of expenditures (table 2.3). Historically (through the mid-1980s) central
fixed revenues were considerably less than central government expendi-
tures, and the difference was made up with net transfers from the local
government sector. For example, in 1984, the local governments ran a
“collections surplus” of about 17 billion yuan while the central govern-
ment ran a “collections deficit” of about 23 billion yuan (table 2.3). The net
transfer was from local to center. However, by 1988, the local government
sector was regularly running a collection deficit. In fact, in 1992, there was
a net transfer from the center to the subnational sector (excluding any ear-
marked grants) equivalent to about 19 percent of central collections. It is
this change in the division of revenues that prompted the 1994 reform of
the revenue-sharing system.
Central government macroeconomic policy

Prior to the 1994 reform, central officials complained that the intergovern-
mental system compromised the effectiveness of macroeconomic policy.
Because subnational governments retained too large a share of the pie, the
central government did not have the resources to invest adequately in
national projects, to compensate poor regions for inadequate taxable
capacity, or to control the aggregate level of expenditure and investment
in the country. Moreover, during periods of expansion, enterprises were
prone to increase their rate of investment because of the availability of
funds and because the automatic stabilizer built into the enterprise income
tax was nullified by the practice of contracting. The result was that aggre-
gate demand grew, the government revenue share of GNP fell, and a fiscal
deficit resulted. There is an analogous procyclical effect in times of reces-
sion. Under the contract system, the central government could do very
little to control this situation.
It was also the case in the pre-reform period that the ability of the central
government to pursue discretionary fiscal policy was limited because of
the possibility of offsetting policy reactions by the local governments. For
Table 2.3. Tax collection and expenditure of central and local governments, 1980–95 (billions of yuan)a
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995.
Central government
Collectionb 28.4 31.1 34.7 49.0 66.5 77.0 77.8 73.6 77.5 82.3 99.2 98.8 98.0 95.8 290.7 325.7
Expenditure 66.7 62.6 65.2 76.0 89.3 79.5 83.6 84.6 84.5 88.9 100.4 109.1 117.0 131.2 175.4 199.5
Collection surplus/ ⫺38.2 ⫺31.5 ⫺30.5 ⫺27.0 ⫺22.8 ⫺2.6 ⫺5.8 ⫺10.9 ⫺7.0 ⫺6.6 ⫺1.2 ⫺10.3 ⫺19.1 ⫺35.5 115.2 126.1
deficit
Local government
Collectionb 87.5 86.5 86.5 87.7 97.7 123.5 134.4 146.3 158.2 184.2 194.5 221.1 250.4 339.1 231.2 298.6
Expenditure 56.2 51.3 57.8 65.0 80.8 120.9 136.9 141.7 164.6 193.5 207.9 229.6 257.2 333.0 403.8 482.8
Collection surplus/ 31.3 35.2 28.7 22.7 17.0 2.6 ⫺2.5 4.7 ⫺6.4 ⫺9.3 ⫺13.4 ⫺8.5 ⫺6.8 6.1 ⫺172.7 ⫺184.3
deficit
Notes:
a
The difference between the local government’s surplus and the central government’s deficit is the central government’s foreign borrowing and
domestic budget deficit.
b
Central and local government collections are before transfer from local governments to the central government, and before earmarked grant
distribution to the local governments.
China 61
example, the central government might levy an extra tax to cool down an
overheated economy, but local governments could react to this by renego-
tiating their contracts with enterprises to stimulate investment, or they
could give preferential tax treatments to avoid the consequences of the
central tax.
Another problem with the revenue-sharing system was that it compro-
mised the implementation of the government’s industrial policy. Qiang
(1995) makes the point that the sharing of income taxes from local enter-
prises and the sharing of indirect taxes collected in local areas encourages
local governments to favor development of projects that show high profits
and sales. It is a problem in all countries that local officials, facing a rela-
tively short political life, favor investment projects that have a high visibil-
ity and offer a substantial short-term return. It is an especially acute
problem for China because of the substantial amount of resources in the
hands of local governments and locally owned enterprises (Chinese
People’s Bank of Construction, 1993).
This leads to a number of problems. First, it diverts resources away from
the more productive social investments that the central government would
prefer. In effect, it dramatically limits the ability of the central government
to control the direction of investment in the economy. Second, it has
encouraged some governments to erect barriers to trade in order to protect
local investments (and implicitly to protect the local tax base). Third, it is
not clear that decentralization in the budgetary and extrabudgetary
accounts has led to a more rapid economic growth rate (Zhang and Zou,
1996).
The validity of the argument that the center was losing its ability to
control effectively the fiscal sector rests on the argument that an increasing
share of total revenues was accruing to the local government sector.21 This
concern is borne out by the evidence presented in table 2.3. Moreover, the
situation worsened through the early 1990s. The elasticity of central
government revenue collections with respect to GNP during the 1985–93
period was 0.24, which is less than the 0.72 GNP elasticity of subnational
government collections.22 The collections’ disparity in favor of local
governments accelerated with the growth of the contracting system from
1988. During the 1985–87 period, the ratio of local government to total
government budgetary collections averaged 54.2 percent. During the
1988–92 period, the average was 59.5 percent. The nexus of collections had
clearly shifted toward the local governments.
The bigger issue is that the subnational governments’ proportion of rev-
enues, after sharing, was increasing. There was a net transfer of resources
from the center to local governments after 1986 (“net transfer” is measured
62 Roy W. Bahl
here as the sum of local government budgetary collections and earmarked

transfers from the center to the local governments, less shared tax transfers
from the local governments to the center). By 1990, the net transfer was
about 2 billion yuan from the center to the local governments. As recently
as 1985, there had been a significant net transfer to the center.
There are a number of reasons why this happened. First, the sharing for-
mulas implemented in 1988 were negotiated to favor the local govern-
ments. Second, poor economic performance and natural disasters
prevented local governments from reaching their negotiated amounts and
forced the central government to forgive part of the planned transfer, and
to increase earmarked grants. Third, local governments used their author-
ity to grant tax concessions and favorable contracts, and this reduced the
flow of revenue to the central government (Cullen and Fu, 1996: 7).
For whatever reasons, the central government’s ability to use the fiscal
system for macroeconomic policy purposes was quite limited by 1992. The
center collected only 28 percent of total revenues in the country, and paid
out more in transfers to the local government sector than it received back
(table 2.3). The enterprise income tax had become a much less important
fiscal instrument, largely because the local governments had more or less
replaced it with negotiated contracts.
Evaluation of the 1994 reform

An evaluation of the success of the 1994 reform in improving intergovern-
mental fiscal relations is problematic, for two reasons. The first is that the
objectives of the Chinese government were different from those pursued
in most reforms of the system of central–local finances. The central govern-
ment of China was attempting to wrestle control of fiscal and macro-
economic policy back from the local government sector. The evaluation of
the reform must begin with an assessment of how well they did this. While
it is also appropriate to consider the more usual issues that arise when one
evaluates an intergovernmental reform – equalization and local autonomy,
for example – it is necessary to recognize that these were not the driving
forces behind the Chinese reform.
The second problem is that data are not yet available to carry out a quan-
titative analysis of the new system of revenue-sharing. In part, this is
because the Chinese government has not yet made available the detailed
data on the outcomes in the post-reform period. The data now available for
1994 and 1995 may show only muted effects of the reform on the inter-
governmental fiscal system because of the provisions to maintain previous
levels during a “transition” period.
China 63
Central control
Will the 1994 reform recentralize the fiscal system? The shift of the VAT to
a central revenue source and the elimination of contracting would seem to
guarantee this result. Note from table 2.3 that in 1994 and 1995, the central
government collected 52 and 56 percent of total revenues, and was in a
surplus position.23 The local governments, on the other hand, were in a
deficit position and forced to rely on grant distributions from the center.
China has moved to the top-down revenue-sharing approach which is
used by most countries in the world.
In fact the central government’s gain of control over the aggregate
budget is understated here because of a transitional revenue-sharing
arrangement. In the longer term, fiscal recentralization will be occurring in
three important ways:
• The VAT is the most productive tax in the system, is fully administered
by the central government, and is no longer a shared tax.
• Revenue-sharing in China is now top-down rather than bottom-up, and
the central government can more easily control the allocation of funds.
• There is no longer a contracting mechanism that will allow local govern-
ments and enterprises to negotiate tax liabilities. This means that the tax
measures undertaken for stabilization purposes now have a better
chance to achieve their desired objectives.
The move to an assignment system for local government revenues raises
two issues. First, the assignment of the enterprise income taxes to local
governments may present difficulties in the future. As and when Chinese
enterprises begin to operate in multiple provinces, the problem of allocat-
ing profits will arise and will bring complication and possibly undesirable
competition. The US experience is a case in point (McLure, 1981). It is
unlikely that the provinces can build their long-run fiscal health on an
enterprise income tax. The assignment of the VAT to the central govern-
ment is appropriate, because the VAT is best administered centrally. But
sharing the VAT on a derivation basis, as is still done, encourages protec-
tionist-type activities by provinces to preserve their tax base. A formula-
based system for sharing VAT revenues with the provinces would avoid
this problem.
Resources available to local governments

The resources available to local government under the reformed system
will depend on whether the overall rate of revenue mobilization is
64 Roy W. Bahl
increased, and on the willingness of the central government to make trans-

fers to local government from its new-found wealth.
With respect to total revenues available, the government expects the net
effect of the reform to be revenue-neutral in the short run. On the one hand,
there are expected negative revenue effects, ceteris paribus from lowering
the top enterprise income tax rate from 55 percent to 33 percent; from elimi-
nating the surtaxes; and from correcting part of the understatement of
labor and capital costs in calculating enterprise profits. A potential short-
run revenue loss could be anticipated from the switch to the credit invoice
method under the VAT. This is a completely different approach to VAT
administration, and some transition costs could be incurred in the first
years of operation.24
Other elements of the reform are expected to produce revenue increases.
The elimination of contracting could increase revenues significantly.
Revenues from the enterprise income tax will also increase as a result of
the elimination of the loan repayment deduction that will broaden the
base. The World Bank (1990) estimated the revenue cost of the loan repay-
ment deduction to be equivalent to about 3 percent of total revenue in 1986.
If enterprise income tax revenues were to be restored to 1985 levels, rela-
tive to GNP, total government tax revenues would be higher by about 6
percent of GNP.
Short-run revenue-neutrality has not yet been achieved. Total revenue
collections continued to fall in 1994 and 1995 relative to GNP (table 2.2).25
The long-run revenue consequences of the reform, however, are likely to
be more significant. The value-added tax is a more efficient revenue-
raising instrument than the product tax, in that it reaches a much greater
number of transactions and it has a self-policing component. If the
administration of the Chinese sales tax was deficient, the shift to a credit-
invoice VAT could bring significantly greater long-run revenue growth
than the present system. Wong, Heady, and Woo (1995: 52), point out that
the buoyancy of the Chinese VAT in the pre-reform period was low rela-
tive to that in other countries. The elasticity of the income tax could also
increase because the incentives for locals to offer preferential income tax
relief have been reduced. Moreover, the local governments have an inti-
mate familiarity with local enterprises and can enhance the efficiency of
local collections.
There are aspects of this reform that suggest diminished local govern-
ment revenues. The more income-elastic VAT was assigned fully to the
central government. Revenues from the VAT are now shared with local
governments, but this seems to be a transition measure, and in the long
run, derivation-based VAT-sharing may be phased out. The reform assigns
China 65
all income taxes to the local governments. If the long-run elasticity of the
income taxes is less than that of the VAT, and surely it will be, then the rela-
tive growth potential of local government taxes will be less. Moreover, the
enterprise income tax is less stable over the business cycle, hence local
government revenue streams will become less certain.
The longer-run revenue position of local governments would appear to
depend on the central government assigning another significant source of
taxation to the local governments or establishing a buoyant intergovern-
mental grant program. The individual income tax, with an expanded base
and piggybacked on to the central tax, would seem the ideal revenue
source for Chinese provincial governments. As to increased transfers, a
more productive VAT would increase the central pool available for tax-
sharing and for earmarked grants. Whether local governments would
actually benefit from an increased revenue pool depends on the willing-
ness of the central government to allocate more resources to the subna-
tional level.
Local government autonomy

The 1994 reform did not increase the autonomy of local governments in
taxation. The power to set rates and define bases, for all taxes, remained
with the central government. In the past, local governments had taken
some autonomy using “back-door” approaches such as providing prefer-
ential tax treatments in return for infrastructure investment, or to pursue
industrial policy. The 1994 reform eliminated the practice of enterprise tax
contracts with local governments. This was unquestionably good tax
policy. The elimination of contracting substituted a transparent tax system
for a negotiated one, and placed all enterprises on a more equal footing.
However much this measure improved the functioning of the economy,
it did reduce the fiscal discretion of local governments. Local governments
in China had made the granting of tax incentives an integral part of their
industrial policy. Moreover, they had used a combination of tax incentives
and discretion in matters of tax administration to leverage more social
overhead investment by enterprises. The elimination of these preferential,
discretionary tax treatments was necessary for macroeconomic purposes,
but it clearly weakened fiscal decentralization.
The changes in the tax administration system may also be having a
centralizing effect. Local governments now collect only income taxes and
the smaller “local taxes” and no longer exert control over the assessment,
collection, and audit of the VAT. An important component of local govern-
ment discretion in tax administration was thus eliminated.
66 Roy W. Bahl
Fiscal disparities
Evaluation of the equalization features of the revenue-sharing system is
not straightforward, since the government’s objectives are not clear on this
matter. We can, however, address the question of whether the reform will
increase or reduce fiscal disparities. Four questions could be answered on
an a priori basis, and supporting empirical evidence might be provided:
• Has the spread in the revenue collection rate between rich and poor
provinces changed?
• Have expenditure differences between provinces changed?
• Is there evidence of greater equalization than in the pre-reform period?
• Do provinces still exert the same pattern of tax effort as in the pre-reform
period?
Revenue capacity
Revenue collections are determined by income level and tax effort. One
would expect wealthier provinces to have a significant advantage, and the
data seem to bear this out in the pre-reform period. In 1992, the five
wealthiest provinces accounted for 23 percent of GDP and 13 percent of the
national population but about one-fourth of all revenue collections. In
1995, with local revenue collections now primarily income taxes, the
revenue share rose to 30 percent (table 2.4). The five poorest provinces
accounted for about 13 percent of national income but 19 percent of the
population. The provinces collected about 12 percent of total revenue and
the share fell between 1992 and 1995. The disparity in per capita revenue
collections among the provinces in 1995 ranged from 1,551 yuan in
Shanghai to 139 yuan in Anhui.
The relationship between income level and collection rate can be tested
in a more systematic way to determine if it generally explains the pattern
of variation in per capita collections across the twenty-eight provinces.26
We have estimated a linear regression of per capita collections by local
governments in 1990 and 1995 on per capita GDP and population size. Per
capita GDP is included to measure the taxing power of the province, and
population gives the impact of a size effect on local collections.
The results presented in table 2.5 show a strong significant relationship
between per capita revenue collections and per capita income in both
years. The cross-sectional income elasticities are about unity: in 1995, a 10
percent difference in per capita income tended to be associated with about
a 10 percent difference in per capita revenue collections. Population size
did show the expected negative relationship with revenues.27
China 67
Table 2.4. Fiscal and economic concentration in rich

and poor provinces
1990 1995
Five richest provincesa

Percentage of GDP 22.8 23.9
Percentage of population 12.7 12.4
Percentage of revenue collections 26.0 30.3
Percentage of local government expenditures 19.8 25.5
Five poorest provincesb
Percentage of GDP 12.7 10.6
Percentage of population 18.9 19.0
Percentage of revenue collections 12.3 10.5
Percentage of local government expenditures 14.0 12.4
Notes:
a
Shanghai, Beijing, Tianjin, Guangdong, and Zhejiang.
b
Guizhou, Gansu, Sha’anxi, Jiangxi, and Henan. Tibet and
Hainan are excluded because of special circumstances and
data availability.
Sources: Computed from data provided in the Statistical
Yearbook of China and data provided by the Chinese Ministry
of Finance.
Table 2.5. Regressions of per capita local government

revenues and expenditures on selected variablesa
Per capita Per capita

revenues expenditures
1990 1995 1990 1995
Constant ⫺2.7636 ⫺2.9755 2.4332 1.1178

(3.64) (4.32) (4.10) (1.63)
Population 1.1207 1.0736 0.5600 0.7235
(12.25) (14.01) (7.84) (9.46)
Per capita GDP ⫺0.0828 ⫺0.1037 ⫺0.3442 ⫺0.2865
(1.58) (2.06) (8.40) (5.71)
R2 0.88 0.95 0.89 0.86
Nb 28.0 28.0 28.0 28.0
Notes:
a
All variables expressed in logarithms with t-statistics
shown in parentheses below the regression coefficients.
b
Tibet and Hainan are not included.
68 Roy W. Bahl
There has been some weakening in the relationship between collections

and income. Virtually all of the higher income provinces had growth rates
in budgetary collections that were below the 1987–92 national average
(Bahl, 1994). The regression estimates presented in table 2.5 show that the
cross-sectional income elasticity of collections fell from 1.12 in 1990 to 1.07
in 1995.
Expenditure variations
Disparities are much less pronounced on the expenditure side of local
budgets, because of the equalizing effects of the transfer system. In 1995,
per capita expenditures varied from highs of 1,837 yuan in Shanghai and
1,234 yuan in Beijing to 228 yuan in Henan and 225 yuan in Anhui.28 The
five highest-income provinces, with 13 percent of the population, account
for 25 percent of the expenditures, whereas the five lowest-income ones,
with 19 percent of the population account for 12 percent of expenditures.
We can conclude that per capita expenditure variations in China are very
great indeed, even though less than those for collections.
What are the determinants of the variations? Higher-income provinces
spend significantly more on a per capita basis because of the greater
demand for public services by their citizens and their enterprises, their
ability to raise more “local fixed” revenues, their ability to attract more
grants, and very importantly, their ability to slow the flow of revenues to
the center. More heavily populated provinces spend less on a per capita
basis, because the fixed component of their cost is spread over a larger
population base. The expenditure regression results presented in table 2.5
confirm these expectations. About three-quarters of the interprovince vari-
ations in per capita expenditures can be explained by variations in per
capita income and population size. Both variables are significant and have
the expected sign. The expenditure income elasticity is smaller than the
revenue elasticity in both years, suggesting some degree of equalization in
the system. In 1995, a 10 percent higher level of per capita income was, on
average, associated with about a 1 percent higher level of per capita expen-
ditures and a 1 percent higher level of per capita collection. The equaliza-
tion of the system appeared to lessen between 1990 and 1995.
Redistribution and the 1994 reform

One could make the case that the 1994 fiscal reform will not reduce dispar-
ities in per capita expenditures, nor will it compensate for uneven taxable
capacity among the provinces. An unintended effect of the fiscal reform of
1994 is a possible widening of the fiscal disparities among provinces. By
giving the local governments all revenues raised from income taxes, on a
China 69
derivation basis, the central government has given an advantage to the

higher-income provinces with more profitable enterprises. And by giving
local governments responsibility for collecting these taxes, advantage has
been given to the provinces that have the means to mount better adminis-
trative efforts. Again these are likely to be the better-off provinces.
The empirical analysis presented above shows that per capita expendi-
ture disparities are less pronounced than per capita revenue collection dis-
parities, suggesting some degree of equalization in the system. However,
this pattern is less true in 1995 than it was in 1990, suggesting that the
reformed system may be less equalizing.29
Tax effort
A criticism of the present intergovernmental fiscal system is that there is
insufficient incentive for revenue mobilization, and therefore tax effort on
the part of the local governments is lower than it should be. This should
have been expected in the pre-reform system. The tax-sharing formula
was based on a retention rate where a percentage of collections was paid
to the central government. This provided a significant incentive to move
funds from the budgetary accounts to the extrabudgetary ones. Moreover,
the local administration still saw itself partly as tax collector from the
state-owned enterprises, and partly as owner. Jun Ma (1996) shows that
the central government’s failure to commit to pre-announced revenue-
sharing formulas tempted local governments to reduce their tax collection
efforts.
Can local governments influence tax effort? In the pre-reform system,
the answer was that they could, in many ways. The local administration
could negotiate “revenue-losing” contracts with local enterprises, could
give tax incentives to them, and could urge the local tax collection service
to be less vigorous in their efforts. As Casanegra (1987: 25) has pointed out,
“. . . tax administration is tax policy.” Even in the reformed system, there
may be inducements to lower levels of tax effort. Though they may no
longer write tax contracts, local governments now control the administra-
tion of all income taxes. Though the giving of tax incentives now imposes
a dollar-for-dollar loss to local budgets, relief from the enterprise income
tax still may be provided for industrial policy purposes. The revenue cost
may be made up with a central grant, which, without conditions, might be
seen locally as a substitute for higher income tax collections.
We have carried out a tax effort analysis, following the approach
described in Bahl and Wallich (1992) and updated in World Bank (1993).
The first step is to estimate an OLS regression of per capita local govern-
ment revenue collections on per capita income and population size. The
70 Roy W. Bahl
Table 2.6. Tax effort indices and rankings: 1990 and 1995
1990 1995
Tax effort Tax effort Per capita Tax effort Tax effort Per capita
Province indices rankings GDP ranking indices rankings GDP ranking
Beijing 1.00 15 2 1.16 5 2

Tianjin 1.03 13 13 1.03 12 13
Hebei 0.92 20 13 0.92 21 14
Shanxi 1.14 15 15 1.30 13 18
Inner Mongolia 0.99 16 16 0.96 18 16
Liaoning 1.12 16 14 1.12 7 16
Jilin 1.07 7 19 0.99 16 13
Heilongjing 1.02 14 18 0.95 19 11
Shanghai 1.40 12 11 1.30 14 11
Jiangsu 0.93 19 17 0.68 27 17
Zhejiang 1.07 18 16 0.67 28 15
Anhui 0.84 23 24 0.93 20 22
Fujian 1.03 12 11 1.01 14 18
Jiangxi 0.88 21 22 1.08 9 26
Shandong 0.77 26 19 0.77 26 10
Henan 0.94 18 26 0.97 17 25
Hubei 0.87 22 12 0.85 23 15
Hunan 0.97 17 18 1.06 10 20
Guangdong 0.75 27 15 1.39 12 14
Guangxi 1.06 9 28 1.03 13 19
Sichuan 1.05 10 25 1.05 11 23
Guizhou 1.32 13 29 1.11 18 30
Yunan 1.78 11 21 1.60 11 24
Sha’anxi 1.05 11 20 1.00 15 27
Gansu 1.27 14 23 1.13 6 29
Qinghai 0.81 25 14 0.79 25 17
Ningxia 0.82 24 17 0.88 22 21
Xingjiang 0.68 28 10 0.83 24 12
results of this analysis are presented in table 2.5. This equation is then used
to obtain an “estimated” value of per capita collections for each province.
This might be interpreted as a measure of taxable capacity, that is, it is the
amount that a province would raise if it used its tax base to the same extent
as other provinces. For example, in 1995 Hebei is estimated to have a
capacity to raise taxes of 203 yuan per capita (predicted by the equation),
but actually raises only 186 yuan per capita. Hebei, then, has a tax effort
index of 0.92, which is 8 percent below the average and ranks Hebei
twenty-first among the provinces in tax effort in 1995.
The estimated indices of tax effort are presented for each province for
1990 and 1995 in table 2.6. There seems to be no clear pattern to the distri-
bution of tax effort indices. There is no significant correlation with either
per capita income, the ratio of expenditures to collections, or the rate of
China 71
urbanization. The results do not support the hypothesis that the present
revenue-sharing system reduces the rate of revenue mobilization.
Summary appraisal of the reform

China’s 1994 reform was not promoted primarily as one to encourage fiscal
decentralization – in fact, the announced objective was precisely the oppo-
site. But the reform program has had profound effects on the balance of
powers between central and local governments in one of the most impor-
tant countries in the world. Though data are not yet available to do a full
quantitative evaluation of the outcome, the following are some stylized
conclusions that might be drawn.
First, China is not a federal country, but its public finance system has fea-
tures of fiscal federalism. Using the Western model for evaluating reforms
of fiscal federalism, we might conclude that the Chinese reforms were
neither efficiency-enhancing nor equalizing. However, they were aimed
primarily at redressing what was perceived as too great a revenue distri-
bution to the local governments, and at regaining control over industrial
policy.
Second, one cannot separate tax policy, tax administration, and inter-
governmental fiscal structure in evaluating fiscal decentralization in China.
The 1994 tax structure reform was good tax policy. But in assigning enter-
prise income taxes to local governments, and continuing a derivation-based
VAT, the reform was not good intergovernmental policy. The division of tax
administration powers between local and central governments was good
policy in that it overcame the problem of divided loyalties on the part of
local tax administrators, but local governments are left with full
responsibility for a tax whose rate and base they cannot influence.
Third, there is little in the 1994 reform that will increase the fiscal auton-
omy of subnational governments in China. The new budget law does give
some discretion over spending composition, but the revenue constraint is
still imposed by the center. Moreover, anticipated future tax structure
changes in the enterprise income tax may further tighten this constraint.
Yet by moving the structure away from a negotiated model toward a more
objective system, the center has put in place the basis for a future fiscal
decentralization. The former negotiated system with divided tax
administration responsibilities would never support a Western-type fiscal
decentralization.
Fourth, the long-term impact of the new revenue-sharing system and the
new tax structure is a recentralization of revenues. The local share will
diminish in future years if the present arrangement is maintained. Either a
72 Roy W. Bahl
new source of local government tax revenue must emerge, expenditure

reassignment to the central government must take place, or the grant
system must be expanded.
Fifth, the new revenue-sharing formula is more transparent than the pre-
vious negotiated system, and this is a major improvement. However, it is
not equalizing, it does not provide incentives for increased revenue
mobilization, and, in the transition period, it retains some bad features: the
VAT continues to be shared on a derivation basis and the earmarked grants
are still distributed on an ad hoc basis. As Bird (1984: 217) so correctly noted
in the context of the Colombian reform, “more good ideas have probably
come to naught because of transitional difficulties than for any other
reason except the political opposition of those who would lose as a result
of the change.” The final test of the success of the revenue-sharing reform
in China – at least from a Western perspective about what is good decen-
tralization – is whether the transitional arrangements can be abandoned in
favor of a fully objective system.
Grading China’s decentralization

Large countries tend to decentralize their government because the dis-
economies of central management are great, because large countries tend
to be ethnically diverse, and because large countries have wide variations
among their provinces in the demand for government services. China is
something of a paradox in that it is the largest country in the world but con-
tinues with a highly centralized fiscal structure. Predictably, however, local
governments, sometimes acting independently and sometimes acting with
the consent of the central government, have taken “back-door” approaches
to fiscal autonomy. The 1994 reform eliminated some of these surreptitious
routes to local autonomy, and took the first steps toward establishing a
modern, assignment-based intergovernmental fiscal system.
Because the 1994 reform represented such an important policy cross-
roads, it would seem a good time to take stock of the Chinese version of
decentralization. The criteria for effective decentralization shown in the left
column of table 2.7 provide a checklist. The conclusion from this evaluation
is that China is in the very early stages of decentralization, and many of the
features required for successful local governance are not yet in place.
The biggest difference between China and the decentralized systems in
the West is the absence of popular representation. Local councils must be
popularly elected and local chief officials must be locally appointed for the
efficiency gains from decentralization to occur (Bahl and Linn, 1992: ch. 3).
The other major difference from international practice in decentralized
China 73
Table 2.7. Requirements for effective decentralization
Benchmark Situation in China
Elected local council Elections, but not popular elections.

Locally appointed chief officials Chief provincial and local government
officials are appointed by the next highest
level of government.
Locally approved budget Local authority budgets do not require
approval, but total revenues are fixed by the
center.
Absence of mandates as regards local Guidelines provided by the central
government employment and salaries government.
Local governments may exert control on the Local governments may choose whether or
level of some revenue sources not to levy certain taxes or charges. The
rates and bases of nearly all taxes, however,
are beyond their reach.
Local governments have some powers to Local governments may not borrow
borrow directly.
The grant system is transparent and local The grant system is ad hoc. The central
governments understand their entitlements government decides on the retention rates
under the sharing system, and on the
distribution of earmarked grants.
Expenditure assignment is clear Expenditure assignment is relatively clear
to local government officials.
Local governments have the capacity to Local governments vary widely in their
collect taxes and deliver services efficiently capacity to deliver services effectively.
Problems arise mostly because of limited
revenues and shortage of capital and skilled
employees.
Local governments keep adequate books of The professional and technical ability of
account staff in many local authorities is below the
level needed to produce accurate and
meaningful accounts. In some provinces the
capacity is quite strong.
The central government has the ability to There is a fiscal planning unit that deals
monitor the progress of effective fiscal with intergovernmental fiscal relations in
decentralization the Ministry of Finance, but there is no
strong underlying data system to support
the work of the staff.
countries is on the financing side: local governments in China may not set
tax rates or borrow for capital projects. Again, determination of the size of
local government is largely a matter of central decision. These are the
major structural impediments to an effective system of local government
financing.
There are also capacity problems to be overcome. There is a wide varia-
tion in the ability of local governments to absorb more fiscal responsibility.
74 Roy W. Bahl
Some provincial and municipal governments in China have well-

developed systems of fiscal administration, while others (especially in the
rural areas) have very weak ones. It is also a problem that the central
government presently has only a limited ability to monitor local govern-
ment fiscal behavior.
On the other hand, some significant recent changes in the fiscal system
could set the stage for fiscal decentralization. The 1994 reform systema-
tized the situation by eliminating the practice of “every province for itself”
autonomy. The first steps toward decentralization have been taken on the
expenditure side, which is the right place to begin. Local governments
have some budget autonomy and are subject to relaxed guidelines as
regards public expenditure mandates, and expenditure assignment seems
relatively clear.
Looking forward, one could take two differing views about the future of
decentralization in China. One is that the 1994 reform clearly was a
recentralization of the fiscal system and weakened the hand of the local
governments. The other view is that the reform replaced a negotiated
system with a structure on which a decentralized system could be built. If
decentralization were a goal of the Chinese government, and it is not clear
that this is the case, then the following reforms are now more within reach
than in the period before 1994: a transparent grant system, based on
formula rather than ad hoc distributions and derivation; taxation by an
assignment system and possibly with some local discretion to add surtaxes
to national bases; an individual income tax as an important source of local
government revenue; and local borrowing powers, at least for the largest
and most prosperous provinces. If such measures are to be part of the next
steps in Chinese fiscal reform, the central government would do well to
begin by strengthening its capacity to provide technical leadership and to
monitor local government fiscal behavior.
Notes
1 Throughout this chapter we use the term “local government” to refer to all sub-
national governments.
2 An evaluation of the tax structure implications of the reform is in Bahl (forth-
coming). In the tax and revenue sections of this chapter, I draw liberally on that
earlier analysis. Baoyun Qioa provided excellent research assistance for both
this chapter and the earlier paper.
3 The Statistical Yearbook of China, which is the data source used for most of this
analysis, does not disaggregate indirect tax revenues by type, nor does it
present detail on the smaller local levies. It does, however, provide official sta-
tistics on Chinese public finance. Other studies, however, have provided more
China 75
detail relying on unpublished data provided by the Ministry of Finance and the
State Tax Bureau. These reports have shown the VAT, product tax, and business
tax to account for about equal shares of indirect tax revenue in 1992 (Bahl, 1994).
4 The tax contracts varied widely, from lump-sum payments to marginal rates
that varied with the level of profits. For a description, see World Bank (1990:
annex 2).
5 All income tax revenues collected from centrally owned enterprises belong to
the central government. All other enterprise income tax revenues, regardless of
ownership, revert to the provincial government.
6 For a good description of sales taxes in China, see World Bank (1990: 28–34).
7 In total, the local tax service collects fifteen different taxes.
8 The evolution of the tax-sharing formulas is discussed in Bahl and Wallich
(1992).
9 This description is elaborated in Qiang (1995).
10 In fact, in the initial years the revenue-sharing may be different from the 75/25
target because local governments were promised their revenues would be kept
at least at 1993 levels in the first years of the reform.
11 World Bank (1993: 80–81) and Bahl (1994).
12 Data are not available for more recent years.
13 The simple correlation between per capita earmarked grants and per capita
income is 0.425, and between per capita earmarked grants and population is
⫺0.662 (Bahl, 1994).
14 For a discussion of the long-term growth of extrabudgetary revenues, and the
composition of this growth, see Wong, Heady, and Woo (1995: appendix I).
15 The ratio of extrabudgetary to budgetary revenue increased with GNP over the
1983–92 period, as indicated by the following regression results:
EXB/B ⫽ 69.8 ⫹ 0.00093 GNP R2 ⫽ 0.43
EXB/B ⫽ 69.8 ⫹ 0.0009(2.45)
where EXB ⫽ extrabudgetary revenue, B ⫽ budgetary revenue, and the figure
in parentheses is the t-statistic.
16 The decline in the revenue share is exaggerated by the Chinese definition of rev-
enues and expenditures, notably the treatment of debt issued as a revenue, debt
repaid as an expenditure, subsidies to cover enterprise losses as a negative
revenue, and the exclusion of government “departmental” revenues from the
budgetary accounts. When adjustments are made to conform these data to stan-
dard international classifications, the results show that the Chinese revenue
share of GNP still declined substantially, but less than is reported in table 2.2
(World Bank, 1994).
17 Various analysts estimate the tax ratio to be higher by 1 to 2 percent of GNP,
depending on the data source used and the definition of revenues (Xu, 1995;
World Bank, 1994; IMF, 1994). Data from the Statistical Yearbook of China, used in
this chapter, give consistently lower estimates. All analyses, however, show
much the same declining trend.
18 Neither is a comparatively low tax ratio an indicator of “bad” fiscal per-
formance. Many of China’s social services, for example, are provided as tax
expenditures rather than through tax financing.
76 Roy W. Bahl
19 The consolidated fiscal deficit in China is not large by international standards
(about 6 percent of GDP) but it has grown significantly since the late 1980s
(World Bank, 1994).
20 Estimated from a linear regression of tax revenues on GNP with both variables
expressed in logarithms. Technically, this is a buoyancy rather than an elastic-
ity coefficient because no adjustments have been made for discretionary
changes.
21 This argument was also made by Xu (1995: 1–25).
22 The elasticities were estimated from linear regression of tax collections on GNP
with both variables expressed in logarithms, but no adjustments were made for
discretionary changes.
23 The data in table 2.3 also indicate a local surplus in 1993, but this is an
“announcement effect.” In anticipation of the 1994 reform, local governments
increased revenues in 1993 to establish a larger base for the new revenue-
sharing system (Xu, 1995: 18).
24 Xu (1995, tables 2 and 3) estimates a reduction in both direct and indirect taxes
as a result of the reform.
25 In fact, a decline in the tax share of GDP may not be inconsistent with revenue-
neutrality, because the pre-reform system was already showing a declining
trend relative to GDP.
26 Tibet and Hainan have been dropped from this analysis as special cases.
27 These relationships also held for the period 1985–90 in work reported by Bahl
and Wallich (1992), Bahl (1994), and a study by Hofman reported in World Bank
(1993). Hofman shows that virtually all of the variation in per capita collections
can be explained by a squared per capita income term.
28 The range in 1991 per capita state and local government expenditures in the
United States (excluding Alaska) was from $6,525 in New York to $2,715 in
Arkansas. The range in China was more than three times that in the United
States.
29 It is not entirely correct to draw conclusions about the new system based on
1995 results, because it was still in a transition period where part of the VAT was
shared with local governments. Moreover, the sharing arrangements were
irregular because of a provision whereby provinces were guaranteed support
for their basic level of expenditures in 1993.
References
Bahl, Roy (1994), “China: tax structure and intergovernmental fiscal relations,”
paper prepared for the World Bank, Washington, DC (March).
(forthcoming) “Central-provincial-local relations: the revenue side,” in Donald
J. S. Brean, ed., Taxation in Modern China, London: Routledge.
Bahl, Roy W. and Linn, Johannes F. (1992), Urban Public Finance in Developing
Countries, New York: Oxford University Press.
Bahl, Roy, and Wallich, Christine (1992), “Intergovernmental fiscal relations in
China,” Working Paper No. 863, Country Economics Department, The World
Bank, Washington, DC (February).
China 77
Bird, Richard (1984), Intergovernmental Finance in Colombia, Cambridge, MA:
Harvard Law School.
(1989), “The administrative dimension of tax reform in developing countries,” in
Malcolm Gillis (ed.), Tax Reform in Developing Countries, Durham, NC: Duke
University Press.
Casanegra, Milka (1987), “Problems in administering a value-added tax in devel-
oping countries: an overview,” Report No. DRD246, Development Research
Department, World Bank, Washington, DC.
Chinese People’s Bank of Construction (1993), “Investment-related background
materials,” unpublished paper, Planning Department (July 3).
Cullen, Richard, and Fu, H. L. (1996), “China’s ambitious intergovernmental fiscal
reform agenda,” unpublished paper, City University of Hong Kong
IMF (International Monetary Fund) (1994), Economic Reform in China: A New Phase
(Occasional Paper 114), Washington, DC: World Bank.
Ma, Jun (1996), Intergovernmental Relations and Macroeconomic Management in China,
New York: St. Martin’s Press.
McLure, Charles E., Jr. (1981), “The elusive incidence of the corporate income tax:
the state case,” Public Finance Quarterly 9: 395–413.
Qiang, Gao (1995), “Problems in Chinese inter-governmental fiscal relations and
the main difficulties in future reform,” in Ehtishem Ahmad, Gao Qiang, and
Vito Tanzi, eds., Reforming China’s Public Finances, Washington, DC:
International Monetary Fund.
Wong, Christine, Heady, Christopher, and Woo, Wing T. (1995), Fiscal Management
and Economic Reform in the People’s Republic of China, Manila: Asian
Development Bank.
World Bank (1990), China: Revenue Mobilization and Tax Policy, Washington, DC.
(1993), Budgetary Policies and Intergovernmental Fiscal Relations, Washington, DC.
(1994), China: Country Economic Memorandum, Macroeconomic Stability in a
Decentralized Economy, Washington, DC.
Xu, Shanda (1995), “Recent fiscal and tax reform in the People’s Republic of China,”
Fifth Symposium on Tax Policy and Reforms in the Asian and Pacific Region
(October 24–26).
Zhang, Tao, and Zou, Heng-Fu (1996), “Fiscal decentralization, public spending,
and economic growth in China,” Policy Research Department, World Bank.
Zhongli, Liu (1996), “Report on the implementation of the central and local budgets
for 1995 and on the draft central and local budgets for 1996,” paper delivered
at the Eighth National Peoples’ Congress (March 5).
3
India: intergovernmental fiscal relations in a

planned economy
M. GOVINDA RAO
There has been a resurgence of interest in fiscal decentralization in virtu-

ally every part of the world in the last three decades. Interest has not
merely been confined to constitutionally declared federations but has also
been seen in unitary countries; it spans both the ideological spectrum and
levels of development across countries. While the emergence of the
European Union as an economic entity has underlined the advantages of
having an enlarged common market with a distinct regional identity, the
economic collapse of the former Soviet Union has demonstrated the
dangers of economic and administrative centralization. In general, admin-
istrative and fiscal decentralization has been prompted by the emphasis
placed on providing efficient and responsive public services. However,
fiscal decentralization has often been warranted by market decentraliza-
tion arising from economic liberalization. In India too, in keeping with the
general trend, decentralization has been a much debated issue.
India represents a classic case of a federation with constitutional
demarcation of functions and finances between the center and the states.1
The 920 million people in the federation are spread over twenty-five states
and seven centrally administered territories (two with their own elected
governments). Separate legislative, executive, and judicial arms of govern-
ment are constituted at both central and state levels. The upper house, or
Rajya Sabha, in the federal parliament is the House of States, to which
members are elected by an electoral college from each of the states. The
Seventh Schedule to the Constitution specifies the legislative domains of
the central and state governments in terms of union, state, and concurrent
lists. The Constitution also requires the president of India to appoint a
Finance Commission every five years (or earlier) to review the finances of
I am grateful to Richard Bird, K. P. Kalirajan, Rohinton Medhora, Ric Shand, Nirvikar Singh,
and François Vaillancourt for useful discussions and comments on an earlier draft of this
chapter. Thanks are also due to other participants in the conference on “Fiscal decentraliza-
tion in developing countries” and particularly to Ms. France St-Hilaire for extremely useful
comments.
78
India 79
the center and the states and recommend devolution of taxes and grants-
in-aid for the ensuing five years.
Historical factors played an important part in the adoption in India of a
federal constitution with strong unitary features. During British rule,
administrative and fiscal centralization was a colonial necessity. At the
same time, the difficulty of administering a large country with a number
of principalities with divergent languages, cultures, and traditions did call
for some degree of decentralization. Indeed, for a period of about two
decades in British India, prior to the enactment of the Government of India
Act 1935, the system required the provinces to make a contribution to the
center. Although there were strong arguments for decentralization at the
time of independence in 1947 and even though the United Kingdom
Cabinet Mission Plan envisaged limited powers for the center in a three-
tiered federal structure, the constitution that was eventually adopted by
the Indian Republic closely followed the Government of India Act 1935,
with pronounced “quasi-federal” features. The shift probably occurred for
two reasons: first, once the Muslim majority areas opted out of India to
form Pakistan, the major reason for a loose federal structure had vanished;
second, a strong center was found desirable to safeguard against fissipar-
ous tendencies within some of the constituent units (Chelliah, 1991).
The centralization inherent in the constitutional assignment was
accentuated by the adoption of a planned development strategy.
Centralized decision-making in relation to production and distribution
activities and the disposal of resources in the “national interest” implicit in
central planning are the negation of the very principle of federalism.
Although in India economic planning has been implemented in a mixed-
economy framework, the strategy of a public-sector-dominated, heavy-
industry-based, import-substituting industrialization required the
Planning Commission to allocate resources according to the envisaged
priorities. The central government had to issue industrial licenses, allocate
credit by controlling the financial institutions, and ration the scarce foreign
exchange available among the different industries to conform to the envis-
aged priorities. It also had to introduce a host of measures to control and
regulate the private sector to ensure that those who were given the license
to produce and import did not exercise their oligopoly powers. This auto-
matically concentrated economic as well as administrative powers
(Chelliah, 1991).
Recent events, however, have called for a greater degree of decentrali-
zation to make the administrative and fiscal regimes compatible with
political and economic changes. On the political front, the decline in the
influence of the Indian National Congress, the political party which ruled
80 M. Govinda Rao
the country for over four decades, and the emergence of a number of
regional political parties in the ruling coalition at the center have shifted
the balance of political power to the regions from the center. In the eco-
nomic sphere, a compelling reason for a greater degree of fiscal decentrali-
zation has been the reassessment of the role of the state in resource
allocation. The economic crisis of 1991 highlighted the inherent weak-
nesses in the dominant development strategy. The fiscal arrangements and
institutions developed in the context of interventionist strategy need to be
reoriented to provide public services according to the diversified demands
of different regions.
This chapter reviews the federal fiscal arrangements in India with a view
to identifying the areas for reform in the emerging economic environment.
The starting point for such a review is the assignment of fiscal responsibil-
ities. The assignment of taxes and expenditures between the center and
individual states should not only be in accordance with the principle of
comparative advantage but also ensure a clear linkage between revenue-
raising and expenditure decisions at the margin. However, even the most
efficient system of assignments cannot clearly match the revenue and
expenditure powers of individual governmental units. When the fiscal
powers of different governmental units overlap, free-riding behavior
among them causes center–state and interstate fiscal disharmonies. The
discussion on assignments and fiscal overlapping in the Indian federation
leads us to the analysis of vertical and horizontal mismatches between
revenue and expenditure powers, or what are known as fiscal imbalances.
These imbalances can be a source of inequity and inefficiency and, there-
fore, intergovernmental transfer mechanisms must be designed to resolve
them. The analysis of federal fiscal arrangements in India should lead us
to identify critical areas for reform.
The assignment question
Tax and expenditure assignments – principles and practices

Assignment of functions and sources of finances among different layers of
government is an important step in the efficient organization of federal
fiscal systems. For analytical convenience, the layer-cake perspective on
federalism in the Musgravian tripartite division of governmental func-
tions assigns stabilization and redistribution mainly to the central govern-
ment and the allocation function is shared among the hierarchical layers
depending upon their comparative advantage in carrying out different
functions.2 The “decentralization theorem” (Oates, 1972) suggests that, so
India 81
long as there are no scale economies, the subcentral provision of public ser-
vices results in welfare gains. The gains will be greater, the larger the vari-
ation in the preferences for public services (Oates, 1977b). The above
principle implies that progressive and mobile tax bases should be assigned
primarily to the central government, and the subcentral units should raise
revenues mainly through user charges, benefit taxes, and taxes on rela-
tively less mobile bases (Musgrave, 1983). At the same time, to reap welfare
gains, the lower-level jurisdictions would have to provide all non-national
public services, and this would create a serious asymmetry between
revenue sources and expenditure functions, or what is termed “vertical
fiscal imbalance” (Hunter, 1977).
The above problem can be alternatively stated in terms of the literature
on “competitive federalism.”3 Successful interjurisdictional competition
requires that the assignment of functions should be done according to
comparative advantage. Given that many tax bases are mobile across
jurisdictions, the more senior governments have a comparative advantage
in levying taxes since they can more effectively control “free-riding” (or
enjoying the benefits of public services without making commensurate
payments). Free-riding can take the form of tax avoidance and evasion,
interstate tax exportation, or benefit spillovers. The lower-level govern-
ments, on the other hand, have a comparative advantage in reducing the
welfare costs of providing public services. For, at lower levels, the mis-
match between goods supplied and demanded would be smaller and the
bundling of public services more flexible.4 Welfare costs can be reduced
either by moving to the jurisdictions providing the preferred bundle of
public services (“exit”) or by consumers influencing policies in order to
obtain the preferred pattern of public services (“voice”). Thus, assign-
ments made according to comparative advantage would result in revenue
concentration and expenditure decentralization. Intergovernmental trans-
fers are necessary to resolve this problem of vertical imbalance (Breton and
Fraschini, 1992).
Analysis of the actual fiscal assignments in different federations,
however, brings out three important features. First, vertical fiscal imbal-
ance is a feature in all federations.5 This has occurred because the actual
assignments of tax and expenditure powers are made broadly according to
the principle of comparative advantage. Of course, we cannot conclude
from this that greater vertical fiscal imbalance necessarily denotes greater
efficiency in assignments. Second, an assignment of powers which mini-
mizes concurrence or overlap can only be made in a de jure sense. De facto
overlapping of tax and expenditure powers between different jurisdictions
is unavoidable. Third, concurrency in tax and expenditure powers is not
82 M. Govinda Rao
necessarily undesirable if there are mechanisms to coordinate the policy

actions of different governmental units, and the benefit of coordination
exceeds its cost.
Tax and expenditure assignments in India

The tax and expenditure powers of the central and the state governments
are specified in the Seventh Schedule to the Constitution. The functions
required to maintain macroeconomic stability, international relations,
and activities having significant scale economies have been assigned
exclusively to the center or have to be carried out concurrently with the
states. The functions which have a statewide jurisdiction are assigned to
the states. Most broad-based and progressive tax bases have been
assigned to the center. The center also has residual tax powers. A number
of taxes have been assigned to the states as well, but from the point of
view of revenue productivity, only the tax on the sale and purchase of
goods is important. The states can borrow from the central government.
They have the powers to borrow from the market as well, but if a state is
indebted to the central government, the borrowing has to be approved by
the center.
Notably, the tax powers are assigned on the basis of the principle of
separation and the tax bases are assigned exclusively either to the center
or to the states. However, the separation is only in the legal sense, not in
the economic one, and this gives rise to anomalous situations. Thus, taxes
on production (excise duty) can be levied by the center, whereas the tax on
the sale of goods is leviable by the states. Similarly, the taxes on agricul-
tural incomes and wealth are leviable by the states, whereas those on non-
agricultural incomes and wealth are leviable by the center. The states have
found taxing agricultural incomes politically infeasible besides being
administratively difficult. At the same time, the separation of the tax base
has opened up a floodgate for both avoidance and evasion of the personal
income tax.6
The Constitution also recognizes that the states’ tax powers are inade-
quate to meet their expenditure needs and, therefore, provides for the
sharing of revenues from personal income tax (Article 270) and Union
excise duty (Article 272). States in need of additional assistance can be also
be given grants-in-aid (Article 275). The tax devolution and grants-in-aid
are to be determined by the Finance Commission, an independent body
appointed by the President every five years (Article 280).
The shares of the central and state governments in revenues and expen-
ditures, summarized in tables 3.1 and 3.2, bring out their relative roles. It
India 83
Table 3.1. India: revenue receipts of the central and state governments
(percentages)
Revenue share Revenue share Revenue share

1985–86 1990–91 1994–95a
Percentage
Item of revenue Center States Center States Center States of total
Tax revenue 49.0 51.0 49.1 50.9 45.5 54.5 82.7

Exclusive central taxes 100.0 — 100.0 — 100.0 — 24.0
Corporation tax 100.0 — 100.0 — 100.0 — 7.6
Custom duties 100.0 — 100.0 — 100.0 — 15.2
Other 100.0 — 100.0 — 100.0 — 1.3
Exclusive state taxes — 100.0 — 100.0 — 100.0 31.2
State excise duties — 100.0 — 100.0 — 100.0 4.4
Sales taxes — 100.0 — 100.0 — 100.0 16.4
Taxes on transport — 100.0 — 100.0 — 100.0 2.6
Other — 100.0 — 100.0 — 100.0 7.8
Shared taxes 51.6 48.4 51.4 48.6 49.1 51.9 27.4
Personal income tax 26.5 73.5 23.4 76.6 22.2 77.8 6.3
Union excise duty 56.6 43.4 57.5 42.5 55.9 44.1 21.1
Non-tax revenue 62.1 37.9 54.3 45.7 80.8 19.2 17.3
Net contribution from ⫺875.9 975.9 ⫺288.1 388.1 119.2 ⫺19.2 2.0
public enterprises
Administrative receipts 47.1 79.2 33.8 66.2 18.8 81.2 8.9
Interest receiptsb 66.6 33.4 59.6 40.4 161.2 ⫺61.2 5.8
External grants 100.0 — 100.0 — 100.0 — 0.7
Grants to states — 100.0 — 100.0 — 100.0 —
Total revenue accrual 38.2 61.8 37.7 62.3 38.4 61.6 100.0
Total revenue collections 65.6 34.4 63.9 36.1 65.5 34.5 100.0
Notes:
a
Revised estimates.
b
Netted for the interest paid to the central government.
Source: Based on data from Ministry of Finance, India.
is seen that the states on average raise about 35 percent of current revenues
and disburse about 57 percent of expenditures. The revenues derived from
exclusive central taxes constitute about 24 percent of total revenue; those
from exclusive state taxes 31 percent, the sharable sources constitute 27
percent, and the remaining 17 percent is from non-tax revenues. The major
taxes levied exclusively by the center are customs duty (15 percent of total
revenue) and corporation tax (8 percent). Among the state taxes, the
revenue from sales tax constitutes about 16 percent. Other state taxes indi-
vidually contribute less than 6 percent of the total revenue.
On the other hand, the share of the states in spending is about 57 percent,
Table 3.2. Share of state governments in total expenditures (percentages)
1985–86 1990–91 1994–95a Percentage

of total
Item of expenditure Current Capital Total Current Capital Total Current Capital Total expenditure
Interest payment 34.6 0.0 34.6 35.5 0.0 35.5 37.2 0.0 37.2 20.1
Defense 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 9.1
Administrative service 85.2 0.8 78.0 76.4 1.8 73.4 66.2 14.7 65.8 13.7
Social and community services, of which 94.8 67.0 92.7 78.9 74.1 78.7 86.4 71.2 85.4 21.5
Education 84.8 79.5 84.7 84.5 61.7 83.9 86.1 43.4 84.9 11.8
Medical and health 92.5 94.8 92.8 90.3 95.3 90.7 91.4 96.1 92.0 4.3
Family welfare 93.4 90.4 93.1 92.2 100.0 92.7 93.5 100.0 93.8 0.6
Others 98.1 40.7 88.2 61.1 64.0 61.4 81.6 65.2 79.9 4.8
Economic services, of which 78.1 46.3 62.9 50.2 53.1 51.1 56.8 66.0 59.5 27.4
Agriculture and allied servicesb 99.9 82.1 96.7 77.8 94.8 78.6 45.8 72.4 46.7 12.3
Industry and minerals 36.7 9.8 18.1 40.7 44.1 41.9 50.7 47.4 49.7 1.9
Power, irrigation and flood control 94.7 65.4 73.9 86.2 62.9 69.1 91.2 77.9 83.7 7.2
Transport and communication 68.3 68.3 68.3 70.4 32.1 47.3 67.6 52.7 60.1 3.8
Others 24.9 18.0 23.9 16.6 51.3 19.7 66.1 49.2 58.9 2.2
Other 80.0 14.7 33.2 57.2 0.0 57.2 85.9 0.0 85.8 5.1
Loans and advances 0.0 51.7 51.7 0.0 51.1 51.1 0.0 62.9 62.9 3.1
Totalc 55.2 43.0 52.1 55.0 44.5 52.9 58.1 53.7 57.4 100.0
Notes:
a
Revised estimates.
b
Includes food and fertilizer subsidies.
c
Excludes appropriation for reduction and avoidance of debt.
India 85
and in current expenditure, 58 percent. However, state control over spend-

ing is lower than indicated by these figures because about 15 percent of
total expenditure is incurred on central sector and centrally sponsored
schemes, which are the specific-purpose transfer schemes. State expendi-
tures on these items actually increased from about 7 percent of the total in
1985–86 to about 13 percent in 1994–95.
The pattern of expenditures shown in table 3.2 indicates that the central
government plays a major role in defense and industrial promotion. On
the other hand, the states have a high share of expenditures on internal
security, law and order, all social services, and economic services like agri-
culture, animal husbandry, forestry, fisheries, irrigation and power, and
public works. The states’ share in expenditure on administrative services
is about two-thirds; on social services they spend over 85 percent and on
economic services almost 60 percent.
Over the last decade, while the share of the states in raising revenues has
remained stable, their expenditure share has increased by about five per-
centage points, particularly since 1990–91. This has occurred because fiscal
reforms initiated in 1991 have led to deceleration in central government
expenditures, but have not reduced central transfers to states; conse-
quently, state expenditures have continued to grow as in the past to
increase their share in the total. Interestingly, the expenditure share has
increased for both current and capital expenditures.
This analysis of constitutional assignment in India brings out the fol-
lowing:
(1) The Constitution exhibits a clear centripetal bias in the distribution of

tax and regulatory powers. In addition to the expenditure functions
assigned, the center can also influence the expenditure decisions of the
states. The assignment of most of the broad-based taxes and residuary tax
powers to the center, its overriding powers in regard to the functions in the
concurrent list, and domination through economic planning and control of
virtually the entire financial sector are only some instances of the center’s
dominance in the economic sphere.
(2) The assignment of tax powers follows the principle of “separation,”

in contrast to that of “concurrence” followed in federations like the United
States and Canada. This, however, cannot avoid de facto overlapping. The
problem is particularly severe in the case of indirect taxes.
(3) The Constitution allows the levy of certain taxes which create severe
impediments to the interstate movement of goods. The levy of tax on inter-
state sale of goods by the exporting state (subject to a ceiling rate of
86 M. Govinda Rao
4 percent), besides creating resource distortions, has caused perverse trans-

fer of resources from the poorer consuming states to the more affluent pro-
ducing states. Similarly, the states can levy a tax on the entry of goods into
a local area for consumption, use, or sale (octroi). This has created impedi-
ments to the free movement of goods and also erected a number of tariff
zones coinciding with the localities in the country (Rao, 1993).
Tax and expenditure overlapping
Intergovernmental competition and fiscal disharmony

The pursuit of common objectives by different levels of government and
the spillover effect of policies at one level on another can create conflict in
jurisdictions. Often, the fiscal policies of subnational governments may
conflict with that of the central government (Perloff, 1985). The subcentral
governments may have their own ideas on redistributive and employment
policies which the central fiscal action has to take into account. Besides,
policies implemented at one level may spill over to affect the effectiveness
of those at another level. An expansionary fiscal policy by the central
government can enhance the unit cost and thus constrain the states’ abil-
ities to deliver public services. An increase in salary levels by the central
government may itself have a demonstration effect on the states. Similarly,
the central government may compress its own expenditures simply by
reducing transfers to the states. Increases in the administered prices of
central public enterprises could enhance the unit cost at the state level and
vice-versa. A reduction in central subsidies may force the states to intro-
duce them for electoral reasons. Similarly, increases in the excise duty by
the center may cause the states to reduce taxes on these items.7 These are
just a few examples of the vertical interdependence of policies.
The most glaring instance of vertical fiscal overlapping is seen in the
taxation of the same base by different levels of government. The
Constitution assigns separate tax powers to the central and state govern-
ments, but, as mentioned earlier, the separation is done in the legal (admin-
istrative), not the economic, sense. Thus, while the center can levy tax on
the production of goods (excise duty), the states can levy sales taxes on the
sale of these goods. The states can also allow an urban local body to levy
octroi on the same goods if they are transported into its jurisdiction.
The adverse allocative implications of the prevailing system of tax
assignments are notable. First, the splitting of the power to levy taxes on
income and capital between the central and the state governments on the
basis of whether they are derived from agricultural or non-agricultural
India 87
sources has led to distortions, tax avoidance, and tax evasion, besides vio-
lating the principle of horizontal equity (Chelliah, 1993). Second, although
in a legal sense, the central excises and state sales taxes and octroi are separ-
ate levies, these fall virtually on the same base causing serious vertical tax
overlapping. The levy of tax on tax and margins on taxes have created a
divergence between producer and consumer prices greater than the tax
element.
The Indian experience brings out three important lessons. First, tax
assignment should not be done merely on legal considerations; economic
consequences of such assignments must be taken into account. Second,
avoidance of concurrency in a de jure sense does not prevent de facto over-
lapping. Third, overlapping by itself cannot be considered undesirable.
What is important is that the tax policies of different levels should be prop-
erly coordinated and harmonized to minimize distortions. In the United
States and Canada, even when the federal and state (provincial) govern-
ments enjoy concurrent tax powers, they have been able to achieve a
greater degree of coordination and harmonization than India (Rao and
Vaillancourt, 1994).
Interstate competition and fiscal disharmony

It is argued that efficiency under fiscal decentralization is enhanced
because of the matching of the supply of public services with demand. The
decentralization model, however, does not say much about production
efficiency. Nevertheless, efficient and stable interstate competition in pro-
viding public services can help the governments to operate on the produc-
tion frontier and even help them shift the frontier to the right.8 However,
the two important preconditions for efficient intergovernmental competi-
tion are “competitive equality” and “cost-benefit appropriability” (Breton,
1987). Competitive equality ensures that the larger/stronger units do not
dominate or coerce smaller/weaker units, and this has to be enforced
through either regional policies or intergovernmental grants. Cost-benefit
appropriability ensures that the residents of a state are not able to consume
public services without making commensurate payments.
Thus, in ensuring fair and efficient competition among the states, the
central government has an important monitoring role. First, it should acti-
vate the process of competition by ensuring the free flow of factors and
products across the country. Second, it should ensure “competitive equal-
ity” among the states through regional policies and by giving intergovern-
mental transfers to offset the fiscal disabilities of the poorer states. Third,
it should initiate the process of harmonization in tax policies among the
88 M. Govinda Rao
states to minimize interstate tax exportation and tax competition. Finally,

it should, through cost-sharing programs, ensure optimal standards of ser-
vices having a high degree of interstate spillovers.
A number of studies have examined the welfare implications of inter-
jurisdictional tax competition but consensus on its desirability has yet to
emerge. The conventional view is that interjurisdictional competition is a
source of distortion. According to this view, cut-throat competition by
states to attract trade and industry into their jurisdictions at the cost of other
states can distort the allocation of resources (Netzer, 1991) and the race to
the bottom in reducing tax rates would result in the less than optimal pro-
vision of public services. The alternative view is that interjurisdictional
competition is a beneficent force. This view has found particular favor
among the “leviathan” theorists who argue that such a competition can
work as an effective constraint on the government’s monopoly power to
maximize revenue (Brennan and Buchanan, 1980). Oates and Schwab
(1988) also show that when communities are homogeneous, where the costs
and benefits are clearly perceived, and where public decision reflects the
preferences of the residents of respective jurisdictions, interjurisdictional
competition is efficiency-enhancing. However, even in this model, if the
jurisdictions are constrained to tax capital for want of more efficient tax
instruments and if public decisions deviate from the will of the electorate,
tax competition may not lead to efficient outcomes.
It is difficult to find a situation where the assumptions of the model of
benign competition will be satisfied in a developing country. Acute inter-
state inequalities in the level of development endow different states with
differing degrees of competitive strength. The delinking of tax and expen-
diture decisions prevents the costs and the benefits of public decisions
from being clearly perceived. Free-riding by the states can cause significant
interstate tax spillovers.
When sales taxes predominate in state tax systems and when the states
levy origin-based taxes, additional problems are created. First, the strategy
of imposing different tax rates on items consumed by residents and on
those exported to non-residents increases rate differentiation within each
of the states. Second, tax competition can cause wide differences between
the states’ tax systems, depending on the structure of production and
consumption and the type of strategy followed to maximize revenue and
to attract capital. Third, origin-based consumption taxes which tax inputs
and capital goods can result in cascading and may cause significant inter-
state tax exportation with adverse effects on both equity and efficiency. In
particular, the levying of tax on interstate sales enables interstate tax
exportation. Besides creating impediments to the free movement of goods
India 89
across the country, the interstate sales tax causes significant resource trans-
fers from the poorer to the richer states (Rao and Vaillancourt, 1994).
Thus, interstate tax competition results from attempts to indulge in free-
riding behavior. The competition may take various forms:
• reducing nominal tax rates to maximize revenue by attracting cross-
border purchases;
• levying selectively lower nominal tax rates and giving incentives to new
industries to attract capital into their jurisdictions;
• adopting strategies to export the tax burden to non-residents by choos-
ing appropriate tax systems and tax rates on both factors and products.
The strategy adopted by the states to divert trade into their jurisdictions
and to export the tax burden to non-residents has been one of the causes
of excessive differentiation in tax rates. The prevailing sales tax systems
have a mix of first point (manufacturers), last point (retail), and multi-
point (cumulative) taxes. In addition, to raise more revenue, the states have
resorted to levying surcharges, turnover taxes, and additional sales taxes,
resulting in acute differentiation in effective tax rates and making the tax
system both complex and irrational. The attempt to attract capital through
liberal tax concessions to new industries has further added to the complex-
ity. The taxation of inputs and capital goods, in addition, has resulted in
cascading.
Fiscal imbalances: trends and issues

Fiscal imbalance refers to the mismatch between the revenue-raising
capacity and the expenditure need of different governmental units. Such
imbalances can arise vertically, between different levels of government, or
horizontally, between different jurisdictions. In most federations, given
that the central government has a comparative advantage in raising rev-
enues and the states are relatively better placed to deliver public services,
vertical imbalance is implicit in the assignment itself. The interjurisdic-
tional differences in income due to various historical and institutional
factors as well as variations in resource endowments create horizontal
imbalances.
Vertical fiscal imbalance

In India, the consequence of constitutional assignment as well as develop-
ments over the years has been to create a high degree of fiscal centraliza-
tion and vertical fiscal imbalance. As shown in table 3.3, the state
Table 3.3. Trends in vertical fiscal imbalance
States’ own current States’ current expenditure States’ own current States expenditurea States’ own revenue
revenues as percentage as percentage of total revenues as percentage of as percentage of as percentage of states’
Period of total current revenues current expenditure states’ current expenditure total expenditurea total expenditurea
(1) (2) (3) (4) (5) (6)
1955–56 41.2 59.0 68.9 61.7 48.5

1960–61 36.6 59.9 63.9 56.8 45.4
1965–66 32.6 55.6 63.5 53.3 42.7
1970–71 35.5 60.2 60.6 53.9 50.3
1975–76 33.5 55.1 70.4 47.6 54.2
1980–81 35.6 59.6 60.1 56.0 43.7
1985–86 35.5 56.0 57.7 52.6 45.5
1990–91 36.6 55.2 53.5 53.1 44.8
1991–92 37.6 58.3 54.8 56.2 47.5
1992–93 36.6 57.9 53.7 55.1 45.8
1993–94 39.6 57.5 55.3 54.7 47.8
1994–95b 41.1 58.3 57.1 57.5 48.9
Notes:
a
Current and capital expenditure
b
Revised estimates
India 91
Table 3.4. Vertical fiscal imbalance in selected countries
Coefficient of
vertical imbalance
Country Year ending 1988 1989 1990
Australia June 30 0.51 0.46 0.47

Brazil December 31 0.24 0.20 0.31
Canada March 31 0.24 0.23 0.21
West Germany December 31 0.20 0.17 0.20
India March 31 0.54 0.55 0.55
United States September 30 0.14 0.14 0.15
Note: The degree of vertical imbalance refers to only state/

provincial/regional governments and has been computed as
(1 ⫺R/E ), where
R⫽total revenues including capital receipts of the states
but excluding intergovernmental transfers
E⫽total expenditures (current and capital) of the states.
Source: International Monetary Fund (1993), Government Finance
Statistics.
governments in 1994–95 collected only 41 percent of total revenues

(column 2), but their share in total current expenditure was 58 percent
(column 3). From the revenue sources assigned to them, they could finance
only about 57 percent of their current expenditure (column 4) and 49
percent of their total expenditure (column 6). Thus, although the states’
revenue collections were only about 40 percent, they had to depend on
transfers and loans amounting to over 26 percent of total revenues to meet
their spending requirements. It is also seen that for India, the states’ depen-
dence on the center for financing their expenditure, or the coefficient of
vertical fiscal imbalance, was the highest among the federations compared
in table 3.4. The revenue from own sources contributed only about 45
percent of expenditure and 55 percent of their expenditure was financed
through central transfers or borrowing.
The relative shares of the central and state governments in revenues and
expenditures presented in table 3.3 bring out the trends in fiscal centraliza-
tion over the years from 1955–56. Interestingly, even when the states’ rev-
enues grew at a rate faster than those of the center, their fiscal dependence
on the latter showed an increase. The states’ share in raising revenue has
increased marginally, particularly since the mid-1980s. However, their
share in total expenditure increased at a faster rate, by almost 10 percent-
age points since the mid-1970s (column 5 in table 3.3) indicating increas-
ing fiscal dependency. Thus, the states’ share in current expenditure
remained more or less at the same level, but the faster growth of current
92 M. Govinda Rao
expenditures than revenues resulted in a declining ratio of revenues to

current expenditure. At the same time, even the broad constancy in the
share of state government expenditure in the total indicates an increase in
expenditure centralization. This is because the proportion of specific-
purpose transfers for various central sector and centrally sponsored
schemes in total transfers increased from 12 percent in the Fifth Plan (1969-
74) to over 18 percent in the Seventh Plan (1985–90); presently, almost 15
percent of the states’ expenditures are on these schemes over which the
states have little control.
Horizontal fiscal imbalance

An important feature of Indian fiscal federalism is the wide interstate
differences in revenue capacity and, consequently, per capita expendi-
tures. There are fifteen relatively homogeneous general category states, but
even these have wide differences in size, revenue-raising capacities,
efforts, expenditure levels and fiscal dependence on the center. In addition,
in terms of economic characteristics and endowments, the ten mountain-
ous states of the north and the north-east differ markedly from the rest and
therefore are considered “special category” states.
The differences in per capita revenue and expenditure among the states
shown in table 3.5 brings out several important features. First, there are wide
interstate variations in revenues both in per capita terms and in the ratio of
revenue to state domestic product (SDP). Second, the variations indicate
differences in revenue capacity and, partly, differences in states’ revenue
efforts. Third, the tax–SDP ratios in the special category states are lower than
in the general category states even when their per capita SDP is higher. This
is partly because in these states there is not much production activity and
government administration is the major determinant of the SDP. Further, the
size of the tax base is smaller than indicated by the SDP, because a significant
proportion of government spending spills over the jurisdictions. Fourth,
although the revenue bases in the special category states are low, their
average per capita current expenditure is higher than not only the all-state
average but also the average of high-income states.9 Fifth, in the case of
general category states, the fiscal dependence on the center is not only high
but varies inversely with the per capita income. Per capita expenditures in
high-income states are higher than the all-state average by 45 percent and
those in low-income states are lower by 16 percent. Nevertheless, the system
has led to significant equalization as may be seen from the fact that while the
per capita taxes in low-income states are about a third of those in high-
income ones, per capita expenditure in the former is close to 60 percent.
Table 3.5. Revenues and expenditures of the states, 1993–94
Own revenue as
Per capita SDP Per capita own Own revenue as Per capita current percentage of current
Major states (rupees)a revenue (rupees) percentage of SDPa expenditure (rupees) expenditure
High-income states 10 211 1 278.63 12.5 1 680.87 76.1

Gujarat 7 600 1 233.82 16.2 1 601.92 77.0
Goa 11 658 2 632.4 22.6 3 499.84 75.2
Haryana 10 359 1 680.61 16.2 1 951.10 86.1
Maharashtra 10 984 1 213.82 11.1 1 578.65 76.8
Punjab 12 319 1 214.61 9.9 1 915.74 63.4
Middle-income states 6 661 765.59 11.5 1 238.84 61.8
Andhra Pradesh 6 651 744.43 11.2 1 151.37 64.7
Karnataka 7 029 970.42 13.9 1 325.3 73.2
Kerala 6 242 884.13 14.2 1 422.7 62.1
Tamil Nadu 7 352 960.32 13.1 1 527.72 62.9
West Bengal 6 055 447.78 7.4 959.89 46.7
Low-income states 4 674 438.97 9.4 969.79 45.3
Bihar 3 650 288.13 7.9 800.22 36.0
Madhya Pradesh 5 485 585.01 10.7 1 077.75 54.3
Orissa 4 726 384.07 8.1 1 048.61 36.6
Rajasthan 5 220 673.13 12.9 1 267.67 53.1
Uttar Pradesh 4 744 401.48 8.5 911.47 44.1
Special category states 5 607 437.56 7.8 1 939.48 22.6
Arunachal Pradesh 7 904 964.11 12.2 4 330.91 22.3
Assam 5 916 530.94 6.9 1 223.0 33.2
Himachal Pradesh 6 519 693.16 10.6 2 489.53 27.8
Jammu and Kashmir 4 244 439.81 10.4 2 162.23 20.3
Manipur 5 362 238.37 4.5 2 243.74 10.6
Meghalaya 5 519 399.53 7.2 2 528.58 15.8
Mizoram 6 599 462.84 6.7 5 399.42 8.6
Nagaland 5 870b 311.42 4.8 5 015.25 6.2
Sikkim 5 416b 868.70 15.6 3 916.40 22.2
Tripura 3 781 202.16 5.1 2 089.88 9.7
All states 6 287 653.5 10.4 1 158.24 56.4
Notes:
a
SDP=State Domestic Product.
b
Estimated.
Sources: Reserve Bank of India Bulletin (December 1995) based on data from Ministry of Finance, India.
94 M. Govinda Rao
Table 3.6. Coefficients of variation in per capita state government expenditures

(percentages)
Expenditure/SDP items 1975–76 1980–81 1985–86 1990–91 1993–94
General administration 23.4 21.9 25.0 29.3 31.1

Education 32.9 31.7 26.4 20.3 26.9
Health 28.8 24.3 27.6 25.8 22.6
Total social services 35.2 29.6 31.1 26.0 24.7
Total economic services 37.4 34.0 41.0 36.7 31.6
Total current expenditure 26.0 23.5 24.8 23.2 27.1
Total capital expenditure 38.7 28.1 54.3 40.2 43.1
Total expenditure 26.6 23.0 28.3 24.2 28.1
Total per capita net state 29.9 31.7 31.7 34.1 36.2
domestic product
Sources: Calculated from budget documents of state governments, various years.
Interstate disparities in India, even among the relatively more homo-

geneous general category states, are not only high but have also shown an
increasing trend. In 1980–81, the per capita SDP in the richest state, Punjab
(Rs 2,674), was about 2.9 times that of the poorest, Bihar (Rs 919). By
1993–94, this difference had increased to 3.4 times, with per capita SDPs in
the two states of Rs 12,319 and Rs 3,650 respectively. The coefficients of vari-
ation in states’ per capita own revenue as well as in per capita SDP
increased over the period 1975–76 to 1990–91 (table 3.6). As interstate differ-
ences in the ability to raise revenues increased over the years, and as federal
transfers did not entirely offset the fiscal disabilities of the poorer states, the
coefficient of variation in expenditures also increased over the time period.
Intergovernmental transfers in India
Economic rationale for transfers

Intergovernmental transfers have been employed to fulfill a variety of
objectives, and the design of a transfer scheme depends on the purpose for
which transfers are given. In the literature, federal transfers are recom-
mended for:
closing the fiscal gap
equalization
spillovers and to foster the provision of “merit” goods
Sometimes, the center also gives transfers to allow agency functions to
be performed.
India 95
Closing the fiscal gap

An important reason for making transfers is to enable sub-central govern-
ments to undertake their functions satisfactorily when the revenues
assigned to them are found to be inadequate. This is because the center has
a comparative advantage in raising revenues, while the States have one in
spending. The resulting vertical fiscal imbalance would have to be offset
through a system of transfers from the center to states.10
Equalization
The imbalance between revenue capacity and expenditure need varies
across states depending on the size of their tax base, the size and the
composition of their population, and other factors affecting the cost of pro-
viding public services. The richer states, owing to their higher revenue
capacity, can provide better standards of public services. To offset this
fiscal disadvantage, equalizing transfers are required.
The argument for equalizing transfers is based on the horizontal equity
grounds initially advanced by Buchanan (1950) and later developed by
Boadway and Flatters (1982). With comprehensive income as the index of
well-being, it has been argued that the federal income tax as presently
structured cannot ensure horizontal equity as its base does not take into
account the redistributive effect of states’ fiscal operations. The states’
fiscal operations cannot be distributionally neutral except in the unlikely
case of their levying benefit taxes. When the quasi-public services pro-
vided by them are financed by resource rents or source-based taxes as
against residence-based taxes, the net fiscal benefits (NFBs) will systemati-
cally vary. The residents in the resource-rich (high-income) regions will
have higher NFBs and their higher public consumption will not be
included in determining the tax base of the central government.
Boadway and Flatters (1982) define horizontal equity in two alternative
ways. According to the broad view, the fiscal system should be equitable
nationwide vis-à-vis the actions of all governments. Two persons equally
well off before central and state actions must also be so afterwards. To fulfill
this concept of horizontal equity, it is necessary to give transfers so that each
province is enabled to provide the same level of public services at a given
tax rate (as in a unitary state). In contrast, the narrow view of horizontal
equity takes the level of real incomes attained by the individuals after a
state’s budgetary operation as the starting point and the central fiscal action
will be directed to ensuring horizontal equity after the state’s fiscal system
has been established. The central budget need not offset the inequalities
introduced by the operation of the state budgets per se, but takes the
income-distributional effects of the states’ fiscal operations as a given.
96 M. Govinda Rao
Transfers to correct spillovers

When there is no perfect “mapping,” the provision of public services by
subcentral governments may spill over the jurisdictions and such external-
ities may result in the non-optimal provision of public services. A Pigovian
subsidy is required to “set the prices right.” To be cost-effective, specific-
purpose transfers made to the states to ensure optimal provision of public
services require matching contributions from them.
The design of intergovernmental transfers

General-purpose transfers, as mentioned earlier, are given to ensure hori-
zontal equity. As the objective of these transfers is to enable the subcentral
governments to provide a given level of public services at a given tax rate,
the transfers should offset the fiscal disadvantages arising from lower
revenue capacity and higher unit cost of providing public services.
This is achieved by giving unconditional transfers equivalent to the
“need–revenue” gap (Bradbury et al., 1984). The “need–revenue” gap mea-
sures the difference between what a state ought to spend to provide speci-
fied levels of public services and the revenue it can raise at a given
standard level of tax effort.
Specific-purpose transfers, on the other hand, are intended to compen-
sate spillovers or are given for merit good reasons to ensure optimal pro-
vision of public services by the states. The transfer system, therefore,
should be specific-purpose and open-ended with matching ratios varying
with the size of spillovers. As the responsiveness of the states to a given
matching rate could vary with their level of income, equalizing matching
ratios are also recommended (Feldstein, 1975).
Thus, in an ideal system, there should be an optimal combination of
general- and specific-purpose transfers. General-purpose transfers would
enable all the states to provide a given normative standard of public ser-
vices for a given tax effort. The specific-purpose transfers would ensure a
given standard of outlay on the aided services.
The Indian experience

Interregional transfer of resources is achieved through a number of policy
interventions, both explicit and implicit. This is particularly true in a
planned economy like India. The important interstate resource transfer
mechanisms besides intergovernmental fiscal transfers include targeting
investments in specific locations by the central government (regional poli-
cies), lending to the states at below market rates of interest, allocating a
India 97
certain proportion of the resources of the banking system and financial

institutions to different states,11 lending to the states at below-market inter-
est rates and making subsidized “priority sector” loan allocations.
Resource transfers may result from interstate tax exportation as well.
We have already referred to the inequitable transfer of resources due to
interstate tax exportation (Rao and Vaillancourt, 1994). The analysis of
regional policies and loan allocations to the state governments and the
private sector similarly indicates that the implicit transfer of resources
through these means are sizable and inequitable. Table 3.7 presents
population shares as well as the interstate distribution of outstanding
loans from the government and the market, the priority sector credit
allocations by the banking system, assistance by All India Financial
Institutions (AFIs), and investments in central government enterprises in
different states. The rates of interest on these loans are subsidized and
there has also been a periodic rescheduling and write-off of central loans
as well. The analysis shows that with respect to each of these sources the
low-income states have received allocations much lower than their
population percentages. The high-income states with a population share
of about 19 percent received 24 percent of the outstanding loans and their
share of central government loans alone was about 27 percent. They also
received 35 percent of the priority sector credit allocations and 43 percent
of the assistance given by AFIs; even the share of investments in central
enterprises was 24 percent. The allocation from these sources to the low-
income states, in contrast, was significantly lower than their share of
population. Thus, the horizontal fiscal imbalance is exacerbated by
inequitable financial flows on account of other policies.
A notable feature of intergovernmental transfers in India is the existence
of multiple channels of transfer from the central government to the states.
The Finance Commission every five years makes an assessment of the
fiscal resources and needs of the center and individual states and recom-
mends the shares of personal income tax and Union excise duty and
grants-in-aid to the states. However, with development planning gaining
emphasis, the Planning Commission became a major dispenser of funds to
the states by way of grants and loans to meet their plan requirements. In
addition to these two channels, various central ministries give specific-
purpose transfers with or without matching requirements.
The trends in the relative shares of the three channels of central trans-
fers12 to states since the Fourth Five-Year Plan, as shown in table 3.8, show
some interesting features. First, the share of statutory transfers in the total
declined from 65 percent during the Fourth Plan (1969–74) to a little over
60 percent during the Seventh Plan. Second, the proportion of formula-
Table 3.7. Interstate redistribution through regional, loan, and credit policies (percentages)
States’ share of outstanding liabilities, Share in bank credit, Share in capital

March 31, 1994 March 31, 1994 Share in AFIs stock of central
Population assistancea enterprises
States share (1995) Internal debt Central loans Other debt Total Priority sector Total (1994–95) (1990–91)
High-income states 18.7 14.4 27.7 18.8 23.6 35.4 29.4 42.8 24.1
Goa 0.1 0.3 0.8 0.3 0.6 0.3 0.3 0.7 negl.
Gujarat 4.9 3.4 6.2 3.4 5.2 4.9 5.4 13.2 5.0
Haryana 2.0 2.5 2.2 4.2 2.6 1.5 2.8 2.6 0.8
Maharashtra 9.4 5.4 10.3 6.3 8.7 25.3 15.8 23.5 17.6
Punjab 2.4 2.9 8.2 4.6 6.6 3.3 5.1 2.8 0.7
Middle-income states 30.9 34.4 28.9 26.2 29.5 33.4 37.3 29.7 25.3
Andhra Pradesh 7.8 9.3 7.2 4.1 7.1 6.8 8.9 7.5 10.1
Karnataka 5.3 4.9 4.6 4.8 4.7 6.0 6.8 6.1 2.3
Kerala 3.4 5.8 3.2 8.4 4.6 3.4 4.3 2.0 1.5
Tamil Nadu 6.4 7.1 5.6 5.4 5.9 10.3 11.1 9.3 5.2
West Bengal 8.0 7.2 8.3 3.4 7.2 6.9 6.2 4.8 6.3
Low-income states 43.6 43.1 34.5 46.9 38.3 14.4 24.2 21.9 34.1
Bihar 10.3 8.7 7.3 10.5 8.1 2.3 4.1 2.1 8.5
Madhya Pradesh 7.9 4.7 4.6 11.4 5.8 3.2 4.9 5.2 11.1
Orissa 3.7 6.3 3.3 5.8 4.3 1.3 2.2 2.3 5.3
Rajasthan 5.3 6.2 4.7 7.9 5.5 2.1 3.3 4.4 1.5
Uttar Pradesh 16.4 17.3 14.6 11.3 14.5 5.5 9.7 7.9 7.8
Special category states 5.4 8.1 8.2 8.1 8.1 1.9 2.8 2.2 7.5
Union territories 1.4 0.0 0.7 0.0 0.5 14.9 6.4 3.4 9.0b
All states 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Notes:
negl.=negligible.
a
All-India Financial Institutions
b
Unallocated 4.9 percent
Sources: Report on Currency and Finance, 1994–95, Reserve Bank of India; Public Enterprise Survey, Ministry of Industry, Government of India.
Table 3.8. Current transfers from the center to the states (Rs. billion)
Finance Commission transfers Plan grants
Plan periods/years Tax devolution Grants Total State plan schemes Central schemes Total Other grants Total
Fourth Plan (1969–74) 45.6 8.6 54.2 10.8 9.7 20.5 9.3 83.9
(54.2) (10.2) (64.6) (12.8) (11.6) (24.4) (11.0) (100.0)
Fifth Plan (1974–79) 82.7 28.2 110.9 29.1 19.3 48.4 5.4 164.7
(50.2) (17.1) (67.3) (17.7) (11.7) (29.4) (3.3) (100.0)
Sixth Plan (1980–85) 237.3 21.4 258.7 73.8 69.0 142.8 15.1 416.5
(57.0) (5.1) (62.1) (17.7) (16.6) (34.3) (3.6) (100.0)
Seventh Plan (1985–90) 494.6 62.7 557.4 155.2 165.1 320.3 35.2 913.1
(54.2) (6.9) (61.0) (17.1) (18.0) (35.1) (3.9) (100.0)
Eighth Plan
1991–92 172.0 34.5 206.4 57.2 55.4 112.5 10.2 329.4
(52.2) (10.5) (62.7) (14.2) (16.8) (34.4) (3.1) (100.0)
1992–93 205.2 26.4 231.7 78.4 65.2 143.9 7.2 383.4
(53.5) (6.9) (60.4) (20.4) (17.0) (37.5) (1.9) (100.0)
1993–94 223.9 20.7 244.6 107.7 74.1 181.8 9.3 435.7
(51.4) (4.8) (56.1) (24.7) (17.0) (41.7) (2.1) (100.0)
1994–95 248.5 24.3 272.8 99.0 94.5 193.5 5.3 471.6
(52.6) (5.2) (57.8) (21.0) (20.0) (41.0) (1.1) (100.0)
Note: Figures in parentheses are percentages of total transfers.

100 M. Govinda Rao
based transfers given by the Finance Commission and the Planning

Commission has declined and that of discretionary transfers has increased
over the years. Third, within the Finance Commission transfer, the propor-
tion of tax devolution has increased.
Finance Commission transfers

The Constitution specifies that the Finance Commission is required to
• distribute the shares of personal income tax and Union excise duty
between the center and the states and among the states,
• recommend grants to the states in need of additional assistance, and
• address any other matter referred to them.
Although the Constitution does not place any restriction on the scope of
the Commission, with the Planning Commission dispensing a significant
share of transfers, the scope of the Finance Commission was restricted in
practice to meeting the non-plan current requirements of the states.13
The Finance Commission’s approach to making federal transfers con-
sists of:
• assessment of overall budgetary requirements of the center and states to
determine the volume of resources that can be transferred during the
period of its recommendations;
• forecasting states’ own current revenues and non-plan current expendi-
tures;
• distributing assigned taxes, broadly on the basis of origin;
• distributing sharable taxes – the personal income tax and Union excise
duties – between the center and the states and among the states; and
• filling the post-devolution projected gaps between non-plan current
expenditures and revenues with the grants.
This is known as the “gap-filling” approach. The relative shares of the
center and the individual states and the criteria adopted for their distribu-
tion among the States according to the recommendations of the Tenth
Finance Commission are summarized in table 3.9.
An important feature of the tax devolutions recommended by the
Finance Commissions is that, while the criteria adopted for distributing
them are different from the principles of grants-in-aid, nowhere is it made
clear that the economic objectives of the two instruments are different
(Rao, 1987). The tax devolution is recommended mainly on the basis of
general economic indicators, and grants are given to offset the residual
fiscal disadvantages of the states as quantified by the Commissions. Even
India 101
Table 3.9. Criteria for tax devolution: Tenth

Finance Commission, 1990–95 (percentage)
Income tax Excise duty
States’ share 77.5 40 ⫹ 7.5a

Criteria for distribution
Tax effort 10.0 10.0
Population 20 20
Per capita SDP
Distance formulac 60 60
Areab 5.0 5.0
Index of infrastructure 5.0 5.0
Notes:
a
7.5 percent is given only to the states with post-
tax devolution deficits.
b
subject to each state receiving the minimum of
2 percent and the maximum of 10 percent of the
sharable taxes.
c
Distance formula ⫽(Yh⫺Yi )Pi/⌺(Yh⫺Yi )Pi where
Yi and Yh represent per capita SDP of the ith and
the richest state, Pi is the population of the ith
state, (Yh⫺Yi ) for the state h is taken to be
equivalent to the value of the second highest per
capita SDP state.
Source: India, Report of the Tenth Finance
Commission, 1994.
in the case of tax devolution, until recently, the criteria adopted for the dis-
tribution of personal income tax and excise duties have been different. The
Tenth Finance Commission also assigned 10 percent weight to the tax effort
factor, but the measurement of tax effort itself is based on the assumption
that the SDP is the sole determinant of taxable capacity and taxable capac-
ity increases with increases in the SDP by an exponential of 2.14 Further,
assigning weights to contradictory factors like “contribution” and “back-
wardness” in the same formula for distribution has rendered the achieve-
ment of the overall objective of transfers difficult.
Over the years, attempts have been made by the successive Commissions
to improve the redistributive element in the transfer scheme by assigning
higher weight to per capita SDP, in either the “inverse” or “distance”
forms.15 Yet population has continued to receive the largest implicit and
explicit weight. The redistributive effect has been further blunted by the
terms of reference which require the Commission to use the 1971 popula-
tion figures wherever population is used in the transfer formula. The
purpose of this is to penalize the states with higher population growth
102 M. Govinda Rao
rates. The important questions are, first, whether the federal transfer
mechanism should be employed as an instrument of population policy and,
second, even if it is, why those states with high population growth due to
migration should be penalized.
The “gap-filling” approach outlined above has been subjected to criti-
cisms. First, none of the Finance Commissions assessed the overall
resource position of the center and the amount of resources required to
meet its commitments on any objective basis. Second, the transfers made
by the Finance Commissions were not designed specifically to offset fiscal
disadvantages of the states arising from lower revenue-raising capacity
and the higher unit cost of public services. While the tax devolution is
determined on the basis of general economic indicators, grants are given
on the basis of projected post-devolution budgetary gaps. Third, the
design of the grants has serious disincentive effects on fiscal management
of the states. Nor are the fiscally disadvantaged enabled to provide a given
level of public service at a given tax-price.
The more recent Commissions tried to reduce the post-devolution gaps
by substantially enhancing the share of tax devolution in total transfers.16
At the same time, they tried to target the transfers by including different
elements of backwardness, besides the “inverse” and “distance” variants
of per capita income, thereby complicating the tax devolution formula.
Also, the more recent Commissions introduced selective norms for the
center and the states by targeting the rates of growth of revenues and
expenditures and by assuming certain rates of return on their loans and
investments. The Ninth Finance Commission actually estimated the
revenue capacities and expenditure needs of the states. However, the prac-
tice of having different approaches for making tax devolution and grants
has reduced the relevance of such exercises.
Plan transfers
The Planning Commission gives both grants and loans to the states to
finance their plans. In earlier years, both the volume and the loan and grant
components were project based, but since 1969 the assistance has been allo-
cated on the basis of a formula (the Gadgil formula).17 At present, 30
percent of the funds available for distribution is kept apart for the special
category states and distributed among them on the basis of plan projects
formulated by them, with 90 percent of the assistance given by way of
grants and the remainder as loans. The 70 percent of the funds available to
the major states is distributed with 60 percent weight assigned to popula-
tion, 25 percent to per capita SDP, 7.5 percent to fiscal management, and
India 103
Table 3.10. Formula for distributing state Plan assistance
Variable Weight
Population (1971) 60.0

Per capita SDP, of which 25.0
Deviation from the average to the states below average per capita SDP 20.0
Distance from the highest per capita SDP for all the general category states 5.0
Fiscal performance, of which 7.5
Tax effort 2.5
Fiscal management 2.5
National objectives 2.5
Special problems 7.5
Total 100.0
Note: The formula is applied to general category states. They receive 70 percent of the total
Plan assistance, of which 30 percent is given as grants and the remainder as loans. The
special category states receive 30 percent of total Plan assistance and 90 percent of the
assistance is given as grants and the remaining 10 percent as loans. The distribution among
them is based on the plans approved.
Source: Planning Commission, Government of India.
the remaining 7.5 percent to special problems of these states (table 3.10). In
these states, 30 percent of the resources is given by way of grants and the
remainder as loans. Thus, plan transfers and their grant and loan compo-
nents are determined independently of the required plan investments,
their sectoral composition, the resources available to the states, or their
fiscal performances.
The tenuous relationship between the required plan investments and
plan transfers becomes abundantly clear when we examine per capita plan
outlays during the Seventh Plan (table 3.11). In all the major states except
Maharashtra, the resources available for plan investments from the states’
resources, before any central transfers were given, were negative. It is also
seen that the deficits were higher in the poorer states. As the richer states
had access to larger non-plan loans and as they could get greater central
plan assistance, per capita plan outlays in high-income states were almost
twice as much as in middle- and low-income states.
Assistance to the central sector and centrally sponsored schemes

This is the third component of transfer and is given for specified purposes
with or without matching provisions. Grants for the central sector schemes
are given to the states to execute central projects and are entirely funded
by the center. Centrally sponsored schemes, on the other hand, are shared-
Table 3.11. Per capita federal fiscal transfers and Plan outlay in the states during the Seventh Plan (1981–82 rupees)
Per capita Index of States’ plan States’ plan

annual SDP, taxable resources Finance resources Plan assistance
1982–85 capacity, before statutory Commission Non-plan after statutory (including Total plan
States (current prices) 1984–85 transfers transfers loans transfers central schemes) outlay
(1) (2) (3) (4) (5) (6) (7) (8) (9)
(7)⫺(6⫹5)
High-income states 3 340 146.30 ⫺134.24 321.43 534.83 722.02 533.18 1 255.20
Punjab 4 013 169.18 ⫺459.28 280.45 318.05 139.23 1 131.83 1 271.06
Maharashtra 3 384 142.75 ⫺229.72 316.24 509.77 1 055.73 233.52 1 289.25
Haryana 3 043 151.11 ⫺175.07 344.39 570.99 740.31 463.18 1 203.49
Gujarat 2 929 122.16 ⫺132.35 344.62 740.53 952.79 304.20 1 256.99
Middle-income states 2 206 112.82 ⫺271.46 439.65 255.78 423.96 227.88 651.84
Karnataka 2461 117.68 ⫺49.98 389.70 112.04 451.76 213.36 665.12
West Bengal 2 230 76.09 ⫺421.11 483.04 278.40 340.34 140.56 480.90
Kerala 2 144 117.60 ⫺521.60 440.26 380.98 299.65 308.19 607.84
Tamil Nadu 2 142 138.64 ⫺186.56 439.21 316.60 569.25 229.51 798.76
Andhra Pradesh 2 053 114.04 ⫺178.07 446.02 190.87 458.82 247.77 706.59
Low-income states 1 689 50.06 ⫺265.69 472.19 171.11 377.60 287.94 665.55
Madhya Pradesh 1 860 58.14 ⫺139.69 422.13 227.32 509.75 200.00 700.76
Rajasthan 1 820 67.46 ⫺380.23 389.99 291.74 301.50 421.77 723.27
Orissa 1 728 37.72 ⫺250.75 582.07 126.74 458.07 310.56 768.63
Uttar Pradesh 1 713 54.14 ⫺256.19 440.86 143.54 328.21 272.18 600.39
Bihar 1 323 32.85 ⫺301.61 525.89 66.20 290.49 235.21 525.70
Average of 14 states 2 345 99.97 ⫺211.92 428.94 261.35 478.36 276.90 755.27
Sources: Columns 1 and 2: Second Report of the Ninth Finance Commission, Ministry of Finance, Government of India, 1990; other columns:
finance and planning departments of the state governments
India 105
cost programs falling within the states’ ambit with matching ratios which
are uniform across the states, but which vary with the projects. There were
262 such schemes in 1985, and more were added in subsequent years.
These transfers have attracted the sharpest criticism owing to their discre-
tionary nature and the conditions attached to them. They accounted for
about 36 percent of the total plan assistance, and about 20 percent of total
current transfers was given to these schemes in 1994–95 (table 3.7).
In summary, the design and implementation of intergovernmental
transfer schemes in India suffer from a number of shortcomings. First,
multiple agencies with overlapping jurisdictions have blurred the overall
objectives of transfers. Second, accommodating different interests has
unduly complicated the transfer formula. Third, the design of the transfer
system is not well targeted to achieve equalization and to ensure minimum
service levels in the states. Fourth, transfers have disincentive effects on
the fiscal management in the states. While there is certainly a role for
specific-purpose transfers in the Indian federation, the design and imple-
mentation of the centrally sponsored schemes have not served the need.
They have tended to multiply state-level bureaucracy and distort states’
own allocations in unintended ways.
Equalizing effects of intergovernmental transfers

The correlation of central transfers with per capita SDP and their elastic-
ities with respect to SDP clearly show that the transfers recommended by
the Finance and Planning Commissions were equalizing (table 3.12). It is
also seen that the equalizing effect was the highest under the award of the
Ninth Finance Commission (1989–92). This finding is also confirmed by
the analysis of interstate inequalities in per capita revenue collections and
per capita revenue accruals shown in table 3.13. It is seen that:
• the aggregate transfers tend to equalize but the degree of equalization
as measured by the reduction in the Gini coefficient during the period,
1989–92, was only about 0.107;
• the equalization achieved by the Finance Commission transfers was
about twice as much as that of Plan transfers;
• equalization of transfers by both the Finance and Planning Com-
missions marginally increased over time;
• within the former, tax devolution has shown the highest degree of
equalization.
In spite of the fact that the transfer system as a whole has appreciable
equalizing impact, per capita revenue accruals, and hence per capita
Table 3.12. Equalizing effect of transfers
Correlation coefficients with per capita SDP Income elasticities
VI F.C. VII F.C. VIII F.C. IX F.C. VI F.C. VII F.C. VIII F.C. IX F.C.
Transfers (1974–79) (1979–84) (1984–89) (1989–94) (1974–79) (1979–84) (1984–89) (1989–94)
Shared taxes ⫺0.167 ⫺0.706** ⫺0.849** ⫺0.809** ⫺0.024 ⫺0.195* ⫺0.507* ⫺0.564*
Non-plan grants ⫺0.240 ⫺0.289 ⫺0.110 ⫺0.286 ⫺0.716 ⫺0.070 ⫺0.302 ⫺0.054
Total finance Commission transfers ⫺0.272 ⫺0.551* ⫺0.664** ⫺0.765* ⫺0.201 ⫺0.280* ⫺0.403* ⫺0.514*
Plan grants: state plan schemes ⫺0.263 ⫺0.524* ⫺0.010 ⫺0.425** ⫺0.243 ⫺0.426** ⫺0.029 ⫺0.557**
Plan grants: central schemes ⫺0.342 ⫺0.101 ⫺0.162 ⫺0.278 ⫺0.460 ⫺0.066 ⫺0.095 ⫺0.070
Total plan grants ⫺0.091 ⫺0.327 ⫺0.092 ⫺0.417 ⫺0.072 ⫺0.236 ⫺0.060 ⫺0.282
Gross current transfers ⫺0.194 ⫺0.519* ⫺0.663** ⫺0.716** ⫺0.115 ⫺0.268** ⫺0.277* ⫺0.408**
Note: * Significant at 1 percent level. ** Significant at 5 percent level.

Elasticity coefficients relate to cross-section of 14 major states.
F.C.=Finance Commission.
Source: Estimated from data taken from the budget documents of the state governments.
Table 3.13. Effect of federal transfers: Gini coefficients of fiscal variables
Sixth Finance Seventh Finance Eighth Finance Ninth Finance

Commission Commission Commission Commission
Variable (per capita) (1974–79) (1979–84) (1984–89) (1989–94)
(1) Own revenue ⫺0.2262 ⫺0.2355 ⫺0.2329 ⫺0.2575

(2) Own revenue ⫹ shared taxes ⫺0.1805 ⫺0.1718 ⫺0.1640 ⫺0.1842
(3) Own revenue ⫹ Finance Commission transfers ⫺0.1599 ⫺0.1615 ⫺0.1576 ⫺0.1742
(4) Own revenue ⫹ state plan transfers ⫺0.2092 ⫺0.2138 ⫺0.2167 ⫺0.2350
(5) Own revenue ⫹ transfers for central schemes ⫺0.2186 ⫺0.2184 ⫺0.2126 ⫺0.2350
(6) Own revenue ⫹ plan transfers ⫺0.2030 ⫺0.1994 ⫺0.1993 ⫺0.2154
(7) Own revenue ⫹ total current transfers ⫺0.1490 ⫺0.1417 ⫺0.1394 ⫺0.1508
(8) Degree of equalization: shared taxes (1–2) ⫺0.0457 ⫺0.0637 ⫺0.0689 ⫺0.0732
(9) Degree of equalization: Finance Commission transfers (1–3) ⫺0.0662 ⫺0.0740 ⫺0.0753 ⫺0.0832
(10) Degree of equalization: state plan grants (1–4) ⫺0.0170 ⫺0.0217 ⫺0.0162 ⫺0.0224
(11) Degree of equalization: central scheme transfers (1–5) ⫺0.0076 ⫺0.0170 ⫺0.0203 ⫺0.0225
(12) Degree of equalization: total plan transfers (1–6) ⫺0.0232 ⫺0.0361 ⫺0.0336 ⫺0.0421
(13) Degree of equalization: total current transfers (1–7) ⫺0.0772 ⫺0.0938 ⫺0.0935 ⫺0.1067
Note: Interstate Gini coefficients correspond to 14 major states only.

Source: Financial accounts of the state governments, various issues.
108 M. Govinda Rao
Table 3.14. Regression results
Independent variable
Dependent variable Constant Per capita SDP R̄ 2
Per capita own revenue ⫺296.3650* 0.1760* 0.8524

(2.4714) (8.7200)
Per capita total revenues 165.2228 0.1500* 0.7965
(1.3369) (7.2019)
Per capita revenue expenditure 208.520 0.1580* 0.6500
(1.100) (5.0146)
Per capita total expenditure 341.4971 0.1801* 0.6395
(1.5679) (4.9050)
Notes:
The regressions have been estimated by employing the log–linear
model; N ⫽14. Figures in parentheses are t-statistics of the regression
coefficients.
* Significant at 1 percent level.
Source: Author’s estimates.
expenditures, have a significant and positive relationship with per capita

SDP. As may be seen from table 3.14, the elasticity of revenue accruals, as
well as of per capita current expenditures, with respect to per capita SDP
is 0.15, which is only marginally lower than that of the states’ own rev-
enues (0.18).
Federal fiscal arrangements in India: major issues

The preceding analysis brings out the important features of federal fiscal
arrangements in India. The analysis highlights a number of shortcomings
which are due not merely to constitutional arrangements, but also to the
conventions, methods, and workings of institutions. To an extent, these
have been shaped by the developmental strategy, and economic liberaliza-
tion would call for a review of the arrangements. Therefore, in reorienting
the federal fiscal arrangements to complement the needs of a market
economy, reforms are needed in both policies and institutions.
Federal fiscal arrangements in a market economy should aim at the effi-
cient delivery of public services, ensure a nationwide market unfettered by
impediments, and activate welfare-improving competition among the
governmental units. Unlike in a planned economy where economic power
is centralized, fiscal functions in a market economy call for decentraliza-
tion. Further, be it from a dynamic developmental perspective or from the
viewpoint of merely exercising preferences for public services through
India 109
“exit” and “voice,” human resource development plays a critical role in a

developing country. This primarily belongs to the domain of subcentral
governments, and in the new environment, their relative role is likely to
increase. The recent initiative to empower the urban and rural local bodies
below the state level through the 73rd and 74th Amendments to the
Constitution is likely to give a further thrust to fiscal decentralization.
Giving a larger role to the market in allocating resources would affect
inter-regional equity as well. This is not to say that the planning process in
the past succeeded in promoting balanced regional development. We have
referred to the inequitable resource transfers due to financial policies, the
center’s own investment decisions, and interstate tax exportation. These,
and the inability of the federal transfer system to adequately offset the
fiscal disabilities of the poorer states, have led to inequitable distribution
of plan outlay and, consequently, levels of infrastructure. Thus, in the liber-
alized environment, the richer states with better physical and social infra-
structures, proximity to the markets, and responsive bureaucracy would
attract a larger share of private investments than the poorer states unless
corrective steps were taken to improve the relative competitive ability of
the poorer states.
The reform of the federal fiscal system should begin with the re-exami-
nation of assignments. It is necessary to review whether the distinction
between income and wealth from agricultural and non-agricultural
sources serves any useful purpose. Again, assigning power to levy sales
tax only on goods and the exclusion of taxation of services altogether from
the purview of the states is clearly anomalous and has denied the states
access to this growing tax base. Further, given that, in any case, concur-
rency in the commodity tax base of the center and the states exists, there is
no reason why concurrent consumption tax powers cannot be given in the
legal sense along with the institution of a mechanism to ensure coordi-
nated levies by the center and the states. This, in fact, would give the states
access to a wider tax base and reduce vertical fiscal imbalance. Of course,
this could also accentuate horizontal imbalance, but the solution to this
would lie in better targeting of the transfers to offset the fiscal disabilities
of poorer states.
It must be admitted that the intergovernmental transfer system has, over
the years, achieved a significant degree of equalization in the levels of
public services across the states. Equally notable is the attempt to reduce
discretion in the transfer system by increasingly resorting to formula-
based transfers. Of course, the formulas used to distribute transfers still
leave a lot of room for improvement in terms of both equity and incentives.
It is therefore necessary to redesign the transfer system to improve
110 M. Govinda Rao
accountability, incentives, and equity. In a more liberalized environment,

inter-state inequality in the standards of public services is likely to
increase. The general-purpose transfers should, therefore, be better tar-
geted. Similarly, the centrally sponsored schemes must be designed to
ensure minimum outlay on specified services throughout the federation.
The states might perhaps be allowed to choose from among a number of
priority schemes instead of establishing specific conditions for each
scheme. Further, consolidation of the large number of centrally sponsored
schemes and the introduction of broad conditionalities could improve
flexibility for the states and reduce resource distortions.
Reforms are needed in the institutional mechanism as well. First, over-
lapping in the functions of different institutions should be avoided. The
Finance Commission can assess and recommend transfers to cover the
entire current needs of the states, and the Planning Commission can assess
the requirements for physical infrastructure and give the required loans.
The working of the Finance Commission, and the methodology adopted
by it, should be changed so that disincentives to fiscal management are
avoided. This requires them to function on professional lines. The appoint-
ment of professionals to the Finance Commission, the strengthening of its
research capacity, a permanent secretariat undertaking continuous
research, enabling greater interaction between governmental units, and
imparting a greater degree of transparency to the functioning of the
Commission are some of the other reforms urgently required.
Notwithstanding the weaknesses, it must be noted that the system of
intergovernmental fiscal arrangements in India has served reasonably well
for over fifty years. It has achieved significant equalization over the years,
instituted a workable system of resolving the outstanding issues between
the center and the states and among the states, and adjusted to the chang-
ing requirements. It has thus contributed to achieving a degree of cohesive-
ness in a large and diverse country. Nonetheless this analysis has brought
out several areas of reform; and it is important to be aware that it is emi-
nently possible to reform the system.
Notes
1 The recent (73rd and 74th) Amendments to the Constitution have sought to
specify the domain of the governmental units below the state level (rural and
urban local bodies) as well. However, they exercise their functions concurrently
with the state governments, with the latter enjoying overriding powers. For an
analysis of the rationale, achievements, and challenges in regard to local
governments, see Singh (1996).
2 There are, however, some disagreements with such a functional allocation.
India 111
Gramlich (1987) and, more recently, Inman and Rubinfeld (1992) have argued
that the subcentral units can make effective contributions to counter-cyclical
policy. Similarly, Pauly (1973) has argued that redistribution should be consid-
ered a local public good and subnational units do have an important role to play
in poverty alleviation; see Rao and Das-Gupta (1995).
3 “Competitive federalism” views governmental systems as competing entities
analogous to firms. It combines the results of electoral competition in a demo-
cratic polity with the federal form of government. The competition analyzed in
this context is not restricted price competition, but Schumpeterian entrepre-
neurial competition; see Breton (1987).
4 Bundling of public services on a “take it or leave it” basis increases welfare
costs; see Breton (1987).
5 A possible exception is that of China and the former socialist countries of
Europe where the local taxes are higher than their expenditures, thus requiring
upward rather than downward transfers. However, the locally collected taxes
are not really local taxes because their base and/or rates are decided at the
central level; see Prud’homme (1995: ch. 2).
6 There are, of course, administrative difficulties in taxing agricultural incomes,
and the expert committee that went into this issue actually recommended a pre-
sumptive method of taxation based on land holdings classified broadly accord-
ing to productivity differentials (India, Ministry of Finance, 1972). However, the
point made here is that in the prevailing situation, even in principle, it is impos-
sible to have a comprehensive concept of income as the tax base, and differen-
tial treatment of income from agricultural and non-agricultural sources has
formally opened up an easy avenue for evasion of the tax through misclassifica-
tion.
7 Recently, increase in the prices of petroleum products led to reduction in the
sales taxes on these products by a number of states.
8 See Breton (1987). Prud’homme (1995), however, asserts that decentralization
could cause a downward shift in the production frontier owing to lower admin-
istrative efficiency. His argument ignores the fact that administrative efficiency
of subnational governments itself is a function of assigning the responsibilities
and resources to them.
9 Of course, the higher than average per capita expenditures in special category
states cannot be entirely attributed to their inherent cost disability. It may also
be due to bad fiscal management.
10 It is also possible to restore the balance through the unbalanced level cutting
expenditures or raising more revenues, or by changing the assignment itself.
However, if the imbalance has occurred even after efficient assignment, these
methods of correcting the balance have an additional welfare cost.
11 In India, the statutory liquidity ratio stipulated for the commercial banks
requires them to keep 35 percent of their resources in the form of government
securities.
12 There is a considerable amount of confusion over the term “transfers.” In
Indian federal finance literature, central loans to states are also characterized
as transfers. Such transactions involve transfers only to the extent of any
interest subsidy or write-off of loans. Sometimes, on the recommendation of
112 M. Govinda Rao
the Finance Commissions, the central loans to the states are rescheduled, the
rate of interest reduced, or even a portion of the loan itself is written off. A
proper estimate of the transfers should include the implicit transfers arising
from interest subsidy and loan write-off, which calls for a separate study.
Here, we have taken only tax devolution and grants as transfers. Transfers
arising from interest subsidy guarantees and loan write-off are not taken into
account.
13 The grants (Gi ) receivable by the ith state are given by:
Gi ⫽ Ei ⫺ (Roi ⫹Rai ⫹ Rsi )
Gi ⱖ 0.
where Ei denotes projected non-plan current expenditures of the ith state,
Roi denotes projected own revenues of the ith state, Rai denotes projected share
of assigned revenues of the ith state, and Rsi denotes shared taxes of the ith state.
14 Tax effort (␩) ⫽ Ti/Yi2, where Ti is the per capita tax revenue collected by the ith
state and Yi is the per capita state domestic product of the ith state.
15 The “inverse” formula is given by:
(Pi/Yi )/⌺iPiYi
and the “distance” formula is given by:
(Yh ⫺Yi )Pi/⌺Yh ⫺Yi )Pi
where Yi and Yh represent per capita SDP of the ith and the highest income state
respectively and Pi is the population of the ith state.
16 The Seventh Finance Commission, for example, increased the states’ share of
Union excise duty from 20 to 40 percent.
17 The formula and its modifications from time to time are evolved on the basis
of consensus within the National Development Council (NDC). The NDC
comprises the cabinet ministers at the center, chief ministers of the states, and
the members of the Planning Commission, and is chaired by the prime minis-
ter.
References
Boadway, R. W. and Flatters, F. (1982), Equalization in a Federal State: An Economic
Analysis, Ottawa: Canadian Government Publishing Center.
Bradbury, K. L., Ladd, H. F., Perrault, M., Reschovsky, A., and Yinger, J. (1984),
“State aid to offset fiscal disparities among counties,” National Tax Journal, 37:
151–70.
Brennan, G. and Buchanan, J. M. (1980), The Power to Tax, Cambridge: Cambridge
University Press.
Breton, Albert (1987), “Towards the theory of competitive federalism,” European
Journal of Political Economy, 3(1&2): 263–328.
Breton, Albert, and Fraschini, Angela (1992), “Free-riding and intergovernmental
grants,” Kyklos, 45: 347–62.
Buchanan, J. M. (1950) “Federalism and fiscal equity,” American Economic Review,
40: 421–32.
India 113
Chelliah, R. J. (1991), Towards a Decentralized Polity (Fourth L. K. Jha Memorial
Lecture), New Delhi: Fiscal Research Foundation.
(1993) “Agenda for comprehensive tax reform,” C. N. Vakil Memorial Lecture at
the Indian Economic Association Conference, Bombay.
Feldstein, Martin S. (1975), “Wealth neutrality and local choice in public educa-
tion,” American Economic Review, 65: 75–89.
Gramlich, Edward (1987), “Subnational fiscal policy,” Perspectives on Local Public
Finance and Public Policy, 3: 3–27.
Hunter, J. S. H. (1977), Federalism and Fiscal Balance, Canberra: Australian National
University Press.
India, Ministry of Finance (1972), Report of the Committee on Taxation of Agricultural
Wealth and Income (Chairman: K. N. Raj), New Delhi.
Inman, R. P. and Rubinfeld, D. L. (1992), “Can we decentralize our unemployment
policies? Evidence from the United States,” paper presented at the
International Seminar in Public Economics, Tokyo (August 31-September 1).
King, David (1984). Fiscal Tiers: The Economics of Multilevel Government, London:
Allen & Unwin.
Musgrave, R. A. (1983), “Who should tax, when and what?” in Charles E. McLure,
Jr. (ed.), Tax Assignment in Federal Countries, Canberra: Australian National
University Press.
Netzer, Dick (1991), “An evaluation of interjurisdictional competition through eco-
nomic development incentives,” in D. A. Kenyon and J. Kincard (eds.),
Competition Among States and Local Governments: Efficiency and Equity in
American Federalism, Washington DC: The Urban Institute.
Oates, W. E. (1972), Fiscal Federalism, New York: Harcourt, Brace, Jovanovich.
(1977a) (ed.), Political Economy of Fiscal Federalism, Lexington, MA: Lexington
Books.
(1977b) “An economist’s perspective of fiscal federalism,” in Oates (1977a).
Oates, W. E. and Schwab, R. M. (1988), “Economic competition among jurisdic-
tions: efficiency enhancing or distortion inducing?” Journal of Public Economics,
35: 333–54.
Pauly, M. V. (1973) “Income redistribution as a local public good,” Journal of Public
Economics, 2: 35–58.
Perloff, Harvey (1985), “Fiscal policy at the state and local level,” in L. S. Burns and
J. Friedman (ed.), The Art of Planning: Selected Essays of Harvey S. Perloff, New
York: Plenum Press.
Prud’homme, Remy (1995), “On the dangers of decentralization,” World Bank
Economic Review, 10: 201–20.
Rao, M. G. (1987), “Fiscal imbalances in Indian federalism: trends and issues,”
Working Paper No. 30, National Institute of Public Finance and Policy, New
Delhi.
(1993), “Indian fiscal federalism from a comparative perspective,” paper pre-
sented at the Conference on Federalism in Diverse Societies, Center for
Advanced Study on India, University of Pennsylvania, Philadelphia
(November).
Rao, M. G. and Das-Gupta, A. (1995), “Intergovernmental transfers and poverty
alleviation,” Environment and Planning C: Government and Policy, 13: 1–23.
114 M. Govinda Rao
Rao, M. G. and Vaillancourt, F. (1994), “Inter-state tax disharmony in India: a
comparative perspective,” Publius: The Journal of Federalism, 24: 99–114.
Singh, Nirvikar (1996), “Governance and reform in India,” Working Paper No. 356,
Department of Economics, University of California, Santa Cruz.
Tiebout, C. M. (1956), “A pure theory of local government expenditures,” Journal of
Political Economy, 64: 416–24.
4
Indonesia and Pakistan: fiscal decentralization –

an elusive goal?
ANWAR SHAH
Responding to a wide variety of concerns ranging from citizen disenchant-

ment with national government failures to globalization, a large and
growing number of developing countries (seventy at last count) have
expressed a commitment or a desire to achieve a greater degree of decen-
tralization. While there has been reasonable agreement on an economic
framework to guide such reform efforts (see Shah, 1994) and often a rea-
sonable national consensus on directions for reform, progress on such
reform efforts has proceeded at a frustratingly slow pace in the majority of
these countries. This chapter reviews the experiences of two large and
diverse countries in reforming their fiscal systems, with a view to devel-
oping a better understanding of the potentials and perils of such efforts in
developing countries. First, reasons are stated for our selection of the coun-
tries to be discussed. This is followed by discussions of their fiscal systems.
A subsequent section reflects upon why the road to reform has been so dif-
ficult in the two countries. A final section draws some lessons of general
interest to other reforming countries.
Indonesia and Pakistan represent interesting case studies in federalism
for two reasons. First, the reform of fiscal systems is high on the policy
agenda in both countries and innovative approaches to the design of inter-
governmental fiscal relations have been followed. Second, in both coun-
tries, impediments to reform are arising less from a lack of institutional
capacity at the local level and more from “rent-seeking” (the pursuit of
enlightened self-interest) by bureaucratic and political elites. A study of
these countries thus uncovers both the potentials and perils of federalism
in developing countries.
Indonesia and Pakistan have many similarities. Both countries support
large populations. At the end of 1996, Indonesia had an estimated popula-
tion of 203 million and Pakistan 137 million. Both countries have pre-
The views expressed in this chapter are those of the author alone and should not be attrib-
uted to the World Bank. The author is grateful to Richard Bird and François Vaillancourt for
helpful comments.
115
116 Anwar Shah
dominantly Muslim populations. Both countries share a colonial heritage:

the former gained independence from Dutch rule in 1945 and the latter
from British rule in 1947. Both countries inherited well-trained bureau-
cratic regimes from their colonial rulers. Both have shown a remarkable
growth performance over the last two decades. Both countries have well-
articulated systems of intergovernmental relations. While, by constitu-
tional intent, Indonesia is a unitary country and Pakistan a decentralized
federation, in practice, both countries have highly centralized governance
structures. Political systems remain fragile in both countries and military
interventions, political unrest, and threats of secession are facts of life in
both countries. Political and bureaucratic corruption and red tape are
rampant. Both countries are ranked low (“poor governance”) on an index
of quality of governance developed by Huther and Shah (1996) which
embodies political participation, bureaucratic efficiency, social develop-
ment, and economic management considerations.
In other respects, Indonesia and Pakistan are dissimilar. Indonesia is far
more diverse in terms of its geography and the ethnic origins of its popula-
tion. Political power in Indonesia is concentrated at the centre, whereas in
Pakistan, in recent years, there has been a recentralization of power at the
provincial level with feudal lords reigning supreme both at the centre and
in the provinces. Private sector participation in the provision of public ser-
vices is much greater in Pakistan than in Indonesia. Such participation in
Pakistan has been a local response to the dysfunctional public sector. In
contrast, Indonesian public services are reasonably well administered and
the main potential improvement would be simply to match these services
with local preferences. Indonesia faces a greater internal, but a far lower
external, security threat than Pakistan. Indonesia has achieved greater eco-
nomic maturity than Pakistan. End-1996 per capita GDP in Indonesia was
US$ 1,210 compared to US$ 520 in Pakistan (table 4.1).
Given their differing backgrounds, the two countries have developed
systems of intergovernmental fiscal relations to suit their specific circum-
stances. In the following, we review these approaches.
The fiscal system in Indonesia

Indonesia is an archipelago of nearly 14,000 islands, of which 930 are
inhabited; it covers 1.9 million square kilometers and has an ethnically
diverse end-1996 estimated population of 203 million. Since 1954, it has
organized itself as a multi-tier unitary state, with its provinces as the
second tier and local government as the third tier. There are 27 provinces
(Level [Dati] I), 25 cities and towns (Level II – Kotamadya), 241 districts
Indonesia and Pakistan 117
Table 4.1. A comparative impressionistic review of selected aspects of the fiscal

system in Indonesia and Pakistan
Selected Indicators Indonesia Pakistan
Population (end-1996) 203 million 137 million

Area 1.9 million square kilometers 0.8 million square
kilometers
Per capita GDP (1996) US$ 1,210 US$ 520
Fiscal constitution (de jure) Unitary Decentralized federation
Fiscal constitution (de facto) Multi-tiered centralized Multi-tiered centralized
unitary federation
Subnational governments 27 provinces, 297 districts, 4 provinces, 15 municipal
3,837 subdistricts, 5,000 corporations, 156 municipal
villages committees, 301 town
committees, 40 cantonment
boards, 118 district councils,
4,565 union councils
Provincial government Weak Strong
constitutional status
Local government Weak Weak
constitutional status
Actual provincial control of Moderate Strong
local governments
Range of local government Limited Limited
responsibilities
Citizen participation Low to moderate Low (rural) to moderate
(urban)
Bureaucratic orientation Improved service delivery Command and control
Political and bureaucratic Moderate High
corruption
Red tape Moderate High
Private provision of public Limited Extensive
services
Provincial fiscal autonomy Weak Strong
Local fiscal autonomy Weak Weak
Local administrative Weak Weak
autonomy
Local government role in Moderate Constrained
public service provision
Transparency and Excellent Good
predictability of the system
of central–provincial
transfers
Design of provincial–local Excellent Non-existent
transfers
Quality and quantity of Very good and extensive Poor/unreliable or non-
public services existent
118 Anwar Shah
(Dati II – Kabupaten), and 65,544 villages (Desas – ranging in population

size from 200 to 55,000). The military and the civil service are represented
in the national parliament. As the head of the central government, the
President appoints provincial governors. Mayors are elected by local
councils from candidates pre-screened by the center (the President) or the
provinces (the governors). Provincial governors also appoint chief admin-
istrative officers of local governments. Appointment of most public offi-
cials at all levels is a responsibility of the center. Thus the center holds real
and effective power in most public matters and provincial–local govern-
ments act primarily on behalf of the center in functions that are delegated
to them by the center. Indonesia is also exceptional in the degree of depen-
dence of its subnational governments on central revenues. In 1994–95,
subnational governments were responsible for about 20 percent of the
consolidated general government expenditures but raised only 7 percent
of consolidated government revenues. This level of fiscal deficiency for
subnational governments is large by international standards. It should
also be noted that large regional disparities exist in Indonesia. For
example, the per capita GDP of the poorest province (Nusa Tenggara
Timur) is only 6 percent of the per capita GDP of the richest province
(Kalimantan Timur).
The centralization of authority in Indonesia has traditionally been
viewed as a means of preserving national unity and promoting economic
growth. But there is now an emerging perception that a highly centralized
fiscal structure may be imposing large political and economic costs on the
country. This awareness has resulted in the adoption of a policy permitting
gradual decentralization of decision-making. Government Regulation No.
PP 45/1992 (on the implementation of regional autonomy with emphasis
on second-level regions) represents the most serious attempt in recent
years to define a specific course of decentralization. This regulation seeks
to transfer some central and provincial responsibilities to the local level.
State policy guidelines issued in 1994 reaffirm this commitment by estab-
lishing a framework guiding policy formulation over the Sixth Plan
period, 1993–94 to 1998–99.
Expenditure assignment issues: a new strategy for weakening

provincialization and strengthening localization
In Indonesia, nearly three-quarters of expenditure is directly undertaken
by the center, and another 10 percent is effectively controlled by the center
through the conditionality of its transfers to local governments. Even for
the remaining 15 percent of expenditure, the center exercises some influ-
ence through its planning and budget approval process. Thus, overall, the
distribution of responsibilities remains highly centralized.
In Indonesia, Law No. 5 of 1974 provides the overall legal framework
guiding the distribution of responsibilities between levels of government.
Specific regulations to implement this legislation have been slow in
coming. Regulation No. PP45/1992 represented a major attempt to chart a
more clearly defined course for the process of decentralization, and to
accelerate this process. In this regulation, the focus of further devolution
of responsibilities is Level II governments (districts), and the devolution
envisaged is from both central and Level I (provincial) governments. The
following functions are reserved for the center: defense and security
affairs; judicial affairs; foreign affairs; part of general administrative affairs
concerning heads of regions; and other administrative affairs that can be
more effectively and efficiently managed by the central government. All
other functions are to be considered for transfer to lower-level govern-
ments. The same regulation also specifies the following functions for Level
I governments: inter-municipal (Level II) affairs; affairs that are not central
to Level II development; and affairs that are implemented more effectively
and efficiently at Level I. All other affairs of Level I governments are to be
transferred to Level II governments in a gradual and continuous manner.
All transfers are to be accompanied by the transfer of associated budgetary
resources. The transfer of responsibilities will be linked to the capacities of
Level II governments. The assessment of the readiness of these govern-
ments to assume additional responsibilities, and of their capacity-building
needs, will be based on a rating of their existing institutional capacities. It
is envisaged that some functions could in part be transferred initially for a
four-year period, with the transfer completed and made permanent upon
assessment of the Level II government performance during that period.
Indonesia is fast approaching a stage in its economic evolution where
the present degree of centralization in the responsibility for the provision
of public services will become increasingly untenable because of both its
implications for efficiency and the strain imposed on the financial and
institutional capacities of the center. Regulation No. PP45/1992, accord-
ingly, represents a timely initiative. The broad framework it sets out for the
transfer of responsibilities to local governments is consistent with general
economic principles for intergovernmental assignment of expenditures.
Within this framework, the actual allocation of responsibilities for indi-
vidual services will need to be carefully determined. This process is to be
guided by an interministerial body on regional autonomy, and will be sup-
ported by studies, some of which are already underway. For local public
services, the following may serve as general criteria for the assignment of
120 Anwar Shah
responsibilities: policy development, standards of service, and per-

formance evaluation (center); oversight (provincial level); provision
(local/regional level); and production (public or private, depending on the
service). For some services that in principle can be assigned to the local
level but require a minimum scale for cost-effective provision, for example,
sanitation systems, a voluntary association of small contiguous municipal-
ities divided into special-purpose regional bodies, is an option that can be
considered. For most local public services, this would not be necessary for
large, metropolitan municipalities. The assignment of responsibility for
the provision of a public service to a particular level of government does
not mean that government must be directly engaged in its production. In
areas where private production and delivery of a public service would be
feasible and more efficient, and would be consistent with ensuring an equi-
table access to the service, the responsible government needs to promote
private participation. Examples of this are services that can be provided
within a competitive framework, such as transport services and refuse
collection.
Tax base assignment issues

In Indonesia, the major, most productive direct and indirect taxes are all
assigned to the center. Personal and corporate (including oil and gas)
income taxes, the main indirect taxes (namely, VAT, excises, and duties on
foreign trade), and the local property tax are all centrally administered,
with revenues from the last of these being shared with lower-level govern-
ments (16 percent to provinces and 65 percent to districts). Natural
resource royalties for petroleum, mines, and forests are also collected by
the center, but shared with lower-level governments. There are no supple-
mentary (“piggyback”) taxes levied by local governments on the tax bases
administered by the center. Tax assignment in Indonesia emphasizes effi-
ciency of tax administration and uniformity of the tax regime across the
country. The centralization of taxes is also motivated by concerns for
reducing regional disparities in the availability of public resources. While
these objectives are legitimate and important, consideration needs to be
given to how tax options at the subnational level can be broadened, con-
sistent with these objectives. The present high degree of centralization of
tax authority is an important reason why local governments are unable to
self-finance but a small proportion of their spending.
Within the present broad framework of tax assignment in Indonesia,
several possibilities for augmenting local tax authority could be consid-
ered. First, local governments could be given a greater role in the
administration (including rate-setting) of the property tax. Levied on

immobile assets, this tax is suitable for administration at the local level. A
greater local role in its administration would likely improve incentives for
collection. The central government could retain the assessment function
for smaller towns but assume a supervisory role for the same in larger
urbanized jurisdictions (which are already playing a significant role in the
administration of the tax under central direction). Second, excises, which
in Indonesia are used as special single-stage sales taxes (mainly on luxury
items) alongside a general VAT, being residence-based taxes, can in prin-
ciple be assigned to the subnational (provincial) level. Administrative
reasons may necessitate keeping these taxes at the central level in the short
term, but possibilities for instituting revenue-sharing could be considered
in the interim. Third, subnational governments could be allowed to levy a
tax on fuel consumption. At a moderate rate, such a tax could raise sizable
revenues, besides supporting environmental objectives. At present levels
of consumption, a 5 percent sales tax on gasoline and diesel could raise
local own-source revenues by about 25 percent if assigned to provincial
governments, or by 65 percent if assigned to district governments. Fourth,
local revenues from the property tax could be augmented by the institu-
tion of frontage/development charges (tax per front foot of the property in
new developments to pay for development of basic infrastructure). In
industrial countries, frontage charges normally amount to about 5 percent
of property tax revenues. Fifth, possibilities for local taxation of some non-
basic public services, such as the telephone, could be explored.
Increasing revenues from existing regional or shared taxes

Subnational governments in Indonesia can levy a multiplicity of taxes,
around fifty, but only a few of them are significant sources of revenue.
About 90 percent of provincial tax revenue comes from just two taxes
(motor vehicle registration and transfer taxes) and about 85 percent of dis-
trict-level tax revenue comes from only six taxes (hotels and restaurants
tax, street lights tax, entertainment tax, advertisement tax, business
registration tax, and slaughterhouse tax). One desirable local tax reform is
to eliminate a large number of levies that are unproductive, which would
lower the costs of collection and free scarce administrative capacity to
focus on the more productive levies, while also reducing tax-induced dis-
tortions. These “nuisance taxes” remain on municipal books as municipal-
ities try to augment their revenues from all possible sources historically
made available to them. There remains much scope for increasing yields
from the more important taxes through better administration. Overall, it is
estimated that stronger mobilization of revenues from existing and new
122 Anwar Shah
subnational taxes and from shared taxes and fees could roughly double
subnational government own revenues. Even then, however, the share of
such revenues in total government revenues will rise to only about 13
percent.
Making more effective use of local user charges

Appropriately charging for public services serves both to raise revenues
and improve efficiency. User charges are especially important at the local
level because, being closer to beneficiaries, local public services are more
amenable to such charges than services provided by higher levels of
government. In Indonesia, user charges contribute a sizable proportion of
regional government revenues (about 15 percent at the provincial and 50
percent at the district level). However, the utilization of these charges
remains well below potential. At the district level, fees and charges cover
less than 10 percent of the outlay on public services. Heavy dependence on
central transfers has created weak incentives for local cost recovery. Many
public services are subsidized to make them affordable to the poor, but the
subsidies are inefficiently targeted, tending to benefit those who are better
off, for example, urban water supply. The equity objectives of subsidiza-
tion could be met more effectively, and at a smaller cost in terms of fore-
gone revenue, by better targeting the poor, for example, through the
adoption of “lifeline” charges whereby the service is charged at full mar-
ginal cost or higher beyond a level of service use associated with the poor.
Better central guidance to local governments on service pricing policies
would be helpful, as the responsible local authorities often lack adequate
understanding of the concepts of cost recovery. This should be supported
by improved accounting practices at the local level to allow a clearer
determination of the costs of service provision. Within central guidance on
service pricing principles, local governments could be allowed greater
authority in adjusting the level of charges. The present requirement for all
adjustments in local charges to be sanctioned by the higher-level govern-
ment acts as a disincentive to frequent revisions and often creates consid-
erable delays. Appropriately designed schemes to allow retention by the
agencies responsible for service delivery of part of the receipts from user
charges, for direct use on the service, could provide a useful incentive to
more vigorous cost recovery, for example, as currently allowed for some
local health services. Finally, reducing the present proliferation of small
charges would facilitate collection. In some cases, duplication of collection
efforts could be avoided through consolidating charges, for example, a
consolidated water tariff including charges for related sanitation and
drainage services.
Table 4.2. Indonesia: fiscal transfers to subnational governments, 1995–96
Grant program Allocation basis
General-purpose transfers
Provincial Development Grant Equal per province (85 percent); area
(15 percent)
District Development Grant Per capita with a floor
Village Development Grant Equal per village
Less-developed Village Grant Per capita
Specific-purpose transfers to provinces
SDO: Subsidy for Autonomous Regions Public sector wages
Provincial Road Improvement Grant Length of road; condition of road; unit cost of
construction and maintenance
Reforestation and regreening Project review
Specific-purpose transfers to local governments
SDO: Subsidy for Autonomous Regions Public sector wages
District/Town Road Improvement Grant Length of road; condition of road; density;
unit cost of construction and maintenance
Primary School Grant School-age children (ages 7–12); need for
facilities
Health Grant Need for medicine, health centers, and
personnel
Reforestation grant Project review
Intergovernmental transfers and loan finance

Even with feasible increases in local revenues, the gap between local
expenditures and revenues is likely to remain large for some time to come.
This underlines the importance of appropriately designing the system of
financial support to local governments, in the form of transfers from the
central government and loan finance.
Design of intergovernmental transfers

Central grants currently finance about 65 percent of expenditure at the pro-
vincial level and 70 percent at the district level. These transfers are of two
kinds: block grants, for general-purpose local spending subject to some
broad central guidelines; and specific grants, for expenditure on uses
specified by the center and subject to relatively detailed central controls
(table 4.2). The former include INPRES (Instruksi Presiden, or project funds
authorized by the President) block transfers to each of the three main levels
of local government – provinces, districts and villages. The latter include
124 Anwar Shah
SDO, a transfer that covers virtually all local government personnel

expenses, and INPRES sectoral transfers for specific development expen-
ditures on roads, primary schools, public health centers, and reforestation.
As part of its policy of gradual decentralization, the government has incre-
mentally raised the share of block grants in total transfers (from 16 percent
in 1986–87 to 22.5 percent in 1995–96) and has also allowed local govern-
ments somewhat greater flexibility in the use of some of the specific grants.
There are several positive features of the design of the Indonesian inter-
governmental grant system: the distribution of grants is transparent,
determined by formulas utilizing objective criteria; the structure of grants
is simple, as both the grants and the criteria used for distribution are few
in number; and the grants achieve an overall equalizing effect on regional
revenue availabilities. In its transparency and simplicity, the Indonesian
grant system compares favorably with the grant systems typically found
in developing countries.
There are, nonetheless, several improvements that could be considered
in the design of the Indonesian grant system that would allow it to achieve
its efficiency and equity objectives more effectively. First, the recent trend
toward increasing the share of block grants in total grants should continue.
Second, regional disparities in overall fiscal capacities (revenue-raising
potential) could be better reflected in the distribution formulas for block
grants, by including a fiscal capacity equalization factor. The criteria cur-
rently used for distribution – area, population and equal shares – are all
focused primarily on capturing the differential needs of regional
administrations. A better capturing of differential fiscal capacities to meet
those needs would contribute to making the distribution of grants more
equitable.
A third possible improvement is that the SDO grant could be consoli-
dated with the general-purpose block grants to the respective levels of
government. As presently designed, this grant creates strong incentives for
higher government employment, and hence a higher wage bill, at the local
level. The center tries to circumvent this perverse incentive by retaining
major control over government employment at all levels, but this under-
mines local autonomy and flexibility in the allocation of budgetary
resources between personnel and other recurrent (operations and mainte-
nance) expenditures.
Fourth, the main improvement that can be made in the specific sectoral
grants is to continue the shift toward using broad guidelines rather than
detailed controls and physical targets in influencing the use of these grants.
The allocation criteria for these grants are broadly appropriate, as they
adequately serve their main objective (ensuring minimum standards of the
targeted basic services across regions). One improvement would be to

change the allocation of the reforestation grant from a project to a formula
basis, as is the case with the other specific sectoral grants. For the health
grant, a possible option is to redesign it as a per capita block transfer
subject to central guidelines on standards of access and service levels.
A fifth, and final, possible improvement is that consideration could be
given to assigning provinces a role in the allocation of central grants to the
lower levels, by making some of the grants pass through them. The ratio-
nale for doing so is that provinces are better placed than the center, espe-
cially in a large and diverse country, to assess the needs and fiscal
capacities of individual lower-level jurisdictions. A provincial role in grant
administration could also contribute to closer monitoring of the use of the
funds.
The Government has recently introduced (with effect from 1994–95), a
new per capita block transfer to villages identified as lagging behind in
development (INPRES Desa Tertinggal). The proposed grant program,
focused on poverty reduction in targeted concentrations of poverty, is well
structured overall. At the completion of its initial three-year period of
operation, consideration could be given to making block transfers to vil-
lages part of a fiscal capacity equalization program.
Framework for local borrowing

Borrowing currently plays a very small part in financing regional govern-
ment expenditures in Indonesia (less than 5 percent). The bulk of central
financial support to regional governments takes the form of grants, as
noted above. Whatever local borrowing does take place is predominantly
from the central government, including the on-lending of foreign loans,
much of it to finance urban water supply projects. The limited scale of local
borrowing, and the negligible local use of loans from commercial sources,
is a reflection of the local governments’ weak revenue base. A strength-
ening of the local revenue position is a prerequisite for greater local use of
loan finance. As progress is made in that direction, an increasing propor-
tion of central support for local capital projects could, and should, be pro-
vided in the form of loans. This would allow central grants to be focused
increasingly on poorer areas. A shift from grant to loan financing of capital
projects would induce the adoption of stronger cost recovery policies by
the local governments. As local financial capacities grow, non-government
sources of loan finance could also be gradually tapped, for example,
through the issue of municipal bonds, at least by the financially stronger
municipalities. For the near future, however, the use of commercial bor-
rowing by local governments must remain limited.
126 Anwar Shah
Framework for planning, budgeting, and monitoring

Indonesia has an elaborate planning process which integrates “bottom-
up” and “top-down” feedbacks. The same process, however, currently
undermines fiscal autonomy granted to subnational governments
through general-purpose fiscal transfers. As a result, local governments
have insufficient control over budgetary allocations for decentralized
functions and funding for these programs also becomes fragmented. In
addition, local inputs into the central planning process are quite limited
and central planning dominates local priorities. Reform of central–local
fiscal relations, therefore, needs to be underpinned by complementary
adaptations in the institutional arrangements for planning, budgeting,
implementation, and monitoring. An important general need is for
greater clarity in the distribution of responsibilities among the different
levels of government. To support efficient operations and proper account-
ability, regulations should comprehensively define the responsibilities of
each level of local government in terms of the decentralized functions, as
distinct from the deconcentrated and co-administered functions. A related
need is to reduce the fragmentation in the sources of funds for the
decentralized functions (for example, there are as many as ten sources of
funds for health service delivery at the district level, with one of them –
health INPRES grant – further split into nine separately specified ele-
ments). All central financial support to regional governments for the pro-
vision of decentralized services should be in the form of grants, or loans,
administered through their respective budgets. One implication of this is
that direct and separate higher-level government budgetary interventions
in the provision of decentralized services can be phased out. Some initia-
tives toward this end have been taken, under donor-assisted projects, for
the decentralized urban and health services. Improved systems for moni-
toring and auditing regional government performance would facilitate
reduced central reliance on direct, detailed controls on the local use of
funds.
Federalism in Pakistan: charting a treacherous course toward

provincialization
Pakistan has a total land area of 796,095 square kilometers and an esti-
mated end-1996 total population of 137 million. It is a federation with four
provinces, Punjab (56 percent of the population), Sindh (23 percent),
Northwest Frontier Province (13 percent) and Balochistan (5 percent). In
addition to the four provinces, parts of Kashmir which were included in
Pakistan are organized as the State of Azad Kashmir and enjoy self-
government. Federally administered northern areas include the adminis-
trative districts of Daiamir, Ghanche, Ghizer, Gilgit, and Skardu. There are
also seven self-governing federally administered tribal areas where the
laws of Pakistan do not apply, a situation which is consistent with the
tradition fostered under the British Empire. For administrative purposes,
the provinces are divided and each division is subdivided into districts.
Again, each district is divided into tehsils. At the district and tehsil levels,
elected councils and provincial civil administrations have overlapping
responsibilities.
Local governments in Pakistan do not have any distinct constitutional
status; they are established by provincial governments and powers are
determined by provincial statutes. At the present time, there are 5,195 local
government units in Pakistan of which only 512 units are in urban areas.
For local government purposes, urban areas are organized according to
population into metropolitan corporations, municipal corporations,
municipal committees, town committees, and cantonment boards.
Metropolitan corporations have some deconcentration of metro functions
by means of zonal offices. Karachi Metropolitan Corporation has in the
past experimented with a zonal lower tier of local government called
municipal committees. Land development and water and sewerage func-
tions in large urban areas are the responsibility of provincial agencies
called Development Authorities (DAs) and Water and Sanitation Agencies
(WASAs) respectively, and for smaller municipalities development expen-
ditures are undertaken directly by provincial line departments. Karachi
Metropolitan Corporation represents the exception in water and sewerage
with its own Karachi Water and Sewerage Board (KWSB). Rural areas are
organized into a two-tier local government structure with the lower-tier
union council representing several villages, and a larger rural area, called
a district, having a district council comprising several union councils.
Provincial statutes regarding local government organization are fairly
uniform and provide for elected local councils to have broad powers over
a wide range of local public services. In practice, however, for reasons to
be discussed later, the existing system of local government has become
dysfunctional.
In terms of program spending, the federal government dominates in
defense, debt servicing, general administration, fuel and power, trans-
portation and communications, industrial development, population plan-
ning, manpower management, water supply and sanitation, and
subsidies. Provincial government spending dominates in law and order,
food and agriculture, rural development, education, and health. Local
128 Anwar Shah
government spending is pre-eminent in social welfare, parks and recre-

ation, and animal care. Overall in 1994–95, federal government spending
accounted for 67.2 percent of the consolidated program total, with pro-
vincial government spending at 28.8 percent, and local government spend-
ing representing only 4.1 percent. It is also interesting to note that federal
and provincial governments accounted for 73.3 percent of total spending
on local public services in the same year. At the same time, the federal
government collects 90.7 percent of the total tax revenue, with the remain-
ing 9.3 percent collected in roughly equal proportions by the provinces and
local governments.
The 1973 constitution delineates the responsibilities of federal govern-
ment (Federal Legislative List) and the areas of shared responsibilities
(Concurrent Legislative List). All remaining functions are the domain of
provincial governments. Local government functions are defined under
provincial laws through the use of Local Government Ordinances. The
federal government is responsible for such key areas as defense, foreign
affairs, international trade, macroeconomic management, and industrial
development, among others. The provinces and the federal government
share responsibility for population planning, curriculum and syllabus
planning, and social welfare, and the provinces are solely responsible for
roads, highways, police protection and justice, agriculture, and post-
primary education. Local governments are given responsibility for local
public services such as basic education and health, fire protection, parks
and recreation, and water and sewerage.
Despite the extent of decentralization of legislative responsibilities as set
out in the constitution and provincial statutes, the actual assumption of
responsibilities is more centralized. For example, though education is a
subnational responsibility in Pakistan, the federal government neverthe-
less takes responsibility in national policy formulation, planning, curricu-
lum, and managing centers of excellence. Similarly, although health care in
Pakistan is a subnational responsibility and much of it is assigned to the
local level, in practice, it is more centralized. The federal government is
responsible for drug regulation and quality control, regulating the stan-
dard of professional education, undertaking measures for containing com-
municable disease, and interacting with external organizations.
Implications of the existing structure

A quick review of the federal structure in Pakistan suggests that the leg-
islative division of powers in Pakistan as set out by the Constitution and
provincial statutes compares favorably to the assignment based on fiscal
federalism principles. However, the actual assumption of responsibilities

by various levels of government during the past two decades has consis-
tently departed from the de jure assignment. Higher level governments
have increasingly encroached not only upon lower-level government
responsibilities, but also upon traditional private sector functions. There
appears to be a trend towards centralization of responsibilities, with the
federal government assuming either an exclusive role (for example, in
population planning, electricity, curriculum development, syllabus plan-
ning, and centers of excellence) or a dominant role (for example, in social
welfare, vocational/technical training, employment exchanges, and his-
toric sites and museums) for some of the functions on the concurrent list,
and even for some specified to be purely provincial. It should be noted that
while the major proportion of social welfare expenditure (87.7 percent in
1994–95) is carried out by local governments/religious institu-
tions/community groups, these entities simply administer federally
defined and financed programs. Similarly, provincial governments have
gradually assumed a large number of functions previously carried out,
and intended to be carried out, by local governments (for example, cura-
tive health, land development, primary education, farm-to-market roads,
rural development, preventive health, water supply, drainage, and sewer-
age) and/or by the private sector (table 4.3). Superior institutional capac-
ity at higher levels of government and the attendant efficiency in service
delivery were cited by the federal and provincial governments as the main
reasons for the centralization of responsibility for public services. The
reason given for higher-level governments in Pakistan having assumed
increased responsibility for direct provision of private goods and services
has been market failures. Such involvement is quite widespread, and
encompasses key economic areas including manufacturing, wholesaling,
retailing, banking, and financial intermediary functions.
In general, wider interpretations of constitutional mandates by federal
and provincial governments have had the following implications for
public sector management in Pakistan.
The private sector is crowded out

Federal and provincial governments during the 1970s acted aggressively
to curtail private sector participation in education and health services and
in industry, banking, and finance. This resulted in limiting citizens’ choices
and lowering the quality and quantity of private and quasi-private goods
and services. It is useful to note here that recent survey results indicate that
the public sector in Pakistan is less efficient than the private sector in the
provision of education and health services.
130 Anwar Shah
Table 4.3. Pakistan: legislative responsibility and actual provision of services by

different levels of government
Actual allocation of
Legislative responsibility Service functions
Federal governmenta Defense Federal government

External affairs
Posts and telegraphs
Telephones
Radio and T.V.
Currency
Foreign exchange
Foreign aid
Institutes for research
Nuclear energy
Ports and aerodromes
Shipping
Air service
Stock Exchange
National highway
Geological surveys
Meteorological surveys
Censuses
Railways
Mineral oil and natural gas
Industries
Federal provincial Population planning
governmentsb Curriculum development
Syllabus planning
Centers of excellence
Tourism
Social welfare
Vocational technical training Federal provincial
Employment exchange governments
Provincial governments Historical sites and monuments
Law and order
Justice
Tertiary health care and hospitals
Highways Provincial governments
Urban transport
Secondary and higher educationd
Agricultural extension
Fertilizer and seed distribution
Irrigatione
Land reclamatione
Local governmentsc Primary education
Curative health Primarily provincial
Preventive health with minor local
Water supply, drainage and sewage government
Farm-to-market roads involvement
Land development
Rural developments Local governments
Link roads
Intra-urban roads
Table 4.3 (cont.)
Actual allocation of
Legislative responsibility Service functions
Street lighting
Garbage collection
Fire fighting
Parks and playgrounds
Notes:
a
According to Federal Legislative List.
b
According to Concurrent Legislative List.
c
According to Local Government Ordinances.
d
University education is funded by the federal government through the University Grants
Commission (UGC), but is administratively controlled by the provincial governments.
e
Development of irrigation and SCARP projects is a federal subject.
Decision-making is highly centralized

In areas of concurrent responsibility, the higher level of government has
often adopted a dominant position. Decisions on most large capital pro-
jects require federal approval. And, as most local functions are being
assumed by the provincial government, local participation in major public
policy decisions that affect local people is by and large non-existent.
Efficiency and equity in public service provision are undermined

Since decision-making is significantly divorced from the people on whose
behalf decisions are being made, there is a greater likelihood of imperfect
matching of local public services with local preferences, and of inadequate
levels of response to local needs. These problems are particularly acute in
primary education and basic health. In primary education, enrollment
rates are low (42 percent), dropout rates are high (50 percent between
Grades 1 and 2) and female literacy rates are low (22 percent). In rural
areas, fewer than 15 percent of girls complete five years of education, and
only 1 percent of girls remain in school by age 14. Absenteeism by teach-
ers goes unreported. Teachers are not accountable to local communities or
parents or even to school heads, but are transferable at will by provincial
departments of education. Centralization of primary education has also
worked to create “ghost” schools which either exist on paper but cannot
be located or have a presence in the form of school buildings but cannot
find students. In basic health, performance under provincial administra-
tion is no better. Most health centers lack medicines and supplies and, in
many instances, proper sanitary conditions. In rural areas, health centers
often lack both medical personnel and supplies. Medical personnel are
132 Anwar Shah
frequently transferred against their will. There are no channels for local
participation and input in the provision of health care. While no welfare
cost estimates on centralization are available for Pakistan, for industrial-
ized countries such losses on account of simply misjudging preferences are
expected to range over 9–20 percent of expenditures. Once one considers
the lack of local accountability and participation, and the delinking of ben-
efits and costs, welfare losses may be expected to be of a much greater mag-
nitude in Pakistan.
Public sector accountability is impaired

Separation of taxing and spending decisions at the provincial and local
levels compromises accountability as citizens do not know exactly who
is responsible for which services. The level of government raising most
of the revenues, that is, the federal government, may be blamed for
provincial–local service delivery problems. Accountability is also
impaired because of a lack of clarity regarding the roles of various levels
of government in areas of shared responsibility.
The local public sector is weakened

In its early stages, Pakistan had a well-planned and well-developed local
government organization structure. During the 1950s and 1960s, local
governments were a vibrant part of the public sector and delivered most
of the local public services. However, in the 1970s and 1980s, some of the
more important local functions, such as primary education and basic
health, were shifted upward to provincial governments. Currently, the
federal and provincial governments dominate the provision of local public
services. As an example, the federal and provincial governments collec-
tively account for 95 percent of total public sector expenditures for water
and sewerage. This centralization of local service responsibilities has led
to deterioration in service quality, as accountability has been significantly
impaired.
The implication of all these points is that the current degree of centraliza-
tion of expenditure responsibilities has not served the country well and
may not be sustainable in the long run.
The federal–provincial allocation of taxing powers

The allocation of taxing powers to various levels of government has an
important bearing on the character of federalism and also accountability.
For example, if the ability of subnational jurisdictions to raise their own tax
revenues were restricted, this could impose significant constraints on the

ability of these jurisdictions to fulfill their proper expenditure responsibil-
ities, and could therefore compromise the potential benefits to be had from
decentralization. The allocation of taxing powers is determined jointly by
the constitutional assignment of tax sources by level of government, and
by the extent to which each level, especially the federal, has chosen to
exploit the major tax bases. The higher the tax rates chosen by the federal
government, the less room there will be for raising revenues at subnational
levels, and the more dependent will the latter be on the federal govern-
ment for their revenues.
The 1973 constitution specifies the following as areas of taxing
responsibility for the federal government:
• customs duties including export duties
• excise duties including duties on salt, but excluding duties on alcoholic
liquors, opium, and other narcotics
• duties in respect to succession of property
• estate duty in respect of property
• income taxes other than on agricultural income
• corporation tax
• tax on sales and purchase of goods imported, exported, produced, man-
ufactured, or consumed
• taxes on the capital value of assets, not including taxes on capital gains
on immovable property
• taxes on mineral oil, natural gas, and minerals for use in generation of
nuclear energy
• taxes and duties on production capacity of any plant, machinery, under-
taking, establishment, or installation
• terminal taxes on goods and passengers carried by railway, sea, or air
and taxes on their fares and freights tariffs.
All other forms of taxation not specified in the above list fall under the
purview of provincial governments. The provincial governments then
allocate some of these to local governments on the basis of Local
Government Ordinances. The major sources of revenue for provincial
governments are, in fact, transfers from the federal government in the form
of revenue-sharing from several major federal revenue sources. In particu-
lar, the provinces have transferred to them predetermined shares of rev-
enues. These are raised by federally administered taxes on incomes, the
sales of goods and excise duties on sugar and tobacco, federally adminis-
tered royalties on oil and gas, and hydroelectricity profits earned by federal
government enterprises. The President indicates the federal taxes that
134 Anwar Shah
would be subject to sharing when convening the National Finance

Commission (NFC). This commission is an intergovernmental body with
representation from federal and provincial governments and legislatures
and is chaired by the federal finance minister. The NFC develops consen-
sus recommendations regarding the federal and provincial shares, and for-
mulas for provincial allocation. These recommendations are not binding
on the federal government, yet when a consensus decision is reached, it is
almost always followed since federal and all provincial governments are
directly represented on the NFC. The major own-sources of provincial rev-
enues are stamp duties, motor vehicle registration taxes, and entertain-
ment taxes. Local governments rely heavily on provincially administered
real property and motor vehicle registration taxes and on self-administered
import/export (octroi) taxes. The use of these assigned taxing powers by
the various levels of government in Pakistan results in actual revenue-
raising being heavily concentrated at the federal level, much more so than
is required to finance its expenditure responsibilities. The federal govern-
ment collects 90.7 percent of consolidated current revenues and retains
only 59.3 percent for own use. It is also interesting to note that while the
federal government collects income and sales taxes primarily for transfers
to provincial governments, which account for 85.7 percent of the provincial
revenues, the latter governments also reciprocate by collecting capital
values tax and the religious taxes, zakat and ushr, at rates determined by the
federal government, and turning the proceeds over to the federal govern-
ment. Such an arrangement whereby revenues are passed upward is
unusual in federations, the notable exceptions being China (a unitary
country) and Russia, where for historical and institutional reasons the
major revenues are collected by subnational governments and shares are
transferred to the central government. In Pakistan, the aforementioned are
relatively minor revenue sources, and the reason for provincial administra-
tion, at least for the zakat and ushr – taxes which do not form part of the
budget – is their religious nature.
One important and distinguishing feature of the constitutional assign-
ment of taxing powers to the federal government and the provinces in
Pakistan is the tendency to assign major tax bases exclusively to one level
of government or the other. Moreover, those bases that are often the largest
revenue sources, such as sales and excise taxes and income taxes, are
assigned to the federal government. In many federations, such taxes are
co-occupied by the two levels of government so both are able to obtain rev-
enues from them independently, though preferably in a harmonized
fashion. This tendency to exclusive assignment has been a factor leading
to the highly centralized revenue-raising system observed in Pakistan, as
well as possibly to the fact that revenue-sharing is resorted to as a means

of transferring funds to the provinces. As noted later, revenue-sharing has
the disadvantage of not fostering accountability at the provincial level.
This feature of exclusive assignment is not present at the provincial/local
level. Entertainment taxes and taxes on cinemas, hotels, and professions
are co-occupied by provincial and local governments, usually with inde-
pendent collection.
Tax assignment in Pakistan emphasizes efficiency in tax administration
and uniformity of the tax regime across the nation, but as a result, neglects
provincial and local autonomy and accountability considerations. Thus,
provincial governments are almost completely dependent upon revenue
transfers from the higher level of government to finance their own expen-
ditures. This degree of separation of taxing and spending responsibilities
may not be in the interest of either the federal or the provincial level of
government. The federal government has been advised by the NFC award
for 1991 to transfer 80 percent of the proceeds from the more dynamic
revenue sources such as VAT and income taxes, and 98 percent of the pro-
ceeds from natural resource royalties, to the provinces. To raise revenues
for its own purposes, the federal government must first look for greater
effort on non-shared revenue sources such as excise taxes. Thus, its incen-
tives to exploit the major broad-based revenue sources are considerably
weakened. Provincial governments, on the other hand, have no control
over their major sources of revenue, and may not have any incentive for
cost efficiency or for raising revenues from own sources, as additional
efforts may not be worth the political costs.
A possible option for reform would have the federal government retain
exclusive responsibility for taxation of international trade, corporate
income, and for value-added taxation of sales of goods and services. The
federal government would maintain primary responsibility for taxing per-
sonal income, though with the ability of the provinces to piggyback at flat
rates. The current practice of the federal government levying excises on
production and royalties on natural resource exploitation and turning over
the proceeds to the provinces could be reviewed. Provinces could have
control over most excise taxes and over wage taxes, and could levy a
broad-based single sales tax that is parallel to the federal VAT, with its base
harmonized with the federal VAT at retail level. Furthermore, local govern-
ments could have a role in setting property tax rates, and access to electric-
ity duties, school fees, water and sewerage rates, and taxes on cinemas,
hotels, and entertainment services. This tax assignment should be consis-
tent with the fixed constitutional arrangements for sharing fiscal powers.
It is interesting to note that in Pakistan in 1994–95, over 90 percent of all
136 Anwar Shah
taxes were realized through withholding and voluntary payments, and

100 major corporations account for 75 percent of income, 80 percent of
excises and 60 percent of sales taxes. Further, the record of the federal tax
administrative bureaucracy in tax enforcement is not enviable. Surveys of
businesses have indicated that the agency appears to be more zealous in
collecting bribes than taxes (Stone, 1995). Therefore, institutional capacity
for tax administration should not be considered to be a serious impedi-
ment to any changes in tax assignment.
These suggestions in aggregate would lead to a significant change in
own revenues generated at various levels of government in Pakistan and
would result in greater conformity of revenue means with expenditure
needs for these governments.
Federal–provincial fiscal transfers

Matching revenue means as closely as possible to the expenditure needs of
various levels of government serves to strengthen accountability in a
federal system. Stronger tax performance and better cost recovery policies
at the subnational level will help in this task. It is, however, desirable in
federal systems for higher-level governments to have access to more rev-
enues than those dictated by their direct program responsibilities alone.
These additional revenues can be used to further national or provincial
economic objectives such as setting national standards, securing economic
union, and ensuring interregional and intermunicipal fiscal equity. The
design of these transfers, however, is critical to achieving the objectives
sought. The issues pertaining to the design of these transfers are the focus
of this subsection. Federal fiscal transfers to the provinces are the domi-
nant source of financing operating expenditures for provincial govern-
ments in Pakistan, accounting for 87.5 percent of such expenditures in
1994–95 (table 4.4). They financed nearly 99 percent of operating expendi-
ture for the two smaller provinces. The design of these transfers has impor-
tant implications for the fiscal behavior of provincial governments and for
the efficiency and equity of public service provision in Pakistan. Provincial
transfers to local governments, on the other hand, are of minor significance
in local finances as the provinces have provincialized major local public
services such as basic health, education, water, and sanitation. The exist-
ing structure of federal–provincial transfers is described in the following
paragraphs. Subsequently, the structure is evaluated for consistency of its
design with the objectives sought.
Federal government transfers to the provinces have both unconditional
and conditional components. These are discussed in turn.
Table 4.4. Pakistan: federal transfers to provinces, 1994–95: summary

indicators
Punjab Sindh NWFP Balochistan All
Operating transfers 82 84 99 99 87
(as % of gross operating expenditure)
Capital transfers 15 12 9 8 12
(as % of gross capital expenditure)
Total transfers 71 72 75 72 72
(as % of total expenditure)
Development loans 31 26 18 17 25
(as % of gross capital expenditure)
Source: Shah, 1995.
Unconditional transfers
These transfers are advised by an intergovernmental body, the National
Finance Commission, that is appointed every five years to conduct a
review of federal transfers. Federal and provincial finance ministers, the
federal finance secretary, and one additional member from each of the
provinces are represented on the Commission. The NFC has a chequered
history, with many instances of either not meeting, or meeting and not
achieving a consensus view. Over the past two decades, only the 1991 and
1997 NFC recommendations have been made public and implemented. As
a result of the 1997 award, the following federal unconditional transfers
are currently available to the provinces.
The constitutionally mandated revenue-sharing program has three ele-
ments: revenue-sharing by origin, revenue-sharing by population, and
special grants. Table 4.5 shows the details of the first two of these. For the
second element, the NFC defined a divisible pool to be shared on a 62.5: 37.5
basis between federal and provincial governments. The divisible pool con-
sists of the following taxes: taxes on income including corporation tax, but
excluding taxes on income consisting of remuneration paid out of the federal
Consolidated Fund; wealth tax; capital value tax; taxes on the sales and pur-
chases of goods imported, exported, manufactured, or consumed; export
duties on cotton; customs duties; federal excise duties, excluding the excise
duty on gas charged at well-head; any other tax by the federal government.
The 1997 NFC award recommended the following special transfers:
• To Northwest Frontier Province: Rs. 3.3 billion in 1997–98 escalated by
changes in the consumer price index (projected annual rate of 11
percent) for a period of five years.
138 Anwar Shah
Table 4.5. Pakistan: constitutionally mandated revenue-

sharing program, 1997–98 (percentages to be shared)
Tax %
Revenues shared on the basis of collection 14.2 (22 billion rupees)

Excise duty and royalty on gas 3.6
Surcharge on gas 4.3
Royalty on crude oil 1.0
Profits on hydroelectricity 5.5
Revenue shared on the basis of population 81.5 (171.4 billion rupees)
Income taxes 21.1
Sales tax 19.5
Export duty on cotton 0.8
Excise duty on sugar 26.8
Excise duty on tobacco products 13.2
• To Balochistan Province: Rs. 4.1 billion in 1997–98 escalated by changes

in the consumer price index (projected annual rate of 11 percent) for a
period of five years.
Conditional transfers
These transfers are relatively smaller. Nevertheless, about a dozen large
grant programs in excess of Rs. 100 million (1994–95) exist.
Matching grants for provincial resource mobilization: this program pro-
vides federal matching assistance at a 50 percent rate, up to a limit, for pro-
vincial revenue effort in excess of the historical average growth rate of 14.2
percent. The limits on these grants are Rs. 500 million each in the cases of
Punjab and Sindh and Rs. 100 million each for Northwest Frontier
Province and Balochistan. The program recognizes only fiscal effort result-
ing from increases in tax rates, withdrawal of exemptions, imposition of
new taxes, and revision in rates of user charges. Development Grants are
based on approval of provincial annual development plans (ADPs) by the
federal government. Federal Contribution for Social Action Program pro-
vides matching transfers on a 75:25 basis to finance provincial develop-
ment expenditures in education, health, water supply, and sanitation
associated with the Social Action Program. Flood and Disaster Relief
grants are by their very nature ad hoc in character, and are usually given to
the provinces for emergency relief and repair and renovations to basic
infrastructure following damage arising from natural disasters. Physical
Planning and Housing Project Assistance transfers finance to federally
approved provincial projects to upgrade urban infrastructure and housing
projects. Federal assistance is given in the form of conditional non-
matching grants. The Tameer-e-Watan Program provides block fund

allocations to federal legislators (in 1994–95, Rs. 6 million each for
members of the National Assembly and Rs. 5 million each for senators) to
be used for development projects of their choice through provincial
Ministries of Local Government and Rural Development. The Tameer-e-
Sindh Program provides federal financing on an ad hoc non-matching basis
for rural development initiatives in the province of Sindh. Federal line
ministries also fund various provincial projects, usually in the areas of
social welfare, population planning, health, irrigation, and drainage. The
Prime Minister occasionally uses discretionary allocation to provide pro-
vincial/local governments with funds for special programs. Finally, while
higher education in Pakistan is a provincial responsibility, the federal
government through the University Grants Commission (UGC) has tradi-
tionally provided financing for university education. The funding mecha-
nism used by the UGC is ad hoc and is primarily guided by budgetary
needs for salary expenditures.
Almost all of these conditional grants programs are discretionary in
nature rather than being formula-driven. This has the disadvantage of
requiring that the grants be allocated by administrative decision, which
can both be costly and detract from provincial decision-making autonomy.
Also, available funds may be unpredictable, thereby hampering long-term
planning.
An economic evaluation of federal transfers in Pakistan

Revenue-sharing and unconditional transfers
The revenue-sharing program suggested by the 1997 NFC award allocates
selected revenue sources on the basis of collection, whereas others are
distributed to the provinces on the basis of population.
The revenue-sharing by collection component, which distributes
roughly 25 percent of total revenues, returns resource revenues (oil and gas
excises, royalties, and hydroelectricity profits) by point of collection.
Hydroelectricity profits of federal public enterprises are also returned to
the province where the plant is located. The formula used to determine
these profits makes no provision for capital consumption allowances. The
existing program of resource revenue-sharing by origin seems difficult to
justify as the federal government collects revenues at centrally determined
rates and then returns them to provinces on the basis of the point of collec-
tion. The reason for doing so may be partly to give resource-rich provinces
feelings of entitlement, and partly to provide political accommodation
needed to maintain national unity. This treatment remains contentious as
140 Anwar Shah
Table 4.6. Vertical imbalances in Pakistan, 1994–95 (percent)
Tax collection level Revenue share Expenditure share Surplus/deficit
National 90.2 67.1 ⫺23.2

Subnational 9.7 32.9 ⫺23.2
Provincial 4.9 28.8 ⫺23.6
Local 4.8 4.1 ⫺0.7
All levels 100.0 100.0 ⫺ 0.0
Source: Calculated from federal and provincial budget data.
the largest province is now seeking the same treatment for federal royal-
ties/excises on cotton (see The Dawn [Pakistan], Internet edition, January
12, 1997). There seems to be neither an efficiency nor an equity argument
for doing so, unlike the case of population, or needs-based revenue-
sharing. Thus, asymmetric treatment of resource-based revenues is diffi-
cult to justify and could be considered for inclusion in a general
revenue-sharing pool to be distributed on a per capita and/or needs basis.
Better still, such revenues could go to the federal general revenues pool for
financing fiscal equalization transfers to the provinces. Furthermore, the
assignment of hydroelectric profits to the province of location of the facil-
ities could be reassessed as it undermines the financial ability and auton-
omy of the national power enterprise (WAPDA).
About 85 percent of revenue-sharing funds are distributed by popula-
tion. This program is to be commended for its simplicity, objectivity, and
success in transferring a large pool of resources in a predictable fashion
to bridge vertical fiscal imbalances. Moreover, the use of population
implies that the funds are distributed in an equalizing manner. Although
population is the sole criterion used to achieve regional equity, it is a rea-
sonable approximation of the need for funds since many provincial
expenditures increase in proportion to population. Overall, the transfers
are strongly equalizing with respect to own-tax collections (the rank
correlation is ⫺1.0), and mildly redistributive with respect to provincial
GDP (rank correlation is ⫺0.4). The program, nevertheless, is subject to
a number of limitations which are discussed in the following para-
graphs.
Limitations of the existing revenue-sharing program

Large transfers to reduce vertical fiscal gaps have the potential of intro-
ducing unintended adverse incentives (table 4.6). It is desirable to examine
first the possible sources for this imbalance, and alternative means of
dealing with these issues, before agreeing on the best system of inter-
governmental transfers. Vertical fiscal imbalance in Pakistan can be attrib-

uted to a number of causes.
Mismatch of revenue means and expenditure needs is an important
source of the problem, especially the overcentralization of taxing
responsibility. We earlier noted that centralization of expenditures has
been used as a response to correct fiscal imbalances arising at the subna-
tional levels. Tax decentralization on the other hand has remained an unex-
plored option. Such an option needs to be seriously examined to deal with
this issue. While significant untapped potential exists that could raise pro-
vincial revenues from own-sources, these efforts alone may not be suffi-
cient for meeting provincial expenditure needs given the set of tax bases
currently assigned to the provinces. Therefore, both tax decentralization
(for example, excises) and joint occupancy of some tax fields (for example,
personal income taxation) could be looked at to reduce provincial fiscal
imbalances. Existing tax assignments are inadequately used partly due to
excessive dependence on federal transfers, which reduces collection incen-
tives, and to weak tax administration.
Thus, in dealing with vertical fiscal imbalances, reassignment of some
responsibilities, tax decentralization, tax base sharing, and tax abatement
are options which could take precedence over unconditional grants or
general revenue-sharing. The latter options impair accountability by
separating taxing and spending responsibilities, and should be relied on
only to the extent that decentralization of fiscal responsibility is not
appropriate. This is clearly not the case in Pakistan. Here, there has been a
trend to centralize both taxing and expenditure responsibilities and then
rely on revenue-sharing to deal with the fiscal gap. These revenue-sharing
transfers finance up to 95 percent of provincial expenditures. Such an over-
whelming dependence of provincial governments on federal transfers has
tended to undermine federal budgetary flexibility as well as impairing
subnational public sector accountability. Furthermore, these transfers may
discourage the provinces from realizing the full potential of own tax bases.
Citizens do not see a link between taxes paid to a particular level of govern-
ment and public services offered by that same level.
The program ignores fiscal capacity in addressing regional equity issues.
In doing so, it lacks an explicit equalization standard against which
program achievements can be measured. Adoption of a formal fiscal
capacity equalization program, on the other hand, has the potential to
determine total amounts of transfers, and their allocation among provinces
by a formula that sets a specific standard of equalization to be achieved.
Finally, special fiscal need grants are intended to compensate the two fis-
cally disadvantaged provinces for their weak fiscal capacities but higher
142 Anwar Shah
expenditure needs. In the absence of significant tax decentralization and a

formal equalization program, such grants represent a pragmatic approach
to dealing with expenditure need differentials.
Conditional transfers: an evaluation

Conditional transfers to subnational governments are advocated to ensure
certain minimum standards of services across jurisdictions (conditional
block grants); to pursue higher-level government objectives (could be
conditional matching or non-matching); or to address interjurisdictional
spillovers of benefits (conditional matching transfers). With the exception
of a few selected programs discussed below, federal–provincial specific-
purpose transfers in Pakistan are, in general, ad hoc and primarily used for
agency functions or to advance political objectives. Moreover, they are
awarded on a discretionary basis rather than being determined by objec-
tive formulas.
The program of matching transfers for resource mobilization was ill con-
ceived. It rewards provinces for higher tax effort due to any changes in
structure and rates of taxation but provides no incentives for revenue
increases due to improvement in efficiency in tax collection and
administration. Thus, it potentially opens up the possibility for a province
to shift a significant burden of its taxation to non-residents by being lax in
collection of existing taxes while introducing newer (possibly nuisance)
taxes. The program also shows the federal government’s lack of concern
with the additional burden of taxation and the deteriorating quality and
quantity of provincial public services. This is particularly worrisome as the
effective burden of the public sector (taxation inclusive of bribes) in
Pakistan is considered very high and the net (fiscal benefit of public spend-
ing minus the burden of taxation) of the public sector as a whole may be
negative. The limits now set to this program are welcome as they limit
potential abuses and also restore the credibility of federal commitment that
had been tarnished by the unfilled open-ended commitment of a similar
program under the 1991 NFC award.
The federal programs of matching transfers for the Social Action
Program could be reassessed. Consideration may be given to instituting
conditional block (per capita) transfers for primary education and basic
health, whose magnitudes are determined by formula (for example, equal
per capita) rather than by discretion. These transfers may be made directly
from the federal government to metropolitan corporations. For other local
governments, these grants would pass through provincial governments.
The primary school grant could be made conditional on provinces meeting
targets on decentralization of primary education and basic health, ensur-
ing private sector financing on a par with the public sector (based on the
number of graduating students) and overall targets on access to such
education.
The basic health grant could be linked to standards of access to such ser-
vices. Similarly, federal transfers to universities through the University
Grants Commission could be eliminated and replaced by per capita trans-
fers to provinces for post-secondary education. Provinces in turn may be
encouraged to substitute own-transfer programs to both public and
private universities and colleges, utilizing the number of graduates by
program type as criterion. Various programs could be weighted differently
but financing of public and private education would be on an equal
footing.
Subnational government access to capital markets

The 1973 constitution (Article 167) allows unfettered capital market access
to provincial governments guaranteed by provincial Consolidated Fund
revenues (which include federal transfers) and subject only to any limits
imposed by the provincial assembly, provided that provinces do not owe
any debt to the federal government. All the provinces are heavily indebted
to the federal government, so access to capital markets for financing
provincial–local capital projects is not available to subnational govern-
ments in Pakistan. In 1994–95, debt charges accounted for 23 percent of
provincial current expenditures. This amounts to 165.8 percent of own-
source revenues. Most of this debt arises from capital investment in social
sectors and, therefore, accumulation of assets results in little increase in
debt-carrying capacity. Currently, the federal government does not
encourage any additional borrowing power by subnational governments
until they have stabilized their budgets. It is important to ensure, however,
that the federal government neither supplies debt, subsidizes debt nor
guarantees debt. Beyond this, credit market discipline is the key to respon-
sible subnational borrowing.
The 1997 NFC provision which allows provincial borrowing to finance
operating deficits could inadvertently encourage fiscal mismanagement.
Subnational borrowing to finance current expenditures should be pro-
hibited. On the other hand, if the provinces and, quite possibly, large met-
ropolitan area governments meet quantitative guidelines on fiscal
discipline, they could be permitted in the very long run to have access to
private capital markets at their own risk with a clear understanding that
the federal government would not guarantee such debt. To ensure that the
provinces do face market discipline and that their borrowing strategies do
144 Anwar Shah
not run counter to the federal government’s macroeconomic stabilization

objectives, the federal government might use the National Economic
Council or another intergovernmental forum as a consultative panel on
macroeconomic policy coordination and, more importantly, to disseminate
information on provincial and metropolitan finances and capital-spending
strategies to the private sector. This body could also catalyze the private
sector’s development of credit ratings of provincial and metropolitan
governments to facilitate bond and loan finance. Needless to say, unless
the creditworthiness of provincial governments is substantially enhanced
with a strong own-resource mobilization effort, provincial access to capital
markets would remain largely closed. Therefore, strengthening
provincial–local revenue sources should receive priority attention. Access
to private capital finance by smaller municipalities does not appear fea-
sible in the near future. For such governments, provincially directed
access, either through special funds or special arrangements like the FIND-
ETER experiment in Colombia, might be examined.
Why the road to reform remains a field of dreams

Fiscal systems in Indonesia and, especially, Pakistan require significant
restructuring. Indonesia has made slow and steady progress on such
reform efforts. In Pakistan such efforts are stalled and even in a state of
reversal in some areas. A number of factors impede the progress of reform
in both countries to varying degrees.
Political factors
In Indonesia, the authors of the Constitution contemplated a centralized
unitary country and dictated against the establishment of “states within
the state.” These concerns for political unity have dominated the design of
institutions. Well-entrenched roles of the military and the civil service in
political affairs, with a strong belief in command and control from the
center, have sustained centralization of responsibility. Appointment of
governors and mayors has also strengthened centralization and limited
local autonomy. Social development, economic prosperity, and concern for
improving the delivery of public services continue to bring a degree of
accommodation for decentralized institutions.
In Pakistan, political instability and feudal interests have contributed to
setting aside constitutional dictums and the introduction of a system of
centralized governance. Pakistan has been under military rule for most of
its existence and past military regimes did not accommodate decentralized
decision-making. During the periods that political activities have been per-
mitted, feudal influences have dominated the political system and favored
either a centralization or a provincialization of authority. This is because,
while Pakistan has experienced increased urbanization in recent years,
with over 40 percent of the total population urbanized, the electoral system
still operates on the 1981 distribution of population (17 percent urbanized).
In rural areas of Sindh and Balochistan, and to a more limited extent in
Punjab and Northwest Frontier province, feudal lords do not allow effec-
tive political participation. The use of the outdated distribution of popula-
tion allows feudal lords to dominate politics at the federal and provincial
levels. A centralized system allows these lords to have more effective
control than would be possible under a decentralized system, with which
the urban sector would have a more significant voice. To further entrench
feudal powers, local governments are currently disbanded in all metro-
politan areas even though the Supreme Court found this practice to be in
contravention of the law. Grants to members of national and provincial
legislatures for development projects also work against the development
of local governments as these members enjoy a greater degree of auton-
omy in project execution in the absence of a well-functioning system of
local government.
Bureaucratic factors
Both Indonesia and Pakistan have colonial legacies. The Dutch in
Indonesia and the British in Pakistan instituted systems of bureaucratic
control to achieve, with maximum efficiency, the colonial objectives of a
predatory state. These systems created a civil service elite who were
highly educated and dedicated to serving the colonial rulers. Their
loyalty to rulers and detachment from the common man were duly
rewarded by allowing them preferential access to all public services
through elite institutions and by ensuring their financial security
through a system of cash rewards and land grants. Thus, both countries
inherited civil service regimes that were highly centralized, efficient,
accountable, professional, and completely detached from the local
population.
After independence in Indonesia, the civil service over time became
an active political partner with the military in governing the country.
Both partners viewed central control as a key element in holding this
country of 14,000 islands together. A centralized regime was also con-
ducive to capturing rents from private sector development. Over time,
nevertheless, they also discovered that the initial degree of centralization
146 Anwar Shah
was not sustainable. Therefore, allowing a substantial degree of auton-

omy to local governments while keeping a weak structure of provincial
governments, was a necessary step to improving the delivery of local
services. A gradual shift towards local control (localization) is thus seen
as posing no threat to a command-and-control oriented bureaucratic
regime.
In Pakistan, after independence, the civil service retained its “profes-
sional” orientation for a while. It continued benefiting from an increasing
array of perks such as almost costless acquisition of prime real estate, free
membership in sports and entertainment complexes financed by public
funds, and privileged access to elite educational institutions for children.
However, political purges of the civil service in the late sixties and early
seventies set it on the road to administrative decline (see Haque, 1996).
With insecurity over the lack of tenure, areas of public intervention
expanded limitlessly, and perks and bribes mushroomed. Corruption
enabled officials to insure their careers against political risks. Thus the
structure became a highly centralized yet dysfunctional system of
administration. A key feature of this system has a special bearing on local
governance. Key positions in provincial (provincial secretaries) and dis-
trict governance (divisional and district commissioners) are held on
assignment by officers of the elite administrative corps (officially, the
“Central Superior Services – Civil Service of Pakistan”). Although per-
forming duties at subnational levels, these officials remain primarily
accountable to the federal government only. This system negates federal-
ism and reinforces federal control over local decision-making. For local
governments, a particularly worrisome aspect of this system of gover-
nance is that if the local governments function well, the district commis-
sioner’s powers are considerably curtailed. On the other hand, if the local
government is not operative, the district commissioner becomes the sole
discharger of judicial and executive functions at the district level. No
wonder it appears that local governments are not allowed to succeed in
Pakistan.
Institutional factors
Institutional factors also impede effective decentralization. Traditional
institutions, the mechanism of governance, and accountability over time
have withered away but have not been replaced by newer institutions.
Instead, the all-pervasive role of the state has retarded the critical observa-
tion of public policies and institutions. There is almost a complete
monopoly by the government of institutions of critical thought and the
media in both countries. Any critical review of government policies and

programs invites a government backlash. In Pakistan, rural self-govern-
ment worked well in earlier days of its independence. This system was
abandoned in favor of a more centralized system, which has resulted in the
denial of access to basic services for the rural population. While a lack of
institutional capacity was cited as a reason for the disbandment of the par-
ticipatory system, the newer system left a majority of citizens with no polit-
ical voice and no access to basic public services. Indonesia, on the other
hand, is now nurturing self-government in rural areas through its village
development grants.
External participants
External participants may also unwittingly impede development of a
decentralized public sector in developing countries. A multitude of
factors contribute to this development. First, a centralized system lowers
transaction costs for external assistance and enlarges the “comfort zone”
for external participants in terms of monitoring the utilization of their
funds for intended purposes. Second, some external participants have
concerned themselves with the revenue performance (“resource
mobilization”) of developing countries. Such concerns may lead to larger
centralized bureaucracies that pay little attention to the efficient delivery
of public services. For example, in Pakistan, the improved revenue per-
formance of governments has been accompanied by the deteriorating
quality and quantity of public services. Third, centralized systems are
more prone to suffer from an absence of an internal policy agenda owing
to a lack of citizen participation and to be more dependent on external
advice on policy reform. Typically this leads to quick policy fixes with
little sustained reform. For example, in Pakistan, the 1956 constitution
stated the achievement of universal literacy as a goal for the ensuing
decade; forty years later, there has been little change in literacy levels. In
population planning, with US assistance, Pakistan established a goal to
reduce the population growth rate to 2 percent by 1975. Twenty years
later, the growth rate has increased to 3.5 percent. Similarly, public deficit
reduction has been an elusive goal for several decades. External assis-
tance, contrary to its intentions, may have helped Pakistan to avoid facing
difficult choices in reducing public sector interventions in the market-
place. The availability of generous external assistance may have played a
part in motivating the federal government to assume some provincial
responsibilities and the provincial governments to take over local govern-
ment mandates.
148 Anwar Shah
Some lessons for developing countries

The experiences of Indonesia and Pakistan offer important lessons for
reform of fiscal systems in developing countries.
(1) Institutions of citizen participation and accountability must be

addressed in any serious reform of fiscal systems. Even in primitive soci-
eties, such as pre-British India, systems of local governance worked effec-
tively to deliver local services and collect local charges because of the
well-understood mechanisms of citizen participation and accountability.
More modern systems of local governance, such as those run by elite
Pakistani bureaucrats with training in management, including financial
management, have failed owing to an absence of citizen voices and
accountability checks. The reform effort must embody appropriate provi-
sions for holding to account elected officials for negligence or misconduct.
Independence of the judiciary and free media can play an important part
in political and bureaucratic accountability. These elements have not been
addressed in the Indonesian and Pakistani reform efforts.
(2) Civil service reform is also critical to the success of a decentralization

program. Bureaucratic ownership of a reform program is critical, but such
ownership is not forthcoming in most developing countries, where decen-
tralization is seen as an attempt to weaken the power of the central bureau-
cracy. To overcome this, the reform of fiscal systems must include the
reform of the central bureaucracy. Such reform must ensure that the center
has no say in the recruitment and promotion of civil servants at the subna-
tional levels, and that the remuneration of subnational services is competi-
tive with that at the central government level. Further, incentive structures
in the civil service should reward service orientation and performance and
discourage command-and-control and rent-seeking. This can be accom-
plished through performance contracts and the recognition of specialized
skills.
(3) Institutional capacity matters are of secondary importance and

should receive lower priority in reform effort. Institutional capacity to
develop and maintain modern organizational practices such as budget-
ing, auditing, and accounting systems is important but should not be
considered as a barrier to decentralization, provided citizen participation
and transparency in decision-making are ensured. Technical capacity can
be borrowed from supportive higher-level governments, amongst other
places.
(4) Asymmetric decentralization as provided under the Indonesian

decentralization program and under provincial local government ordi-
nances in Pakistan offers a thoughtful approach to decentralization.
Regardless of the availability of help from higher-level governments, lack
of institutional capacity should never be considered as an excuse not to
decentralize. Instead, an objective program of decentralization which rec-
ognizes the nature and type of local government, its clientele, and its fiscal
capacity can be developed. Moreover, various local governments can be
assigned differential powers by taking into account the aforementioned
factors, as was done in Pakistan in the past, and is being done more system-
atically in Indonesia by rating each local government.
(5) A major separation of spending and taxing decisions leads to lack of

accountability in the public sector. In Pakistan, federal revenue-sharing
transfers finance up to 99 percent of expenditures in some provinces. This
delinking of taxing and spending responsibilities has led to accountability
problems at the provincial levels. In the event of such delinking, the role
of conditional (conditional on standards of services and access to such ser-
vices but not on expenditures) block transfers is worth examining in order
to enhance accountability.
(6) Sharing of revenue on a tax-by-tax basis distorts incentives for effi-

cient tax collection. In Pakistan, tax-by-tax sharing of income and sales
taxes has impeded reform of trade taxes which are not shared with the
provinces.
(7) Successful decentralization cannot be achieved in the absence of a

well-designed fiscal transfers program. The design of these transfers must
be simple, transparent, and consistent with their objectives (table 4.7). The
experiences of Indonesia and Pakistan offer important insights on grant
design. For example, Indonesia’s education and health grants use simple
and objectively quantifiable indicators in the allocation of funds, and
conditions for the continued eligibility for these grants emphasize objec-
tive standards for access to these services. Indonesian grants for public
sector wages, on the other hand, represent an example of not so thought-
ful design as they introduce incentives for higher public employment at
subnational levels. Pakistan’s matching grant for resource mobilization
similarly rewards relatively rich provinces for additional tax effort. It also
calls into question the credibility of federal commitment, as the federal
government has not been able to meet the commitment arising from this
grant program.
Table 4.7. Principles and best practices in grant design
Grant objective Grant design Best practices Practices to avoid
To bridge fiscal gap Reassign responsibilities; tax Tax abatement in Canada and tax Deficit grants; tax-by-tax sharing
abatement; tax base sharing base sharing in Canada, Brazil, as in India and Pakistan
and Pakistan
To reduce regional fiscal disparities General non-matching fiscal Fiscal equalization programs of General revenue-sharing with
capacity equalization transfers Australia, Canada, and Germany multiple factors
To compensate for benefit Open-ended matching transfers South Africa grant for teaching —
spillovers with matching rate consistent hospitals
with spillover of benefits
Setting national minimum Conditional non-matching block Indonesia: roads and primary Conditional transfers with
standards transfers with conditions on education grants; Colombia, conditions on spending alone; ad
standards of service and access Chile, and South Africa: hoc grants
education transfers
Influencing local priorities in Open-ended matching transfers Matching transfers for social Ad hoc grants
areas of high national but low (preferably with matching rate assistance, as in Canada
local priority varying inversely with fiscal
capacity)
Stabilization Capital grants provided; Limit use of capital grants and Stabilization grants with no
maintenance possible encourage private sector future upkeep requirements
participation by providing
political and policy risk
guarantee
Source: Shah, 1994.

(8) Finally, contrary to a common misconception, a developing country’s

institutional environment calls for a greater degree of decentralization
than is needed for an industrialized country. For the efficient working of a
centralized bureaucracy, advanced information-gathering and transmittal
networks, an efficient and dedicated civil service, and well-developed
institutions of citizen participation and accountability are needed. This is
possible in the environment of an industrialized country. A more primitive
public sector environment is better suited by a decentralized form of
governance because information requirements and transaction costs are
minimized by moving the decision-making closer to people who are
affected by those decisions. Closeness also serves to enhance participation
and to promote preference matching for public services, transparency, and
greater accountability. The experience of Pakistan demonstrates that
public sector performance is significantly improved by decentralized
decision-making even when the enabling environment is quite weak.
References
Haque, Nadeem-ul (1996), Appropriate Reform Strategies in Pakistan, Washington,
DC: International Monetary Fund.
Huther, Jeff, and Shah, Anwar (1996), A Simple Measure of Good Governance and its
Application to the Debate on the Appropriate Level of Fiscal Decentralization,
Indonesia, Government of (1996), Monitoring Indicators of Repelita VI, Jakarta.
Manor, James (1996), “Political economy of decentralization,” unpublished paper.
Shah, Anwar (1994), The Reform of Intergovernmental Fiscal Relations in Developing
and Emerging Market Economies, Washington, DC: World Bank.
(1995), “Fiscal federalism in Pakistan: challenges and opportunities,” World
Bank, Washington, DC.
Shah, Anwar, Zia Qureshi et al. (1994) “Intergovernmental fiscal relations in
Indonesia: issues and reform options,” Discussion Paper No. 239, World Bank,
Washington, DC (October).
Stone, Andrew (1995), “The climate for private sector development in Pakistan:
results of enterprise survey,” Private Sector Development Department, World
Bank, Washington, DC.
5
Morocco and Tunisia: financing local

governments – the impact on infrastructure
finance
FRANÇOIS VAILLANCOURT
The purpose of this chapter is to present intergovernmental finance

arrangements in both Morocco and Tunisia, focusing particularly on the
financing of local sector investment. These two countries are of interest in
a book of comparative studies since they can be seen as reasonably repre-
sentative of the French tradition in this field, as perpetuated in Africa. Yet,
there are significant differences between them. The chapter has four sec-
tions. The first presents the relevant institutional setting; the second
describes the main characteristics of intergovernmental finance, which in
Morocco requires a presentation of the pre- and post- (1996) reform
systems; the third examines the financing of local investment; the final
section then evaluates this system in comparison to an ideal system of
responsibilities, taxes, transfers, and borrowing mechanisms.
The institutional setting

In this section, the main sociodemographic characteristics of Morocco and
Tunisia are introduced. This is followed by a description of the organiza-
tion of subnational governments in both countries. Finally, there is a com-
mentary on key institutional features.
Sociodemographic characteristics
Table 5.1 summarizes the key sociodemographic characteristics of both
countries. Morocco’s territory encompasses a coastal plain, mainly on the
Atlantic, the Atlas mountain ranges, and a desert region, while Tunisia has
a coastal plain and a desertic region. Both countries depend on seasonal
rain for their agricultural output and substantial parts of their populations
reside on the coast.
I thank participants in the PARADI/ICTS conference and, in particular, Richard Bird and
David Sewell for useful comments on a first draft.
152
Morocco and Tunisia 153
Table 5.1. Morocco and Tunisia:

sociodemographic characteristics
Morocco Tunisia
Area (km2) 710 850 163 610

Population (1994) (thousands) 26 074 8 785
Density (pop./km2) 36.7 53.7
Literacy rate (1990) (percent) 49 65
GDPa (millions) 245 570 15 869
GDP per capitaa 9 418 1 806
GDP per capita (US$) 1 041 1 822
Urbanization (1992) (percent) 47 57
Note: a GDP is for 1992 in Morocco and 1994 in Tunisia,

expressed respectively in Moroccan dirhams and
Tunisian dinars.
Sources: World Bank, World Development Report 1994,
tables 1 and 31; IMF, International Financial Statistics
1996, country tables.
There are two major ethnic groups in Morocco, the Berbers and the
Arabs, who speak the common official language – Arabic – and practice a
common religion – Islam. French is the most widely used European lan-
guage. Morocco has a monarchy whose ruling family can trace its roots to
the prophet Muhammad thus providing it with both a civil and a religious
role. Since independence, the role of the monarchy has been evolving
slowly from a traditional to a constitutional one.
In Tunisia, there is one ethnic group, the Arabs, whose members speak
Arabic, and practice Islam. French is again the most widely used European
language. Tunisia is a republic with a dominant party.
The organization of subnational governments

Table 5.2 shows the organization of subnational governments in the two
countries.
In Morocco, the sixty-five provinces created in 1963 are each headed by
a governor appointed by the central government. Each province has a local
assembly which is elected by members of municipal and city councils.
Provinces are mainly responsible for rural investment and deliver few ser-
vices to citizens.
Municipalities, of which there are 1,544, were created in 1960 and saw
their autonomy and powers increase in 1976 with the Communal
Charter (Charte communale). Some of these duties, listed in table 5.2, such
as urban transit and water, are carried out in some cases by autonomous
154 François Vaillancourt
Table 5.2. Morocco and Tunisia: subnational governments
Main characteristics Morocco Tunisia
First level Provinces: 65 Regions: 23

(Provinces and préfectures) (Conseils régionaux)
Second level Municipalities: 1 544 Municipalities: 257
Urban (cities): 247 Urban (cities): 257
Rural: 1 297 Rural: none
Urban groupings Urban communities: 14 Tunis district: 1
(Communautés urbaines)
Responsibilities of municipalities
Roads and sidewalks Yes Yes
Street lighting Yes Yes
Garbage collection and treatment Yes Yes
Markets Yes Yes
Slaughterhouses Yes Yes
Urban transit Yes (R) No: national government
Water Yes (R) No: SONED
Sewers Yes (R) No: ONAS
Notes: R: Local or regional régies in some cases.

SONED and ONAS are national agencies.
Source: Table ronde sur le financement des investissements municipaux dans des pays du Maghreb
(1995), Washington, DC: World Bank.
establishments (régies). Municipalities are administered by elected

municipal councils (the last elections were in 1997). Each municipality, in
turn, elects a mayor from amongst its members. Municipalities, while
subject to the direction of governors, or the Ministry of the Interior, have
become more independent in recent years. This is a result of various
factors such as the increase in the quality of their personnel and the dif-
ficulty of supervising a larger number of local governments.
The number of municipalities reflects changes in 1992 that increased the
number of cities from 97 to 247 by dividing some existing cities and by
transforming urbanizing rural areas into cities, thus eliminating an inter-
mediate category known as “autonomous centers” (centres autonomies). At
the same time, the number of rural municipalities increased from 760 to
1,297 and the number of urban communities from 2 (Casablanca and
Rabat) to 14, as a result of the growth in urban agglomerations.
Finally, one should note that a 1992 constitutional amendment gave
regions the status of local governments but did not specify their numbers,
powers, or budget. This was clarified in 1996 along with the change in the
legislative system discussed later.
In Tunisia, most regions are made up of both urbanized (communalized)
and non-urbanized (non-communalized) territories: only two are fully

urbanized. In 1994, cities, of which there were 257, included 61 percent of
the population. Regions are headed by a governor named by the central
government and have a regional assembly made up of regional members
of the national parliament, mayors, and appointed members. Populations
in non-urbanized territories are often defined with respect to villages
(agglomerations). Villages have consultative councils used as sounding
boards by the region, but they have no formal recognition as local govern-
ments.
Cities are governed by elected municipal councils that elect a mayor
within their ranks. The number of municipalities has been increasing
through time owing to the decisions made by the central government. In
addition to the responsibilities listed in table 5.2, some Tunisian cities are
involved in the field of welfare through spending on day-care or
allowances to the poor. Following the 1995 municipal elections, there has
been a movement toward decentralization.
Some observations
While the Moroccan and Tunisian communal systems are both inspired by
the French model, there are differences that are worth noting. First, Tunisia
is more centralized than Morocco, and both countries are more centralized
than a North American state or province, or than some European countries.
The greater degree of centralization in Tunisia than in Morocco is perhaps
explained by the following factors: a smaller territory (table 5.1), making it
easier to exercise central control; a greater concentration of population that
yields economies of scale from national programs; and a stronger tradition
of central planning that persists today with the Ninth Plan, to be imple-
mented in 1997–2001 – whereas planning has been abandoned in Morocco.
The greater degree of centralization in both countries than in North
America or parts of Europe is explained, in part, by the fact that lower
overall levels of education and fewer amenities in smaller centers make it
difficult to recruit competent municipal staff. Other factors such as the
impact of colonization may also matter. In both countries, primary and sec-
ondary education and local police services are the responsibility of the
national government as in the European (particularly the French) tradition.
A second noteworthy difference is that Tunisia’s territory is not fully
divided into municipalities whereas Morocco’s is so divided. This is both
a plus and a minus in fostering decentralization. A positive aspect is that
the more homogeneous municipalities in Tunisia are implementing
decentralization policies, since the managerial capabilities of the less
well-endowed municipalities are still fairly good. This matters since the
formal equality of the urban and rural municipalities in Morocco prevents
the urban ones being treated differently by giving them more power com-
mensurate with the greater sophistication of their staff, elected leadership,
and better-informed electorate.
On the negative side, the absence of rural municipalities or counties as
a result of the legal framework means that, even if the central government
wished to devolve some powers to the rural areas, it could not do so at the
present time. As a result, Tunisians living in similar-sized small
agglomerations (less than 5,000 inhabitants) are faced with different
degrees of control and different tax burdens, depending on whether their
agglomeration is a commune or not. If it is a commune, they elect a com-
munal council that collects local taxes, receives subsidies, borrows, carries
out investments, and provides local services. If it is not a commune, there
is no council; therefore, no local taxes, subsidies, or loans are received, and
the region provides local services and carries out investment.
Intergovernmental finance
In this section, we present first, for the two countries in turn, the local tax
system and the intergovernmental transfer scheme. In the case of Morocco,
we also consider the reform of intergovernmental transfers. We then turn
to the relative importance of these sources of funds and finally comment
on both countries.
The funding mechanism: Morocco

Tax revenues
Own revenues in Morocco are collected from three sources:
(1) two national taxes assessed and collected by the central government,
the receipts of which are transferred (less a 10 percent collection fee) to
local governments; these are the urban tax (taxe urbaine) and the busi-
ness tax (patente);
(2) one local tax assessed and collected by the central government, the
supplementary urban tax (taxe d’édilité);
(3) thirty-six local taxes and fees assessed and collected by the local
authorities.
The rates for centrally collected taxes are set by the central government.
The rates for local taxes are set either centrally, locally within a centrally
set rate band, or, in a few cases, just locally.
Table 5.3. Morocco: parameters of urban tax and supplementary urban tax
Urban tax Supplementary urban tax
Rates 0 to 30 percent, depending on value 6 or 10 percent depending on

location (periphery/center)
Taxable value Owner-occupied: 25 percent of Owner-occupied: 25 percent of
assessed value assessed value
Rented: not taxable Rented: 100 percent of assessed
value
Empty units Not taxable Not taxable
New buildings Exempt for five years Not exempt
Source: Information provided by Moroccan officials.
Two taxes (taxe urbaine and taxe d’édilité) are assessed on the rental value
(valeur locative) of taxable units. Leases are used to establish rental values
directly or through comparisons between owner-occupied and leased
units.
For housing units, the parameters listed in table 5.3 are used. For non-
housing units, the urban tax rate is 13.5 percent and the supplementary
urban tax rate, 6 or 10 percent. In addition, these commercial or indus-
trial units must pay a business tax set according to the rental value, the
type of business, and the nature and quantity of inputs used. As a result,
several hundred categories are used in assigning one of six business tax
rates. Note that business tax revenues collected in urban communities are
distributed between cities on the basis of their populations.
The three main local taxes and fees are:
(1) market fees (wholesale, fish, souks, and slaughterhouses) that account
for 28 percent of revenues;
(2) forestry income (rural municipalities) (19 percent of revenues);
(3) construction fees (14 percent).
One notes the variety of tax bases used, for example, sales and opening
hours (beverage houses), number of students (private schools), and so
on.
Transfers
Since the 1984 tax reform, Moroccan local governments have been entitled
to at least 30 percent of national value-added tax (VAT) receipts. This
entitlement replaced an entitlement to other indirect taxes which them-
selves replaced gate fees (droits de porte). Since the implementation of VAT
in 1986 and of the transfers in 1988, this 30 percent has been both the
minimum and maximum share of VAT receipts transferred.
From 1988 to 1995, VAT transfers were divided into current and capital
transfers. Current transfers were calculated as budget-balancing grants
(subvention d’équilibre). Local governments submitted their forecast
revenues, expenditures, and deficits to a joint Ministry of the Interior/
Ministry of Finance committee that examined them, modified them, and
set the subsidy. Not surprisingly, such a system led to ever-increasing
deficits. In particular, expenses included both principal and interest pay-
ments on loans (annuités) leading to costless borrowing with budgetary
priority (inscription d’office) given to repaying the lending agency (Fonds
d’équipement communal [FEC]).
Capital transfers were granted once current transfers had been set. Since
budgets were often approved only toward mid-year, capital spending
often lagged behind receipts by one year. In addition, the promise of 30
percent of VAT revenues quickly proved a strenuous one for the central
government. As a result, some central government capital expenditures,
such as primary school buildings and rural electrification, were designated
(transferred) as local investment (charges transferées) along with some
national expenditures deemed to benefit local governments (charges com-
munes). As a result of this shifting from central to local government of the
central deficit, investment funds available to local governments have
dwindled in recent years.
The disincentives associated with deficit financing quickly became
apparent to the Moroccan government, which decided in 1992 to replace it
with a formula-based apportionment system. The introduction of the
formula, originally planned for 1994, was delayed by the following factors.
First, the creation in 1992 of several hundred new municipalities meant
that 1992 data could not be used to calculate the various parameters used
in the new formula. Since 1993 data would not be available until mid-1994,
a decision was taken in early 1993 to move implementation from January
1, 1994 to January 1, 1995. Second, it was considered appropriate to update
population figures using the 1994 census rather than rely on data from the
previous census in 1982. This made implementation in 1995 difficult, but
not impossible. Third, there was a debate within the Moroccan administra-
tion as to the advisability of this change. Within the Ministry of the Interior,
some officials resisted this change since it reduced the discretionary power
of the government, while others defended it for precisely that reason as
well as with arguments of predictability and simplicity. Within the govern-
ment, there was some resistance from the Ministry of Finance since this
would have an impact on its ability to relabel central expenditures as local
ones. The formula was put in place for an eighteen-month budgetary year
(January 1996–June 1997) made up of the 1996 (transitional six-month
period) and 1996–97 budget years. This change in the dates of the bud-
getary year applies to both central and local governments. It was intro-
duced to provide better forecasting of economic conditions since the
budgetary year now starts after the rainy season. This allows the Moroccan
government to make decisions based on good information gathered on
expected crop yield.
The transfer formula in place in 1996 can be summarized in the follow-
ing steps.
(1) Establish the predicted amount of VAT for the fiscal year and multi-
ply it by 0.3. This yields the amount estimated to be available for local
governments.
(2) Reserve 30 percent of this 30 percent local share of VAT (9 percent of
VAT revenues) for transferred charges (15 percent), common charges
(10 percent) and the adjustment fund (5 percent), that is, unexpected
expenditures on items like natural disasters. This 30 percent reserve
and its composition can vary in the future. For example, transferred
charges are expected to drop in 1997–98. However, one may see a
need to fund a new order of local government as regions appear.
Regions were created by a constitutional amendment in 1992, but
their powers were defined only by a second set of amendments in
September 1996. They number sixteen and are entitled to a share of
the VAT revenues and to shares of the PIT and CIT that remain to be
set.
(3) Allocate the remaining 70 percent as follows: urban municipalities
and communities (UMC), 30 percent; rural municipalities (RM) 20
percent; and provinces and prefectures (PP) 20 percent.
(4a) Provinces and prefectures (PP). Each PP receives its 1995 wage bill.
The sum of these wage bills is subtracted from the 20 percent, and the
remainder (of the 20 percent) is allocated using three weighted crite-
ria: identical fixed amounts for each PP (one-sixth); the area of the PP
(one-sixth with minimum and maximum); and the population of the
PP (two-thirds with a minimum).
(4b) UMC and RM funds are allocated according to the weighted sum of
three criteria: lump sum (UMC: 15 percent; RM: 30 percent), equaliza-
tion (UMC: 70 percent; RM: 55 percent), and tax effort (15 percent for
both). Equalization targets are 1.25 times the average tax potential
and payment is capped at a maximum of 2.5 times the average
population. UMC tax potential is measured by the patente, taxe
urbaine, and taxe d’édilité, while for RM, incomes from the tax on the
sale of forest products (a half) and market fees are also used. In the
case of UMC, this is reasonable since the assessed taxes are calculated
using national tax rates and tax rolls prepared using national norms
and rules. In the case of RM, the use of taxes on forest product sales
and market fees poses a problem because, in both cases, these are real-
ized rather than assessed amounts that thus depend on forest
resources and on the decision to sell forest products and on the tax
and market fee collection effort. Equalization may reduce these
amounts.
Tax effort is measured using other taxes. Municipalities with a tax
effort greater than 0.65 times the average tax effort receive an amount
depending on their share of these extra tax revenues (half per capita,
half gross amount).
(4c) The actual amount of transfers depends on actual VAT revenues.
(5) In 1996–97, transition rules yielded a subsidy at least equal to that of
1995. A three- or four-year transition is planned for the progressive
replacement of the old subsidy by the new one.
The funding mechanism: Tunisia

There are two main sources of own revenues
(1) direct taxes made up mainly of the rental tax (taxe locative or TL), the
business tax (TCL), the hotel tax (taxe hôtelière or TH), and the tax on
unbuilt land (TTNB);
(2) indirect taxes and fees including an electricity tax and market fees.
Direct taxes
The two main taxes (about 40 percent each of direct tax revenue) are TL
and TCL.
The TL (now superseded by the TIB, see below) is set nationally at 20
percent of rental value and is collected by the Ministry of Finance on
housing units (industrial, commercial, and professional buildings are
exempt), with the tax roll prepared by the municipality or commune (CU)
through a triennial or quinquennial census. The law prescribes using rent
as the tax base for rental property, while comparable values are set for
owner-occupied housing. Rental information is to be obtained from the
occupant. It is sometimes checked against information in registered leases,
which itself may be inaccurate. New buildings are exempt for five years.
In practice, communes use either the number of rooms or the surface floor
area of the housing unit along with values per room or per square meter
that vary according to the neighborhood to set the tax base for non-rental
units.
The business tax is payable by individuals who pay income tax on
profits or professional income, partnerships, and corporations. It is usually
collected as one-fifth of 1 percent on local gross business income, subject
to a maximum of 20,000 D per business and to a minimum of the notional
TL that would have been collected if these establishments had been subject
to the TL. Such a minimum requires the establishment of a business TL roll,
something that is done in only a few communes. The main issue with the
TCL is that businesses do not provide a proper breakdown of their activ-
ities by communes. As a result, communes like Tunis, where head offices
are located, benefit unduly from it.
Indirect taxes
Other taxes include the electricity surtax collected centrally with a rate of
2 millièmes (1000 millièmes ⫽ 1 D) per kilowatt-hour and justified as
financing street lighting. It also allows the electricity board to cover some
of the expenses of municipal governments. Other taxes include the
entertainment tax (the rate varies between 3 and 18 percent depending on
the type of entertainment) and the tax on beverage houses with more than
ninety tax rates.
In early 1997, the Tunisian parliament adopted modifications to local
taxes to be implemented from January 1, 1997. The main changes were:
• Introduction of the TIB (taxe sur les immeubles bâtis). It requires com-
munes to use notional values per square meter of floor area to establish
the rental tax roll. The proposed values, to be set by presidential decree,
will vary from 2 to 8 dinars per square meter, depending on the size of
the housing unit and the neighbourhood. The TIB replaced the 20
percent tax rate by four tax rates of 8, 10, 12, and 14 percent. These tax
rates will be applied by the communes on an area basis. Areas will be
classified according to the availability of six services: garbage collection,
street lighting, covered roadway, covered sidewalk, sanitary sewers,
and rainwater sewers. The tax rates will be set as follows: one to two ser-
vices: 8 percent; three to four services: 10 percent; five to six services: 12
percent; five to six services and other or better quality services: 14
percent.
• Raise the maximum TCL payable from 20,000 to 50,000 D and distribute
the TCL (and the TH) more accurately, using square meters if necessary
for multi-establishment enterprises.
• Simplify the entertainment and beverage house taxes.

• Introduce a contribution for new public parking to be paid by builders
of new housing units with insufficient private parking.
Transfers
Current transfers are paid out by the Fonds commun des collectivités locales
(FCCL). The budget of the FCCL is set by the central government and is
not linked to any specific tax source. It is reviewed annually and is allo-
cated as follows.
• A central reserve receives 25 percent while the communes (CU) and
regional councils (CR) combined receive 75 percent.
• The 75 percent is allocated by decree between CUs (86 percent) and CRs
(14 percent).
The total amount for communes is allocated according to three criteria:
their population (45 percent), their tax effort measured by a three-year
average of collections of the TL (45 percent), and a flat-rate amount for each
(10 percent). Similarly, the total amount for regional councils is allocated
according to their population (85 percent) and with a flat-rate amount for
each (15 percent).
The central reserve is allocated to Tunis (Ville, District, CR), to regional
capitals (chef lieu), to civil protection, to ONAS (Office National de
l’Assainissement, responsible for sewers), and to the Caisse des préts et sub-
ventions aux collectivités locales (CPSCL), the lending agency for local
governments.
Sources of funds for local governments

Tables 5.4 and 5.6 present, for Morocco and Tunisia respectively, the rela-
tive importance of own revenues (two kinds) and transfers in current rev-
enues for two years for all local governments, while tables 5.5 and 5.7
present, again for the two countries, own revenues and transfers by type
of local government for one year. These tables show that by 1993, both in
Morocco and Tunisia, transfers accounted for 35–40 percent of local
government revenues. However, they also show different trends in
Morocco (growing) and Tunisia (decreasing) in the importance of trans-
fers. Again, in both countries, municipalities get a greater share of their
revenues from their own sources than from transfers, while provinces and
regional councils depend more heavily on transfers, especially so in
Morocco. More recent data (not shown) indicate that this situation has not
changed in recent years.
Table 5.4. Morocco: sources of funds, local governments, 1991 and 1993
1991 1993
DH millions % DH millions %
Tax revenues (1) 3 810.3 72.2 3 713.1 60.7

Major taxes (2) 2 112.4 40 1 775.9 28.7
Local taxes and fees (3) 1 697.9 32.2 1 957.1 32.0
VAT deficit subsidy (4) 1 467.2 27.8 2 403.1 39.3
Total current revenues (5) 5 277.6 100 6 116.2 100
Note: (1)⫽(2)⫹(3); (5)⫽(1)⫹(4).

Sources: “L’équipement local et son financement: cas du Maroc,” in Table ronde sur le
financement des investissements municipaux dans les pays du Maghreb (1995), Washington, DC:
World Bank; information provided by Moroccan officials.
Table 5.5. Morocco: sources of funds by type of local government, 1993
Sources of funds Provinces Rural municipalities Urban municipalities
Total (DH millions) (1) 832.5 1 131.3 4 152.4

Tax revenue (% of total) (2) 12.7 71.6 56.2
VAT deficit subsidy (% of total) (3) 87.3 28.4 43.8
% of local sector (4) 13.6 18.5 67.9
Note: (1)⫽(4)⫻(5) in table 5.4.

Source: Information provided by Moroccan officials.
Table 5.6. Tunisia: sources of funds, local governments, 1992 and 1993
1992 1993
DI millions % DI %
Tax revenues (1) 116.0 61.6 128.6 63.1

Major taxes (2) 46.2 24.5 51.8 25.4
Local taxes and fees (3) 69.8 37.1 76.8 37.7
FCCL subsidy (4) 72.3 38.4 75.3 36.9
Total current revenues (5) 188.3 100 203.9 100
Note: (1)⫽(2)⫹(3); (5)⫽(1)⫹(4).

Source: Information provided by Tunisian officials.
Table 5.7. Tunisia: sources of funds by type of local

government, 1993
Sources of funds Municipalities Regional councils
Total (DI millions) (1) 183.5 201.4

Own revenues (% of total) (2) 65.1 45.8
Transfers (% of total) (3) 34.9 54.2
% of local sector (4) 90 10
Note: (1)⫽(4)⫻(5) in table 5.6.

Source: Information provided by Tunisian officials.
The size of the envelope

In Morocco, the legal size of the envelope is predetermined at 30 percent
of VAT revenues, with the actual size depending on the state of the
economy, which impacts on VAT revenues and on mandated expenditures
(charges transferées, charges communes) by the central government. In
Tunisia, the amount is set as part of the annual budget process. Both
approaches have the weakness of single-year budgeting, with uncertainty
as to a major source of revenue for local governments. The Moroccan
system has the additional difficulty of requiring the central government to
claw back resources to achieve budgetary equilibrium. The Tunisian
system allows, at least in principle, for changes in overall funding com-
mensurate with both changes in responsibilities and in autonomous
revenue potentials, regardless of the evolution of a given revenue source.
Before 1987, the Tunisian system was similar to the Moroccan one with the
funding of the FCCL, created in 1975, depending mainly on the revenues
from tax on gross business income.
In both countries, the responsibilities and funding of local governments
are set centrally by a combination of laws enacted by parliament and
decrees issued by the government. Since local governments are not seen as
being a level of government in their own right, they do not formally nego-
tiate these matters with the central government. That said, one should note
that in Morocco one-third of the members of parliament are not elected by
the population, but by various groups including local authorities. As a
result, some members have special expertise and interest in local matters.
In a reform implemented in 1997, the unicameral system was replaced by
a bicameral system. All members of the lower chamber of parliament are
directly elected, but two-thirds of the members of the upper chamber are
elected by local and regional bodies.
The allocation formulas

The criteria used to allocate amounts between the rural and urban
municipalities or between regional/provincial and local authorities are
not clearly spelt out in either country. This leads to disparities on a per
capita basis that may be justified by needs.
With respect to allocation among municipalities, the relevant compari-
son is between the post-1996 Moroccan formula and the Tunisian one,
since, before 1996, the Moroccan allocations were settled on a case-by-case
basis. The key difference is that the Moroccan formula now uses both defi-
ciencies in fiscal potential (equalization) and tax effort, while the Tunisian
one uses population (as a needs indicator) and tax effort. The Moroccan
formula partially corrects differences in fiscal potential and thus favors the
poor, while the Tunisian formula amplifies differences in fiscal potential as
measured by collections and favors the rich.
These differences raise two questions. First, why do they exist and
second, which approach is preferable? The answer to the first question is
not obvious. One part of the explanation is that the Moroccan formula was
put in place ab novo, drawing on foreign experiences as a source of inspira-
tion, while the Tunisian one evolved from a situation where tax efforts (roll
preparation and collection) were felt to be insufficient and thus in need of
encouragement. Another part is that in Morocco’s national assessment,
norms for major sources of revenues provided the tax potential informa-
tion required to introduce an equalization formula, whereas in Tunisia
there are no national norms. The local tax reform of 1997 in Tunisia will
generate such a national tax base, thereby allowing the use of fiscal poten-
tial. In 1998, such use was being considered to assist the poorer CUs. As for
which approach is preferable, there is no reason for transfers to replicate
local revenue patterns. If these patterns are deemed appropriate from a
distributional perspective, then central government revenues should be
lowered to allow local authorities to increase their tax effort so as to collect
these additional revenues themselves.
The nature of own revenues

There are several factors worth noting here. First, in both Morocco and
Tunisia, local governments have very little freedom in setting tax rates or
user fees and are not legally responsible for tax collection, although they
may encourage it and provide resources to facilitate it. They also play a
more limited (Morocco), or small (Tunisia), role in establishing their tax
rolls. Taking into account national and local taxes, we estimate that in
Morocco local governments receive 70 percent of their own revenue from
taxes with centrally set tax rates, while in Tunisia this share is 65 percent.
As a result, they do not control their own revenue and cannot therefore
increase or decrease their provision of local services freely in accordance
with the wishes of their electorate (they can neglect to collect taxes and to
provide services, but this will create tax arrears and unmet needs).
Second, there are similar taxes levied on a single residential tax roll that
could be amalgamated to simplify collection, but which remain separate
because of tradition or small differences in the tax statutes. One reason for
this lack of reform is that these taxes are relatively unimportant in the
national perspective and thus do not attract the attention of key policy-
makers in the Ministry of Finance, where the rate and base setting author-
ity resides.
Third, the two business taxes have shortcomings. The Moroccan patente
tax roll is constructed by acquiring information on the types of goods sold
and on the nature and amount of input used. Each establishment is then
assigned to one of six classes of businesses with tax rates varying from class
to class. To give an example, a dentist’s office will see its tax rate vary
between 15 and 30 percent of rental value according to the number of chairs
and the availability or not of prosthesis-making facilities. This creates micro-
economic distortions in the choice of input mixes but does not take into
account the environmental impact of industrial and commercial activities.
As for the Tunisian TCL, it is levied on gross business income. It does not
take into account the use of land or public infrastructure. As a result, high-
yield square meters in luxury shops generate more TCL than low-yield
square meters in a trucking/warehousing complex, even if the nuisance
and impact of the latter are higher.
Financing local investment

In the third part of this chapter, we present the various sources of funds
available to local governments in both countries and then examine them in
turn.
Sources of funds
The funding mechanisms are summarized in table 5.8. In general,
Moroccan municipalities have more freedom in their investment and bor-
rowing decisions than their Tunisian counterparts and bear a greater share
of the cost. However, the fit between national and local priorities is
stronger in Tunisia. This is achieved, in part, by the subsidy scheme
described in table 5.9.
Table 5.8. Morocco and Tunisia: main characteristics of local government

investment finance, 1995–96
Morocco Tunisia
Own funds available Yes. Excess of current revenues Yes. Excess of current revenues
over expenditures ⫹ sales of over expenditures ⫹ sales of
assets assets
Alternative use Held by government treasury Held by government treasury
Low-yield deposit Low-yield deposit
Loans agency FEC – monopoly CPSCL – monopoly
Interest rate Market 12–13% Non-market 6.5%
(inflation ⬄ 6%) (inflation ⬄ 5%)
Subsidy Before 1995: yes (VAT share) Yes. By CPSCL or ministry with
Since 1996 : no various rates depending on
project (see table 5.9)
Take up condition None Twinned with loan for a
planned project
Project selection By local government on an By consensual process to
annual basis. prepare a five-year plan (PIC).
Loan-financed projects must Projects not in the plan can be
meet repayment capability test. financed by own resources.
Source: Table ronde sur le financement des investissements municipaux dans les pays du Maghreb
Table 5.9. Tunisia: sources of funds for investments by local governments, 1995
Sources of funds (%)
CPSCL Funds
Type of investment Own resources Loans Subsidies
Roads, streets, lighting, sewers 30 37 33

Waterworks, garbage collection centers 20 35 45
Cultural and sporting equipment 18 18 64a
Rehabilitating low-income areas — 30b — 70
Economic investment (slaughterhouses, markets, etc.) 40 60 0
Buildings 50 50 0
Equipment 28 72 —
Green spaces 37 41 22
Technical studies of investment projects 0 100 0
Notes:
a
Not by CPSCL but by the relevant ministries.
b
The exact percentage of own resources depends on the commune concerned.
Source: Table ronde sur le financement des investissements municipaux dans les pays du Maghreb
Table 5.10. Morocco and Tunisia: sources of

investment financing, five-year periods
(percent)
Morocco Tunisia
(1990–94) (1992–96)
Savings 29.3 22.9

Subsidies 45.7a 39.1b
Borrowed funds 25.1 38
Notes:
a
These VAT investment subsidies (ended January
1, 1996) include some transferred investments.
b
22.4 percent of the funds are from the CPSCL and
16.7 percent from various agencies and
departments (mainly ONAS and Youth Ministry).
Source: “L’équipement local et son financement:
cas du Maroc” and “Financement des communes,”
in Table ronde sur la financement des investissements
municipaux dans les pays du Maghreb (1995),
In both countries, the lending agency – FEC or CPSCL – has recently

been the subject of reforms that increased its administrative and funding
capabilities and, in Morocco, its autonomy. Given that, and the changes in
the VAT transfer in Morocco and in local taxation in Tunisia, one may
expect that the share of subsidies reported in table 5.10 will decrease
through time.
Some observations
Insofar as cash balances yield low or even negative real returns, there is an
incentive for local governments to invest in low-yielding projects that pay
a slightly higher rate of return. Cash balances should yield the market rate
of interest, preferably in a competitive market for deposits.
The rate of interest on borrowed funds should also be the market rate of
interest in order to face localities with the real cost of funds. Subsidies
should be paid up front in a lump sum (Pechon, 1995).
The use of a variable subsidy rate to reflect the degree of national prior-
ity, which in turn depends on various factors including externalities, is a
good signaling technique. There is no reason, however, to link the receipt
of a grant to that of a loan for local governments with own funds available.
To do so requires them to borrow at a rate higher than the yield on these
funds.
Overall assessment
In the first three sections, various aspects of the local government systems
in Morocco and Tunisia are presented and examined. In this section, an
overall assessment is provided. Responsibilities and revenues are exam-
ined first, by putting forward three questions, examining what theory and
best practice indicate are sensible answers, and confronting Moroccan and
Tunisian institutions and practice with these answers. A checklist provided
by Bird (1994) is then used to examine the financing of infrastructure.
Have the appropriate functions been devolved to local governments?

There is by now a well-accepted body of guiding principles on what crite-
ria should be used to assess the level of decentralization. They can be
summarized as follows for any function:
• the more important the economies of scale in production, the more
centralized production should be;
• the greater the geographically based heterogeneity in tastes for a pub-
licly provided service, the more decentralized production should be;
• the greater the area of impact associated with the production of a
service, the more centralized it should be so as to avoid unwarranted
spillovers between small jurisdictions;
• the greater the diversity in possible production processes and the larger
the number of varieties of local services, the more decentralized produc-
tion should be to encourage emulation.
These criteria justify the provision of all the services listed in table 5.2 by
local governments. Yet, as shown there, this is not the case in Tunisia. One
can thus state that the appropriate functions have been formally devolved
in Morocco but not in Tunisia. If one takes into account that local govern-
ments remain, in both countries, under the supervision of the Ministry of
the Interior, usually through provincial/regional governors, that their key
staff are named by either the Ministry of the Interior (secretary-general) or
the Ministry of Finance (tax collector) and that staffing levels and wages
are subject to central approach, then effective decentralization is less than
formal decentralization and inadequate in both countries.
Do local governments have appropriate tax revenues?

The taxation literature shows that local governments should finance them-
selves first by user fees for local services, and second, for services which
are not easily priced, by local taxes, such as a property tax, borne by local
users and not exported to residents of other jurisdictions (as business taxes
can be).
Both Moroccan and Tunisian municipal governments receive about 60
percent of their revenues from taxes and fees and 40 percent from current
transfers. This own-revenue share is low by North American standards,
such as the 90 percent observed in Quebec (Vaillancourt, 1995), but not
unusual when compared with various European countries in 1990 (France
65 percent, United Kingdom 40 percent, Germany 55 percent) (Costa,
1996). The fact that, in both cases, most tax rates and fees are not set
autonomously is of greater concern, since it seriously limits local freedom
in establishing the appropriate prices for services.
Are transfer mechanisms appropriate?

Transfers from national to subnational governments have two main roles.
First, they can be used to increase the provision of a service that is under-
supplied by the local governments, usually because they do not take into
account geographical or intergovernmental spillovers. Second, they can be
used to equalize tax potentials between local governments, thus allowing
poor local governments to offer comparable levels of services to those of
the richer ones.
In the case of Morocco, transfers have had an explicit equalization goal
since 1996, although specific spending has not been targeted. In the case of
Tunisia, current transfers increase or at least maintain disparities between
local governments, hence are non-equalizing, while capital transfers are
used to encourage certain kinds of investments that do not, however, have
large spillovers, since drinking water and sewers are nationally provided.
Overall, the Moroccan system is more appropriate than the Tunisian one.
Having examined decentralization in general, let us now turn to the more

specific issue of infrastructure and decentralization. Bird (1994) argues that
decentralization is not necessarily bad for the provision of infrastructure
and that what matters are the incentives. They will be appropriate if they
make “those who make the decision bear the financial (and political) con-
sequences” (Bird, 1994:30). Thus, politicians should bear the costs of their
mistakes, local citizens should pay for local services and local decision-
makers should be accountable for the use of revenues. This requires an
appropriate framework of reporting mechanisms.
Using these criteria, one concludes that the Moroccan institutions are
more appropriate, since Moroccan local authorities must bear the full cost
of their freely chosen investment, while Tunisian local authorities must

negotiate their investment with the central government but benefit from
both upfront and interest rate subsidies.
Conclusion
This chapter has presented and discussed intergovernmental finance
arrangements in Morocco and Tunisia. While both countries have been
influenced by the French system, there are important differences in their
institutions. First, the local sector is less important in Tunisia than in
Morocco (2.5 versus 3.5 percent of GDP: Vaillancourt, 1995) but is less
dependent on transfers for current revenues. Second, the new Moroccan
transfer system is more redistributive than the Tunisian one which is under
review. Third, Moroccan local authorities have more freedom than the
Tunisian ones in choosing investment projects but face market interest
rates.
In some cases, institutional differences are accounted for by these differ-
ences in finance arrangements, while in others, they are hard to justify.
References
Bird, R. (1994), “Decentralizing infrastructure for good or for ill,” Policy Research
Working Paper 1258, World Bank, Washington, DC.
Costa, A. (1996), “Finances locales, une comparaison européenne,” Problèmes
économiques 2481 (August 7 1996): 20–1.
Pechon, F. (1995), “Le rôle du crédit et ses perspectives d’évolution,” in Table ronde
sur le financement des investissements municipaux dans les pays du Maghreb,
Vaillancourt, F. (1995), “Les collectivités locales dans l’économie: analyse et
comparaisons internationales,” in Table ronde sur le financement des investisse-
ments municipaux dans les pays du Maghreb, Washington, DC: World Bank.
6
Colombia: the central role of the central

government in fiscal decentralization
RICHARD M. BIRD AND ARIEL FISZBEIN
As in a number of other countries in Latin America, prolonged and some-

times violent conflict between “centralists” and “federalists” occurred in
Colombia during the latter part of the nineteenth century. As in most of the
continent, the centralists won.1 Despite marked regional differences and
persistent strong regional identities fostered by the difficulties of travel in
a country broken up by mountainous terrain, Colombia’s governing struc-
ture remained highly centralized until very recently. The central govern-
ment (and its many “decentralized” agencies) not only controlled almost
all public revenues and expenditures but also, outside of the largest cities,
virtually supplanted the traditional territorially based governments, the
departments and municipalities, in providing even the most local of ser-
vices. Little scope and little reward existed for local initiative, and the most
successful local politicians were those who could best exploit the
labyrinthine central government system for the benefit of their constitu-
ents.
By the late 1970s, however, this system had begun to break down, for
two reasons. First, and most importantly, central government finances
were increasingly strained by the task of financing the expansion of local
public services to an increasingly urbanized population. This strain was
perhaps felt earlier and more strongly in Colombia than in other Latin
American countries owing to the country’s long tradition of maintaining a
relatively stable and conservative central government fiscal policy.2
Second, the highly centralized system was also contributing to political
unrest, as such critical services as water, education, and health were failing
to reach large segments of Colombian society. In the eyes of some reform-
ers, elected local authorities with resources adequate to meet local needs
appeared increasingly necessary. Whether as a response to these pressures,
Much of this chapter is based on a recent report prepared for the World Bank by the authors
with the assistance of several colleagues (World Bank, 1996a), but it should be emphasized
that the views expressed here are the personal opinions of the authors and are not to be attrib-
uted to any other person or organization.
172
Colombia 173
its own fiscal needs, or a desire to implement its own policies more effec-
tively, the central government looked to decentralization as one way to
deal with its fiscal difficulties, as indicated by its appointment in 1980 of a
Commission on Intergovernmental Finance to study ways in which to
increase reliance on local resources for local purposes.3
The report of this Commission turned out to be only the first of a series
of important measures of fiscal decentralization which have gone on to this
day, especially as one part of a major constitutional reform in 1991.4
Following the continuing and commendably rational path the process of
decentralization has taken to date in Colombia, the law implementing the
fiscal decentralization mandated by this reform (Law 60) contained a
“sunset” clause requiring formal evaluation and, if necessary, revision in
1996. This chapter reflects our participation in this process of evaluation.
Reforming the decentralization law means very different things to
different people in Colombia. The core economic ministries of the central
government – Finance and Planning – are concerned mainly with the loss
of control as a result of the decentralization of revenues to subnational
governments. In their view, these increased transfers lie at the root of the
fiscal difficulties recently experienced by the central government, so their
inclination is clearly to reduce the size of these transfers as well as to
increase central government control of transferred resources. Sector min-
istries such as those in charge of health and education would like to use the
opportunity to reform the law both to increase the amount of resources
available for their sectors and to pursue their own reform agendas.5
Although municipalities and departments do not appear to have a clear
agenda, they would of course like to increase local control over resources
without any reduction in the size of the transfers. Congress, which is in
charge of the reform, similarly does not appear to have a clear agenda and
would rather avoid changes if possible.
In these circumstances, our evaluation of the situation is that no major
overhaul of the existing system is needed. Instead, what is required is a
judicious combination of (1) fine-tuning to eliminate some problems
arising from the present system of transfers, (2) considerable simplification
of the existing rules governing intergovernmental relations, and (3) more,
not less, decentralization. As we argue in this chapter, however, the central
problems relating to decentralization in Colombia require more, not less,
attention by the central government to improve its own performance in a
variety of ways. Once this is done, preferably along the lines sketched here,
the present system of fiscal decentralization as set out in Law 60 should,
we argue, produce broadly acceptable results.
The next three sections of this chapter briefly outline the current system
174 Richard M. Bird and Ariel Fiszbein
of intergovernmental finance in Colombia and our suggestions for change.

The remainder of the chapter then considers several aspects of fiscal decen-
tralization that have given rise to much concern in recent years in
Colombia – as in other countries – with respect to its effects on the pattern
of local expenditures, the level of local taxes, and, more broadly, macro-
economic policy.6 A brief concluding section considers what lessons
Colombia’s on-going experience with fiscal decentralization may have for
other countries.7
The system of intergovernmental finance

Although far from complete – and even further from perfect – the process
of fiscal decentralization has gone a long way in Colombia, and seems
likely to go further, in part at least because of the strongly regional char-
acter of the Colombian economy and a series of recent political and
constitutional reforms that have made this regionalism fiscally real in a
way that has not been true for a century. Unlike many other countries in
Latin America, Colombia has never been dominated economically by its
capital, Bogotá. Four cities have over a million inhabitants, and there are
thirty cities with populations over 100,000. With the election of municipal
mayors and, more recently, departmental governors, regions such as those
centered around the other major metropolitan areas (Medellín, Cali, and
Barranquilla) are no longer as politically subordinate to Bogotá as in the
past.
In total, Colombia has 33 intermediate-level general governments (32
departments and the special district of the capital city), several national
territories, and over 1,000 municipal governments. The assignment of
expenditure responsibilities is by no means sharp: cities, departments, and
the national government, for example, all have sections in charge of areas
such as education, health, and transport and there are often other public
decentralized agencies involved in the provision of the same services. In
practice, however, municipal governments are responsible for local ser-
vices (streets, water, and garbage removal), and share responsibility for
health and education with the regional (departmental) governments.8
Variations from place to place in population density and most relevant
economic and fiscal variables are considerable. Although there are over
700 municipalities with less than 15,000 inhabitants, over 70 percent of the
population is classified as “urban.”9 About 30 percent of the population of
about 35 million is concentrated in the three most prosperous and urban-
ized areas of Bogotá, Antioquia (Medellín), and Valle (Cali). Since these
three areas also account for close to half of Colombia’s gross domestic
Colombia 175
product (GDP), it is not surprising that there are wide differences in

regional incomes, with ten departments having average per capita GDP of
less than half the national average (which is about US$1,400). In contrast,
per capita GDP in the sparsely populated area of Arauca is well over
US$5,000 as a result of oil production. Generally, however, the ratio of per
capita incomes in rich and poor regions is closer to two, or at most three,
to one. Within each region, however, income dispersion between urban
and rural and within each category is considerably greater. The very
poorest people in Colombia are found in rural areas, but close to half the
poor are in large urban areas.10
Although most of the statistics just cited come in the form of “depart-
mental” numbers, the role of departmental governments, as table 6.1 indi-
cates, is much less important than that of municipal governments, for two
reasons. First, the general departmental governments have few and inelas-
tic resources at their disposal (mostly from beer, liquor, and tobacco taxes)
– in 1994, for example, departments accounted for less than 9 percent of
total government revenues, compared to over 12 percent for municipalities
– and few functions on which to spend them. Superficially, this may not
seem to be true in view of the important role of departments in providing
health and education services: the same source, for example, shows
departmental expenditures as 19 percent of the total (compared to 21
percent for municipalities).11 As noted below, however, departmental
governments as such have until now played little real role with respect to
these important functions, despite their formal responsibilities.
This situation is in the process of change, however, in part because the
second reason why departmental governments have been unimportant,
namely, that governors were appointed by the central government,
changed in 1991 with the first regional elections of governors. The election
of municipal mayors since 1986 has undoubtedly been a principal reason
for the growing importance of municipalities in the Colombian fiscal and
political scene. Similarly, the future for departmental governments may be
different from the recent past. Nonetheless, for the most part the remain-
der of this chapter will focus on the municipal level, which is where the
major decentralization to date has occurred and where the effects are most
marked.
The importance of subnational governments

Table 6.1 shows the structure of the Colombian public sector in 1985, 1990,
and 1994.12 Somewhat surprisingly, these data suggest that the subnational
sector as a whole has not become more important over this period as a
Table 6.1. Colombia: the structure of

the public sector (percentages)
1985 1990 1994
National
Revenue 73.9 75.1 79.2
Expenditure 63.8 56.4 59.8
Departmental
Revenue 13.6 13.1 8.4
Municipal
Revenue 12.4 11.8 12.4
Source: Colombia, Departamento Nacional

de Planeación.
share of total public sector activity. In terms of GDP, however, there has
indeed been a marked growth in the importance of the subnational sector.
In 1994, total expenditure at the departmental and municipal levels came
to about 12.6 percent of GDP, compared to 9.2 percent in 1985.13 Combined
departmental and municipal own-source current revenue in the same
years (based on the data underlying table 6.1) was only 6.7 and 5.5 percent
respectively. The conclusion seems obvious: most of the substantial recent
increase in subnational expenditures must have been financed by national
transfers. National transfers rose significantly over this period, from about
2.5 to 5.0 percent of GDP, so that about 60 percent of the expansion of the
subnational sector was accounted for by additional transfer financing.
Another 10 percent was financed by expanded borrowing, with the
balance being accounted for by increased own-source revenues.
From 1988 to 1994, according to the more detailed data in World Bank
(1996a), total municipal government expenditures approximately
doubled, from 2 to 4 percent of GDP, with most of the increase taking place
in 1993–94. Current expenditures rose significantly, from 1.2 to 2.2 percent
of GDP, but capital expenditures increased even more, from 0.7 to 1.8
percent over the period. In real per capita terms, Colombian municipal
governments were spending over three times more on investment in 1994
than in 1988, compared to a doubling of current expenditures (excluding
debt service) over the period.
Over this period, only 37 percent of the expansion of municipal expen-
diture was accounted for by increased national transfers, with another 21
percent being accounted for by increased borrowing, especially in 1994.
Colombia 177
The single most important source of finance for the expanded municipal
sector was thus increased own-source revenues. Municipal governments
have access to potentially lucrative revenues of their own in the form of
property taxes and (at least in the case of the larger urban areas) a tax on
local business (the “industry and commerce tax,” which is mainly levied
on estimated gross receipts).
An important aspect of the municipal government universe in
Colombia, however, is that the big cities are both very important and very
different from the rest. In 1994, for example, the five largest cities
accounted for 69 percent of total local current revenues, 49 percent of
current expenditures, and 34 percent of local investment. They also
accounted for 58 percent of all local borrowing in that year, but they
received only 17 percent of national transfers. In contrast, the 582
municipalities with less than 20,000 inhabitants accounted for only 5
percent of own revenue and less than 4 percent of debt, but they received
30 percent of national transfers and accounted for 12 percent of local
current expenditures and almost 20 percent of local investment.
Colombia’s smaller municipalities have always been very heavily depen-
dent on transfers to sustain their level of activities. The recent constitu-
tional reform definitely increased this dependence on transfers, but the
dependence itself has long been a feature of Colombian local finance.14
Finally, it should be noted that the numbers set out above may be inter-
preted rather differently. It is true that from 1988 to 1994, subnational
current spending (including transfers) increased by about 2.5 percent of
GDP. However, 60 percent of this apparent increase resulted from account-
ing changes in the education expenditure financed by national transfers
and did not reflect increased subnational control over expenditure alloca-
tion.15 For the subnational sector as a whole, increased own revenues over
the period essentially financed the balance of increased current expendi-
ture. In other words, all the remaining increase in national transfers during
the period, like the additional borrowing, went to finance the very sub-
stantial increase that took place in subnational investment (amounting to
1.4 percent of GDP).16 As noted later, this is precisely the result the national
government appears to have wanted.
Intergovernmental transfers
The key to understanding intergovernmental fiscal relations in Colombia
is the system of intergovernmental transfers. This system has three basic
elements: the situado fiscal (SF), the participaciones municipales (PM), and the
sistema nacional de cofinanciación (SNC). The SF consists of 24.5 percent of
national current revenues (2.7 percent of GDP in 1995); it is transferred to

departments (and districts) to finance education and health in part in equal
(per unit) shares and in part on a population basis. The PM also consists of
a percentage of national current revenues, increasing annually to a sched-
uled maximum of 22 percent in 2002 (and amounting to 1.7 percent of GDP
in 1995); it is transferred to municipalities for “social investment,” on the
basis of a complex formula which on the whole clearly favors the smaller
and poorer municipalities. Finally, the SNC, which finances specified sub-
national projects on a “matching” basis amounted to a further 0.8 percent
of GDP in 1995.17 The principal characteristics of these three transfer pro-
grams may be set out briefly.
The situado fiscal in 1995 was 23.5 percent of national current revenues
(increasing to 24.5 percent in 1996). These revenues are supposed to be
divided as follows: 15 percent of the total is divided in equal parts among
the departments, the Capital District (Bogotá) and the Special Districts of
Cartagena and Santa Marta, and the remaining 85 percent largely in
accordance with an estimate of “user” population (of public education and
health facilities). The actual formula is quite complex, including, in princi-
ple, small components for “fiscal efficiency” and “administrative effi-
ciency,” although these elements appear to have little effect on the actual
distribution. (The detailed formulas for both SF and PM [below] take two
pages to set out in World Bank (1994).) Indeed, several departments receive
amounts in excess of those given by the formula in order to cover actual
expenditures – something which is legally possible because the depart-
ments in question are not “certified” to manage the SF. At least 60 percent
of SF must be spent on pre-school, primary, and secondary education and
20 percent on health, with the remaining 20 percent on either (in practice,
it almost always all goes to pay teachers’ salaries).
The participaciones municipales in 1995 were 15 percent of national current
revenues, increasing by 1 percentage point a year to 22 percent in 2001.
These revenues are supposed to be distributed to municipalities (and res-
guardos indígenas) as follows. First, 5 percent of the total is distributed to
municipalities with less than 50,000 inhabitants (in accordance with the
formula set out next) and an additional 1.5 percent goes to municipalities
along the Magdalena River, in proportion to the length of river frontage.
Next, of the remainder, 40 percent is distributed on the basis of the number
of inhabitants with “unsatisfied basic needs” (as measured by an index of
necesidades básicas insatisfechas, or NBI), 20 percent in proportion to the
degree of relative poverty of the municipality, 22 percent in proportion to
population, 6 percent in proportion to “fiscal efficiency,” 6 percent in pro-
portion to “administrative efficiency,” and 6 percent in proportion to the
Colombia 179
change in the NBI. PM transfers must be spent on “social investment,”

with at least 30 percent to education, 25 percent to health, 20 percent to
water and sewerage (if less than 70 percent of population currently
covered), and 5 percent to recreation, sport, and culture. The remaining 20
percent is de libre asignación but must still be spent on social investment. In
all cases, rural areas must receive a share of expenditure at least equal to
their proportion of population (and if the rural share is greater than 40
percent, they must get an extra 10 percent).
The sistema nacional de confinanciación consists of four funds: the Social
Investment Fund (FIS), the Rural Development Fund (DRI), the Fund for
Urban Infrastructure (FIU), and the Fund for Road Infrastructure (FIV).
These funds help to finance projects by subnational governments through
matching grants, with different matching rates applied for different types
of projects.
The design and functioning of these three transfer programs is clearly of
critical importance in determining the outcome of fiscal decentralization in
Colombia. Each of these transfers generates incentives that may be more
or less conducive to the maintenance and expansion of the coverage and
quality of such critical services as education, health, or water. In addition,
some have argued that the cumulative impact of these extensive central
transfers – amounting to about 40 percent of total central government
expenditure – has been to weaken macroeconomic policy by discouraging
local efforts to raise own revenues and encouraging frivolous and irre-
sponsible spending by transfer recipients. These points are discussed later
in the chapter.
Policy objectives and transfer design

At present, five distinct policy objectives are addressed with the three
transfer instruments outlined above (table 6.2). The first two objectives
involve the provision of minimum service levels in education and health,
respectively. These objectives are being pursued to varying degrees by all
three transfer programs. A third important objective is to finance, at least
in part, the cost of building the physical infrastructure necessary to expand
the coverage of key services (water, education, and health); this objective
is addressed through both the PM and the SNC. Apart from these specific
“output” objectives, the system of intergovernmental transfers pursues
two additional objectives. First, it attempts to increase the fiscal capacity of
territorial entities to finance a variety of other decentralized or local ser-
vices (for example, roads and recreation). The main instrument employed
for this purpose is a percentage (20 percent) of the PM that in principle can
Table 6.2. The system of intergovernmental transfers
Instruments
Sistema nacional de
Objectives Situado fiscal Participaciones municipales confinanciación
Finance education X X X
Finance health X X X
Finance infrastructure X X
Increase local capacity to X
finance other services
Interregional redistribution X X X
be allocated freely (transferencias de libre asignación). At the same time, the

system tries to compensate for differences in fiscal capacity or needs by the
use of distribution formulas for both the SF and the PM that transfer pro-
portionally more resources to the poorest entities.18
Using the same instruments both to finance the provision of education
and health services and to redistribute resources with the purpose of
increasing fiscal capacity and compensating for interregional differences
may generate important cost inefficiencies because costs are not directly
associated with differences in income or wealth. For example, in the case
of the PM, the per capita transfers received by any municipality are
strongly influenced by a poverty indicator, with the result that some local
governments receive a transfer that, in per capita terms, is three to four
times higher than that received by others in the same department.19 An
important part of these transfers is used to cover the recurrent costs of pro-
viding education and health services.20 Since these costs are mainly estab-
lished by national policies, they do not vary between municipalities by
such large factors, and the variations that do exist cannot logically be
linked directly to a poverty indicator.
Similarly, using the SNC to finance such recurrent costs of providing
health and education services as teachers’ salaries limits the administra-
tive capacity of subnational governments, creates dependency on the
national level, and thus likely reduces the efficacy with which services are
provided. In 1995, for example, we estimate that two-thirds of the SNC
funds directed to education corresponded to recurrent costs and only one-
third to capital expenditures.21
Finally, assigning the function of financing infrastructure (new water
systems, roads, schools, or hospitals) to an automatic and permanent
transfer such as the PM (whose value is not linked to coverage needs)
generates two types of problems. First, the national government loses a
Colombia 181
potentially important tool to implement the goals of its development plan

and to direct resources to those regions with the most important coverage
gaps. Second, even if one does not worry too much about this, given the
way the role of central planning is commonly downplayed these days, the
present system does not recognize the obvious fact that construction of
infrastructure is a discrete event. Since, as noted below, other national poli-
cies virtually guarantee that local governments will lack the flexibility to
allocate these resources to other uses, this rigidity in funding could lead to
even more allocative distortions.
The basic principle that should guide the design of a system of inter-
governmental transfers is that the purpose of transfers is not to finance par-
ticular governmental entities but rather to contribute to an effective
provision of services to the population. To the extent that services are pro-
vided by subnational governments without the fiscal capacity to finance
them fully, that there are externalities associated with the services in ques-
tion, and that interregional differences in needs are important, transfers
may be needed. But it is important to ensure that those responsible for the
provision of a service have a clear mandate, resources to finance it (includ-
ing, whenever needed, own resources), and flexibility to make decisions,
and are held accountable for results. In addition to providing incentives for
agents (recipient governments) to act as the principal (the central govern-
ment) wishes, other critical elements of transfer design are simplicity,
objectivity, and transparency.22
Financing education and health

In Colombia, arguably the principal objective of the transfer system is to
guarantee the provision of minimum service levels of education and health
to the population. Given the important externalities associated with those
services and their special status (in accordance with the Constitution and
economic development policy), the central government has assumed, in
effect, the responsibility of financing the provision of these services. The
magnitude of the minimum service level and the extent of its coverage
among the population, however, constitute variables to be determined by
policy considerations and resource availability. Education and health are
by far the most important sectors affected by the system of intergovern-
mental finance in Colombia. The amount and design of all major transfers
are dominated by concern for these two sectors.23
To put this another way, Colombia has initiated a process of decentrali-
zation in the education and health sectors at least in part because it con-
siders that subnational governments are more effective in providing these
services. Nonetheless, to date, there has been considerable confusion

arising from the unclear allocation of expenditure responsibilities. In the
case of education, for example, it is not at all clear who is in charge of what.
The relative roles and responsibilities of the national, departmental,
municipal, and school-level authorities need to be disentangled and clari-
fied. Nonetheless, it is clear that both levels of subnational government
will require resources from the national level in order to finance this
service adequately. In the case of health, the concurrent implementation of
social security reform makes the process of decentralization even more
complex. It is not just subnational governments that receive transfers from
the national level for health but also quasi-public enterprises (that is, hos-
pitals) and companies offering health insurance packages to the poor. The
complexity of this scheme precludes further discussion here.24
In general, however, the main guiding principle with respect to financ-
ing these basic services should clearly be that money should flow to the
entities (even if they are not part of the public sector) that are actually
responsible for the provision of these minimum service levels. The magni-
tude of the funds needed to finance the proposed new system of capitation
grants for education and health is a function of unit costs (which could
vary according to geographic and demographic conditions) and coverage.
The main change needed in the existing system is to establish a direct link
between the guaranteed service and its financing that is not a function of
either current revenues or complex formulas attempting to measure
poverty, effort, or other variables that do not represent the cost of pro-
viding the service.25
A problem of particular importance concerns the transition from the
existing to a new system: how do we get there from here? More policy pro-
posals probably founder on the rock of inadequate consideration of the
transitional problem than any other. The major transitional problem with
the revised transfer for education proposed in World Bank (1996a), for
example, is that it implies a shift of teachers away from those localities with
an “excess” of teachers towards those with a “deficit” (in terms of the stan-
dard per student allocation). As with all transitions, three dimensions need
to be taken into account: (1) the pace at which deficits are rectified; (2) the
pace at which excesses are corrected; and (3) the resolution of the imbal-
ance if (1) does not equal (2). As a rule, it is easier to build up than to run
down (although it is important not to dump too much new money too
quickly into any governmental unit since it will always be spent, but not
necessarily efficiently). Reducing government expenditures is never politi-
cally easy, particularly when, as in this case, it may involve moving per-
sonnel from more desirable locations to less desirable ones.
Colombia 183
The reality is thus likely to be that it will be difficult to move from the
present inappropriate distribution of official teachers to one more in accord
with needs without putting additional funding into the system to finance
the slower pace of reducing the excess. In other words, it may be that the
only way to get “there” from “here” is by putting more money into the
system – an ironic outcome, if true, given that one of the principal factors
motivating decentralization has clearly been the desire to “off-load” some
social expenditures from the hard-pressed central budget. Change to a
better decentralized system may save funds in the long run, and should
certainly produce more services for the same money, but the transition
process itself is likely to cause considerable friction and to require that
additional resources flow into the system, at least for some period.26
Financing infrastructure
The national government has two reasons for being interested in what
local governments do in financing infrastructure. First, some local infra-
structure projects may involve significant externalities. Second, some such
projects may constitute essential elements of national development pro-
grams. National support of infrastructure related to the provision of basic
education and health services, for example, may qualify for both reasons,
as may national support of projects improving the level and quality of
water supply and sewerage. Support of local roads and some rural
development projects may be justified as part of the national effort to
improve the economic productivity of poor rural areas.
Whatever the rationale for national interest in local investment in physi-
cal infrastructure, concern for the provision of services and the creation of
new infrastructure is inappropriately mixed in the present transfer system.
The amounts allocated through the existing transfer mechanisms bear no
relation to any rational investment policy in critical areas such as educa-
tion and health. Moreover, many of the investment projects currently sup-
ported (for example, with respect to sports and recreation) should likely
not be eligible for national support in any case.
In a decentralized system, in principle, subnational governments should
identify infrastructure needs and execute projects. Financing large infra-
structure projects from local resources alone may of course not always be
possible, given the scanty current revenues of most local governments.
Moreover, small localities seldom have much access to private capital
markets. If small local governments are to carry out costly public works,
they must as a rule rely heavily on grants from higher-level governments.27
In theory, a matching grant, in which the central government pays part
of an expenditure made by a local government, is the best way to finance

projects in which some of the benefits from the local activity in question
spill over to other localities. The share paid by the central government (the
matching rate) should be related to the size of the spillover and may also
depend in part upon the financial position of the local government (that is,
by altering the matching rate in accordance with local capacities one can,
in principle, stimulate similar responses in different localities).28 Properly
designed matching grants also have the political advantage of introducing
an element of local involvement, commitment, accountability, and
responsibility for the aided activities.
Money alone will not do the job, however; it must be provided in the
right framework, in the right amounts, and to the right recipients under
the right conditions. Revising the present SNC drastically along the lines
sketched above and removing the requirement to spend most of the PM on
investment are needed reforms. For such a system to work, however, the
national government must have both clear objectives and an operational
system that can work efficiently with subnational governments interested
in having access to these resources. It is not clear that either condition is
adequately satisfied at present, although some of the agencies now
involved in the SNC appear to have operated quite successfully in some
respects (World Bank, 1995).
General-purpose grant
As noted earlier, it appears at present that every transfer program in
Colombia is intended to achieve every objective. In no area is this problem
more obvious than with regard to the redistributive aspect of transfers.
Although every transfer program seems to be intended primarily to give
more to those who have less, the result can hardly be said to be a pattern
of redistribution that makes much sense in terms of either efficiency or
equity.29 Essentially, the argument to this point has been that the two major
growth-related programs with strong interpersonal redistributive ele-
ments – education and health – should be separated out and funded by a
system of national capitation grants, thus emphasizing that the central
distributional issue relating to these programs relates to redistribution
among persons rather than regions. Similarly, the other main direct national
interest in what subnational governments do (the provision of infra-
structure related to these two programs and other projects with significant
spillovers to broader regions) should be separated out more clearly and
financed on a matching grant basis.
Such a system will not work properly, however, unless a final critical
Colombia 185
component of the system of intergovernmental transfers, which may be

called a “general-purpose grant” (GPG), is put into place. A GPG has two
important, and distinct, rationales. The first, and the most important ratio-
nale from a systemic perspective, is to provide the necessary underpinning
for decentralization in general, and for the other transfers proposed above
in particular, by equalizing (to some level) the fiscal capacity of territorial
entities, thus putting all on the same footing with respect to incentives. The
second rationale is to provide sufficient resources to enable all local
governments, even the smallest and poorest, to provide a basic package of
local services other than health, education, or construction of infra-
structure.30
From a purely economic point of view, the second of these may appear
to make little sense in some respects. However, in Colombia at present, and
no doubt for years to come, most smaller rural areas will not be able to
provide any significant local services solely from their own resources. This
lack of local resources should not be confused with a lack of local capacity
to make and implement suitable expenditure decisions, since there is
strong evidence that even the poorest areas in Colombia have managed
surprisingly well once they have been enabled and encouraged to do so.31
In any case, a critical question in transfer design is often how to provide
the basic resources such municipal governments need to survive while
maintaining adequate incentives for them to do what they can in terms of
raising their own revenues. Fortunately, it turns out that the transfer
design needed to make the capitation systems for education and health
and the matching grant system for infrastructure function properly will
also accomplish the politically necessary task of ensuring the survival of
even the fiscally weak.32 And, as just mentioned, preliminary indications
in Colombia suggest that such localities seem, on the whole, to spend the
money as well, or better, than would occur under a more centralized
system (World Bank, 1995).
Any sound design for intergovernmental transfers requires explicit
incorporation in some way not only of an appropriate measure of “need”
but also of “revenue capacity.”33 At one extreme, the aim might be to
provide each local government with sufficient funds (own-source revenues
plus transfers) to deliver a (centrally) predetermined level of services.
Because such capacity-based transfers are in principle based on measures
of potential revenue-raising capacity (not on actual revenues), no disin-
centive to fiscal effort is created by this approach. Differentials in the cost
of providing services (for example, in rural or less densely populated areas)
may or may not be taken into account.34 Of course, transfers based solely
on capacity measures do nothing to ensure that the recipient governments
will in fact use the funds they receive as the central government might wish
– unless receipt is conditioned on performance, and compliance is moni-
tored in some way.
Although it is difficult to measure either “capacity” or “need” in any
fully satisfactory way, some general-purpose grant along these lines is
clearly needed to replace the confusing use of redistributive measures in
the SF and PM transfers. Such a grant will enable all local governments to
operate at some minimum level and provide other services besides health
and education. At the same time, and more importantly, it will partially
compensate for the large differences in fiscal capacity that exist and hence
place all on a more level playing-field with respect to the (essentially) equal
(per service unit) capitation grants proposed above for basic education and
health. The key aspects of the GPG are that the use of the funds thus pro-
vided are at the discretion of the recipient (no earmarking – see next
section) and, especially, that they are distributed according to need and
fiscal capacity.35
Earmarking and expenditure mandates

As Colombian central officials see matters, the main problem with the
present system of intergovernmental transfers is not, as we have argued,
that its design provides inappropriate incentives but rather that its sheer
size gives rise to three main problems: (1) the recipient governments do not
spend the money correctly; (2) they have also slackened off in terms of col-
lecting their own taxes; and (3) the growing burden of transfers on the
central budget is a major obstacle to restoring macroeconomic balance.
This and the next two sections discuss each of these problems in turn. We
find little evidence in support of any of these arguments.
Owing to the perceived profligacy of local governments with the “easy
money” flowing in from transfers, Congress has attempted to control local
spending in three ways: first, by requiring almost all the PM to be spent on
“social investment”; second, through such administrative controls as
establishing fixed percentages of transfers that must be spent on certain
sectors; and, finally, by limiting even the so-called “free” allocation of 20
percent of PM transfers. Unfortunately, such measures have created addi-
tional inefficiencies. The use of fixed percentages may easily lead to over-
spending in certain sectors. For example, it is unlikely that 15 percent of
the PM can possibly be the optimal amount to be spent on health subsidies
in all municipalities, as the current system implies. Moreover, the general
restriction against using PM resources for recurrent expenditures36 not
only precludes some gains in allocative efficiency that may theoretically be
Colombia 187
expected with decentralization, but may even lead to an increase in the

volume of current expenditures financed through credit, as this is one of
the few means local governments have to escape such limits. Finally, the
emphasis on administrative controls distracts the national government
from its responsibility to monitor and evaluate the effectiveness with
which subnational governments implement programs and provide the
services for which they became responsible.37
The inefficiencies and distortions generated by the rigidity implicit in
the transfer system itself, with its various rules as to how the funds may be
spent, are further exacerbated by a proliferation of norms and mandates
imposed on local governments. Laws such as those on public services,
environment, and sports earmark local resources or impose significant
costs on local finances. For many municipalities earmarking and unfunded
mandates imply costs that surpass the 20 percent of the PM that is not
already earmarked (table 6.3). Local governments now clearly have more
resources than they did before Law 60; but in at least some cases they may
now have even less freedom than before in terms of deciding how much
should be spent on what.
Colombian public finances have long been characterized by extensive
earmarking, that is, the assigning of revenues from specific sources to par-
ticular expenditures. Earlier studies found such earmarking to be particu-
larly important at the municipal and, especially, departmental levels.38
Although earmarking at the national level has been reduced in recent
years, no similar changes have been made at the subnational level: most
departmental own-source revenues, and a substantial fraction of munici-
pal own-source revenues are still required by national legislation to be
spent on particular activities. In addition to such revenue earmarking, and
increasingly in recent years, almost all national transfers to subnational
entities are earmarked in the sense that they must be spent on specific
activities in amounts (or proportions) designated by national legislation.
Finally, Colombia, like many other countries, offers many examples of
unfunded legislative “mandates,” that is, expenditure obligations
imposed on subnational governments with no additional funding being
attached.39
The importance of the earmarking and mandating of local own revenues
may be indicated by reference to a recent detailed examination of a small
number of specific municipalities.40 As shown in table 6.3, all three of the
municipalities examined, which varied widely in location, size, and
sophistication, appeared to be spending as much as, or in some cases more
than, required by Law 60 on the various areas of “social investment” des-
ignated there. Moreover, although some of the transfers received from the
Table 6.3. Earmarking and mandates in specific

municipalities (million pesos)
Zipaquirá Cajicá Polícarpo

1995 budget 1994 1994
Law 142/93 73.0 12.9 0.1

Law 99/93 20.9 19.7 4.7
Law 3/91; Law 9/89 96.6 73.3 23.8
Law 136/94 53.8 31.9 9.2
SISBEN 20.0 17.0 10.0
Anticorruption 6.0 2.0a 0.0
Stratification 12.0 4.0a 0.0
Internal control 2.0 1.0a 0.0
Contraloría 69.0 23.0a 0.0
Personería 54.5 18.0a 0.0
UMATA 11.9 1.0 0.0
Total 419.7 203.8 47.8
As percentage of transfers 15.6 23.7 14.0
Note: a Estimated, based on actual expenditure on Zipaquirá,

assuming that cost is proportional to population.
The items identified in this table are (some of) those for which
national laws require municipal governments to spend either
specific percentages of their own resources (including the 20
percent de libre asignación of the PM transfer). Some of these items
– contraloría, personería – relate to nationally mandated increases
in the salaries of local officials. Others relate to major reforms in
other areas, such as the SISBEN requirement to register citizens
for purposes of health insurance. Still others simply reflect the
desire of national legislators to control municipal expenditures by
designating specific percentages or amounts to be spent on such
items as agricultural extension (UMATA), and the funds to be
spent on housing (Law 3/91 and Law 9/89), water and sewerage
(Law 142/93), environmental protection (Law 99/93), and the
training of municipal officials (Law 136/94).
Source: World Bank, 1996a, vol. II, p. 57.
national government were supposedly de libre asignación (that is, could be

spent as the municipalities wished), in fact most of the transfers, and in
some cases some own-source revenue in addition, were at least in princi-
ple also supposedly budgeted as specified by national laws.41
For the municipal sector as a whole, the evidence suggests that as a rule
total expenditures on “social investment” substantially exceeded the man-
dated requirements. That is, not only were the designated proportions of
the transfers spent on the designated activities but some additional local
resources were also spent. For example, in 1994 50 percent of transfers were
to be spent on designated activities, with 30 percent of the 50 percent, or 15
percent in total, going to education: the actual proportion of transfers spent
Colombia 189
on education in 1994 was over 28 percent, or almost twice the required

amount.42 In short, if local governments are spending transfer funds incor-
rectly in some sense, the responsibility seems to lie with those who set the
rules, namely, the central government.
Fiscal effort and fiscal flexibility

A recurrent theme in Colombia has been concern with the effort, or lack of
effort, demonstrated by departmental and municipal governments in
mobilizing local fiscal resources. This theme was, for example, emphasized
in varying degrees by two major reviews of fiscal decentralization under-
taken in 1981 and 1992 respectively.43 More recently, the effect on local
fiscal effort of the increased fiscal transfers to municipal governments
resulting from Law 60 of 1993 has given rise to much concern and to a
series of relatively inconclusive empirical studies in Colombia.44
Despite all the fuss, there is little or no evidence to support the appar-
ently widespread belief in Colombia that the expansion of national trans-
fers to date has significantly reduced local fiscal effort. The studies that
support this conclusion, like those that do not, are far from robust,45 and
the available data on the whole suggest that the major sources of depart-
mental and municipal own revenue have in recent years continued to
expand at, or above, the rates observed in earlier years when transfers
were lower.46 Local governments on the whole appear both to have main-
tained their own efforts to collect revenues (as discussed below) and to
spend the money sensibly and in accord with the wishes of local people
(World Bank, 1995).
In any case, the constant focus on “fiscal effort” in recent Colombian dis-
cussion seems misguided conceptually, empirically, and in policy terms.
Conceptually, while it is not easy to define fiscal effort, it is probably most
meaningfully understood as the ratio of actual taxes collected to “poten-
tial” taxes estimated on the basis of some standard measure of fiscal capac-
ity and some standard (for example, national average) tax rate.47 Even
when so defined, the general absence of reliable empirical estimates of
fiscal capacity renders the concept non-operational. Moreover, given the
very limited flexibility Colombian departments and municipalities have to
alter their revenues through their own actions in any case, it is far from
clear to what extent it is meaningful to interpret the behavior of revenues
as reflecting their “effort.”
With respect to most major departmental and municipal taxes, for
example, both the tax base and the tax rate (or, in some cases, a maximum
rate) are determined by national legislation. In the case of the local property
tax, the single most important source of local own-source revenues, in most
of the country the tax base can be increased only by the action of a national
agency (the Instituto Geográfico “Agustín Codazzi” for periodic reassess-
ments and the central government for annual inflation indexing). The pro-
vision for automatic adjustment of cadastral values in Law 14 of 1983, for
instance, has been consistently determined by the central government to
mean that such values should be increased by less than the increase in the
consumer price index (CPI).48 When combined with the fact that regular
reassessments in much of the country occur at much longer intervals than
the recommended three years, the result is that the only way that local
governments which levy the maximum allowed rate can increase property
tax collections is by collecting the taxes assessed more rapidly.49
It is common to note the very low level of property taxes in countries
such as Colombia and to argue that there is obviously much room for
increased local “fiscal effort” in this respect. Again, there is of course some
truth in this, although experience everywhere suggests that it is neither
quick nor easy to obtain additional revenues from property taxes
(Dillinger, 1991). In Colombia most local governments have little possibil-
ity of doing so anyway because they control neither the tax rate nor the tax
base. In the larger cities the most “flexible” local tax is a tax on business –
the (not particularly desirable) industry and commerce tax.50 However,
there is little scope for this tax in the smaller municipalities, many of which
have little of either industry or commerce. Indeed, in the smaller commu-
nities it is hard to see any way in which local governments could increase
revenues quickly other than by borrowing – a fact which may explain at
least some of the increase in recourse to this source that has occurred in
recent years.
On the whole, the evidence to date in Colombia appears to support more
the so-called “flypaper effect” – that is, that transfers tend to increase total
expenditures by about the same amount as the transfers because local
taxes are not reduced – than the “fiscal effort hypothesis” that increased
transfers result in reduced local taxes. If so, this casts doubt on the good
sense of following the frequent advice to place still greater weight on
rewarding fiscal effort by increased transfers – unless, for some reason that
is far from clear in the context of Colombia today, a larger public sector is
in itself assumed to be a major policy objective.
The correct focus of attention with respect to subnational revenues in
Colombia is thus not the arguable and, it appears, so far unimportant
deterrent effect of increased transfers on local revenues but rather the more
important questions of the effects of these transfers on local expenditures
and the inadequate design of both the major transfers and of most local
Colombia 191
revenue sources. It would seem advisable for the central government to

focus on problems which are entirely within its own policy domain rather
than on what seems to be the non-problem of inadequate fiscal effort by
the local governments – a problem which, even if it exists, cannot readily
be dealt with by national government policy, and which arguably should
not be even if it could be.
On the other hand, it should perhaps be noted that there is an alterna-
tive and more meaningful way in which to interpret the strong emphasis
in the work of such authors as Wiesner (1992, 1995) on the importance of
taking “fiscal effort” more explicitly into account in designing transfer pro-
grams. From this perspective, the reason for the concern with fiscal effort
is not some technical worry about the substitutability of national transfers
for local resources. Rather, it arises from the importance of requiring local
citizens to pay, in some meaningful sense, for what they get so that those
who make local expenditure decisions will be held accountable (through
local political institutions) for their actions. So long as local governments
spend “other people’s money,” according to this plausible argument, they
are unlikely to be under much local pressure to spend this money effi-
ciently.
While studies such as World Bank (1995) suggest that this may be an
unduly pessimistic reading of Colombian experience to date, it is surely
correct to say that experience everywhere indicates that people are more
careful in spending money they have to earn (pay themselves) both
because they are aware of the pain of taxation as well as the pleasure of
expenditure and because they will feel more “ownership” of the activity as
a result. From this perspective, careful attention should indeed be paid to
local resource mobilization as an essential component of any successful
decentralization exercise. Unless increased transfers are matched by some
local contribution, however small it may be in the poorest communities, it
is unlikely that the full efficiency benefits of decentralization can be real-
ized. People do not, it seems, take ownership of what is given to them in
the same way as they do of things they have to pay for themselves, at least
in part. And without local ownership, expenditure efficiency seems
unlikely to be enhanced by decentralization.
Transfers and deficits

An interesting aspect of the recent move to decentralize public sector activ-
ities in many countries around the world has been the revival of an old
worry – that subnational governments, left to their own devices, will act in
a macroeconomically “perverse” fashion.51 In the case of Colombia, as in
other Latin American countries such as Argentina and Brazil, the basic
worry is that increased national transfers will induce subnational govern-
ments to cut their own taxes while expanding expenditures both through
increased transfers and through “leveraging” increased borrowing of their
new (transfer) revenue base. Subnational deficits, and hence total public
sector spending and the overall public sector deficit, will hence expand.
Although little or no evidence of such destabilizing effects in Colombia is
visible to date, even those whose studies produce such reassuring news are
quick to point out that the future may be worse in this respect, as transfers
continue to expand and local governments become more irresponsible.52
When the national government transfers revenues to subnational
governments, it also, as a rule, transfers expenditures. This has certainly
been the case in Colombia over the last few years: the entire increase in the
SF, for example, and (as discussed above) the entire transfer to municipal-
ities has been explicitly earmarked, in effect, for specific expenditures. In
other words, subnational expenditures in principle should, in the absence
of behavioral reactions, increase by exactly the same amount as subna-
tional revenues. That is, revenues available for expenditure at the national
level decline and those available at the subnational level rise by the same
amount as subnational expenditures. In this framework, the overall effect
of this accounting change obviously depends not on the amount of the
transfers but rather on two other factors: what happens to national expen-
ditures and what happens to non-transfer (own) subnational revenues.
Consider the latter first. One of three things may happen to subnational
revenues as a result of increased transfers: they may decline (“fiscal lazi-
ness,” decrease in fiscal effort, or “substitution effect” of transfers); they
may remain the same as in the absence of transfers (“flypaper effect”); or
they may – as some seem to desire – actually increase local own revenues
(“stimulative effect” of grants). As discussed above the evidence, such as
it is, suggests that, at least up to now, there is no evidence of a decrease in
fiscal effort. In other words, the local reaction to transfers suggests that
there has been no change in the size of public spending and deficits (“fly-
paper”) or even an increase in the size of local spending but not in the
deficit (“stimulation”).
Now consider national spending. Unless national spending falls by at
least the amount of the transfer, the result will definitely be an increase in
both the total size of the public sector and the size of the public sector
deficit. Table 6.4 illustrates the various possible outcomes. In this table,
national expenditures and subnational revenues are the critical exogenous
variables, national revenues are assumed to be given, and subnational
expenditures are assumed to increase by the amount of national transfers.
Colombia 193
Table 6.4. Macroeconomic effects of increased transfers
Gn ⫺ ⫺ ⫺ 0 0 0 ⫹ ⫹ ⫹
Gs ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹
Gt ? ? ? ⫹ ⫹ ⫹ ⫹⫹ ⫹⫹ ⫹⫹
Ts ⫺ 0 ⫹ ⫺ 0 ⫹ ⫺ 0 ⫹
Dn ⫺ ⫺ ⫺ 0 0 0 ⫹ ⫹ ⫹
Ds ⫹ 0 ⫺ ⫹ 0 ⫺ ⫹ 0 ⫺
Dt ? ⫺ ⫺⫺ ⫹ 0 ⫺ ⫹⫹ ⫹ ?
Key:
Gn National government expenditure (excluding transfers)
Gs Subnational government expenditure (including transfers)
Gt Total public sector expenditure (Gn⫹Gs )
Ts Subnational own revenues
Ds Subnational deficit including transfers (Gs⫺Ts )
Dn National deficit excluding transfers (Gn⫺Tn ); Tn assumed constant
Dt Total public sector deficit (since Tn constant, varies as Gt⫺Ts )
⫹⫽increase; ⫺⫽decrease; 0⫽no change; ⫹⫹⫽strong increase; ⫺⫺⫽strong decrease;

?⫽unknown “depends on relative magnitudes of offsetting changes”.
Under these circumstances, total spending – the size of the public sector –
will increase as a result of increased transfers unless national spending
declines by the same amount (or, most improbably, subnational taxes and
hence own-revenue financed subnational spending fall by more than trans-
fers!). Similarly, the total public sector deficit will increase with increased
transfers unless national spending declines proportionately, unless (again
most improbably) offset by a large enough increase in local taxes. In short,
given that on the whole the evidence suggests that local taxes will remain
steady or increase a little, rather than decline (or increase a lot), the key to
the impact of increased transfers on macro stability lies in the behavior of
national (non-transfer) spending.
How does this analysis work out in Colombia? One view is that most of
the marked adverse swing in public sector current saving in recent years
is attributable to increased transfers, and some have interpreted this to
mean that increased transfers to regions are the villain on the macro scene.
This seems unwarranted. In the first place, the total magnitude of the real
increase in regional transfers is much smaller than the swing in the deficit.
Most of the apparent increase in central government transfers is attribut-
able to (1) the reform of the social security system, which involves a transi-
tional increase in (non-regional) transfers in order to make explicit a deficit
that was always there but not previously accounted for explicitly, and (2)
the redefinition of the SF to encompass a wider range of educational
expenditures. These expenditures were previously shown as national
expenditures; they are now shown as national transfers and subnational

expenditures. In reality, however, they are still national expenditures (to
pay nationally set wages of teachers).53
Even apart from these accounting factors, there has clearly been an
increase (about 1 percent of GDP) in national transfers to municipalities –
and as mentioned earlier this amount is scheduled to increase steadily in
the future. Until now, all of these increased transfers have effectively
flowed to nationally mandated expenditures. In effect all that is happen-
ing is that a portion of national revenues is being indirectly spent on
national objectives through the agency of subnational governments. But
there has been no corresponding reduction in direct national expenditures.
In short, if there is a villain in this picture it would appear to be the national
government which has failed to reduce its own program expenditures in
proportion to the extent such expenditures are now being carried on at the
local level.54
On the whole, given their relatively inflexible tax base, Colombian local
governments have probably done as well as could be expected in raising
real resources in the last few years. Of course, some have done better than
others; and all could no doubt do better, but that is not the point. Given the
basically unpromising subnational revenue base, unless expenditure is
actually to decline (in relative terms) either transfers must be increased, or
borrowing must be expanded.55 Is it surprising that some local govern-
ments have chosen the latter path? Of course, from the point of view of
strict fiscal virtue they should have bitten the bullet and reduced expendi-
tures at once, but even local politicians, like central politicians, are human
and tend to take the easy way out, if there is one.56
Indeed, a final comment on this general area is that national fiscal
virtue in budgetary matters may in part have been achieved by creating
local villains by pretending that the downward shift of revenues to local
governments that has clearly taken place in Colombia has not been
accompanied by an equally strong downward shift of expenditures.
Matters may get worse in this respect in the future. Local governments in
Colombia are now responsible for virtually all the expenditure functions
(for example, health and education) which account for expenditure
growth in countries all around the world. At the same time, they are pri-
marily restricted to an antiquated and unresponsive revenue basis, which
even at its best – and of course it is very far from its best, particularly in
the less advanced areas – probably could not provide adequate resources
to maintain the required local share of funding. Intergovernmental
finance seems likely to be a controversial subject in Colombia for years to
come.
Colombia 195
Lessons from Colombian experience

Decentralizing the public sector in any country is, as some recent authors
have emphasized,57 a process that is not without danger. As yet, however,
the alleged “dangers” of decentralization are not very visible in Colombia.
Indeed, even under the distorted system of incentives facing local govern-
ment in Colombia, the early evidence in support of the developmental ben-
efits of decentralization in terms of public sector efficiency is surprisingly
positive (World Bank, 1995). With a better-designed incentive structure,
there seems good reason to hope that even more favorable outcomes may
emerge over time. What is needed to improve efficiency is not so much a
radical redirection of Colombia’s policy of decentralization as some rede-
sign of transfers combined with better-informed and more consistent
implementation on the part of the national government than has been
evident to date.
Of course, improving the incentives for effective and efficient provision
of services in a decentralized environment in any country is likely to
require reforms in addition to those affecting the transfer system. In the
case of Colombia, for instance, it is necessary to clarify the distribution of
responsibilities between levels of government and address institutional
bottlenecks in several sectors (for example, education; see World Bank,
1996a). Second, changes are needed to ensure that subnational govern-
ments have sufficient flexibility to use resources, eliminating earmarking
(and mandates) that do not respond to very clearly established needs.
Third, changes are needed to give subnational governments more flexibil-
ity in altering tax rates to mobilize resources.58 Finally, changes are needed
to promote responsibility and accountability in public sector performance
at all levels of government. Many of these points seem much more widely
applicable than just in Colombia.
Whatever is done on the revenue side, many of the constraints currently
imposed on local government finance by the extensive use of earmarking
and expenditure mandating should clearly be removed in Colombia. The
constant extension, and frequent change, of these various attempts by the
national level to direct subnational expenditure patterns in detail is not
only basically inconsistent with the stated general intent of decentraliza-
tion but is also likely to guarantee that decentralization in the sense of
developing effective and efficient subnational governments will fail. An
essential ingredient of strengthening local fiscal capacity therefore seems
to be some weakening of national control over what local governments do
with the money.59
Moreover, if decentralization is to work, those charged with providing
local infrastructure and services must be accountable both to those who

pay for them and to those who benefit from them. Enforcing accountabil-
ity at the local level requires not only clear incentives from above but also
the provision of adequate information to local constituents and the oppor-
tunity for them to exercise some real influence or control over the service
delivery system. Informal community organizations almost by definition
must be structured like this or they cannot exist, but it can be a real chal-
lenge to introduce a similar degree of responsiveness into formal govern-
mental organizations.
Accountability is the key to improved public sector performance, and
information is the key to accountability. The systematic collection, analy-
sis, and reporting of information that can be used to verify compliance
with goals and to assist future decisions is a critical element in any decen-
tralization program. Information is needed not only on financial aspects
(budgeting and expenditure reporting) but also on inputs, outputs, and,
where possible, outcomes to ensure accountability. Such information is
essential both to inform public participation through the local political
process and for the monitoring of local activity by central agencies respon-
sible for supervising and (usually) partially financing such activity. Unless
local “publics” are made aware of what is done, how well it is done, how
much it costs, and who paid for it, no local constituency for effective
government can be created. As a recent study of local government capac-
ity in Colombia makes clear, a stronger community voice is often an essen-
tial ingredient for improved government performance, both through
making it easier to identify local preferences and through strengthening
accountability.60
Since the election of mayors was introduced in 1986, a reduction in
clientelism and more transparent and fair electoral practices have con-
ferred more legitimacy on the leadership role of mayors and made the
position more attractive. A renovation in municipal leadership has
resulted. Many current mayors are political outsiders with backgrounds
either in the private sector or in civic movements independent of political
parties.61 With decentralization, municipal political life has become more
clearly local in nature, increasing demands on new local leaders to respond
to their communities.
In addition to devolving some degree of real political responsibility to
local residents, a sound decentralization program must also be accompa-
nied by an improvement in the evaluation capacity of the central govern-
ment. Decentralization and evaluation are not substitutes; they are
complements. An essential element of the “hard budget constraint” system
needed to induce efficient local decisions is thus adequate central enforce-
Colombia 197
ment capacity in the shape of credible information-gathering and evalua-

tion. The “carrot” of central financial support of local efforts must be
accompanied by the “stick” of withdrawn support if performance is inad-
equate, which of course requires both some standard of adequacy and
some way of knowing how performance measures up. To mention perhaps
the most obvious and simple example: if a recipient of transfer funds does
not submit an adequate financial report, it should not receive such funds.62
In sum, Colombia’s ongoing experience with fiscal decentralization sug-
gests that success requires the central government to pay more attention to
its own actions and to worry less about what local governments are doing,
or not doing. Specifically, the central government should: (1) decide more
clearly what it is trying to do; (2) redesign its policy instruments to achieve
its objectives more efficiently; (3) free local governments from many of the
present restrictions on both revenues and expenditures that hamper them
from operating efficiently; (4) ensure that it has sufficiently accurate and
timely information to know what is going on with respect to subnational
finance; and (5) alter its policies if, and only if, a change is necessary to
achieve some important national policy objective. Once it has got its own
act together, the central government can and should let decentralization
take its course without constant changes in direction and policy.
If the rules of the game are set up properly so that local governments
account properly for funds they are expending on behalf of the national
government, and there is at least a modicum of local political accountabil-
ity for what they do with their own funds, there should be no need for the
present cumbersome system of interventionist measures – many of which
in effect attempt to compensate for the lack of clear national policies but
fail in this task and instead succeed mainly in making it impossible for
anyone to run a subnational government efficiently. Getting the rules right
does not, of course, guarantee success, certainly not in every locality, but it
is clearly an essential first step to realizing the potentials (and avoiding the
dangers) of decentralization in any country.
From this perspective, Colombian experience over the last few decades
suggests at least three important lessons for other countries. First, there is
much to be said for proceeding down the path of decentralization with
deliberation and care. To illustrate: the provision in Law 60 requiring a
thorough, public examination of the effects of decentralization after an
initial five-year period is commendable and deserves emulation else-
where. Second, democracy matters: local elections have proved particu-
larly important in improving local capacity and “ownership” of public
sector programs and hence accountability. Third, whether there is democ-
racy or not, to a large extent central governments get the local governments
they deserve by virtue of the incentives their policies create for local
governments. The central role of central governments in determining the
outcome of fiscal decentralization is clearly evidenced by recent
Colombian experience.
Notes
1 A brief account of this history and the situation at end of the 1970s described in
the rest of this paragraph may be found in Bird (1984). The next two paragraphs
closely follow portions of a previous update in Bird (1990a).
2 For an account of fiscal policy in the early postwar years, see Bird (1970); a
comparative analysis of more recent years may be found in Perry (1992).
3 The Commission report was published as DNP (1981); Bird (1984) is an English
version of this report.
4 For an outline and discussion of the relevant constitutional changes, see
Wiesner (1992).
5 For example, the Ministry of Health wants to adjust the decentralization law to
help implement a reform of the social insurance system.
6 The last of these problems is, for example, at the centre of recent critical papers
on fiscal decentralization by Prud’homme (1995), and Tanzi (1995); our own
views on the pros and cons of decentralization are on the whole closer to those
in two subsequent papers responding to Prud’homme – McLure (1995) and
Sewell (1996).
7 Unfortunately, it sometimes seems that all anyone outside Colombia knows
about the country is its problems with drugs and political violence. Both of
these problems exist, and are important, but since neither of them affects the
issues discussed in this chapter in any fundamental way, they are not further
discussed here.
8 See the discussion of the institutional structure for providing these services in
World Bank (1996a: annexes 1 and 2), as well as the earlier discussion in Bird
(1984).
9 Most of the figures in this paragraph are taken from Ferreira and Valenzuela
(1993). It should perhaps be noted that, as is common in Latin America, a
municipio includes both any urban area (cabecera) and the surrounding rural ter-
ritory, so some counted as “urban” may live in very small communities indeed.
Nonetheless, there is no question that the bulk of the Colombian people now
live in urban areas, rather than in rural areas as was the case only thirty years
ago.
10 See World Bank (1994) for a detailed study of poverty in Colombia.
11 Data from Wiesner (1995). The difference between the revenue and expenditure
figures cited reflects the importance of central government transfers, as dis-
cussed below.
12 A word of caution seems warranted with respect to these (and most other)
public sector figures in Colombia, namely, that it is not difficult to find differ-
ent figures that seem to refer to the same concepts for the same years. To illus-
Colombia 199
trate: the figures in table 6.1 were obtained from the Departamento Nacional de
Planeación in September 1996 as a result of a request to duplicate and update
the similar figures in Wiesner (1995: 128). Nonetheless, the figures in table 6.1
for 1985 (the only year of overlap) are significantly different from those in
Wiesner, and no clear trends are shown in the table. Still, one has to work with
what one has, and these were the latest authoritative data available at the time
of writing this chapter.
13 However, on the basis of data in World Bank (1996a: annex 3), which net out
transfers within the subnational sector, the 1994 figure for the subnational
sector is only 10.6 percent of GDP; again, it is difficult to reconcile figures from
different sources. Unless otherwise indicated, the remaining data in this section
come from the source just cited.
14 This is, for example, one of the major themes in Bird (1984).
15 See the discussion in DNP (1995a).
16 The excess of transfers plus borrowing over investment is accounted for by
various “balancing” items that have not been explicitly discussed here.
17 All these figures are as estimated in World Bank (1996a). For discussion of these
transfers, see also Wiesner (1995).
18 Through the use of quotas (cupos) and differential matching rates, SNC is also
used as a redistributive instrument, but this aspect is not further discussed here.
19 The poverty indicator used – the NBI (necesidades básicas insatisfechas) index –
has been shown to be quite deficient (World Bank, 1994).
20 For example, 30 percent of the PM spent on education was used to pay teach-
ers’ salaries (DNP, 1995a).
21 This estimate is based on unpublished data kindly made available to the
authors by DNP.
22 These guiding principles are widely accepted in the international literature, for
example, Shah (1994) and Bird (1990b). Note that we are not discussing the
arguably quite distinct case of “truly federal” countries; on this, see Bird (1995)
and Bird and Chen (1996).
23 Unfortunately, it is not possible within the limits of this chapter to describe the
complex present arrangements for financing education and health or the
equally complex current proposals for reform in these areas, so readers may
have to take some of the following in faith; for detailed analysis, see World Bank
(1996a): annexes 1 [on education] and 2 [on health].
24 See World Bank (1996a: annexes 1 and 2), for further discussion.
25 Specific ways of implementing such a system are discussed in detail in World
Bank (1996a), which considers such questions as the unit cost of provision, the
number of units provided, the rate of extension of coverage, adjustments to unit
cost over time, the division of transfers between departmental and municipal
levels of government, the rate of transition, and the monitoring and enforce-
ment of the system.
26 More positively, it should be noted that this cost is already being paid since the
current system is not conducive to expenditure rationalization and cost control.
Under the proposed system, this additional cost would have not only a purpose
but also, more importantly, a termination date.
27 Subsidized loans could be used instead, but this common practice is undesir-
able. Rather than (in effect) mixing grants and loans in this way, it is better to
keep them separate: give a grant where warranted, but otherwise require all
lending to be done on commercial terms, with no explicit or implicit national
subsidies or guarantees.
28 This is the concept of “wealth neutrality” popularized by Feldstein (1975).
29 See, for example, the discussion in Wiesner (1995), Sánchez and Gutiérrez
(1995), and World Bank (1996a: annex 3).
30 The economic case for such equalization transfers is often confused with
questions more properly related to interpersonal redistribution. The basic
argument is that since it has been decided, for whatever reason, to charge
local governments with the delivery of an adequate “minimum bundle” of
local public services to citizens, and all citizens of Colombia should be enti-
tled to some basic level of such services regardless of where they happen to
live, equalizing transfers to localities (not persons) are needed to achieve this
goal.
It should perhaps be noted that this objective of providing similar public ser-
vices regardless of location may conflict with the desirability of migration from
less (privately) productive to more productive locations. Although this subject
has been discussed extensively (if not very conclusively) in the literature, it is
not further considered here on the assumption that in Colombian conditions,
the relatively small differences in location-specific public service bundles
(excluding education and health, which are assumed to be portable) that might
result in different locations from an equalization program are unlikely to be sig-
nificant factors in migration decisions.
31 See the extensive analysis of the surprising degree of local “capacity” to manage
efficiently in World Bank (1995).
32 For a related argument, see Bird (1993).
33 See, for example, Shah (1994) and Ahmad (1997).
34 Caution is necessary in introducing such differentials because it is easy to turn
a simple, transparent formula into an obscure and manipulable one by intro-
ducing too many refinements in the definition of the elements of the formula.
35 Those interested in the details of how the general scheme sketched in this
section would work in Colombia are referred to World Bank (1996a), especially
annex 5.
36 In a clever bit of wordplay “social investment” has been interpreted to cover the
salaries of education and health workers. But one cannot get the streets cleaned
this way.
37 The tendency to “over-regulate” expenditure in advance, in a futile attempt to
make up for the failure to monitor adequately what is actually done is an old
and pervasive problem in Colombia; see, for example, the discussion in Bird
(1992).
38 McCleary and Uribe (1990), for example, estimated that two-thirds of munici-
pal, and four-fifths of departmental, current income were earmarked if net
public enterprise profits were included. This ratio fell to only about 20 percent
of current revenues for municipal governments if net profits of their enterprises
Colombia 201
were excluded; the similar figure for departments was over 75 percent. For a
similar analysis for earlier years, see Bird (1984).
39 A particularly egregious recent example is the requirement in a recent “anti-cor-
ruption” law that all local governments, no matter how small, must establish
and maintain a free “800” number for citizen complaints. The cost of comply-
ing with this provision in some of the more remote localities which do not even
have telephone service could easily exceed the entire PM transfer!
40 For details, see World Bank (1996a: annex 4).
41 Note that this is in addition to any earmarking that may have been imposed by
the local governments themselves; the practice seems as popular at the local
level as elsewhere in Colombia, but of course the implications of local councils
voluntarily choosing to limit their expenditure choices are quite different from
those of having such limits imposed upon them.
42 Although based on data from DNP (1995d), our conclusion differs because the
DNP analysis considers only the amount of the inversión forzosa portion of the
transfer that is so spent rather than the total amount of the transfer that is spent
on education.
43 See Bird (1984), as subsequently updated in part in World Bank (1987) and
World Bank (1989), and Wiesner (1992); see also Wiesner (1995).
44 See, for example, Wiesner (1995), Steiner and Correa (1994), Ferreira and
Valenzuela (1993), Fainboim, Acosta, and Cadena (1994), Vargas et al. (1994),
Junguito, Melo, and Misa (1995), Maldonado (1995), Sánchez and Gutiérrez
(1994, 1995), CGR (1995), and DNP (1995b).
45 The empirical evidence of an adverse effect on effort in studies such as those
cited in note 44 is mixed – some studies find such an effect and some do not.
Moreover, many studies appear to be flawed either in terms of data (few studies
seem to have taken much care in assembling or understanding the data used)
or methodologically. To cite only two examples from the “effort” literature:
Slack and Bird (1983), in an early econometric study, found an adverse effect on
fiscal effort, but this study relates the SF to departmental own revenues even
though the SF is not received by the departmental governments and there was
therefore no reason to expect their fiscal behavior to react to changes in it. More
recently, Steiner and Correa (1994) contended that they found “fiscal apathy” at
the subnational level, but this conclusion rests on regressing local tax revenues
on national transfers, and it is hard to see how one could expect any other result,
given the inelasticity of most local taxes and the rapid increases in national
transfers as a result of the constitutional reform. To interpret such results as
telling us anything meaningful about local fiscal effort seems hardly legitimate.
46 See above and World Bank (1996a: annex 3).
47 See Bird (1976) and Bird and Slack (1990), and sources cited therein, for earlier
discussions of this issue.
48 See the discussion in World Bank (1989); matters have not changed in this
respect since this report was written.
49 To put this point another way, as Linn (1980) shows, the effective rate of
property tax is a function of the collection ratio, the statutory rate, exemp-
tions, and the assessment ratio. Let TC ⫽ taxes collected, TL ⫽ taxes assessed,
AVT ⫽ taxable base, AV ⫽ assessed base (including exempt property), and

MV ⫽ market value. Then the effective tax rate, TC/MV ⫽ TC/TL*TL/
AVT*AVT/AV*AV/MV. Of these ratios, only the first is clearly under munici-
pal control, so any assessment of “effort” with respect to property taxes should
focus on changes in this ratio only. Unfortunately, no information is available
on this subject.
50 For a critical discussion of this tax, see Bird (1984).
51 The literature on the alleged “fiscal perversity” of subnational governments
from the point of view of macro stabilization dates back to the early Keynesian
era, when it was argued that subnational governments would tend to accentu-
ate recessions by cutting back on their expenditures exactly when the economy
needed stimulus; see the review and appraisal of this argument in Rafuse (1965)
and Oates (1972).
52 See, for example, the studies by Sánchez and Gutiérrez (1995) and Junguito,
Melo, and Misa (1995).
53 For a detailed discussion of how education is financed and controlled, see
World Bank (1996a).
54 Of course, some of the increased national expenditures in the last few years have
arisen for other, quite legitimate, reasons such as increased security costs; but in
the present context what this means is that other national expenditures should
have been reduced still further if expansionary effects were to be avoided.
55 There is of course no direct correlation between the amount of money spent on
a function and the services delivered, so that considerable care must be used in
asserting that local expenditures are too important to be cut.
56 Given the very limited taxing powers available to local governments in
Colombia and the impossibility of widespread local government bankruptcy,
the only way to reduce the “moral hazard” of subnational borrowing resulting
in central government bail-outs may be to institute a credible review/control
system for debt work-outs along the lines discussed for Argentina in World
Bank (1996b), but this subject cannot be further developed here.
57 See Prud’homme (1995) and Tanzi (1995).
58 Specifically, localities should be able to set tax rates on residential property and
perhaps determine certain other tax rates as well, although their freedom to tax
non-residents (for example, through taxes on exporting sectors) should be
restricted. For further discussion, see Bird (1993).
59 This recommendation assumes that, as mentioned in note 56, localities will
have no recourse to national funding on a discretionary basis – no bail-outs.
60 The following paragraph is based on World Bank (1995).
61 In the 1988 and 1990 municipal elections, the two major political parties con-
trolled almost 90 percent of the municipalities (Gaitán and Moreno, 1992). In the
1992 elections, however, candidates from non-traditional parties gained control
of about one-third of the 1,000 municipalities.
62 Interestingly, this is not now done in Colombia, where over 100 small
municipalities apparently never submit any information to the central govern-
ment but nonetheless regularly receive transfer payments. The apparent
explanation is a political one: for the most part, these municipalities are those
Colombia 203
in which the writ of the central government does not run because they are
largely controlled by various guerilla movements (and/or drug dealers).
References
Ahmad, Ehtisham, ed. (1997), Financing Decentralized Expenditures, Cheltenham:
Edward Elgar.
Bird, Richard M. (1970), Taxation and Development: Lessons from Colombian
Experience, Cambridge, MA.: Harvard University Press.
(1976), “Assessing tax performance in developing countries,” Finanzarchiv, 34:
234–65.
(1984), Intergovernmental Finance in Colombia: Final Report of the Mission on
Intergovernmental Finance, Cambridge, MA.: Harvard Law School,
International Tax Program.
(1990a), “Fiscal decentralization in Colombia,” in R. Bennett (ed.),
Decentralization, Governments and Market, Oxford: Clarendon Press.
(1990b), “Intergovernmental finance and local taxation in developing countries:
some basic considerations for reformers,” Public Administration and
Development, 10: 277–88.
(1992), “Budgeting and expenditure control in Colombia,” Public Budgeting and
Finance, 2: 87–99.
(1995), “Fiscal federalism and federal finance,” Anales de las 28 Jornadas de
Finanzas Públicas, Facultad de Ciencias Económicas, Universidad Nacional de
Córdoba, Argentina.
Bird, Richard M. and Chen, Duanjie (1996), “Federal finance and fiscal federalism:
the two worlds of Canadian public finance,” in D. Perry et al., “Essays on fiscal
federalism and federal finance in Canada,” Discussion Paper No. 6,
International Centre for Tax Studies, University of Toronto.
Bird, Richard M. and Slack, Enid (1990), “Equalization: the representative tax
system revisited,” Canadian Tax Journal, 38: 913–27.
CGR (Contraloría General de la República) (1995), La situación de las finanzas del
estado 1994: Informe del Contralor General de la República, Bogotá.
DNP (Departamento Nacional de Planeación) (1981), Finanzas intergubernamentales
en Colombia, Bogotá.
(1995a), “Política de consolidación de la descentralización,” CONPES 2788 (June
15).
(1995b), “Impacto de las transferencias sobre los recursos propios de los munici-
pios: una contribución a la discusión” (October 7).
(1995c), “Estudio de casos municipales” (October 5).
(1995d), “Evaluación de la destinación de la participación de los municipios y de
los resguardos indígenas en los ingresos corrientes de la nación para la vigen-
cia 1994” (September 20).
Dillinger, William (1991), Urban Property Taxation in Developing Countries,
Fainboim Yaker, Israel, Acosta Navarro, Olga Lucia, and Cadena Clavijo, Hector
José (1994), “El proceso reciente de descentralización fiscal en Colombia y sus
perspectivas,” Coyuntura Económica.
Feldstein, Martin (1975), “Wealth neutrality and local choice in public education,”
American Economic Review, 65: 75–89.
Ferreira, Ana Maria, and Valenzuela, Luis Carlos (1993), Decentralización fiscal: el
caso colombiano, Santiago, Chile: CEPAL.
Gaitan Pavia, Pilar, and Moreno Ospina, Carlos (1992), Poder local. Realidad y utopia
de la descentralización en Colombia, Bogotá: Tercer Mundo.
Junguito, B. Roberto, Melo, B. Ligia, and Misa, A. Martha (1995), “La descentral-
ización fiscal y la política macroeconómica,” Banco de la República, Bogotá.
Linn, Johannes F. (1980), “Property taxation in Bogota, Colombia: an analysis of
poor revenue performance,” Public Finance Quarterly, 8.
Maldonado Copello, Alberto (1995), “Estudio sobre transferencias y esfuerzo fiscal
municipal,” Departamento de Investigaciones Colciencias, Fundación
Universidad Central Bogotá (August 31).
McCleary, William A. and Uribe Tobon, Evamaria (1990), “Earmarking govern-
ment revenues in Colombia,” WPS 425, Country Economics Department,
World Bank, Washington, DC (September).
McLure, Jr., Charles E. (1995), “Comments on Prud’homme,” World Bank Research
Observer, August 10: 221–26.
Oates, Wallace E. (1972), Fiscal Federalism, New York: Harcourt, Brace, Jovanovich.
Perry, Guillermo (1992), Finanzas públicas, estabilización y reformas estructurales en
América Latina, Washington, DC: Inter-American Development Bank.
Prud’homme, Remy (1995), “The dangers of decentralization,” World Bank Research
Observer, August 10: 201–20.
Rafuse, Robert (1965), “Cyclical behavior of state–local finances,” in R. A.
Musgrave (ed.) Essays in Fiscal Federalism, Washington, DC: Brookings
Institution.
Sánchez Torres, Fabio, and Gutiérrez Sourdis, Catalina (1994), “Descentralización
fiscal y transferencias intergubernamentales en Colombia,” Coyuntura
Económica.
(1995) “Colombia,” in Ricardo Lopez Murphy, (ed.), Fiscal Decentralization in
Latin America, Washington, DC: Inter-American Development Bank.
Sewell, David (1996), “The dangers of decentralization according to Prud’homme:
some further aspects,” World Bank Research Observer, February 11: 143–50.
Shah, Anwar (1994), Perspectives on the Design of Intergovernmental Fiscal Relations,
Policy Research Paper, Washington, DC: World Bank.
Slack, N. Enid, and Bird, Richard M. (1983), “Local response to intergovernmental
fiscal transfers: the case of Colombia,” Public Finance/finances publiques, 38:
429–39.
Steiner, Roberto, and Correa, Patricia (1994), “Decentralization in Colombia: recent
changes and main challenges,” Lehigh University, Lehigh, PA.
ciency and macroeconomic aspects,” in Michael Bruno and Boris Pleskovic
(eds.), Annual Bank Conference on Development Economics, Washington, DC:
World Bank.
Colombia 205
Vargas, Cesar, et al. (1994), “Financiamiento del desarrollo regional: situación
actual y perspectivas,” Planeación & Desarrollo.
Wiesner Duran, Eduardo (1992), Colombia: Descentralización y federalism fiscal:
Informe final de la Misión para la Descentralización, Bogotá: DNP.
(1995), La descentralización, el gasto social y la gobernabilidad en Colombia, Bogotá:
DNP and Asociación de Instituciones Financieras.
World Bank, (1987), “Colombia: a review of recent fiscal decentralization mea-
sures,” Report No. 6631-CO (April).
(1989), “Colombia: decentralizing revenues and the provision of services: a
review of recent experience,” Report No. 7870-CO (October).
(1994), Poverty in Colombia, Washington, DC.
(1995), Local Government Capacity in Colombia: Beyond Technical Assistance,
Washington, DC.
(1996a), “Colombia: reforming the decentralization law: incentives for an effec-
tive delivery of services,” Report No. 15298-CO (April).
(1996b), “Argentina: provincial finances study: selected issues in fiscal federal-
ism”, 2 vols., Report No. 15487-AR (July).
7
Argentina: fiscal federalism and decentralization

ERNESTO REZK
From a fiscal viewpoint, Argentina must still be regarded as a formal

federation, rather than a truly federal country, in so far as the experience
of the last sixty years shows that the provinces shifted the responsibility
for the main taxes (income, sales, excise and fuel taxes) into the central gov-
ernment’s hands, despite their constitutionally ample independent taxing
and spending powers. The resulting revenue concentration (at the national
level) was matched by a spending decentralization process whereby the
responsibility for key areas, such as education, health, and housing, was
transferred to the provinces.
In relation to the above-mentioned features, this chapter, after some brief
theoretical considerations, examines the analytic aspects of decentraliza-
tion in Argentina. Both macro and micro issues of intergovernmental fiscal
relations are reviewed and scrutinized. The chapter also sheds light on the
likely evolution of federal fiscal arrangements; in particular, whether the
situation will evolve in the direction of delegation (strengthening of the
principal–agent relationship) or, on the contrary, toward devolution, in
which mechanisms of collective decision-making closer to the public
choice approach can be expected to prevail.
An outline is also given of the key elements of the institutional setting in
Argentina, as well as a description of changes over time of assignments of
expenditures and taxes between the central and the subnational govern-
ments. The relevant quantitative aspects of Argentine federal finances and
intergovernmental fiscal arrangements are shown with statistical data for
the period 1983–95.
Understanding fiscal decentralization

Theoretical and policy-oriented interest in fiscal decentralization rose
during the eighties as a result of various unprecedented situations faced
by countries all over the world.
Bird (1993: 207) based the revival and new strength of decentralizing
206
Argentina 207
processes on developing countries’ ad hoc policies purposively aimed at

escaping from ineffective and inefficient governance, macroeconomic
instability, and inadequate economic growth. Tanzi (1995: 296), in turn,
mentioned the role played by developments in the European Union which
are in favor of decentralization.1 He assigned a similar influence to the
growing disenchantment with powerful central governments (especially
in industrial countries), which have proved to be incapable of stabilizing
their economies or of improving the distribution of income, and have
created a presumption toward giving more power to both the market and
subnational and local governments.
Fiscal decentralization is also found in countries like Argentina and
Brazil, where the major macroeconomic problems of the eighties (mainly
inflation and fiscal deficits) required important fiscal adjustment programs
to be implemented at the central level, subject to constraints imposed by
subnational governments’ own constitutional fiscal powers and the exist-
ing arrangements among different government levels.2
Even though the growth in importance of decentralizing processes can
be explained at least in part on allocative and efficiency grounds,3 the
notion of decentralization is far from being a simple one. This is demon-
strated in the three categories4 distinguished by Bird (1995):
• deconcentration: this is where decentralization occurs within the national
level and limits itself to simply scattering fiscal responsibilities among
regionally established branch offices;
• delegation: this is the situation typically depicted by the principal–agent
model, the rationale being that the higher government level chooses to
achieve its allocative goals more efficiently by transferring spending
functions and funds to subnational and local governments on the under-
standing that the latter will carry out their spending commitments in
line with the central government’s policy objectives;
• devolution: in this case, fiscal decentralization means not only lower
levels’ legal authority to carry out expenditure assignments (which they
also have under delegation) but also, and far more importantly, their
control of decisions concerning their spending functions.
As Bird (1995) emphasizes, devolution not only stresses allocative effi-
ciency but also political values such as enhanced governance (local
responsiveness) and political participation.5 The concepts of administrative
and fiscal decentralization, used in the existing literature, imply similar
notions according to Tanzi (1995: 297). The former embodies situations in
which most taxes are levied by the central government, although funds
are transferred to subnational governments to carry out their spending
208 Ernesto Rezk
activities as agents, following the guidelines and controls imposed by the

national level (delegation). The latter is more akin to the situation in most
federations where the subnational levels have constitutional or legal
powers to raise taxes and perform spending functions in which they have
complete decision-making authority, that is, local autonomy.
Delegation and devolution variants of fiscal decentralization can also be
analyzed in terms of the theoretical approach underlying collective deci-
sion mechanisms, as well as in terms of the scope for independent deci-
sion-making at different government levels. As mentioned above,
delegation entails the use of the principal–agent relationship to constitute
contracts whereby the principal (the central government) appoints agents
(subnational governments) to act on its behalf. Within this approach, the
central government transfers funds to lower levels (either conditional
grants or conditional matching grants), expecting them to deliver public
goods or services according to predetermined standards and quantities,
while, on the other side, local governments accept this delegation, expect-
ing to maximize their own economic and equity goals.6 As Wiesner (1994:
181) and others make clear, given that local governments are expected to
satisfy presumed central government priorities and national policy objec-
tives, the notion of fiscal responsibility means here that subnational and
local governments are accountable to their principal and not to local tax-
payers or residents.
The well-known public choice approach7 relates to devolution by requir-
ing that the different government levels finance most of their spending
functions out of their own taxes and revenues (the principle of financial
autonomy). It also shows that the accountability of subnational govern-
ments is to their taxpayers and not to the central level; the acknowledged
link between taxing and spending ensures, in turn, what Bird (1995)
defined as “the maintenance of a legitimate system of governance in terms
of public support of government.” Public choice fiscal models and devolu-
tion raise another interesting issue, stressed by Wiesner (1994), that is the
fact that subnational and local governments’ ample autonomy for spend-
ing must be exercised subject to a budget-constrained choice which
induces efficient resource allocation; in this respect, taxpayers’ demands
for goods and services resemble those of efficiency-maximizing buyers.
The setting in Argentina

Argentina is constitutionally a three-tiered federation composed of the
national government, twenty-three provinces and the autonomous
government of the city of Buenos Aires8 (which houses the federal capital),
Argentina 209
and over a thousand municipalities. In all cases, elected officials exercise

executive powers (president, governors, and mayors), whereas legislative
powers are in the hands of elected senators and representatives9 (at the
national and provincial level) and of elected municipal councillors (in the
case of local governments). All elected executive and legislative members
remain in office for four years, except for national senators whose mandate
lasts six years. The members of the national and provincial Courts of
Justice are nominated by the Executive Powers and respectively appointed
by the Congress and the provincial legislatures. The Senate is, at least in
theory, the main and most important federal body. Not only does each
province, regardless of its population, have three senators but the
Constitution rules that the Senate must be the chamber of origin for certain
laws involving relations among jurisdictions (for example, those setting up
revenue-sharing systems) or policies checking economic disparities
among provinces or regions.
As Argentina is a federal country, besides political powers, subnational
and local governments constitutionally have ample independent fiscal and
spending functions. Hence, the validity of the argument that the country
has behaved more like a “formal” rather than a “true” federation (follow-
ing Bird’s [1995: 295] definition of both terms) stems from the subnational
and local governments falling short of properly implementing their
autonomous broad powers for taxing and expenditure. In this respect, the
national constitution dictates that provincial constitutions must ensure the
administration of justice and primary education in their jurisdictions. At
the same time, provinces are recognized in their precedence in that they
are entitled to have their own political institutions and may keep all
powers not expressly delegated to the central government. Municipal
autonomy and regimes are also constitutionally guaranteed, provinces
being responsible for establishing the institutional, political, administra-
tive, economic, and financial content and scope.
With regard to the assignment of spending functions, except for defense
(a central government commitment) and law and order (the central and
subnational governments have exclusive authority), there are practically
no fields for which responsibility is constitutionally limited to a single
level.10 Thus, for most types of expenditure all three levels exhibit concur-
rent competencies, although in practice certain fields are in the domain of
a single government level (social security services, predominantly in
central government’s hands, illustrate this assertion).
The tax assignment issue, as dealt with by the Constitution, reveals
another feature of the highly federal-like fiscal setting in Argentina. Only
export and import duties can be exclusively imposed by the central
210 Ernesto Rezk
government; both the central and subnational levels share the faculty to
raise indirect taxes whereas the former can levy direct taxes only for
limited time periods and on grounds of defense, common security, and
general welfare.11 Put plainly, however, the implementation of the main
direct tax in Argentina (income tax) amounts in practice to provinces yield-
ing their control of this tax to the national government.12 Local govern-
ments’ fiscal powers, as described by provincial constitutions, normally
apply to user charges for services rendered to residents; municipalities are,
however, empowered to collect taxes on property, business, and consump-
tion,13 and contributions for improvements.
The actual tax assignment in Argentina, as shown in table 7.1, shows that
the assignment of revenue sources to various government levels is broadly
consistent with the widely accepted Musgravian economic criteria.
Fiscal instruments are more efficiently and equitably administered by
the central government in the case of:
• tax bases unevenly distributed among jurisdictions
• taxes on mobile production factors
• taxes aimed at stabilization and redistribution goals
Fiscal instruments fare better at the subnational and local levels in the
case of:
• taxes on totally immobile factors and assets
• residence-based taxes on consumption
Fiscal instruments such as benefit contributions and user charges are
suited to all levels.
Table 7.1 already hints at what will become clear in the quantitative over-
view of the following section; that is, that overlapping tax systems (rather
than separate revenue sources)14 founded on jurisdictions’ fiscal faculties
led to revenue-sharing as the preferred method of tax coordination in
Argentina. Starting in 1935, revenue-sharing has thus become the main
intergovernmental fiscal arrangement, whereby the central government
and the provinces intended to achieve the goal of avoiding tax competition
among jurisdictions. They also aimed to ensure that both levels had direct
access to tax revenues that were sufficient to close the fiscal gap. In other
words, the rationale for the system was that transfers from shared taxes,
plus their own revenues, should in principle permit provinces to finance
their spending functions.15
Table 7.1 shows that the primary distribution16 of shared national taxes
involves not only the central and subnational levels but also national and
provincial17 social security systems and diverse funds accruing to
Table 7.1. Argentina: tax assignment – responsibility for administration, collection and disposition of revenues (percentage
shares in tax revenues)
A. Tax levied by the central (national) government
Provinces and
the city of National social Provincial social Various provincial
Nation Buenos Aires security system security systems fundsa
Personal and corporate income tax 22.82 35.04 29.60 .— 12.54

Value-added tax 31.74 43.16 22.15 2.20 0.75
Excise taxes (internal taxes) and others 35.66 48.49 15.00 .— 0.85
Tax on assets 17.83 24.25 7.50 .— 50.42
Tax on personal goods .— ..— 90.00 10.00 .—
Tax on fuel sales 29.00 ..— .— .— 71.00
Energy tax .— — .— .— 100.00
Tax on salaries .— ..— 100.00 .— .—
Stamp duty 100.00 ..— .— .— .—
Import duties 100.00 ..— .— .— .—
Export duties 100.00 ..— .— .— .—
Statistics tax 100.00 ..— .— .— .—
Note: a Existing funds are mainly conditional transfers to provinces for their use in the following fields: energy, infrastructure, roads, public
works, housing, and education. There also exists a subsidiary fund for regional compensation and another for unconditional Treasury grants to
provinces.
B. Taxes levied by the provincesb
Provincial social Various municipal

Provinces Municipalities security systems fundsa
Turnover tax 80.00 16.10 .— 3.90

Urban and rural property tax 80.00 16.10 .— 3.90
National tax-shared revenues 80.00 16.10 .— 3.90
Tax on provincial and municipal payrolls .— .— 100.00 .—
Stamp duty 100.00 .— .— .—
Tax on motor vehicles .— 100.00 .— .—
Social infrastrucure tax 100.00 .— .— .—
Tax on gambling 100.00 .— .— .—
Tax on turf activities 100.00 .— .— .—
Fees and user charges 100.00 .— .— .—
Note: b Notwithstanding marked similarities, there are in Argentina twenty-three provincial tax-sharing systems. Revenue disposition has been
illustrated here by the prevailing scheme in the province of Córdoba, which fairly represents the rest.
C. Municipalities fiscal revenuesc
Property tax
Contribution on improvements
Business tax
Advertisement and publicity tax
Tax on raffles and other games of chance
Tax on energy sale
Tax on cemetries’ occupancy
Various fees and user chargesd
Notes:
c
This fiscal revenue pattern belongs to the city of Cordoba’s municipal government but, again, most local governments in Argentina share the
same revenue structure.
d
Fees received by municipalities are normally collected for streets, pavements, and public spaces’ occupancy, and for applications or requests to
the local government for an administrative action. User charges are normally linked to sanitary protection and inspections in public places, as
well as to the approval, supervision, and inspection of gas networks, and the construction of houses and buildings within municipal boundaries.
Argentina 213
provinces in the form of conditional transfers on a principal–agent basis.

The percentages of the primary distribution have undergone continual
changes since the system came into being in 1935, although each time a
new intergovernmental arrangement was enacted (table 7.2), the resulting
percentage pattern generally favored subnational governments as these
were left with a bigger portion of shared taxes.18 Thus, after the last modi-
fication occurred in 1990, the provinces retained 57.05 percent of the total
net shared amount (that is to say, the balance remaining after funds for
national and provincial social security systems and other special funds
have been deducted) while the central government kept only the remain-
der (42.95 percent).19
Twenty-three independent revenue-sharing systems also exist at the
subnational level, whereby each province and its municipalities partici-
pate in the distribution of the main provincial tax and of national shared
taxes. As in the national case, primary distribution also entails the
establishment of several (now municipal) funds for specific purposes.
Regarding secondary distribution, objective-devolving and equalizing
parameters ceased to exist after 1988 (table 7.2) and revenue disposition
among provinces has been thereafter performed on the basis of Law
23 548’s fixed coefficients, agreed upon by means of political negotiations
and whose value has remained constant ever since.20 However, explicit
distribution criteria exist in almost all provincial revenue-sharing schemes,
partly based on devolution principles such as municipal population, area,
own revenues, and efficiency in tax collection, and partly on equalizing
principles that may allot part of the funds equally or relate distribution to
local governments’ social spending, expenditure budgets or theoretical
salary needs for minimal public service levels.21
One point worth mentioning in this connection is that the widespread
use of revenue-sharing (since 1935) for channeling intergovernmental
fiscal relations has so far been based upon “contract laws” promulgated by
Congress, so called because they require provincial legislatures’ explicit
ratification before their content becomes mandatory for both government
levels. The 1994 Constituent Assembly changed this situation by giving
constitutional status to revenue-sharing and dictating that a new system
(entailing objective distribution criteria and built-in fund transfers to
provinces and the city of Buenos Aires) be implemented from January 1,
1997, in which primary and secondary distribution is directly related to
subnational jurisdictions’ services, functions, and competencies.22
The Constituent Assembly also introduced a clause whose expected
positive impact on decentralization is certain to occur since it favors better
future decisions concerning delegation or devolution from the center to the
214 Ernesto Rezk
Table 7.2. Tax-sharing systems (1935-95) – criteria for secondary distribution

(percentage shares)
Equalizing criteria
Spending needs
Total for
Devolution Current Minimum Supply Tax equalizing
criteria spending spending cost capacity criteria
Period (1935–53)
Law 12139 (1935)
Internal taxes on 91 — 9 — — 9
consumption
Internal taxes on 100 — — — — 0
production
Law 12143 (1935) 40 30 30 — — 60
Law 12147 (1935) 40 30 30 — — 60
Decree 14342 (1946) 40 30 30 — — 60
Law 12956 (1947) 36 27 27 10 — 64
Law 14060 (1951) 100 — — — — 0
Period (1954–72)
Law 14390 (1954) 16 — 82 — 2 84
Law 14788 (1959) 25 25 25 25 — 75
Period (1973–84)
Law 20221 (1973) 0 — 65 10 25 100
Period (1985–95)
Law 23548 (1988)a — — — — — —
Note: a Existing laws, for this period, do not specify any method for the calculation of
distribution indices.
Source: Own estimates based on mentioned legal regimes.
provinces. The clause states that no transfer of competencies, functions, or

services is to be accomplished unless the previous revenue reassignment
took place by means of a Congress law which was in turn ratified by the
province concerned.
Performance of federal fiscal arrangements and of

decentralization23
Despite the fact that constitutional provisions suggest that the federal
finance24 perspective characterizes intergovernmental fiscal relations, the
empirical evidence that has been gathered clearly shows that Argentina is
one of the more fiscally concentrated federations in the world, in which
local governments are by far the political system’s weakest link, and that
interjurisdictional fiscal arrangements (apart from revenue-sharing) are
Argentina 215
Table 7.3. Distribution of tax revenue by source and level of government, 1995
Indirect tax
Income Social security Property Other
tax contributions tax domestic internation. taxes Total
Central government 100.00 89.30 11.70 73.80 100.00 53.00 77.30

Provinces 0.0 10.70 67.90 21.30 0.0 41.50 18.90
Municipalities 0.0 0.00 20.40 4.90 0.0 5.50 3.80
% of total 11.30 24.00 6.80 42.30 9.00 6.60 100.0
Source: Own estimates based on figures from the Informe Económico 1995, Secretary for
Economic Programming, Argentina, April 1996.
better depicted by the delegation variant of decentralization and by an

extensive use of the principal–agent analytical model. The above assertion
on concentration finds empirical support not only in tables 7.3 and 7.4,
which show each government level’s actual tax-raising and spending
assignments, but also in tables 7.5 and 7.6, for 1991 and 1993 respectively,
indicating that the national government increased its participation in two
of the four major spending groups.25
As previously noted, the marked tax centralization makes itself evident
through the actual distribution of fiscal powers in Argentina (table 7.3).
The central government was responsible in 1995 for more than 77 percent
of tax yield, while provinces and municipalities respectively accounted for
almost 19 percent and 4 percent of the total. This concentration results from
taxes exclusively raised by the central government (like non-shared import
and export duties, and taxes on salaries on personal goods) while the other
important taxes in terms of GDP points (VAT 6.0 percent, income tax 2.34
percent, excise taxes 0.7 percent and liquid fuels tax 0.64 percent) are also
collected by the central government and transferred to provinces both by
the revenue-sharing system and through various conditional grants that
reassert the federal government’s preferences for delegation rather than
devolving decentralization (table 7.1). Table 7.3 not only stresses the domi-
nance of consumption taxes (42.3 percent)26 but also the role they play in
actual own provincial revenues (around 48 percent). At this level, indirect
taxation is overwhelmingly represented by the turnover tax, whose
replacement by a consumer tax at the retail level, owing to its cascading
and pyramiding effects and to its negative impact upon exporting
sectors,27 is being discussed. This is despite its important contribution to
provincial budgets and the relative ease of collection. Table 7.3 also shows
that property taxes are mainly collected at subnational levels, where their
yield of 24.4 percent makes them the second most important tax in terms
216 Ernesto Rezk
Table 7.4. Vertical imbalance, 1995
Before transfers After transfersa
Central Provincial Central Provincial Central Provincial

governm. governm. governm. governm. governm. governm.
Revenue shareb 81.80 (18.20 58.06 (41.94 54.26c (39.20c

Expenditure share 64.30 (35.70 56.39 (43.61 56.39 (43.61
Surplus/deficit 17.50 (17.50) 1.67 (1.67) 2.13 (4.41)
Notes:
a
Includes transfers due to any existing interjurisdictional fiscal arrangement between the
central government and the provinces.
b
It includes revenues of all kinds (excludes borrowing).
c
When the overall fiscal deficit is allowed for (see note 28).
Source: Own estimates based on figures from the “Informe Económico 1995,” Secretary for
Economic Programming, Argentina, April 1996 and on data from the Secretary for the
Assistance to the Provincial Economic Reform.
of contribution to provinces’ own revenues. As for Argentinian municipal-

ities, this government level, contrary to other federations’ experience, has
so far played a minor role in consolidated tax collection (only 3.8 percent
relative to 8 percent and 14 percent reached by German and Brazilian local
governments respectively), the main part of which consisted of property
and business taxes, diverse fees, and user charges which cover, on average,
less than 50 percent of their spending needs.
Fiscal centralization is only partially reduced by intergovernmental
transfers of all kinds (table 7.4), as almost 60 percent of total tax revenues
are still retained by the central government. As will be noticed, the rela-
tively greater decentralization observed on the expenditure side requires
corrections in present intergovernmental arrangements for revenue
assignment and distribution, if one wants the 1.67 percent computed sub-
national governments’ deficit to disappear.28
Tables 7.5 and 7.6, showing spending data for 1991 and 1993 respec-
tively, by function and by jurisdictions’ percentage share, give empirical
support to the assertion that the revenue centralization pattern did not
hold in the case of some expenditures in which sustained decentralization
processes occurred, although mainly in the form of delegation. Such were
the cases of primary and secondary education and welfare and, to a minor
extent, health, for the provision of which, provinces assumed exclusive or
very important roles.29 In support of this argument, table 7.7 shows that
unconditional grants to provincial governments dwindled from around 80
percent of the total from 1983 through 1991 to 73 percent in the period
1991–95. At the same time, conditional grants grew from their traditional
Argentina 217
Table 7.5. Total spending by level of government, 1991 (percentage shares)
All government National

levels government Provinces Municipalities
Total expenditure 100.00 55.64 35.67 8.68
Administration and law and order 100.00 47.12 35.39 17.49

General services 100.00 31.95 32.78 35.27
Justice 100.00 43.59 56.41 0.00
Defense 100.00 100.00 0.00 0.00
Public order and safety 100.00 43.20 56.80 0.00
Social expenditure and spending 100.00 53.37 38.37 8.26
on human resources
Education, science, and 100.00 33.63 63.99 2.38
technology
Health 100.00 50.40 43.27 6.33
Housing 100.00 8.06 91.94 0.00
Welfare 100.00 10.91 63.64 24.45
Social security 100.00 76.04 23.96 0.00
Labor 100.00 100.00 0.00 0.00
Other expenditure 100.00 0.00 0.00 100.00
Infrastructure for economic affairs 100.00 57.27 39.76 2.97
and services
Primary sectors 100.00 10.53 89.47 0.00
Fuel and energy 100.00 79.07 20.16 0.78
Manufacturing 100.00 71.43 14.29 14.29
Services 100.00 58.82 36.97 4.20
Public debt 100.00 93.75 6.25 0.00
Source: Based on data provided by the Secretary for Economic Programming.
20 percent to 27 percent of overall transfers. Finally, funds for decentral-

ized services (mainly education) increased to around 10 percent of total
grants received by provinces and amounted to 34.5 percent of conditional
transfers. Given that the responsibility for important social services
(education and health) was transferred to provinces in the last four years,
it is clear that agency relationships (and not mechanisms akin to public
choice) underlie decentralization, to the detriment of both the principles of
financial autonomy and fiscal responsibility (towards their residents) of
subnational governments. It is, however, an encouraging feature concern-
ing local decision-making, that automatic transfers (shared revenues, dis-
equilibria funds, and royalties to oil-producing provinces) gradually
became the almost exclusive components of unconditional grants while
the non-automatic discretionary Treasury grants (ATN in table 7.7) saw
their share reduced to 2.0 percent of overall transfers by 1995.
218 Ernesto Rezk
Table 7.6. Total spending by level of government, 1993 (percentage shares)
All government National

levels government Provinces Municipalities
Total expenditure 100.00 57.13 35.12 7.75
Administration and law and order 100.00 51.72 32.40 15.88

General services 100.00 39.15 29.36 31.49
Justice 100.00 54.55 45.45 0.00
Defense 100.00 100.00 0.00 0.00
Public order and safety 100.00 45.13 54.87 0.00
Social expenditure and spending 100.00 55.60 37.69 6.71
on human resources
Education, science, and 100.00 25.21 72.49 2.29
technology
Health 100.00 48.03 45.79 6.18
Housing 100.00 12.50 87.50 0.00
Welfare 100.00 5.81 79.07 15.12
Social security 100.00 80.65 19.35 0.00
Labor 100.00 100.00 0.00 0.00
Infrastructure for economic affairs 100.00 54.62 40.16 5.22
and services
Primary sectors 100.00 26.47 73.53 0.00
Fuel and energy 100.00 78.13 20.83 1.04
Manufacturing 100.00 66.67 8.33 25.00
Services 100.00 53.01 39.76 7.23
Public debt 100.00 92.77 7.23 0.00
Source: Based on data provided by the Secretary for Economic Programming.
In assessing whether spending functions (whether by delegation or

devolution) fulfilled Oates’ decentralization theorem,30 aggregate figures
in tables 7.5 and 7.6 permit conclusions that, although limited in scope,
reveal the trend followed by expenditure assignment in Argentina. The
percentages shown indicate a certain degree of overlapping of spending
functions, which is undoubtedly attributable to concurrent powers
constitutionally granted to government levels in almost all spending
fields. A more careful inspection of tables 7.5 and 7.6 shows for administra-
tion and law and order, but not for defense spending (which is exclusively
national), a relatively even participation of central and subnational
governments, which can be explained by the typical duplication of execu-
tive, legislative, and judiciary powers, and police forces in a federation.
The massive social expenditure delegation to provincial governments
which, except for social security payments and services, are now involved
Table 7.7. Unconditional and conditional transfers to provinces and the city of Buenos Aires, 1983–95 (percentages)
Unconditional transfers Conditional transfers
Non- Non-
Automatic automatic Automatic aut.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Total
1983 27.0 0.0 18.1 45.8 0.0 4.3 0.0 0.0 0.0 0.0 0.0 3.4 0.3 19.7 1.3 100
1984 28.9 0.0 18.8 40.4 0.0 4.4 0.0. 0.0 0.0 0.0 0.0 2.3 0.4 13.8 1.0 100
1985 65.3 0.0 10.9 11.7 0.0 6.4 0.0 0.0 0.0 0.0 0.0 3.7 0.8 19.7 1.5 100
1986 60.9 0.0 11.6 14.6 0.0 5.6 0.0 0.0 0.0 0.0 0.0 3.8 0.9 10.7 2.0 100
1987 55.5 1.9 18.9 15.7 0.0 4.6 0.0 0.0 0.0 0.0 0.0 6.2 0.7 15.2 1.3 100
1988 57.9 7.5 11.8 11.0 0.0 5.8 0.0 1.8 0.0 0.0 0.0 1.5 0.6 11.2 0.9 100
1989 60.3 8.8 17.6 11.7 9.4 2.1 0.0 1.2 0.0 0.0 0.0 1.3 0.0 17.4 0.3 100
1990 65.6 2.0 18.1 10.9 3.8 1.7 0.0 2.1 0.0 0.0 0.0 2.5 0.1 11.5 1.7 100
1991 73.8 2.2 14.9 11.2 1.7 0.7 0.0 0.6 0.0 2.9 0.3 1.9 0.2 18.6 1.1 100
1992 66.3 1.5 14.1 11.7 0.0 1.7 8.9 0.9 2.2 2.1 0.9 3.4 0.6 15.6 0.2 100
1993 62.7 3.8 13.8 12.5 0.0 1.9 9.0 1.4 4.2 1.1 1.0 1.6 0.7 15.8 0.6 100
1994 61.7 3.6 13.4 13.5 0.0 1.9 8.7 1.2 5.4 0.8 1.0 1.8 0.5 16.2 0.2 100
1995 63.5 3.8 13.8 12.0 0.3 2.0 9.2 0.8 6.1 0.4 0.8 1.2 0.6 15.2 0.4 100
Key to columns:
1. Tax-sharing system 19. Income tax (infrastructure)
2. Disequilibria fund 10. Educational fund
3. Royalties 11. Infrastructure transfer
4. National Treasury grants (ATN) 12. Others
5. Anticipated taxes 13. Energy fund (FEDEI)
6. Road construction 14. National housing fund (FONAVI)
7. Decentralized services 15. Regional Disequilibria fund
8. Social security
Source: Own estimates based on figures from the Secretary for the Assistance to the Provincial Economic Reform.
220 Ernesto Rezk
in 70 percent of spending in this field, may be taken as an indication that

the subnational level is in a position to internalize geographically the costs
and benefits of services and also to check possible spillover effects. In
Argentina, the provinces and the city of Buenos Aires are practically
responsible for primary and secondary education (post-secondary educa-
tion is still a central government responsibility) as well as housing and
welfare programs. Shared with the central government are health and
social security services (in the latter case only for pension payments and
other services to provincial and municipal personnel). Subnational
governments receive various conditional revenues to finance these ser-
vices, as detailed in table 7.7. Subnational governments are also active with
respect to infrastructure and economic expenditure, and their share of 40
percent takes the form of direct social overhead capital provision and of
sectoral services to productive activities.
In analyzing the rationale whereby most social and human resources
expenditure is being decentralized to provinces in Argentina, Oates’
theorem must be quoted again in that provinces are generally considered
to be (save perhaps for the case of some provinces’ exclusive provision of
complex or specialized medical attention)31 the political jurisdictions with
respect to which spillover effects can be identified and controlled.
Nevertheless, it must be borne in mind that for the main decentralized ser-
vices (such as education and housing), what has been transferred is the
provision or the respective program operation while policy objectives and
design mostly remain in central government’s hands. This confirms rather
than denies the emphasis placed on the attainment of efficiency goals. The
underlying rationale here is that subnational governments will be keener
to identify communities’ needs and preferences and, for this reason,
decentralization will be conducive to allocative and productive effi-
ciency.32
When decentralization is assessed in relation to municipalities, table 7.6
stresses not only that their importance is minor relative to other levels (7.75
percent of total public spending) but also that their role is mainly confined
to performing general administrative functions (including diverse control
and inspection services and supervision of land use). Additional roles are
in providing traditional “community” services as well as some still incipi-
ent performance in social areas such as primary education, preventive
health care, and social welfare. It is thus concluded that in Argentina,
actual decentralization in relation to local governments still falls short of
both constitutional powers and real economic possibilities, in part because
localities have failed to take a more active role with respect to club-like
local public goods.33 Such provision by municipalities is consistent with
Argentina 221
economic principles, as these goods allow technical excludability and

benefit and cost appropriation to local residents only.
The fiscal gap in Argentina

Existing tax and spending function assignments among the central and
subnational governments have in part been the cause of the country’s
serious vertical fiscal imbalance. According to table 7.4, it appears that
while revenues in 1995 reached 96 percent of spending at the national level,
resources from all sources (after intergovernmental transfers have been
accounted for) only made up, in the same year, 90 percent of total pro-
vincial expenditures.
The measurement of the fiscal gap was here performed by using alter-
native versions for the vertical fiscal imbalance coefficient34 whose formu-
las and yields for the period 1983–95 are shown in table 7.8.
Notwithstanding criticisms raised by Bird (1986), this coefficient’s
values, and their evolution in the period considered, allow some inter-
esting conclusions to be drawn. It is to be noted, in the first place, that
provinces’ own fiscal resources finance less than half of provincial spend-
ing (as measured by Vce when shared revenue is regarded as a transfer). It
is, however, worth pointing out that despite marked cyclical behavior,
since the Stabilization Program started in 1991 values have been close to
the traditional historic floor of 40 percent.35 Nonetheless, this still leaves
Argentinian provinces well behind the fiscal situation of subnational
governments in other federations, like Brazil, in which states and
municipalities levy on average more than 60 percent of total tax revenues.
The extreme values shown in table 7.836 (31 percent and 49 percent) are
explained by the effect of the 1989 hyperinflation upon tax revenues and
the fact that revenues recovered more than expenditures in 1993.
The acute inflation endured by the country during the second half of the
eighties,37 the climax of which was reached with the 1989 and 1990 hyper-
inflations, also damaged provincial finances, as the calculated values of V
show for the period 1986–89. This illustrates the uneven way in which
national and subnational governments, linked by inter-jurisdictional tax
arrangements, bore the fiscal revenue losses caused by inflation differently.
In fact, whereas the yield of the formal tax system in 1988, 1989, and 1990
was respectively 17.3 percent, 18.1 percent, and 20.8 percent of GDP, the
figures were 21.6 percent, 28.5 percent, and 25.1 percent when the inflation
tax is included within the estimation of the tax yield.38 In other words, the
inflation tax (exclusively appropriated by the central government)
reached, in these years, 19.7 percent, 36.7 percent, and 17.1 percent of total
Table 7.8. Vertical fiscal imbalance coefficients, 1983–95
1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995
(millions of 1995 pesos)
To 4 090 5 445 5 637 6 793 6 092 5 308 4 359 5 101 6 081 8 374 9 443 9 959 8 999
Ts 3 352 3 070 6 836 7 652 7 479 6 639 6 693 6 471 8 960 9 871 9 629 9 807 9 082
R 1 067 1 235 1 007 1 583 1 381 1 220 1 256 985 1 070 1 604 4 836 1 989 1 967
Go 6 688 5 227 1 321 2 027 2 226 2 324 3 047 1 463 1 210 1 082 1 553 1 659 1 422
Gc 2 359 2 323 2 315 2 895 3 764 2 495 1 368 1 925 1 975 3 939 4 178 4 421 3 800
B ⫺605 2 318 1 003 1 272 2 718 3 372 1 172 3 080 1 774 639 ⫺217 1 947 3 734
E 16 951 19 617 18 120 22 222 23 660 21 359 17 895 19 025 21 070 25 509 29 422 29 783 29 004
V 0.47 0.62 0.80 0.78 0.75 0.77 0.75 0.82 0.85 0.80 0.81 0.80 0.82
Ve 0.50 0.50 0.74 0.72 0.63 0.62 0.69 0.66 0.76 0.78 0.81 0.73 0.69
Vce 0.30 0.34 0.37 0.38 0.32 0.31 0.31 0.32 0.34 0.39 0.49 0.40 0.38
Key to rows:
ce
V ⫽ 1 ⫺ [(Go ⫹ Gc)/E]
V ⫽ 1 ⫺ [(Go ⫹ Gc ⫹ 〉)/E]
c e
Vce ⫽ 1 ⫺ [(Ts ⫹ Go ⫹ Gc ⫹ 〉)/E]
where:
Go Unconditional national grants to provinces
Gc Conditional grants to provinces
B Net borrowing by provinces
Ts Provinces’ revenue from shared national taxes
To Provinces’ own tax revenue
R Provinces’ other non-fiscal revenue
E Provincial expenditure
Argentina 223
national tax revenues, which is in line with the enlargement of the fiscal
gap (table 7.8) and with the trend followed by V in the second half of the
decade.39
Values for the measures V (when shared revenue and net borrowing are
both considered part of subnational governments’ own revenue) also
exhibited a cyclical pattern throughout the period considered, their respec-
tive maximum and minimum figures being 85 percent for 1991 and 47
percent for 1983. The 80 percent floor achieved since 1990 (in spite of
important increases in overall provincial public expenditure) responded to
improvements in the collection of national taxes (mainly VAT and income
tax) and this was immediately reflected in revenue-sharing and in the
other transfers to provinces (including the conditional ones for decentral-
ized services).
In sum, while Vce figures depict in general a poor provincial per-
formance, as regards their own fiscal base40 (excluding shared taxes), the
coefficient’s value approached the level of many federations (except
Brazil) when shared revenues were included among the provinces’ own
fiscal resources, and even reflected a smaller fiscal gap than in Australia or
Canada, where V reaches 70 percent on average. This positive feature
points out that despite possible imperfections, revenue-sharing can still be
regarded as the best available option for the system of federal finances in
Argentina.41
Measures aimed at checking the problem of vertical fiscal imbalances
entailed a mix of instruments on the part of the federal government: condi-
tional and unconditional transfers summarized as follows, as already
shown in table 7.7. Conditional transfers, amounting to 26.7 percent in
1995, included automatic grants for financing services decentralized to
provinces and to the city of Buenos Aires (9.2 percent) and a further set of
other automatic transfers for specific use, accruing from diverse funds, of
which infrastructure (6.1 percent), house-building (5.2 percent), road
construction (2 percent), social security (0.8 percent) and energy (0.6
percent) are particularly worth mentioning. The non-automatic condi-
tional transfer was fed from a Regional Disequilibria Fund whose share
(0.4 percent) was almost nominal and far from being decisive in any policy
sense.
As for unconditional transfers, apart from the already mentioned domi-
nant part taken by existing revenue-sharing arrangements (table 7.7), that
made up 63.5 percent of a total of 73.4 percent in 1995, the remaining auto-
matic grants consisted of royalties to oil-producing provinces (3.8 percent)
and of payments from the Fiscal Disequilibria Fund (3.8 percent) set up in
1993 as a part of the Fiscal Agreement42 between the central government
224 Ernesto Rezk
and the provinces. Non-automatic unconditional transfers were in turn

mostly represented by the so-called Treasury grants (ATN) (2 percent).43
The percentages reached in 1995 by conditional and unconditional
grants support the argument that the central government was gradually
exercising more control over recent provincial fiscal development. This
feature was particularly noticeable in the provisions of the Fiscal
Agreements, many of which were oriented toward achieving “tax base
homogeneity” and “tax rate limits and uniformity,” not to mention in the
pressures exerted upon provinces aimed at setting budget spending limits
and doing away with presumably economically inefficient taxes (turnover
tax, stamp duties, and taxes on energy).44
Horizontal fiscal imbalances

Different Argentine provinces do not have comparable fiscal situations.
They exhibit deep fiscal differences arising from existing socioeconomic
disparities whose magnitude is stressed in table 7.9. The table shows that
five jurisdictions (covering 27 percent of the total area of Argentina)
embodied almost 70 percent of total population and 80 percent of GDP in
1995, and that maximum per capita income (for the city of Buenos Aires)
was 8.5 times the minimum one (for the province of Formosa). The same
feature occurs with regard to other indicators such as the unemployment
and activity rates and the poverty indices for the different provinces.45
As expected, regional socioeconomic disparities affected tax collections
in many subnational governments (table 7.10 for 1994) and only the High
Revenue 1 provinces (and most of the High Revenue 2 group plus Entre
Rios and San Luis) had per capita own tax revenues higher than the “all
provinces average.” This statistical conclusion is by no means surprising if
one expects a positive functional relationship between income per capita
(and tax potential) and actual tax yield in each province. On the expendi-
ture side, per capita spending shows both the effect of population density
(economies of scale) and the spending “floor” ensuring administration and
public goods and services provision in each jurisdiction. In this connec-
tion, and contrary to what was noted earlier on the revenue side, the
spending of almost all High Revenue 1 provinces fell short of the “all
provinces average,” while in the other groups it normally exceeded the
average sum.46 Obviously, the persistence of such a disparity between rela-
tive revenue and spending levels has been made possible because the
redistributive feature of most national transfer arrangements (revenue-
sharing inclusive) enables poorer provinces to maintain extremely high
per capita spending levels. This assertion is illustrated by table 7.10, which
Argentina 225
shows that intergovernmental grants of all kinds financed more than 90

percent of per capita spending in Low Revenue 2 provinces (Catamarca
and La Rioja) whereas such grants financed barely 45 percent of High
Revenue 1 provinces’ per capita outlays. Nevertheless, equalization is not
being implemented on equity grounds in Argentina, i.e. with a view to
financing a minimum provision of public goods and services in each
jurisdiction. What really happens is that less favored provinces normally
get resources to finance their disproportionate public employment, which
rules out the possibility of efficiency and effectiveness objectives of public
expenditure being achieved since more public employment does not nec-
essarily mean more and better services. The latter assertion is proven with
table 7.11 which shows that, with the exception of the special case of the
two underpopulated Patagonian provinces of Santa Cruz and Tierra del
Fuego, the Low Revenue 2 provinces (Catamarca and La Rioja) exhibit
both the highest number of public employees per thousand inhabitants
and the maximum per capita public spending in personnel. In spite of this
their governments are permanently subject to criticism for their low stan-
dards of service provision and their poor administrative performance.47
Finally, the intended (though so far unsuccessful) policy to check hori-
zontal fiscal disparities through secondary distribution of shared taxes is
depicted in table 7.2. This shows that ever since revenue-sharing was
implemented for the first time in 1935 a gradual replacement of devolution
by equalizing criteria took place in favor of poorer provinces. This feature
is seen more clearly in table 7.12. While richer and higher tax revenue pro-
ducing provinces had small shares in secondary distribution, the share
going to the poorer provinces followed a consistently rising trend in the
revenue-sharing period 1935–94.48
Conclusions
This chapter has clearly shown that Argentina has not so far taken advantage
of the possibilities arising from its federal institutional setting, despite exist-
ing constitutional provisions whereby subnational and local governments
are given significant independent fiscal and spending functions. In particu-
lar, decentralization has been more marked in spending than in the tax field
thus creating a situation of vertical fiscal imbalance, which in turn has
entailed a resort to diverse conditional and unconditional federal grants.
In several important fields, spending decentralization took place mainly
through delegation49 which, together with the marked tax collection
concentration at the national level, brought about a lesser degree of finan-
cial autonomy on the part of the subnational level and a gradual switch of
Table 7.9. Provincial social and economic indicators
Index of per
capita pro-
Provinces’ vincial GDP Surface Population Unemploy- Activity Poverty
share of (Average ⫽ Population area density for ment rate rate for index for
GDP (%) 100) for 1995 (sq. km.) 1995 for 1995 (%) 1995 (%) 1995 (%)
(1) (2) (3) (4) (5) (6) (7) (8)
High Revenue 1 78.3 114 23 671 686 754 926 31 15.7 44.17 15.8
MCBAa 22.9 265 2 988 006 200 14 940 13.3 49.00 8.1
Santa Fe 9.7 114 2 934 220 133 007 22 18.4 41.00 17.6
Mendoza 4.4 101 1 500 818 148 827 10 6.7 37.30 17.6
Buenos Aires 33.8 88 13 333 670 307 571 43 16.6 45.90 17.1
Cordoba 7.5 89 2 914 972 165 321 18 15.9 38.00 15.1
High Revenue 2 6.5 114 1 971 777 930 423 2 13.2 40.70 20.3
Tierra del Fuego n.a. n.a. 96 917 21 263 5 10.4 40.80 22.4
Santa Cruz 0.7 134 180 115 243 943 1 7.1 38.80 14.7
La Pampa 1.1 135 280 876 143 440 2 10.6 41.90 13.5
Chubut 1.5 131 396 800 224 686 2 14.5 39.00 21.9
Neuquen 1.5 113 460 395 94 078 5 16.5 40.10 21.4
Rio Negro 1.7 106 556 674 203 013 3 13.5 42.40 23.2
Medium Revenue 9.7 72 4 647 860 476 411 10 15.1 36.75 27.5
Tucuman 2.9 83 1 209 716 22 524 54 19.2 37.30 27.7
San Luis 0.7 76 320 109 76 748 4 10.4 38.50 21.5
Salta 1.9 69 952 174 155 488 6 15.7 38.00 37.1
Jujuy 1.1 69 551 804 53 219 10 12.4 33.80 35.5
Entre Rios 2.1 68 1 063 416 78 781 13 13.2 35.10 20.6
San Juan 1.0 63 550 641 89 651 6 14.0 38.50 19.8
Low Revenue 1 4.7 43 3 761 596 426 050 9 10.6 34.26 36.0
Corrientes 1.3 53 852 685 88 199 10 14.7 35.60 31.4
Santiago del Estero 0.9 45 696 092 136 351 5 8.6 32.70 38.2
Chaco 1.1 43 890 548 99 633 9 12.8 33.40 39.5
Misiones 1.0 39 877 904 29 801 29 7.8 35.70 33.6
Formosa 0.4 31 444 367 72 066 6 6.4 33.00 39.1
Low Revenue 2 0.8 52 533 725 192 282 3 11.8 36.37 27.7
Catamarca 0.5 60 287 567 102 602 3 12.2 36.60 28.2
La Rioja 0.3 42 246 158 8 9680 3 11.3 36.10 27.0
All provinces 100.0 100 34 586 644 2 780 092 12 15.0 41.77 19.9
Maximum value 33.8 265 13 333 670 307 571 14 940 19.2 49.00 39.5
Minimum value 0.3 31 96 917 200 1 6.4 32.70 8.1
Average 4.2 84 1 441 110 1 158 837 634 12.6 38.30 24.7
Standard deviation 7.8 51 2 616 918 73 433 2 983 3.5 3.90 8.8
Max./min. value 112.7 9 138 1 538 20 234 3.0 1.50 4.9
Note: a Municipalidad de la Ciudad de Buenos Aires.

Sources:
(1) Fundación de Investigación Económica Latinoamericana, 1992
(2) Own estimates
(3) Statistical Yearbook, Republic of Argentina, 1995 (INDEC)
(6) Rate of open unemployment in urban areas (October 1995) (INDEC)
(7) Rate of activity in urban areas (May 1995) (INDEC)
(8) Percentage of households with Unsatisfied Basic Needs (NBI) (INDEC)
Table 7.10. Per capita provincial revenue and spending, 1994 (1995 pesos)
Total Own Own Other National Shared Total

rev. rev. tax own revenue tax Other spending Current Capital
(1) ⫽ (2) ⫽ rev. rev. (5) ⫽ (6) ⫹ rev. Royalties transfers (9) ⫽ spending spending
(2) ⫹ (5) (3) ⫹ (4) (3) (4) (7) ⫹ (8) (6) (7) (8) (10) ⫹ (11) (10) (11)
High Revenue 1 702 401 360 47 295 188 2 105 726 636 90
MCBA 976 901 837 65 73 55 0 18 969 888 80
Santa Fe 754 334 297 42 415 296 0 119 764 699 64
Mendoza 781 284 257 84 440 270 32 137 832 727 105
Buenos Aires 600 321 287 36 278 160 0 117 616 525 90
Cordoba 791 383 324 65 403 298 0 105 888 772 116
High Revenue 2 1 628 401 248 169 1 210 498 234 478 1 903 1 489 414
Tierra del Fuego 3 626 700 587 147 2 903 734 350 1 819 4 665 3 886 779
Santa Cruz 2 826 513 319 199 2 302 866 479 957 2 917 2 079 843
La Pampa 1 719 556 260 311 1 152 650 22 480 1 607 1 195 408
Chubut 1 106 187 110 92 901 390 144 367 1 376 1 183 195
Neuquen 1 748 489 259 241 1 239 376 527 336 2 152 1 555 597
Rio Negro 1 150 321 252 87 813 441 71 301 1 456 1 227 230
Medium Revenue 912 223 153 83 676 459 5 212 1 033 875 158
Tucuman 704 155 128 28 548 381 0 166 831 675 156
San Luis 1 428 251 248 96 1 084 704 0 379 1 614 1 132 482
Salta 805 201 94 115 596 394 16 186 942 810 132
Jujuy 1 026 253 93 169 766 500 0 266 1 140 982 158
Entre Rios 967 317 247 80 641 444 6 192 957 871 85
San Juan 1 033 181 133 60 838 591 4 243 1 341 1 179 162
Low Revenue 1 870 106 91 39 740 511 3 226 930 783 148
Corrientes 726 91 79 20 627 422 0 205 747 654 93
Santiago del Estero 965 121 96 37 831 569 0 262 1 036 916 121
Chaco 885 112 100 53 732 542 0 190 894 752 141
Misiones 695 106 102 16 578 370 11 197 804 665 139
Formosa 1 312 101 67 95 1 151 805 7 339 1 435 1 112 325
Low Revenue 2 1 769 157 88 75 1 606 885 0 721 1 910 1 738 172
Catamarca 1 499 110 99 14 1 386 933 0 452 1 746 1 647 99
La Rioja 2 087 212 75 146 1 866 829 0 1 037 2 104 1 845 258
All provinces 817 341 292 58 466 288 16 163 874 749 125
Maximum value 3 626 901 837 311 2 903 933 527 1 819 4 665 3 886 843
Minimum value 600 91 67 14 73 55 0 18 616 525 64
Average 1 259 300 223 96 940 501 69 370 1 410 1 166 244
Standard deviation 709 201 175 73 635 222 150 383 866 688 216
Max./min. value 6 10 13 22 40 17 — 101 8 7 13
230 Ernesto Rezk
Table 7.11. Provincial public employment and spending on personnel, 1995
Per capita
Employees Wage spending spending in
per thousand per employee personnel
inhabitants (US$) (US$)
High Revenue 1
MCBA 27 1 631 538
Santa Fe 31 1 080 400
Mendoza 28 1 307 435
Buenos Aires 22 1 003 260
Cordoba 28 1 281 427
High Revenue 2
Tierra del Fuego 53 3 292 2 095
Santa Cruz 76 1 791 1 627
La Pampa 57 1 051 716
Chubut 47 1 317 749
Neuquen 62 1 321 986
Rio Negro 53 1 240 781
Medium Revenue
Tucuman 45 786 427
San Luis 54 1 007 653
Salta 44 1 012 530
Jujuy 51 1 024 625
Entre Rios 41 1 068 529
San Juan 64 816 628
Low Revenue 1
Corrientes 45 842 454
Santiago del Estero 56 826 552
Chaco 41 1 164 569
Misiones 39 897 420
Formosa 72 898 776
Low Revenue 2
Catamarca 81 1 016 984
La Rioja 97 1 031 1 194
All provinces
Maximum value 97 3 292 2 095
Minimum value 22 786 260
Average 51 1 196 723
Standard deviation 18 500 408
Max./min. value 4 4 4
Source: Own estimates based on figures from the Secretary for the Assistance to the
Provincial Economic Reform.
the latter’s fiscal responsibility as accountability followed the revenue

source (that is, to the principal rather than to residents).50 All in all, condi-
tional transfers to provinces represented in 1995 a record figure of almost
30 percent of total transfers. In this connection, the use of principal–agent
relations, rather than public choice-like mechanisms for collective deci-
Argentina 231
Table 7.12. Secondary participation by provinces, selected years (percent)
1935 1946 1958 1972 1973 1984 1994
High Revenue 1 67.5 70.3 65.6 59.5 50.7 51.5 43.7

MCBA 5.0 8.8 11.1 4.7 0.0 5.3 0.0
Santa Fe 12.6 12.6 10.3 8.7 9.1 8.4 8.9
Mendoza 19.9 9.7 5.3 5.2 4.7 4.2 4.2
Buenos Aires 20.7 29.7 31.0 32.9 28.0 25.4 21.9
Cordoba 9.3 9.5 7.9 8.0 8.9 8.2 8.8
High Revenue 2 0.0 0.0 1.8 8.1 8.9 9.8 9.9
Tierra del Fuego 0.0 0.0 0.0 0.3 0.0 0.3 0.7
Santa Cruz 0.0 0.0 0.0 1.6 1.4 1.5 1.6
La Pampa 0.0 0.0 0.0 1.5 1.8 1.4 1.9
Chubut 0.0 0.0 0.0 1.6 1.7 1.9 1.6
Neuquen 0.0 0.0 0.0 1.4 1.7 2.0 1.7
Rio Negro 0.0 0.0 1.8 1.7 2.3 2.6 2.5
Medium Revenue 27.6 20.7 18.4 17.1 19.6 18.6 21.9
Tucuman 9.6 6.2 4.1 3.7 4.6 5.0 4.7
San Luis 0.6 1.4 1.5 1.4 1.8 1.5 2.3
Salta 2.3 1.9 3.4 3.2 3.8 3.9 3.8
Jujuy 2.6 1.6 2.7 2.5 2.2 2.2 2.8
Entre Rios 3.8 5.4 4.3 3.9 4.7 3.5 4.9
San Juan 8.7 4.2 2.4 2.4 2.6 2.4 3.4
Low Revenue 1 3.8 7.0 11.4 12.7 17.2 16.6 19.7
Corrientes 1.9 4.0 3.6 3.9 3.8 3.5 3.7
Santiago del Estero 1.9 3.0 2.5 2.3 4.0 4.0 4.1
Chaco 0.0 0.0 3.0 2.7 4.1 4.0 5.0
Misiones 0.0 0.0 2.3 2.4 3.0 3.0 3.3
Formosa 0.0 0.0 0.0 1.4 2.3 2.2 3.6
Low Revenue 2 1.1 2.0 2.8 2.6 3.7 3.5 4.8
Catamarca 0.5 1.1 1.4 1.4 1.9 1.9 2.7
La Rioja 0.6 0.9 1.4 1.2 1.7 1.6 2.1
All provinces 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Maximum value 20.7 29.7 31.0 32.9 28.0 25.4 21.9
Minimum value 0.5 0.9 1.4 0.3 1.4 0.3 0.7
Average 6.7 6.7 5.6 4.2 4.5 4.2 4.3
Standard deviation 6.5 7.1 6.8 6.3 5.5 4.8 4.2
Max./min. value 41 33 22 110 19 87 32
Source: Own estimates based on figures from the Secretary for the Assistance to the
Provincial Economic Reform and from Nuñez Miñana and Porto (1983).
sions, was accentuated following the Convertibility Economic Program

and particularly since 1993, when Fiscal Agreements were imposed on
provinces to achieve national stabilization goals.51
The revenue-sharing system has been, for more than sixty years, the pre-
ferred mechanism for tax coordination in Argentina, not only among
national and subnational governments but also among provinces and their
municipalities. At the same time, the secondary distribution of revenue-
232 Ernesto Rezk
sharing was one of the instruments resorted to for checking horizontal

fiscal imbalances. This is sufficiently proven by the fact that devolution cri-
teria were gradually and totally replaced by equalizing criteria in the suc-
cessive revenue-sharing.52
The higher per capita transfers received by some backward provinces
resulting from re-distributive policies aimed at favoring poorer provinces
have, however, failed to check horizontal fiscal disparities, since national
grants only permitted larger provincial expenditures whose efficiency and
efficacy were dubious. In other words, the direction imposed on secondary
distribution did not prove to be a successful policy for achieving redistri-
bution goals (as the poverty index figures in table 7.9 show for low-income
provinces) and, what is worse, curtailed incentives for fiscally backward
provinces to develop their own tax potential,53 thus perpetuating the
almost complete dependence on transferred funds and the persistence of
horizontal fiscal imbalances.
The 1994 Constituent Assembly introduced important changes in the
fields of fiscal federalism and federal finance by giving, for the first time in
the country’s history, constitutional status to revenue-sharing as the key
interjurisdictional fiscal arrangement and setting up the framework within
which the new regime would supposedly come into being, from January
1, 1997.54
The 1994 constitutional reform also meant a step forward with respect
to the devolution variant of decentralization, and in relation to the likeli-
hood of a “true” rather than a “formal” federation, by mandating that no
future transfer of services would be possible unless accepted by the con-
cerned province and accompanied by a matching transfer of funds or
revenue sources.
In sum, the present concern of subnational governments regarding
federal finance issues shows increasing provincial awareness of their role
in a federal context and their real desire to regain the exercise of taxing
powers previously yielded. The Federal Congress is at the very root of this
change, as for the first time in decades both senators and representatives
are reluctant to endorse the national government’s new fiscal faculties in
so far as they may collide with provinces’ constitutional faculties or shrink
the proportion of national revenues shared with the provinces.
Notes
1 Tanzi (1995: 296) argued that the creation of a central entity transcending the
European member states in some economic functions opened up a debate
which was similar to those analyzed in discussions relating fiscal federalism
and fiscal decentralization.
Argentina 233
2 Argentina is, however, a good example for showing that fiscal decentraliza-
tion is progressing but is subject to important setbacks: the central govern-
ment, on the grounds of achieving macroeconomic stabilization goals, has
enforced since 1992 a Fiscal Agreement whereby provincial governments
have had to limit their total spending and to revise and adapt (following
national guidelines) their tax systems in exchange for a guaranteed shared
revenue floor.
3 Although a more precise concept would be that policies favoring decentralized
decision-making enable governments to perform their assigned tax and spend-
ing activities subject to efficiency and equity objectives.
4 Each depending on the degree of independent decision-making exercised at the
subnational and local level.
5 Bird (1994) argues that outcomes, in this variety of fiscal decentralization, are
presumed to be satisfactory simply because the process itself is intrinsically
desirable, insofar as it institutionalizes the participation of those affected by
local decisions.
6 By assuming a straight application of the approach, with implied symmetric
information and agents’ fair behavior by agents, agency problems (adverse
selection and moral hazard) are here ruled out.
7 Public choice refers to the economic study of non-market decision-making
mechanisms and is also defined as the application of economics to political
science. This theory is embedded in the Wicksellian link between a public good
and the tax used for its financing and in the viewpoint that the collective choice
process amounts to voluntary exchanges among individuals from which every-
body benefits.
8 By the last constitutional reform, in 1994, the city of Buenos Aires changed its
political status. The designated mayor was replaced in 1996 by an elected head
of government, and a Chamber of Representatives replaced the former munici-
pal council. However, the jurisdiction cannot totally be equated to a province
as the national government still retains responsibility for some functions and
services (for instance, the police force) while the Congress keeps some legisla-
tive faculties. Transfer of justice to the city is now favored by the national
government.
9 The Argentine Congress is bicameral: senators represent the provinces and the
city of Buenos Aires while representatives are chosen in proportion to popula-
tion, taking the country as a single district; direct election applies in both cases.
Unicameral and bicameral legislatures coexist at the provincial level; again,
senators represent provincial departments (in provinces having this chamber)
while provinces also stand as single districts for the election of representatives.
10 As an interesting case, provinces can provide post-secondary education.
However, all non-public universities in Argentina are funded by budgetary
transfers from the central government.
11 This in turn implies that no restrictions exist for subnational governments to
raise direct contributions.
12 The income tax was created in Argentina in 1935, practically together with, and
included in, the first revenue-sharing system between the central government
and the provinces. In order to fulfill what the Constitution prescribed, the
234 Ernesto Rezk
income tax law was established for a limited time period and has since been
permanently prorogued.
13 Municipalities have been applying taxes to the consumption of energy (levied
on the basis of electricity and gas sales to households and firms). As from 1993,
provinces signing the Fiscal Agreement agreed to ensure that local govern-
ments will do away with these contributions.
14 Separation of tax sources was actually the prevailing situation until 1935, when
the first revenue-sharing scheme was implemented.
15 This balance seldom occurred in reality mainly because of provincial tax admin-
istrations’ failure to reach even modest tax yield targets and because provinces
partially bore the financial burden of transferred services as accompanying
grants did not fully make up the cost of programs.
16 The terms primary and secondary distribution refer to the apportionment
between the central and the provincial governments and among the provincial
governments, respectively, of tax revenues included within the national
revenue-sharing system.
17 Provincial social security systems only provide services and pay pensions to
provincial and municipal employees.
18 The evolution of the primary distribution since 1935 clearly shows the share
gained by provinces (figures are percentages):
Year National government Provinces Others
1935 82.5 17.5
1947 79.0 21.0
1973 48.5 48.5 3.0
1988 54.7 42.3 3.0
1990 57.0 43.0
19 Changes in percentages occurred not only because of political negotiations
through which governors exerted pressures on the central government but also
because of changes in the rates and in the taxes included in the revenue to be
shared. Nevertheless, as table 7.1 shows, the percentages mentioned can be
deceptive as they apply after the amount of revenue shared has been reduced
by the amount destined to provincial social security systems and other specific
funds which cannot be spent freely by the provinces.
20 As from 1992, the central government guarantees a “floor” revenue to provinces
signing the Fiscal Agreements; that is to say, fixed distribution parameters
apply over actual tax revenues or over an agreed amount, whichever is greater.
The term Fiscal Agreements refers to fiscal instruments whereby provinces
accept a curtailment in their tax and spending powers (elimination of some dis-
torted duties, limits on rates, replacement of turnover tax, and limits on annual
spending) for which they would be granted the above-mentioned “floor”
shared revenue, the elimination of the tax on assets and effective reduction (to
firms) of social security contributions (50 percent on average, but varying
according to the region).
21 In some provinces the inverses of population and fiscal capacity are used as
equalizing criteria, thus favoring low-density and poorer municipalities.
Argentina 235
22 Equalizing criteria are not just included but, rather, emphasized, as the
constitutional mandate literally states that the system “will be fair, solidarity-
framed and will give priority to the achievement of similar degrees of develop-
ment and quality of life, and of equal opportunities throughout the country.”
23 This section is largely based on Rezk (1995) and Rezk, Capello, and Ponce
(1996).
24 This situation of federal finances is taken to occur, in Bird’s sense (1995: 294),
when the relationship between the central and subnational governments (both
having independent fiscal powers) is one of negotiation among equals; that is,
between government levels that are each, within a sphere, coordinate and inde-
pendent.
25 Figures for 1993 indicated that the spending for all government levels
amounted to 26.9 percent of the Argentinian GDP, 18 percent of which belonged
to social expenditure and spending on human resources, 4.7 to administration
and law and order, 2.5 percent to infrastructure for economic affairs and ser-
vices, and 1.7 percent to public debt.
26 One of the principal problems with fiscal decentralization in developing coun-
tries, in contrast to developed ones, arises because of the high degree of depen-
dence of the former on consumption taxes and the apparent difficulty of
working out adequate “two-level” consumption tax systems. This comment is
also applicable to latent problems in the Argentinian national and provincial tax
structure.
27 Despite successive postponements, this replacement is one of the main condi-
tions imposed upon subnational governments signing Fiscal Agreements.
28 Percentages 1.76 and ⫺1.67 come up when revenues and expenditures, respec-
tively are consolidated for all government levels, assuming that no national
deficit exists. Given that this assumption was not realistic for 1995, the last two
columns of table 7.4 include a new vertical fiscal imbalance estimation where
the national fiscal deficit is allowed for.
29 Suffice it here to point out that while provincial spending in elementary educa-
tion went from 92.3 percent (in 1991) to 95 percent (in 1993) of the total consoli-
dated spending for Argentina, total provincial figures for secondary education
rose from 40.8 percent to almost 100 percent of the total in the same period.
30 Oates’ classical argument (1972: 55) is that each public service should be pro-
vided by the jurisdiction which has control over the minimum geographic area
that would internalize benefits and costs of such provision.
31 An example of this could be public hospitals in the city of Buenos Aires and in
the provinces of Buenos Aires, Córdoba, Santa Fé and Mendoza.
32 Nevertheless, Prud’homme, in a recent article (1995), challenges the merits of
decentralization for allocative efficiency on the following two grounds: for
assuming hypotheses not easily defensible in developing countries and for
placing emphasis mainly on demand efficiency instead of supply or productive
efficiency.
33 The politico-institutional situation of Argentinian municipalities is similar to
neither Brazilian local governments (direct central–municipal relations) nor
Canadian ones (only provincial–municipal relations). However, it is closer to
236 Ernesto Rezk
the latter apart from strong revenue-sharing links with provinces and their
dependence on certain provincial transfers. Local governments do not have in
practice constitutionally legal impediments to seek aid at the central govern-
ment level, for instance the financing of determined public works or the
control and management of infrastructure (for example, the municipality of
Córdoba’s overt lobbying and pressure upon the national government as it
sought to improve its own chances as an interested party in the airport
privatization).
34 Despite its widespread use, the conceptual validity of the coefficient (due to
Hunter [1977]) has been challenged by Bird (1986) on the grounds that, mea-
sured by a single number, balance is taken to be an “unequivocal good” (i.e.
success will depend in any federal fiscal system on whether this balance is
achieved). On the other hand, given that the coefficient is expected to measure
subnational governments’ degree of dependency on central governments,
interjurisdictional fiscal arrangements in each country involve differing
degrees of control not reflected in results obtained by the coefficient; this makes
international comparisons difficult.
35 This is a consequence of both the success in checking high inflation and the
pressure exerted from the central government upon provinces urging them to
improve their tax administration systems.
36 The lowest percentage for the period (30 percent) was in 1983; this year was not
included as it was when the military handed over political power to civilians.
Until then, provinces did not have any real autonomy, as governors were
appointed by the central government.
37 The following percentages are the inflation rates, measured by the consumer
price index, experienced by the Argentinian economy from 1989 to 1994.
1989: 3 079
1990: 2 314
1991: 171
1992: 24.9
1993: 10.6
1994: 3.9
38 Including, of course, those taxes with revenue that was shared among the
national government and the provinces.
39 As from 1991 (and coincidentally with the stabilization program initiated by the
Argentina government), the incidence of inflation tax practically died out as
figures for 1991-94 indicate that it reached 2.1 percent, 0.8 percent, 0.8 percent
and 0.7 percent, respectively, of all fiscal revenues.
40 It is widely accepted that tax evasion in Argentina is higher at the provincial
level (or at least in the majority of provinces) because of weaker fiscal
compliance-inducing mechanisms.
41 The overwhelming politico-institutional acceptance of the revenue-sharing
system in Argentina is certainly an important policy datum.
42 See the considerations on the concept of Fiscal Agreements in note 20 above.
43 These Treasury grants are highly criticized and resisted by most provincial
Argentina 237
governments since they are allocated to provinces in a totally discretionary
manner and following, in some cases, party lines.
44 It was pointed out (Rezk 1995: 8) that Fiscal Agreements reflect the tradeoff
between national stabilization goals and provinces’ financial autonomy.
45 To gain a better understanding of the figures and percentages in table 7.9, note
that total GDP, per capita GDP, the unemployment rate and the activity rate for
the period 1992-94 were:
1992 1993 1994
GDP $226.7 bill. $257.6 bill. $281.6 bill.
Per capita GDP $6 786 $7 620 $8 235
Unemployment rate 6.9%/7.0% 9.9%/9.3% 10.7%/12.1%
(May/Oct.)
Activity rate 37.6%/38.8% 37.6%/37.6% 38.0%/37.6%
(May/Oct.)
46 The economically backward and sparsely populated western province of La

Rioja provides one of the most interesting cases of horizontal fiscal imbalance,
as per capita outlays were 3.4 times greater than in Buenos Aires whereas the
latter had per capita own tax revenue 3.83 times larger than the former.
47 The problem of accountability (to the province’s taxpayers or to the national
government) is also raised here as in both cases national transfers of all kinds
amount to 90–92 percent of their total revenues.
48 The good performance of provinces in the High Revenue 2 category is due to
the royalties received by these oil-producing regions.
49 That is the case of fields of experience in which provinces have a dominant role
relative to the national government (for instance, primary and secondary
education, housing, welfare, and, increasingly, health, according to tables 7.5
and 7.6); as can be seen from table 7.7, conditional transfers include grants for
decentralized services, for housing (FONAVI) and for unsatisfied basic needs
in the conurbation of the city of Buenos Aires. With regard to provincial spend-
ing on infrastructure and economic services, there are also conditional transfers
for energy (FEDEI), road construction and, since 1991, infrastructure projects.
50 This assertion should not, however, be taken in black and white terms as pro-
vincial governments are also accountable to their citizens for transferred ser-
vices. A proof of this is that whenever people demonstrate in provinces
(demanding a greater quantity or higher quality of services such as education)
the national and provincial governments are both subject to criticisms.
51 The relative increase of conditional transfers was acknowledged by the World
Bank report on Argentina (1996: 32) when it stated “since 1991, the primary dis-
tribution system (of shared tax revenues) changed several times, in line with the
need of the central government to cut the overall fiscal deficit and to appropri-
ate some funds from the ‘coparticipable’ pool.” In practice, those changes
amounted to reductions in the unconditional coparticipable pool and increases
in the earmarked funds for provinces and social security systems.
52 This is in turn reflected by the percentages of the secondary distribution for the
period 1935-94, in table 7.12.
238 Ernesto Rezk
53 One negative feature in the provinces, which is particularly severe in the “low-
income” ones, is the extremely poor performance of their tax administrations.
54 Although there exist several versions of the new system, none has reached leg-
islative status (even at the level of Senate committees) as the constitutionally
imposed deadline of January 1, 1997 has proved impossible to meet. Therefore,
the present percentages, for primary distribution, and the percentages of Law
23548 for the secondary distribution, will continue being used in the meantime.
References
Bird, R. M. (1986), “On measuring fiscal centralization and fiscal balance in federal
states,” Environment and Planning C: Government and Policy, 4: 389–404.
(1994), “A comparative perspective on federal finance,” in K. Banting, D. Brown,
and T. Courchene, eds., The Future of Fiscal Federalism, Kingston, Ont.: Queen’s
University School of Public Policy.
(1995), “Fiscal federalism and federal finance,” Anales de las 28 Jornadas de
Finanzas Públicas, Córdoba, Argentina.
Hunter, J. S. H. (1977), Federalism and Fiscal Balance, Canberra: Australian National
University Press and Center for Research on Federal Financial Relations.
Nuñez Miñana, A. and Porto, A. (1983), Distribución de la Coparticipación Federal de
impuestos, Análisis y Alternativas, Buenos Aires: Consejo Federal de
Inversiones.
Oates W. (1972), Fiscal Federalism, New York: Harcourt, Brace, Jovanovich.
Prud’homme, R. (1995), “On the dangers of decentralization,” Policy Research
Working Paper 1252, The World Bank, Washington, DC.
Rezk, E. (1995), “Federal finance in the Argentine, Germany and Brazil,” Annals of
the Arnoldshain Seminar I, J. W. Goethe University, Frankfurt and Anales de las
Jornadas de Finanzas Públicas, Córdoba, Argentina.
Rezk, E., Capello, M., and Ponce, C. (1996), “Arreglos fiscales interjurisdiccionales
en Argentina,” International Seminar on Fiscal Federalism and Federal
Finances, Córdoba, Argentina.
Tanzi, V. (1995), “Fiscal federalism and decentralization: a review of some effi-
ciency and macroeconomic aspects,” in M. Bruno and B. Pleskovic, eds.,
Annual World Bank Conference on Development Economics, Washington DC.
Wiesner, E. (1994), “Fiscal decentralization of the public sector,” in Economic and
Social Progress in Latin America 1994 Report, Special Report: Fiscal
Decentralization, ch. 1, Washington DC: Interamerican Development Bank.
World Bank (1996), Argentina. Provincial Finances Study: Selected Issues in Fiscal
Federalism, Report No. 15487 AR, Washington, DC.
8
South Africa: an intergovernmental fiscal system

in transition
JUNAID K. AHMAD
South Africa is addressing critical questions of intergovernmental fiscal

relations as it undertakes the process of abolishing apartheid and rede-
fining its system of governance. In the process, fundamental changes are
being implemented in the structure of government: the constitution has
been radically redesigned; new tiers of government are being created;
decentralization has been initiated; electoral accountability has been
extended to all irrespective of race; and, finally, the rapid delivery of basic
public goods for the non-white community has commenced.
This chapter discusses how the intergovernmental fiscal system is
being radically altered to support these fundamental changes.1 In
particular, it focuses on the restructuring of the provinces and the metro-
politan areas of South Africa. The success and sustainability of the reform
of the intergovernmental system will depend critically on the coordina-
tion of the change process for these two tiers of government. The first
section of this chapter sets the stage by describing how apartheid laws
influenced the intergovernmental fiscal system in South Africa. The fol-
lowing section of this chapter provides a discussion of how the new
provinces have been constituted. A third section focuses on the structure
of urban governance for the large cities of South Africa. In a final section,
there is analysis of some of the tensions inherent in the changes being
undertaken.2
Initial conditions: the consequences of apartheid

South Africa’s policies of racial separation created dualism in intergovern-
mental fiscal relations, as in social and economic life. Overwhelming fiscal
dependency in black areas coexisted with considerable self-sufficiency and
autonomy in white areas.3 This section describes how the spatial policies
of apartheid affected the fiscal structures of regional and local govern-
ments in South Africa.
239
240 Junaid K. Ahmad
Table 8.1. Shares of revenue, expenditure and transfers by tier of government,

1990–91
Transfers as percentage
Revenue shares Expenditure shares of own revenue
Central 80.1 70.9 .—

Province 2.0 9.7 83.6
SGT/TBVC 6.7 11.8 58.1
Local 10.6 7.6 17.8
Total 100.0 100.0 .—
The national context: the apartheid structure

Before the historic elections of April 27, 1994 the structure of government
in South Africa comprised three tiers: the center, the regions and local
authorities. The regional tier included the Provinces, the Self-Governing
Territories (SGTs), and the TBVC (Transkei, Bophutaswana, Venda, and
Ciskei) group. Local authorities consisted of the Regional Service Councils
(RSCs),4 the Black Local Authorities (BLAs) and the White Local
Authorities (WLAs). The national boundary was formed by the four
provinces and the SGT/TBVC states. Each province was served by several
RSCs. In turn, the RSCs spanned groupings of BLAs and WLAs.
Under apartheid, South Africa was a highly centralized state. According
to 1990–91 budgetary figures, for example, the central government
accounted for approximately 80 percent of the total revenues and 70
percent of the total government expenditures (table 8.1). Local govern-
ments, on the other hand, represented about 10 percent of revenues col-
lected and about 8 percent of expenditures. Given that the provinces were
deconcentrated arms of the center, these figures actually understate the
extent of centralization.
This high level of centralization, however, masked a duality at the sub-
national level: nationwide about 80 percent of the revenues of white
municipalities originated from own-sources. By contrast, at the regional
level, the corresponding figures for the provinces and the SGI/TBVC states
were 16 percent and 42 percent respectively in 1990–91. The dependence
of BLAs on the center was equally high, with fiscal transfers from the
center accounting for over 75 percent of total revenues.
The system of apartheid thus fostered a dual intergovernmental struc-
ture. On one hand, the central government provided the bulk of the fiscal
resources of the TBVC, SGTs, and BLAs, and on the other, it allowed white
local authorities sufficient fiscal and regulatory autonomy to act as an inde-
South Africa 241
pendent tier of government. The origin of this system can be traced directly
to the spatial policies of apartheid (Bruckner, 1994).
Spatial policies and the structure of government

Regional governments
Apartheid reserved 87 percent of the country’s land for the whites, who
currently constitute only 13 percent of the population. The black commu-
nity was physically herded into ten “homelands,” artificial jurisdictions
containing some of the most desolate territory in the country. Four of these
– Transkei, Bophutaswana, Venda, and Ciskei (the TBVC states) – accepted
the political fiction that they were separate nations, while the remaining
six did not, preferring the status of “self-governing territories” (SGTs).
The economic conditions of the homelands resulted in weak fiscal
capacities for most of these racially segregated regions. To ensure a
modicum of services, the central government provided the homelands
with large fiscal transfers and access to centrally guaranteed loans from
capital markets and the public sector. In addition, to foster an economic
base, central tax incentives were provided for economic activities to be
located in the homelands. Even with the fiscal transfers from the center,
the delivery of basic services was meager. This situation was in part due
to the weak administrative capacities of the bureaucracies but primarily a
result of the lack of political accountability of the homeland governments.
With no threat of electoral change, the appointed leaders of the homelands
had little incentive to serve their communities. Neither did the center
provide any incentive: dedicated to sustaining apartheid, the central
government remained willing to finance the growing fiscal deficits of the
homelands.
Local governments
Nowhere is the impact of the spatial policies of apartheid more visible than
in the cities of South Africa (Swilling, 1991). At first, influx control laws
attempted to regulate the growth of the white cities by controlling the
migration of blacks from the homelands. But with the need to ensure a
steady stream of workers for the urban economy, and ultimately with the
difficulty of regulating migration, the controls were replaced by residen-
tial laws. These permitted the blacks to remain in the white urban areas as
long as they were restricted to residential settlements at the urban fringe –
often one to two hours from the centers of employment. As a result of these
policies, the shape and pattern of urbanized land in South African cities
showed the following common characteristics:
242 Junaid K. Ahmad
• the physical separation of black and white cities

• an extreme dispersion and fragmentation of settlements, with high-
density low-income areas located at the urban fringe
• a concentration of jobs and services centered in the white areas contain-
ing most of the retail and service activities.
Eventually, the growth of the black population in the urban areas forced
the central government to give municipal status to the townships. By the
mid-1980s, the metropolitan areas were characterized by municipalities
divided along racial lines, called black and white local authorities (BLAs
and WLAs).
Financing of BLAs
Apartheid policies forced the concentration of lower-income households
in the townships and placed effective barriers to the growth of the formal
manufacturing and commercial sector in the BLAs. These regulations
deprived the black community of ownership of land and housing and
stopped the growth of a property tax system. In the long run, the lack of
property rights also prevented the black community from using land and
housing as a form of collateral to generate both a growing economic and a
taxable income base. User charges and rents from the delivery of services
and housing by the public sector were expected to be the main sources of
revenue for the BLAs. Rent and service charge boycotts were, however,
used as an effective tool to protest at the lack of representative government
structures, thus diminishing the already limited amount of revenue being
collected by the BLAs. Together, these factors ultimately resulted in
growing budget deficits. As in the case of the homelands, these were
financed by fiscal transfers from central government.
The dependence on upper-tier financing is well reflected in the data for
BLAs in the former Central Wits area, which was one of the richest regions
in the country: as table 8.2 shows, own revenues accounted for only a small
portion of total revenues. For central RSCs as a whole, own revenues of
BLAs in 1991–92 were approximately 5 percent of total revenues. The rest
were either central government transfers through provincial administra-
tions, or grants for current expenditures from RSC levies (payroll and turn-
over taxes).
The situation nationwide was similar, with grants from the central
government accounting for over 80 percent of the revenues of the BLAs.
Initially classified as bridging loans and intended for financing capital
expenditure in black communities, the flows were eventually converted by
the central government to explicit grants for financing current expendi-
South Africa 243
Table 8.2. Revenue structure of BLAs in Central Wits: 1991–92 budget (million
rand, percentages in parentheses)
Alexandra Diepmedow Dobsonville Soweto
Own revenue (0.3 (3.5 (0.9 (9.9

(0.86) (6.45) (8.11) (5.46)
Health grant (0.7 (0.6 (0.0 (10.1
(2.02) (1.10) (0.0) (5.57)
Provincial grant (16.8 (28.2 (7.4 (103.4
(48.41) (51.93) (66.67) (57.03)
RSC grant for O&M (16.9 (22.0 (2.8 (57.9
(48.70) (40.52) (25.23) (31.94)
Total ( 34.7 (54.3 (11.1 (181.3
(100) (100) (100) (100)
Table 8.3. Central Wits: contributions to net income

(including interest and grants) (percentages)
1990–91a 1991–92a 1992–93a
Property taxes 40.10 52.67 51.38

of which
Residential 12.10 15.80 15.40
Industrial/commercial 17.20 24.40 23.70
Government 10.70 12.50 12.30
Surpluses from services 23.22 11.19 13.81
Other 4.42 4.46 3.98
Interest 29.25 28.54 27.80
Grants 3.01 3.13 3.03
Total 100.00 100.00 100.00
Note: a Year ending June 30.
ture. In addition, as explained below, even the RSC revenues, intended to

finance capital expenditures in the black communities, were increasingly
diverted to finance the current deficits of BLAs.
Financing of WLAs
Relative to BLAs, the fiscal story of WLAs was one of sharp contrasts. As
the data on Central Wits RSC in table 8.3 suggest, grants from other tier
governments accounted, on average, for approximately 3 percent of the
total income of WLAs; the corresponding figure for BLAs was close to 95
percent. Property taxes, implicit taxes on the consumption of electricity,
244 Junaid K. Ahmad
Table 8.4. Shares of property taxes in total revenue:

actual and required (for electricity and water
surpluses to be generated from property taxes)
Actual share Required share
1989–90 Johannesburg 17.5 32.4

1989–90 Roodeport 22.9 27.6
1989–90 Roodeport 23.0 27.0
1989–90 Roodeport 22.9 28.3
water, and other services (or the surpluses on trading services), and inter-
est income provided the main sources of income for WLAs.
Several important features of the WLA finances should be noted. First,
the implicit taxes on consumption of services represented an important
source of income for WLAs. To ensure access to this source of revenue,
local authorities were given the exclusive right to distribute services such
as water and electricity. As a result, each white municipality had its own
distribution network. In a relatively integrated urban setting such as
Central Wits, for example, there were at least five separate local authority
distribution utilities for water and electricity, and nationwide there were
more than 700 at the local level. These distribution utilities purchased
electricity and water in bulk and distributed it to users within their munici-
pal boundaries. Consumers paid not only full user charges for capital,
operating, and maintenance costs but also an implicit excise tax on water
and electricity. Between 1990 and 1992 the tax on electricity in the various
WLAs in Central Wits averaged around 20 percent. The average for water
was 11 percent. There were large variations around these averages, with
Johannesburg, for example, charging 45 percent implicit tax rate on
electricity in 1991 and Randburg charging 22 percent for water.
The revenues from the high level of taxation of electricity and water
were used to cross-subsidize other services and dampen the increases in
property taxes. In fact, in the major WLAs, taxes on properties were set
residually after the municipalities had decided on the level of profits to be
generated from the distribution of services such as electricity and water.
For example, table 8.4 shows for Johannesburg and Roodeport the
required increases in the share of property taxes in total revenue if the com-
bined amount of electricity and water surpluses were to be generated from
taxes on property.
The cross-subsidy from trading accounts to property taxes, however,
South Africa 245
pointed to a second characteristic of the WLA finances: the role of cross-

subsidy between businesses and residents. Overall, residents in WLAs in
the Central Wits area received a rebate of the order of 35–60 percent of their
property tax payment, which was not available to businesses or commer-
cial property.
The final fiscal incidence and efficiency of this system of finance are not
obvious. Clearly, by zoning businesses within white communities, WLAs
have been able to cross-subsidize their residents by means of property tax
rebates. However, to the extent that businesses were able to pass on the
cost of these excess charges, especially in the case of non-tradables, the
cross-subsidy was proportionately less. On the other hand, given the
highly integrated nature of the labor market in the metropolitan areas, the
potential for exporting part of this tax burden from WLAs to BLAs may
have been high, thus further protecting the cross-subsidy to white resi-
dents. This latter issue has been one of the principal sources of social and
political tensions generated by the policies of apartheid. By zoning busi-
nesses within their municipal boundaries, WLAs forced members of the
black community to work and consume in white areas, thus indirectly
financing part of the cross-subsidy to white residents.
Regional service councils

To supplement the central transfers for BLAs, the national government in
1987 also authorized the creation of RSCs/JSBs (Joint Services Boards).
Given the authority to raise payroll and turnover taxes over a larger met-
ropolitan area comprised of several WLAs and BLAs, the RSCs had the
authority to finance, on a project-by-project basis, capital investments of a
regional nature and also infrastructure in the black townships. Even
though the mandate was to finance capital investments, the RSC budgets
were progressively diverted to fund the current account deficits of BLAs –
thus reducing the need for central government to continually increase
fiscal transfers. For example, in the 1992–93 budget, the Central RSC allo-
cated approximately 40 percent of its revenues to finance the current
expenditures of BLAs in its jurisdictions. The remainder was used to
finance capital investment in the townships.
By creating a fiscal arrangement over a group of WLAs and BLAs, RSCs
formed a geographic boundary that had the potential of capturing some of
the spillovers resulting from the fragmented nature of the conurbations.
For example, by levying turnover and payroll taxes within the boundaries
of WLAs and financing investments in the BLAs, the RSCs returned to the
black community some of the share being captured by WLAs.
In sum, the policies of apartheid left a legacy of centralized government
246 Junaid K. Ahmad
structures in South Africa. The centralization was a result, first, of spatial

policies which forced the black community to be clustered in economically
weak regions. Only large fiscal transfers could sustain these “homelands”
and townships. Second, it was a direct result of the racial policies of
segregation. It may be argued that given the economic consequences of
apartheid, decentralized financing of racial policies would have created,
over time, strong incentives for subnational governments to opt out of the
racial system. Strong central control was therefore necessary to enforce a
common system of segregation and distribute its costs across the regions.
Creating new provinces

The January 1994 interim constitution radically altered the fiscal land-
scape. The homelands were abolished, their territories reintegrated into
South Africa, and nine new provinces were created. The provincial assign-
ment of expenditure and taxation powers and the weak fiscal base of most
new provinces have left them with significant vertical fiscal imbalances
and horizontal fiscal disparities. Most provinces are, therefore, dependent
on large central grants, creating potentially a significant mismatch
between political and fiscal autonomy. The Financial and Fiscal
Commission was established by the Constitution as an independent body
to address, in effect, this tension. Its political and legislative ability to
address the problem of disparities remains to be seen.
Elimination of racial jurisdictions

The elimination of racial jurisdictions – SGT/TBVC states at the regional
level and BLAs at the urban level – has been central to the creation of a non-
racial government in South Africa. It signaled an end to the spatial policies
of apartheid and provided the geographical basis for holding elections. But
more fundamentally, the new boundaries have shaped the fiscal capacities
of the emerging jurisdictions and are influencing the evolution of the inter-
governmental fiscal system.
The debate on drawing the boundaries of the new provinces has been
influenced by several factors. Proponents of decentralization, and certainly
those who feared the emergence of a Leviathan in an ANC-dominated
central government, strongly advocated creating provinces of equal fiscal
capacity. The proposed boundaries made little economic or political sense
– giving rise to potential spillovers, where integrated urban agglomerations
faced the threat of being divided, or possible community clashes, where the
new jurisdictions ignored the historical and cultural roots of the different
South Africa 247
ethnic groups. In the final analysis, the idea of equalizing by reshaping

boundaries rather than using fiscal transfers was rejected.
The level of urbanization in South Africa was also raised as a funda-
mental factor in drawing the provincial boundaries. Unlike the rest of
Africa and more similar to the middle-income countries of Latin America,
South Africa is a highly urbanized economy. Indeed, four metropolitan
areas account for approximately 60 percent of the GDP and about 50
percent of the population.5 In this context, it was argued that large
provinces made little economic sense and proposals were put forward for
the establishment of several – seventeen to twenty – smaller “regions,”
thus giving prominence to the urban nature of the country. It was also
argued that if large provinces were to be established, the metropolitan
areas should receive regional status.
The new interim constitution finally divided the country into nine
provinces. The metropolitan areas were not given regional status and,
according to the new constitution, are very much under the jurisdiction of
the new provinces.6 For the most part, the boundaries between provinces
are based on the lines of the “development regions” established in the
National Development Plan. These lines primarily reflect market forces in
the private provision of goods and services. The provision of public ser-
vices (the geographic scope of benefits, spillovers, and economies of scale)
was rarely considered. Even so, these boundaries created fewer problems
than those found in many federations, which often follow rivers, includ-
ing navigable ones straddled by large metropolitan areas – a sure recipe for
spillovers of costs and benefits.
Table 8.5 shows the correspondence between the nine new provinces
and the fourteen former administrative entities (the four administrative
provinces, the four TBVC homelands, and the six SGT homelands).
Overall, nine new provincial capitals, nine new parliaments, and nine new
executives (“premiers”) were chosen, and nine new provincial adminis-
trative bodies were created. The problem, however, was not only one of
subdivision of the four existing provinces into nine. Since the homelands
had to be reabsorbed, their governments were eliminated and absorbed
into the new provincial governments (in the case of civil servants). In each
of these cases it was necessary to divide the fiscal affairs of the previous
administrative entities among the new provinces. Only in KwaZulu/Natal
did the mergers entail combining only two of the previous entities (Natal
and KwaZulu), both of which fell entirely within the new province. Even
now, several years after the creation of the new provinces, the practical
problems of the merger of the previous jurisdictions continue to challenge
the newly elected leaders.
248 Junaid K. Ahmad
Table 8.5. Jurisdictional changes
New province Former administrative entity
Gauteng Transvaal
Northern Province Transvaal
Venda
Gazankulu
Lebowa
Mpumalanga Transvaal
Bophuthatswana
KaNgwane
KwaNdebele
North West Cape
Transvaal
Bophuthatswana
Free State Orange Free State
Bophuthatswana
QwaQwa
KwaZulu/Natal Natal
KwaZulu
Eastern Cape Cape
Ciskei
Transkei
Northern Cape Cape
Western Cape Cape
Source: Development Bank of South Africa, 1995,

table 22, p. 88.
Expenditure assignment
In addition to the creation of a new regional tier with new political and
administrative units, the restructuring of the intergovernmental system
has led to a significant decentralization of expenditure assignments. In
the previous system, the provinces were, in effect, administrative arms of
the center and responsible for implementing the mandates of the center.
In 1993–94 provincial governments accounted for approximately 22
percent of total public expenditure. By 1995–96, expenditure responsibil-
ities of the provinces had increased to approximately 40 percent of all
public expenditures and included major items such as health, education,
and housing.
Tax assignment
Under the new constitution, the provinces can impose taxes, other than
income tax, value-added tax (VAT), and other sales taxes, and they can
South Africa 249
impose surcharges. But exercise of this authority requires an Act of

Parliament passed after consideration of recommendations by the
Financial and Fiscal Commission. The provinces can share in revenues col-
lected nationally from the individual income tax, the VAT, and the fuel tax.
Experience from other countries suggests that the prohibition of pro-
vincial income taxes, VAT, and other sales taxes is appropriate. Unless
implemented as surcharges, provincial income taxes would entail costly
duplication of administration and compliance procedures, and perhaps
undesirable complexity and economic distortions. Brazilian experience
shows that a provincial VAT is difficult to implement without investment
in a significant administrative apparatus (McLure, 1994). Since the central
government levies a VAT, provincial sales tax could impose unacceptable
burdens of compliance and administration. As a result, provincial sur-
charges on the income tax base of the central government provide the only
acceptable source of substantial amounts of own revenue for the provinces
(provincial surcharge on excises was, technically, another possibility, but
was barred under the new constitution). Such charges should be adminis-
tered by the central fiscal authorities; in the case of the company income
tax, uniform rules could be used to apportion the income of firms oper-
ating in more than one province. Surcharges would provide some
provinces with substantial fiscal autonomy. By choosing higher or lower
surcharge rates, they could have larger or smaller budgets and respond
better to the preferences of the citizenry.
Government has, however, decided not to devolve any major tax instru-
ments to the new provinces. As a result, in the new South Africa, while the
provinces have achieved political autonomy and the new political and
administrative leaderships have inherited increased expenditure
responsibilities, their revenue autonomy remains curtailed. In the past,
provinces’ own revenues were approximately 5 percent of the revenues
collected by all tiers of government. In the new intergovernmental system,
they are approximately 2 percent. Central government thus continues to
have access to the rate and base setting authority of the three main taxes –
personal income tax, corporate income tax and VAT – leaving provinces to
rely on grants from central government to overcome vertical fiscal imbal-
ances. In this respect, the fiscal structure of South Africa resembles the
highly centralized structure of Australia, rather than the relatively
decentralized systems found in Canada and the United States.
Horizontal inequality
No matter which taxes are assigned to provincial governments, there are
going to be substantial differences in the abilities of the various provinces
250 Junaid K. Ahmad
to finance public services from their own tax revenues. This is an inevitable
result of the economic disparities among the provinces and the concentra-
tion of economic activities in a handful of metropolitan areas. Whereas the
estimated GDP per capita for the country in 1991 was R5,042, it was as
much as R10,949 for Gauteng region, but only R2,317 in Eastern Cape and
R1,266 in Northern Province. The national average for personal income per
capita for the same year was R2,566, compared to R4,993 for Gauteng,
R1,358 in Eastern Cape, and R725 in Northern Province. These figures
suggest that in addition to the need for grants to overcome the vertical
imbalance, equalization transfers will be needed to reduce the fiscal dis-
parities that currently exist.
The Financial and Fiscal Commission

The Constitution has established the Financial and Fiscal Commission
(FFC) and entrusted it with a variety of powers related to intergovern-
mental fiscal relations. In a preliminary draft of its Framework Document
(1995), the FFC defined itself as
an independent and impartial statutory institution, accountable to the legisla-
ture, with the objective of contributing towards the creation and maintenance of
an effective, equitable, and sustainable system of intergovernmental fiscal rela-
tions, rendering advice to legislatures regarding any financial and fiscal matter
which has a bearing on intergovernmental fiscal relations.
The Constitution requires the national legislature to consider the
recommendations of the FFC on a variety of issues, including
• the sharing of revenues with the provinces
• taxes, surcharges, and user charges to be imposed by provincial govern-
ments
• borrowing by provincial and local governments
• guarantees of loans to the provinces.
In addition, the FFC is supposed to make recommendations on taxes
proposed by local governments, the sharing of provincial revenues with
local governments, and provincial guarantees of loans to local govern-
ments. Thus, in principle, the FFC could be an extremely important part of
the system of fiscal relationships in South Africa. Yet, it is an appointed
body, with no independent power base. Its eventual role remains to be
seen.
Under the Constitution, each province is entitled to “an equitable share
of revenue collected nationally to enable it to provide services and to exer-
cise and perform its powers and functions.” The FFC has articulated three
South Africa 251
objectives of intergovernmental grants: efficient resource allocations; fiscal

equity; and development of fiscally sound and democratically responsive
provincial government. It proposes using an objective and relatively
simple grant formula to achieve this, explicitly rejecting the use of ad hoc
grants. The grant system explicitly addresses the issues of vertical and
horizontal fiscal balance.
Vertical balance
Reflecting the constitutional allocation of functions of government and
its beliefs that the most pressing service deficits are at the subnational
level and that most activities of the central government do not grow with
population, the FFC in 1996 proposed a freeze in real spending of the
central government for the following three years; thus all public rev-
enues made available by growth (conservatively estimated to increase
from 2.5 percent per year to 3 percent over five years), except those
needed to reduce the budget deficits from 5.8 percent to 4.5 percent of
GDP, should be available to the provinces via grants. Under the FFC’s
proposals, the central government’s share of financial resources would
drop from 56.5 percent in the fiscal year 1995–96 to 53.1 percent in
2000–01.
Horizontal balance
The FFC has proposed a three-part grant formula. The unconditional
“basic grant” is intended to provide for each province “the fiscal resources
necessary to function as an efficient and accountable government for its
people.” This part is based on population figures; in recognition of the
special problems of the rural areas (substantially greater poverty levels,
higher unemployment, and the greater cost of providing services) each
rural person has been counted as equal to 1.25 urban persons.
The second part is intended to allow provision of a “national standard”
level of public services. This part comprises grants for primary and sec-
ondary education (a given amount per student aged 5–17) and primary
health care (a given amount per qualifying member of the population).
The third part of the formula focuses on “fiscal capacity equalization” of
the provinces. FFC has proposed a surcharge rate on the personal income
tax base as a source of own revenues for the provinces. On the basis of this
tax instrument the FFC has proposed an equalization formula. The
formula equalizes around the national average level of tax capacity and
ensures that the equalization component is self-financing without impos-
ing any additional claim by the provincial governments on the private
economy.
252 Junaid K. Ahmad
Funding
The FFC proposes to fund grants from a pool of revenues from the indi-
vidual income tax, the VAT, and the fuel tax, instead of having different
sharing rates for each. Besides being simpler, this avoids the risk of the
central fiscal authorities concentrating resources on the taxes with the
lowest sharing rates. Funding of “national standards” grants would be
given first priority, with the pool of funding available for “basic grants”
determined as a residual.
Phase-in
This grant structure would result in reductions in shares of grant funds (in
real grants in two cases) going to four of the provinces (Western Cape,
Eastern Cape, Northern Cape, and Free State) and increases in shares
going to the other five. To allow for orderly retrenchment and efficient
expansion, the FFC proposes to phase in its formula over a five-year
period, with floors (a fall of 3–4 percent) and ceilings (a rise of 5–6 percent)
on annual changes in real grants.
Structure of urban governance

As new provinces were being created, a similar process was unfolding
at the urban level, with the creation of new metropolitan governments.
In deciding the organization of the metropolitan systems, the urban
political leadership faced several options. One was to maintain the
status quo of jurisdictional fragmentation,7 or a structure of urban
governance under which the responsibility for the same functions lies
with many local governments operating in the area. In the case of South
Africa, the proposal was to merge adjacent BLAs and WLAs into new
municipalities – a twinning of sorts – and allow several of these “twins”
to operate as independent municipalities. A second option was to create
a centralized metropolitan government with the BLAs and WLAs in
each urban area being merged into a single local government. A third
option, one that was not actively debated, was the model of functional
fragmentation, under which the provision of services is area-wide and
is shared between general-purpose local government and autonomous
agencies. Applied to South Africa, this model implied that service util-
ities could serve several municipalities (the “twins”) or a metropolitan
government.
The process of urban restructuring is continuing. The optimal structure
of urban governance in South Africa will eventually depend on how the
political process in South Africa ranks the objectives of fiscal stability,
South Africa 253
income distribution, and efficiency as the dominating concerns of public

policy. In practice, the political process has given equal importance to each
of these factors and the emerging structure of the urban governance in the
metropolitan areas reflects these tensions.8 This section discusses some of
the criteria that are influencing the structure of urban governance and uses
the examples of Johannesburg and Cape Town to describe the process of
change at the urban level.
The organizational structure of metropolitan areas in South Africa will
be largely influenced by political considerations, different dimensions of
efficiency in the delivery and financing of services, distributional objec-
tives, and the concern of central government that financing of local ser-
vices does not undermine the macroeconomic stability of the economy
(Ahmad, 1997).
Political considerations
Abolishing the racial basis of the local government system is a funda-
mental political consideration in the reform of urban governance in South
Africa. As in the case of the provinces, a redrawing of jurisdictional bound-
aries and a merger of BLAs and WLAs either into a metropolitan structure
or as “twins” in a model of jurisdictional fragmentation is, therefore,
inevitable.
Political considerations also include the creation of a system of local
government through which local council members can be held account-
able and responsible for their decisions. The primary instrument to achieve
this objective is the electoral system. However, efficient markets, especially
for land and capital, can complement the influence of the electoral process
in holding in check the behavior of local officials. Similarly, accountability
is enhanced through access to direct fiscal instruments by local authorities.
Officials can be held responsible for their actions if, at the margin, decisions
to alter the level of services are reflected in increases or decreases in tax
rates directly under their control.
In considering the political dimensions of structuring urban gover-
nance, therefore, policy-makers have the difficult task of choosing an
organization for financing and delivering services that is best able to
strengthen the political institutions of democracy while enabling markets
and fiscal instruments to hold local officials accountable. How this inter-
play between political institutions, markets, and fiscal instruments evolves
will certainly influence the choice between jurisdictional and functional
fragmentation and a centralized metropolitan form of government for the
cities of South Africa.
254 Junaid K. Ahmad
Efficiency considerations
Efficiency can be defined along several dimensions. It includes the extent
to which the preferences of the constituencies can be reflected in the local
budgets and also the extent to which technical efficiency can be achieved
by taking advantage of the economies of scale inherent in the delivery of
services.
In the case of South African cities, locational efficiency is an equally
important issue. With the end of apartheid’s zoning policies, the distribu-
tion of population and business activity over the urban areas is in a state
of flux. Already, in several of the major metropolitan areas, there has been
a decentralization of employment from the core cities into the surround-
ing white municipalities. In addition, in some cities, there has been an
outflow of white residents from the city center and an inflow of black resi-
dents from adjacent townships and regions. In this context, it will be
crucial that any strategy for financing urban infrastructure does not distort
the locational choices facing individuals and firms. In the case of South
Africa’s metropolitan areas, drawing of narrow municipal boundaries
may result in unequal fiscal burdens in the financing of urban services. The
potential for distorting locational decisions would therefore be high.
The different models of urban governance involve tradeoffs over the
various dimensions of efficiency. Consumer preferences are best reflected
in homogeneous and small local government structures but could lead to
overall losses in consumer welfare if fragmented local authorities attempt
to deliver and finance services whose benefits and costs are area-wide. On
the other hand, locational distortions are minimized and technical effi-
ciency is realized under a metropolitan structure.
Income redistribution
The mobility of factors at the local level is an important consideration in
limiting the responsibility for income redistribution to upper-tier govern-
ments. Cognizant of this important constraint on implementing redistrib-
utive policies at the local level, South Africa has begun to manage the issue
of income redistribution through the intergovernmental system by coor-
dinating redistributive policies among different levels of government.
Urban governments are therefore expected to play a part in this process.
In particular, to avoid the locational distortions that would be associated
with redistributive measures, a more consolidated urban governmental
structure such as a metropolitan system is being advocated. A metro-wide
structure would not only minimize the locational impacts, but could also
South Africa 255
achieve efficiency objectives by accommodating wide variations in levels

of urban service demand without imposing a uniform level of service pro-
vision. In other words, instead of completely passing on the responsibility
for redistribution to the central government, centralization at the local
level through some form of metropolitan governance structure provides
local authorities with the flexibility to implement redistributive measures
without significantly reducing efficiency.
Fiscal stability
Access to significant long-term finance will be required to improve the
standard of municipal services for the black community. However, bor-
rowing by city governments, indeed by any large local government, raises
the potential problem of moral hazard. The sheer economic size of the met-
ropolitan areas of South Africa suggests that central government would
face considerable political and financial resistance to allowing such cities
to default and face bankruptcy in a situation of financial duress. South
Africa’s cities may be “too big to be allowed to fail.” In this context, capital
markets would perceive the debt of urban governments in South Africa to
be an implicit obligation of central government and, consequently, instead
of acting as a disciplining device may, in fact, be “too generous” in financ-
ing their investments.
Central government in South Africa is debating alternative mechanisms
for addressing this policy problem of enabling access to capital markets by
local governments while avoiding the pitfalls of moral hazard. Two differ-
ent mechanisms are being assessed. First, government is exploring the
possibility of legislating clear rules of “bankruptcy” for local government.
International examples being discussed include the US model of a finan-
cial control board and the New Zealand model of court-appointed receiv-
ership. Under both, a worsening fiscal situation of a local government is
addressed by the imposition of a control process that takes away mana-
gerial autonomy from the elected officials and passes it to an independent
board. A buffer is thus created between different tiers of government. To
be credible the process will need to impose the cost of “bankruptcy” on
both the elected officials (for example, loss of autonomy) and creditors
(uncertainty about payment).
Second, government is looking at the possibility of centralizing the bor-
rowing powers of local government whereby the center would either
borrow directly and on-lend to the local governments or allow a public
financial intermediary (PFI) to perform this on-lending operation. For
several reasons, however, centralization of borrowing powers may fail to
256 Junaid K. Ahmad
provide macroeconomic control over the fiscal affairs of government as a

whole (Ahmad, 1997). International experience from several countries –
and indeed the past history of South Africa – suggests that the allocation
of credit through public institutions inevitably gets embroiled in a political
process. Capital does not necessarily flow to the most productive but often
to those who are politically the most astute. In such cases, government bor-
rowing may be inefficient and the subsequent investments generally
unproductive. This has the potential of contributing to a gradual loss of
fiscal control as more and more borrowing is required to deliver a
minimum level of public services. In addition, the institutional incentives
of PFIs are such that they may not be immune to the moral hazard syn-
drome. In fact, more than in the case of subnational governments, which
generally have access to tax bases and whose politicians have to face an
electoral process and could, therefore, face the consequences of their
myopic decisions, PFI employees are seen as responsible to central author-
ities and their debt perceived as an off-budget obligation of central govern-
ment. In fact, by removing the direct relationship between capital markets
and subnational governments, PFIs can undermine a potential instrument
for ensuring the efficiency and accountability of local government.
An alternative, but also a complement, to the process of control boards
described above is to structure urban governance in such a way as to mini-
mize the problem of moral hazard. A possible solution would be to priva-
tize public goods, or alternatively, as Baumol and Lee (1991) suggest, create
“contestable markets” for the participation of the private sector in the
financing and delivery of public goods. Simply put, changes in technology
worldwide now permit the privatization of the delivery and financing of
what are still considered public goods in many countries including South
Africa. These include, for example, distribution of water and electricity at
the retail level, transportation services, and garbage disposal. By placing
these in the private domain, governments may avoid the problem of moral
hazard: bankruptcy of these entities would enable other private sector
parties to bid for their assets. This lack of competitive pressure and threat
of takeover in the public delivery of municipal services leaves open the
issue of the implicit financial obligation of central government and the
assumption that one tier of government under financial duress would be
supported by another tier.
The privatization route has the additional advantage of enabling service
deliverers to access financial equity as well as borrow in the capital
markets. Because equity-holders stand to lose in the case of bankruptcy,
the participation of financial equity provides an incentive internal to the
structure of the service provider to monitor debt, creating another layer of
South Africa 257
protection against moral hazard that is not present in a public system of

delivery. In addition, the availability of private sector bond insurance pro-
vides an added mechanism for ensuring that central governments do not
take on the role of “banker of last resort.” In the event of an issuer default,
the bond insurers guarantee the timely payment of interest and principal
in accordance with the issuer’s original payment schedule. The insurer
will often work with the issuer to address the financial problems and,
therefore, minimize its own losses. Generally, the guarantee is irrevocable
and the bonds are guaranteed for their life regardless of what may happen
to the issuer.
The model of private sector delivery and the availability of bond insur-
ance agencies suggest that devolving financial and delivery responsibil-
ities of governments, if possible, onto the private sector is an important
component of ensuring macroeconomic stability. This implies that choos-
ing the functional fragmentation model of metropolitan governance, with
the area-wide delivery of services in the private sector, is an alternative tool
for achieving the fiscal stability objective of central government.
Johannesburg and Cape Town: two examples

The metropolitan areas of Johannesburg and Cape Town epitomize the
changes occurring at the city level in local government’s fiscal affairs. The
situation in both cities suggests that a hybrid structure of urban gover-
nance is a model that may be applicable for the large metropolitan areas.
But different political processes in the two cities may lead to very different
models being finally implemented.
A metropolitan government has been created in Johannesburg and Cape
Town. Political considerations of ensuring a non-racial structure and the
recognition that urban governments will have responsibility for imple-
menting some measure of income redistribution were two important
justifications for the creation of a metropolitan government. The merger of
adjacent BLAs and WLAs – the twinning option – would have only par-
tially fulfilled the political objectives of creating a non-racial local govern-
ment structure. In all of the major urban centers, some fiscally strong
WLAs were physically not close enough to any BLAs to justify a merger.
In addition, in many cases there would be significant fiscal disparities
between the newly merged BLAs and WLAs. A larger metropolitan
boundary was necessary to avoid the potential locational distortions that
might emerge from local financing of services within narrow administra-
tive boundaries.
Fiscal autonomy and accountability were also important in the move
258 Junaid K. Ahmad
towards a metropolitan structure. Intergovernmental transfers could, of

course, be used to compensate for the fiscal disparities between the differ-
ent “twinned structures” and support a more fragmented local govern-
ment structure – the jurisdictional fragmentation model of urban
governance. There was sufficient evidence from the past, however, to
suggest that reliance on intergovernmental transfers might reduce the
fiscal autonomy and accountability of local governments – two important
elements in ensuring the benefits of fiscal decentralization. In addition,
there was significant concern about the predictability of such transfers and
the uncertainty it would create in local government planning. Finally,
political leadership at the central government level was convinced that the
fiscal base of the urban sector would be sufficient to support the expansion
of municipal services into the black community while freeing central
government fiscal resources to fund national public goods. Such horizon-
tal cross-subsidy between municipalities could take place within a metro-
politan system.
While a metropolitan tier was perceived as necessary for achieving fiscal
autonomy and accountability of local government vis-à-vis the central
government, it was nevertheless viewed as too far removed from the com-
munities. To further enhance the accountability of officials, bring govern-
ment closer to the constituencies and, therefore, improve economic
efficiency, Johannesburg and Cape Town adopted a two-tier metropolitan
structure: a first, metropolitan tier overseeing a group of “twinned”
municipalities in the second tier. In addition, a second tier provides a
political check on attempts by the metropolitan tier to engage in extensive
redistribution. It offers, therefore, a level of protection for the former WLAs
and may have been an important necessary concession to the white com-
munity to allow the formation of a metropolitan tier in the first place.
While the two-tier model was initially politically accepted and was in
the process of being implemented, the issue of the allocation of fiscal
responsibilities – expenditure and tax assignments – remains unresolved,
and is still in a state of flux. If implemented, privatization of municipal ser-
vices will significantly influence the assignment of expenditure between
the metro and the municipal tier and, subsequently, the assignment of
taxes.
Expenditure responsibilities
It was suggested that concerns about fiscal stability and operational effi-
ciency favored the privatization of municipal services. The extent of the
economies of scale in the delivery process will be an important determi-
South Africa 259
nant of whether the municipal service should be privatized at the metro-

politan or municipal level and whether there will be one or multiple pro-
viders in an urban jurisdiction. Where economies of scale do not exist (for
example, in garbage removal) several private providers could operate in
one metropolitan area, each serving a different municipality. In either case,
private providers facing competitive pressures would have the incentive
to respond to the willingness-to-pay of the customers and tailor the stan-
dard of delivery to demand. For local public goods, on the other hand, eco-
nomic efficiency suggests that the division of labor between the
metropolitan tier and the municipalities would depend on which jurisdic-
tion benefits from the expenditures. Street lighting, for example, could well
be a municipal responsibility. Roads, to take another example, with met-
ropolitan-wide spillovers in costs and benefits would be better financed by
the metro tier.
If privatization of municipal services is accepted as part of the reform
process, a critical element of urban governance will be the regulatory func-
tion of government. This function is needed to ensure, in particular, that a
private monopoly does not emerge in delivery of services, especially
where direct competition in the market is not possible. The general issue
of whether regulatory functions should be central or local or whether they
are a concurrent responsibility of different tiers of government is beyond
the scope of this chapter. However, given that most of the municipal ser-
vices should be privatized at the metropolitan level, it is logical that the
metro tier should have the regulatory responsibility of overseeing the per-
formance of the privatized entities and managing the privatization
process.
In addition to the role of regulator, a fundamental function of the metro
tier would be in the area of income redistribution. Given the fiscal dispar-
ities between the different municipalities, the metro tier could have the
responsibility for equalizing their fiscal capacities.9 However, if the model
of functional fragmentation is accepted and municipal services are deliv-
ered through the private sector, an additional component of the metro’s
distributional responsibility could be to target fiscal transfers directly to
low-income households, for example, through vouchers, to ensure access
to a basic minimum level of services.10 Because this type of fiscal assistance
would be to individual households, access to a minimum level of service
within a metropolitan area would not be location-specific. Locational deci-
sions of households would therefore be undertaken on the basis of eco-
nomic costs and benefits rather than fiscal considerations. In addition, by
allocating the responsibility for distributional objectives directly to the
public sector, the service deliverer would have the flexibility of focusing
260 Junaid K. Ahmad
on operational efficiency and pricing services according to economic cri-

teria. This separation of distributional and efficiency objectives in the
pricing and delivery of municipal services would also promote locational
efficiency.
Tax assignments
The discussion on tax assignment takes as its starting point the expendi-
ture assignments discussed above and that, in the short to medium run, the
tax instruments originally assigned to the RSCs, WLAs, and BLAs will
become available to the new urban governments.11
Assigning the payroll and turnover taxes

Zoning policies separated the residences of the blacks from their places of
employment and purchase of consumption goods. In addition, for admin-
istrative convenience, payroll taxes were collected at the source of employ-
ment and turnover taxes in the jurisdiction in which the taxed transactions
occurred. In both cases, this was primarily in the WLAs. To the extent that
each metropolitan jurisdiction will include both BLAs and WLAs and,
therefore, the residences and the places of employment and consumption
of the black community, it would be more efficient and equitable to assign
the turnover and payroll taxes to the metro tier.12
In addition, while the incidence of these taxes is on the residents of both
the former WLAs and BLAs, most of the revenues will be used to fund
expenditures for low-income black households. Given its potential redis-
tributive function, the assignment of the payroll and turnover taxes to the
metro tier is, therefore, appropriate. At the same time, given that low-
income communities are part of these tax bases, there would be an incen-
tive for metropolitan authorities not to be excessive in pursuing their
redistributive objectives.
Central fiscal transfers

Central transfers have traditionally been used for redistributional pur-
poses. It is expected that in the new fiscal system the transfers will continue
to be directed to achieve distributional objectives. The metro tier would,
therefore, be the appropriate tier at the local level to manage these trans-
fers.
Property taxes
Property taxes are a mainstay of finance in urban areas and an efficient tax
instrument for financing local public goods. In addition, autonomy in
South Africa 261
setting property rates can help establish the accountability of local officials.
Residential property taxes should, therefore, remain at the level of the
municipalities to ensure that local officials can respond to the preferences
of their constituencies.
However, there is an important redistributive issue regarding the use of
commercial property taxes. As in the case of payroll and turnover taxes,
the issue arises from the location of residences and businesses. As dis-
cussed earlier, apartheid policies zoned economic activities in WLAs and
provided these jurisdictions with exclusive access to commercial property
taxes. Given that economic activities will remain “locked-in” in the former
WLAs for some time to come, a case can be made on redistributive grounds
for the sharing of revenues from commercial property between the
municipalities and the metro tier.
Inman (1996) also provides an argument based on fiscal efficiency for
transferring the non-residential property tax to the metropolitan tier. Based
on the experience of several metropolitan areas in North America, Inman
suggests the possibility of municipalities using the commercial property tax
base to indulge in a “beggar-thy-neighbor” type of tax competition. To
avoid this outcome, Inman suggests a Minneapolis-St. Paul model of
sharing the commercial tax base at the metropolitan level. The objective
would be to levy a common commercial property tax at the metropolitan
level, pool the revenues, and transfer it back to the municipalities on a
formula basis (e.g. capacity equalization or poverty targeting).
Surcharges on municipal services

If current municipal services are privatized, local authorities could pre-
serve their access to revenues from water and electricity by converting the
implicit surcharges into explicit excise taxes on utility services. To mini-
mize locational effects and tax avoidance and given that the services will
be delivered at the metropolitan level, such taxes should be assessed and
collected on a uniform basis at the metropolitan level rather than at the
municipal level. If fiscal transfers to low-income households are financed
by explicit taxes on these services, it may be administratively easier to have
the service providers act as the administrators of taxes and subsidies.
Alternatively, the service provider could provide cross-subsidies between
households at different income levels through the pricing of the services –
a less preferable option given the potential for distorting prices and, once
introduced, the tendency of the public sector to “tax” private entities
through greater cross-subsidies.
In sum, in a two-tier metropolitan system it would be efficient and equi-
table to assign central fiscal transfers, payroll and turnover taxes, and
262 Junaid K. Ahmad
excise taxes to the metropolitan tier. Residential property taxes would

remain at the municipal level while taxes on commercial property could be
shared between the two tiers. This tax assignment is based on the alloca-
tion of expenditure suggested earlier which included giving the metro tier
the primary role in income redistribution while privatizing the basic
municipal services. The assignment is also based on minimizing potential
locational inefficiencies and providing a check on the extent to which met-
ropolitan governments can undertake redistribution of income.
The structure of urban governance in the major metropolitan areas of

South Africa is undergoing a dramatic transformation. The model of separ-
ate municipalities has given rise to a hybrid metropolitan system in
Johannesburg and Cape Town. The abolition of the racial basis of the cities,
distributional concerns, and the legacy of apartheid’s spatial polices were
major determinants of the choice of a metropolitan system. Establishing
greater fiscal autonomy – for local authorities from central government
and for municipalities from the metropolitan tier – was an important
consideration in favoring a two-tier structure.
However, the two-tier system is being questioned. In the case of
Johannesburg, metropolitan government has used an inefficient and ad hoc
rule for implementing interjurisdictional equalization. It has imposed a
uniform property tax rate across the metropolitan area and shifted any
resulting fiscal surplus from one jurisdiction to another. Not only has this
resulted in sharp increases in property taxes in many jurisdictions – of the
order of 100–200 percent – but shifts in fiscal flows have also subsidized
the deficits of municipalities whose credit control is weak. In addition,
municipalities with surpluses now have an incentive to ensure their
elimination. In the case of Cape Town, the RSC revenues have been used
to expand the metropolitan bureaucracy and provide ex post budget relief
for the municipalities. As a result, in both cities, the ad hoc fiscal system has
produced inefficient redistribution and political fighting between
municipalities over the use of funds. As a consequence, delivery of local
goods to the black community has fallen far below expectation.
In the face of this crisis, central government has adopted the position of
favoring the model of a centralized metropolitan government – the “mega-
city model,” as it has been named in South Africa – with municipalities
being converted into administrative wards. The assumption is that a
centralized model will facilitate more rational systems of redistribution and
accelerate delivery of services. In the coming weeks and months, the politi-
cal debate will decide whether the mega-city model will be implemented
nationwide or whether cities will be given the local autonomy to go their
South Africa 263
own route. At this stage, Johannesburg seems to be moving toward a mega-

city model. Cape Town, on the other hand, is debating the possibility of a
two-tiered structure integrated through a fiscal system as described above.
Unresolved tensions: the evolution of intergovernmental fiscal

relations
Four important issues are shaping the evolution of the intergovernmental
system in South Africa: the extent of fiscal autonomy and accountability of
provinces and cities, the level of capacity at different levels of government
to manage decentralized responsibilities, the role of financial markets, and
the relationship between provinces and the metropolitan areas. How these
issues will be addressed will determine the extent of decentralization of
the governmental tiers and the relative role provinces and metropolitan
areas will play in the economic affairs of the country.
Accountability, fiscal control, and cooperative governance

A major tension that is emerging in the fiscal affairs of the different tiers of
government is between accountability and fiscal stability. One perspective,
perhaps that of central ministries, is that fiscal stability and control is best
achieved, especially during a period of transition, by having central
control of tax instruments and by limiting the borrowing powers of subna-
tional governments. This approach assumes that the political process is in
its infancy and markets sufficiently imperfect that control from the top is
necessary to maintain fiscal discipline.
Another perspective, perhaps that of the Financial and Fiscal
Commission, is that political accountability and fiscal control may be
achieved simultaneously by providing subnational governments with
own revenues and direct access to capital markets. In this model, the threat
of having to raise own taxes and face capital market prices for wrong
policy decisions is assumed to be more effective in limiting the potentially
myopic behavior of politicians than the alternative of central controls. The
latter would have to be through administrative and bureaucratic budget-
ing rules and monitoring systems for the flow of funds. The decentralized
model of accountability and fiscal control would also suggest that there are
alternative approaches to managing the risks of decentralization (for
example, legislating a bankruptcy process for the public sector and man-
aging implicit financial obligations explicitly) which do not require full
centralization of fiscal powers. In fact, the current fiscal crisis of
Johannesburg and some of the provinces and the center’s insistence on no
264 Junaid K. Ahmad
“bail-outs” will test whether the center will fall prey to the pressures of
moral hazard or establish the framework for subnational fiscal discipline.13
South Africa has chosen, for the moment, not to provide provincial
governments with a tax base, relying instead on central grants to fund pro-
vincial expenditures. It remains to be seen if this system will enable the
twin objectives of political accountability and fiscal macro control to be
achieved. It may be hypothesized that without the threat of having to use
own taxes, provincial leadership may not have the incentive to spend effi-
ciently. The potential political game that may be played is one of a pro-
vincial leadership’s ability to shift the responsibility for a lack of delivery
onto central authority in the hope of inducing additional funding ex post.
In other words, the problem of moral hazard also exists on the fiscal side
of funding governments.
Grant funding also raises the problem of explicit and implicit condition-
ality and unfunded mandates. Currently, the public sector wage structure
and sectoral conditionality on spending (percentages for health and educa-
tion, in particular) are set from above, leaving provinces with limited
expenditure flexibility. In addition, with central ministries losing expendi-
ture responsibilities but maintaining the ability to set national standards,
there is a possibility that provinces may be saddled with unfunded man-
dates or the exact opposite may happen as central government policies end
up being ignored. In combination with the lack of fiscal instruments, the
conditionality of the grants may be creating a situation where the new
provinces are, like their predecessors, deconcentrated arms of the center.
Cognizant of the fact that elected political bodies will have difficulty in
accepting administrative roles, there is a move towards a concept of “co-
operative governance.” This concept is not well defined but entails a joint
approach, or cooperation, between provincial and central government in
budgeting and delivery implementation.14 What it entails in practice is still
to be determined.
The issue of fiscal autonomy and accountability is also central to the
mega-city debate. It is slowly being recognized that the issue of redistribu-
tion is not at the heart of the choice between a centralized metropolitan
government and a two-tiered model. Policies of redistribution can be
designed equally well under either model of metropolitan government. In
the two-tiered system, for example, an intra-metropolitan fiscal transfer
system can be designed to achieve intermunicipal fiscal equalization as well
as interpersonal equalization through poverty-targeted grants to house-
holds. Instead, what is at issue is accountability and governance. A two-tier
model, with municipalites having access to own taxes, may provide greater
checks and balances than a centralized system. In sum, therefore, the fiscal
South Africa 265
instruments offered to provinces relative to the center and to municipalities

in the intra-metropolitan fiscal system will be important determinants of the
overall level of accountability of the intergovernmental system.
Capacity and decentralized responsibility

Limited capacity in public administration and political management is an
important constraint in the delivery process in South Africa today. This
lack of capacity is particularly evident at the provincial level and has been
presented in policy debates as an important argument in favor of not
giving fiscal responsibilities to provinces. Instead, it is argued that a
program of capacity-building needs to be undertaken before a process of
decentralization can be initiated.
The relationship between capacity constraints and the feasibility of
decentralization is not clear cut. A unique study of decentralization experi-
ence in Colombia suggests that capacity-building actually increased with
decentralization (Fiszbein, 1995). Expressed simply, giving responsibility
to lower-tier governments and opening the electoral process put pressures
on subnational governments to invest in a capacity-building process. This
demand-driven approach seems to have elicited greater demand for
capacity-building when compared to the traditional top-down, supply-
driven system adopted in many countries. In addition, the experience of
Colombia seems to suggest that a demand-driven system leads to a more
diverse supply response. The private and public sectors and community
organizations all have a role in enhancing the capacity of both the political
and administrative leadership.
The issue of capacity also raises the possibility of adopting an asym-
metric approach to decentralization. Not all provinces or local govern-
ments are equal in terms of fiscal and human capacity. These differences
suggest that decentralization of fiscal responsibilities need not be imple-
mented uniformly across the nation and across all levels of governments.
For example, major cities may be candidates for inheriting greater fiscal
responsibilities than smaller towns or rural local governments. Similarly,
one or two provinces may have the capacity to better manage their own
affairs. In these circumstances, the center could adopt a phased approach
to decentralization.
Separating fiscal from financial issues: some general considerations

The role of governments in the financial sector has important implications
for the design of an intergovernmental fiscal system. To begin with, the
266 Junaid K. Ahmad
fiscal base of subnational governments, which is defined by both the ver-

tical and the horizontal dimensions of fiscal decentralization, establishes,
in effect, the “collateral” that will determine their access to financial
markets. Policy uncertainties on the fiscal side, therefore, such as uncer-
tainty in the allocation of tax bases or the unpredictability of intergovern-
mental transfers, will influence the extent to which different tiers of
government can take advantage of financial markets. It may even be that
the uncertainty will be read by capital markets as the national govern-
ment’s way of centralizing fiscal powers. Subnational debt may then be
interpreted as being a direct obligation of central government. Defining the
fiscal rules is therefore a necessary step in enabling access to capital
markets and resolving the moral hazard problem.
More importantly, government’s involvement in the financial sector
dampens the economic role of capital markets in allocating credit and sig-
naling creditworthiness. The financial sector’s ability to promote economic
efficiency and the accountability of local officials, important elements of a
decentralized model of government, are adversely affected. Thus, pro-
moting fiscal decentralization by only devolving tax and expenditure
responsibilities will not allow local governments to take advantage of the
efficiency gains that can be achieved through the interplay of political
jurisdictions and capital markets in the financing and delivery of services
at the local level. To draw an analogy, would a decentralized model of local
government be effective if labor mobility between local jurisdictions was
controlled?
In addition, if the centralization of borrowing powers facilitates access
to finance by local authorities that are not creditworthy or makes capital
more “affordable,” central government will have provided, in effect, an
implicit subsidy through the financial sector. If the objective is to provide
such subsidies, a preferable alternative would be for central government
to convert the implicit financial subsidy into an explicit fiscal grant and
allow local authorities to pledge such transfer as collateral in accessing
capital markets. Such lump-sum transfers would not distort the price of
capital and would enable local authorities to establish a direct relationship
with capital markets. Fiscal subsidies have the added advantage of being
transparent and more easily monitored, and stand a greater chance of
being held in check via the political process.
Centralization of borrowing powers may distort the economic incen-
tives faced by local authorities in other ways. As von Hagen (1991) and
Wildasin (1995) suggest, fiscal restraints on capital market activities are
often circumvented and the resulting behavior – for example, dipping into
pension funds and engaging in off-budget activities – may impact on the
South Africa 267
allocation of resources. Wildasin (1995) suggests that restricting the access

to capital markets may also lead local authorities to reduce their invest-
ments in local goods. In the case of South Africa, the political pressure for
the rapid delivery of local services is sufficiently high that central govern-
ment may be forced to compensate local governments for the restriction on
borrowing by eventually funding local services through intergovern-
mental transfers. In effect, this would mean that central tax bases would be
funding local goods – bringing back centralization through the back door.
In sum, the discussion suggests that keeping the fiscal and financial roles
of government separate and decentralizing borrowing powers may be
fundamental in ensuring an efficient system of intergovernmental fiscal
relations. In particular, such measures would provide the incentive for
local governments to be more efficient in their delivery and financing of
services. Equally important, the separation may be necessary to ensure
that government intervention does not impede the functioning of the
capital markets – a distinct possibility in the context of low-income coun-
tries where financial markets are thin and government is the predominant
borrower in the economy.
Central government in South Africa has established several public finan-
cial intermediaries to assist in the financing of urban and rural infra-
structure. The regulatory framework for determining the relationship
between the public and private financial intermediaries and fiscal and
financial dimensions of intergovernmental relations remains to be clari-
fied.
Provinces and metropolitan governments

In most countries, boundaries between regions and the relationship
between different tiers of government were created in a pre-urban setting.
In such a setting, where the population was largely rural and the economic
base linked to agriculture, it may have made economic sense to provide
regional governments with substantial fiscal powers and to make local
governments subordinate to them. In South Africa’s urbanized setting,
however, where most of the population lives in urban areas and local
governments provide most subnational services, should all local govern-
ments be subordinate to regions? The question is specially relevant where
cities are so large that it seems appropriate to establish some form of met-
ropolitan government.
It may be argued that there is little reason to make all local governments
– especially those of larger metropolitan areas – totally subordinate to
regions. In a modern urban society many public services have either local
268 Junaid K. Ahmad
or national significance, but not regional significance. This line of reason-

ing has important implications for the structure of intergovernmental
fiscal relations. First, it suggests that some functions that are commonly
assigned to regional governments may be best assigned to metropolitan
governments or to the central government. Second, central government
may consider giving some metropolitan governments the status of a
region.
The South African constitution is, however, not clear on whether all
local governments including the metropolitan governments are under the
regulation of provinces or whether local governments are an independent
sphere of government. As a result several tensions are emerging. First,
provinces are politically reluctant to allow fiscal transfers to be sent
directly to the cities. Second, central government has centralized the bor-
rowing authorities of provinces on the correct premise that provinces do
not have the fiscal capacity to enter into a direct relationship with capital
markets and that permitting such a relationship to emerge would have
been fraught with moral hazard. This has left the central government in a
difficult political position for passing legislation enabling cities to go
directly to the market. Third, the decision not to provide a tax base to the
provinces has prevented central government from providing a surcharge
on personal income tax to the metropolitan areas and eliminating the inef-
ficient turnover tax. Fourth, by not providing provinces with a tax base,
the center has left open the possibility of the provinces taking a share of
metropolitan and local government revenues – a possibility not ruled out
by the Constitution. Ultimately, in the case of Gauteng, which is primar-
ily an urban complex, the province could have played the role of a met-
ropolitan tier – the fiscal equalizer as described in the case of
Johannesburg. The establishment of Gauteng province precluded this
option.
The intergovernmental fiscal system in South Africa is still in a state
of flux. It is truly a system in transition. The changes that are taking place
are fundamental. A new constitution has been established which has
catalyzed the process of restructuring the political, economic, and fiscal
relations between different tiers of government. In doing so, the leaders
of South Africa have attempted to create a democracy, begin the process
of delivery of public goods to all irrespective of race, foster economic
growth, and maintain fiscal stability. Few nations have attempted such
a transformation in such a short period and few have achieved such a
level of success as South Africa. Whether the success will persist will be
determined by the changes in the intergovernmental system of the
country.
South Africa 269
Notes
1 The chapter draws on McLure (1994) and Ahmad (1997).
2 The process of reform in South Africa is a very dynamic one and one which is
undergoing rapid changes. Therefore, the story outlined in this chapter is of
necessity incomplete.
3 Differences between white and black areas were not the only ones; apartheid
also imposed differences between the Indian and Colored population. For
expositional economy, this chapter focuses on the differences between black
and white.
4 In Province of Natal these were called Joint Service Boards (JSBs).
5 In fact, the urban sector accounts for over 80 percent of GDP and approximately
62 percent of the population. Agriculture and the rural economy accounts for
only 5 percent of GDP. That approximately 40 percent of the population still
resides in rural areas is due in large part to the spatial policies of apartheid.
6 In this sense, the new constitution adopts a regional approach inherent in
constitutions written during a time when cities did not exist, urbanization was
limited, and economies were predominantly rural. As argued later, it is not clear
from a public finance perspective, whether, in the context of an urbanized
society, externalities and spillovers have a regional basis or are defined on
national and local levels only.
7 Albeit without the racial nomenclature.
8 In their international review of the structure of urban governance, Bahl and
Linn (1992) conclude that while the centralized metropolitan model seems to
dominate, the structure of urban governance in most countries is a hybrid.
9 The experience of Minneapolis-St. Paul in metropolitan tax base sharing pro-
vides an interesting case study of how fiscal disparities were reduced across 188
municipalities (Rusk, 1993).
10 It may be more appropriate for the national government to finance a voucher
program but for a metropolitan government to implement it.
11 See Wildasin (1993) and McLure (1994) for an analysis of the efficiency of exist-
ing local taxes, alternate fiscal instruments for local governments, and the effi-
ciency and equity implications of relying on central tax bases for financing local
infrastructure investments.
12 The incidence of the payroll and turnover taxes will, in practice, be spread more
widely than just the metro area, as in the case of Johannesburg. There may be a
case, therefore, for sharing these revenues nationally, not just on a metropolitan
basis. This might, however, reduce the fiscal autonomy of the metropolitan tier
as the metros would, as a result, rely more on transfers from other tiers. This
tradeoff between efficiency of tax instrument and autonomy of metropolitan
governments is better addressed by replacing payroll and turnover taxes with
alternate fiscal instruments. For a discussion of this point, see Wildasin (1993)
and McLure (1994).
13 The management of subnational fiscal crises including addressing the “bank-
ruptcy” of several provinces and local governments has been an important
part of the evolution of the intergovernmental system during this period of
270 Junaid K. Ahmad
transition for South Africa. It is an issue that has not been discussed in this
chapter. However, it should be stressed that the management of these crises
while a regulatory framework is still being developed is one of the success
stories of government in the field of fiscal management.
14 An important aspect of the intergovernmental system and at the heart of the
concept of “cooperative governance” is the design and implementation of the
budgeting system also known as the Medium Term Expenditure Framework.
This chapter has not analyzed the evolution of this very important budgeting
framework, which is also one of the success stories of government.
References
Ahmad, Junaid (1997), “Structure of urban governance in South Africa,” in David
E. Wildasin, ed., Fiscal Aspects of Evolving Federations, Cambridge: Cambridge
University Press.
Bahl, Roy, and Linn, Johannes (1992), Urban Public Finance in Developing Countries,
Baumol, William, and Lee, Kyu Sik (1991), “Contestable markets, trade and
development,” World Bank Economic Observer, 6: 110–23.
Bruckner, Jan K. (1994), “Welfare gains from removing regional land-use distor-
tions,” Working Paper, AF1EI, World Bank, Washington, DC.
Development Bank of South Africa (1995), South Africa’s Nine Provinces: A Human
Development Profile, Pretoria.
Fiszbein, Ariel (1995), Local Government Capacity In Columbia, Washington, DC:
World Bank.
Inman, Robert (1996), “Provincial grants in South Africa,” unpublished draft.
McLure, Charles (1994), “Intergovernmental fiscal relations in South Africa,”
Working Paper, AF1EI, World Bank, Washington, DC.
Rusk, David (1993), Cities Without Suburbs, Princeton, NJ: Woodrow Wilson Center
Press.
Swilling, Mark, Humphries, Richard, and Subhane, Khehla (1991), Apartheid City
in Transition, Cape Town: Oxford University Press.
von Hagen, Jurgen (1991), “A note on empirical effectiveness of formal fiscal
restraints,” Journal of Public Economics, 44: 199–210.
Wildasin, David E. (1993), “Local finance of urban infrastructure in South Africa,”
Working Paper, AF1EI, World Bank, Washington, DC.
(1995), “Financing local government outlays in South Africa,” Working Paper,
AF1EI, World Bank, Washington, DC.
9
Bosnia-Herzegovina: fiscal federalism – the

Dayton challenge
WILLIAM FOX AND CHRISTINE WALLICH
The challenge of fiscal federalism in Bosnia is perhaps unique in the world:

the Dayton talks held in October 1995, immediately after the ceasefire,
began with a fully blank slate.
• How would the new nation that emerged as a result of the peace talks
be structured from a fiscal perspective?
• What would be the role of the central state and what would be that of
the two subnational units (the “entities”) that constituted it?
• How would the three previously warring communities of Croats,
Bosnian Muslims (Bosniacs), and Serbs work together to form a central
government, what would the entity governments look like, and what
would be the fiscal functions and rights of these entities?
• How would the entities, in turn, be structured internally, and what
would be their fiscal governance?
All these questions were open in October 1995, when the international
community worked together with experts and political leaders to forge, for
Bosnia, a new constitution, and the new intergovernmental fiscal system
that would be set out in it. The new system of fiscal federalism should, it
was agreed, be able to withstand the stresses that would be a natural con-
sequence of the centrifugal forces still present in the country, be economi-
cally sensible, and yet also obtain the consensus of the three parties who
would live with it and implement it. The purpose of this chapter is to
describe Bosnia’s current arrangements in fiscal federalism, to outline the
unique challenges that the Dayton system proposed, and to draw some
lessons for the design of fiscal federal systems in ethnically diverse
economies.
The findings, interpretations, and conclusions expressed in this chapter are entirely those of
the authors, and do not necessarily represent the views of the World Bank, its executive direc-
tors, or the countries they represent. Very useful comments and assistance were provided by
Wei Ding, Sweder van Wijnbergen, Sebnem Akkaya, Luis Alvaro Sánchez, Jennifer Keller,
Richard Bird, François Vaillancourt, and Stanley Winer.
271
272 William Fox and Christine Wallich
The Dayton fiscal and governance challenge

Under the Dayton accords, Bosnia-Herzegovina (hereafter, Bosnia) was
established as a single sovereign state composed of two “entities,” the
Federation (comprising two major ethnic groups, and split between Croat-
majority and Bosniac-majority areas) and the Serb Republic (Republika
Srpska). The Federation has a three-tier fiscal system – the Federation,
cantons, and municipalities. While most cantons in the Federation are
mono-ethnic, there are two multi-ethnic cantons, and many multi-ethnic
municipalities. The Serb Republic is a largely mono-ethnic entity as a result
of the war. It has a two-tier structure, consisting of the Republic and the
municipalities. Freedom of movement is intended to prevail and the right of
return is to be respected under the Dayton agreements.
As a result of the war, the unified fiscal system of the old Republic of
Bosnia-Herzegovina was broken up, and until recently there were three
totally different fiscal systems in the country – one in the Croat-majority
area of the Federation, one in the Bosniac-majority area, and one in the Serb
Republic. While all three systems bore the imprint of their Yugoslav
origins, the system in the Croat-majority area came, over the course of the
four years following Bosnia-Herzegovina’s declaration of independence,
to be harmonized with that of Croatia, while that of the Serb Republic
became harmonized with that of the current Federal Republic of
Yugoslavia – Serbia-Montenegro. The Croat and Bosniac systems have
now begun to be merged, as new Federation structures are put in place.
Customs structures, the tax administration, customs levies, excises, and
several other taxes are now largely harmonized. However, the divergence
between the Federation and the Serb Republic systems remains significant.
Income and spending differences

There are significant income differences between ethnic groups in Bosnia,
and these differences are reflected in income differentials between the
different parts of the Federation and between the Federation and the Serb
Republic. Recent (1995) estimates suggest per capita incomes of about US
$500 in the Bosniac-majority area of the Federation (accounting for 2.5
million of Bosnia’s total current population of 3.5 million),1 about US$1,800
in the Croat areas of the Federation (some 400,000 population), and about
US $1,000 in the Serb area – Serb Republic (700,000 population). This dis-
parity in per capita incomes suggests major differences across these areas
in terms of tax bases and taxable capacities, and in the absence of equaliza-
tion policies, in terms of service provision levels.
Bosnia-Herzegovina 273
Even prior to the war, there were differentials in public service levels
across the regions in Bosnia. The war, which broke the fiscal structure into
three separate systems, has significantly exacerbated spending differences.
In the Bosniac-majority area, for example, the pension level was 11.4
Deutschmarks (DM) per month in 1995, while the scheme operating in the
Croat-majority area paid pensions of about 65 DM per month. Estimates
put the damage from the war at over US $20 billion (World Bank, 1996).
However, the destruction of infrastructure and public utilities, as well as
of the capital stock in the productive sectors, hit the Bosniac areas the
hardest, since industry and most core infrastructure facilities had been
strategically located in central Bosnia since the early days of Yugoslavia.
The tax bases of the Bosniac-majority area and its ability to generate
income have thus been significantly impaired.2 By contrast, the Croat-
majority area’s strategic location near the coast enabled this region to
control trade into the interior of the country during much of the war, and
despite its limited resource endowment, to generate substantial revenues
from customs and other trade-related levies. The Serb Republic’s economy
has been badly damaged by the embargo, and is expected to recover only
slowly.
A role for equalizing policies?

These differences in taxable capacities and public service provision levels
would suggest, other things being equal, that from both equity and effi-
ciency perspectives, inter-entity (and intra-entity, in the case of the
Federation) fiscal equalization policies could have a constructive role to
play in the new fiscal system. Such policies would prevent the large differ-
ences in access to fiscal resources currently available to the subnational
entities and the groups within them from being translated into large dif-
ferentials in the supply of public services and infrastructure.3 From both
perspectives, one might be especially concerned about those public ser-
vices with major externalities for the nation as a whole, such as health and
education, which are the domain, under the Dayton constitution of the
entities (or, in the Federation, the cantons).
Implicit in this view of equity and efficiency is that the relevant domain
of concern (McLure, 1994) for efficiency- and equity-focused redistribution
policies is the nation as a whole, that is, all of Bosnia. Under such a view,
the population of Bosnia sees itself as “Bosnian,” instead of as “Croat,”
“Bosniac,” or “Serb,” and supports policies that address interregional dis-
parities. On the other hand, if people’s primary allegiance is to the national
groups within the entities, there will be a more limited view of the domain
of concern, and differences in the average incomes of groups within an

entity or between entities, or in the average level of public service provi-
sion, may not be the predominant concern of these groups.4
A strong role for the central state?

The vast challenge of recovering from the destruction of war would also
suggest that a strong central government, capable of maintaining macro-
economic stability (see Prud’homme, 1994; Tanzi, 1995), that creates a
“common economic area” throughout Bosnia and mobilizes resources for
reconstruction through effective budgetary and foreign borrowing poli-
cies, could also bring substantial benefit. With the near-complete destruc-
tion of the productive sectors’ capital stock and a major portion of the
country’s infrastructure destroyed, effective use of scarce domestic bud-
getary resources and foreign flows could not be more critical.
However, as this chapter will suggest, the legacy of war, and the legacy
of the intergovernmental fiscal system in Yugoslavia, make the hope for a
strong central government difficult. The state that has been created out of
Dayton, far from being empowered and fiscally robust, is one with no
taxing powers of its own, and therefore is dependent totally on the entities
for resources via “contributions” to the state budget. Its spending author-
ity is virtually nil. From a macro perspective, its ability to manage a
stabilization policy using fiscal instruments is highly constrained. The
state’s discretionary macroeconomic management tools are limited further
by the fact that the central bank is to be run initially as a currency board.5
With a weak fiscal authority at the center, the potential for regional equali-
zation is also limited. And, after the national elections, the state may not
borrow abroad or indebt itself, without the consent of the entities. In a nut-
shell, Dayton set up a “bottom-heavy” system with most responsibilities
and rights in the hands of the entities. Clearly, setting up an intergovern-
mental system that makes sense, especially in this fraught political context,
is a major challenge.
The legacy of Yugoslavia

The legacy of Yugoslavia has significant relevance to the design of Bosnia’s
intergovernmental fiscal system; some would say that it is the specter that
still hangs over the design of such policies. Yugoslavia was always the
most decentralized of the socialist economies, both in terms of its economic
and fiscal management and its political structures. The republics had sig-
nificant autonomy from the very beginning, unlike the rest of the socialist
world, and the power of the republics was further increased in the 1963
constitution, creating the basis for the emergence of centrifugal forces early
on. Tax reforms in 1971 gave greater revenue-raising powers to the
republics and local authorities, and the 1974 Constitution transferred
many functions to the republics, with “areas of common interest” remain-
ing the responsibility of the state. These included military expenditures,
administration, and economic interventions, including support to enter-
prises.
Yugoslavia’s six republics were highly differentiated in income level and
culture. Slovenia was by far the most prosperous, with a per capita income
in 1990 of US $6,500, with Croatia the next best-off. Bosnia-Herzegovina,
Macedonia, and Montenegro were considered the least-developed
republics. Bosnia’s per capita income in 1990 was only 29 percent of that in
Slovenia (and Macedonia’s only 22 percent). Further, the level of per capita
budget expenditures in these republics reflected these large income differ-
ences. In the social sectors, for example, Slovenia’s per capita social spend-
ing was twice the Yugoslav average, while social spending in Bosnia and
in Macedonia was 70 percent of the average (OECD, 1992). These spend-
ing differentials were the result, inter alia, of different initial endowments
and incomes, but equally importantly, of the strong reliance, peculiar to the
socialist world, on “derivation-based” systems of revenue-sharing. Most
fiscal revenues were nationally “owned” and retained by the areas in
which they were collected, so that the socialist revenue systems typically
benefited those areas with the more robust tax bases.
Regional redistribution policies: who gained from cross-subsidies?

Addressing the huge regional imbalances in income and expenditures was
an early objective of Yugoslav federal policy. Federal support to the
republics came through two channels – fiscal and financial. The fiscal
channel had several streams. The “Fund for the Development of the Least
Developed Regions,” established in 1965, introduced explicit support to
the less-developed regions in Yugoslavia – Bosnia-Herzegovina,
Montenegro, Macedonia, and Kosovo.6 At its peak, this fund mobilized
some 2 percent of Gross Social Product (GSP) from the better-off regions
which contributed a fixed proportion of their GSP to the fund.7 Federal
budgetary resources were also directed to raising the level of social outlays
in the poorer republics. The financial channel operated through the repub-
lic-level commercial banking systems and the central bank,8 and contrib-
uted both to Yugoslavia’s hyperinflation and to its external indebtedness.
Many attempts have been made to estimate the overall magnitude and
net effects of these fiscal and financial transfers between republics.

Vodopivec’s (1991) empirical estimate of the net redistribution among
republics of all the subsidies and taxes suggests that the three better-off
republics (Slovenia, Croatia, and Serbia) were net “taxpayers” and the
three less-developed republics (Bosnia-Herzegovina, Macedonia, and
Montenegro) were net beneficiaries. Such estimates, however, can only be
suggestive at best, given that – as in much of the socialist world – the inter-
republic redistributions through the fiscal and financial systems were, in
the end, a series of ad hoc, bargained, and very non-transparent agree-
ments, whose effects and incentives were not at all well understood, and
whose very murkiness gave rise to much dissatisfaction among the
republics.
Structural vulnerabilities
The better-off republics’ dissatisfaction with federal redistributive policies
was compounded by another feature of Yugoslavia’s intergovernmental
fiscal system – the “bottom-up” system of intergovernmental revenue
flows, which Yugoslavia had in common with most other socialist coun-
tries, and which still holds in Bosnia today (Bird, Ebel, and Wallich, 1995).
This system conferred a very special vulnerability on the federal budget,
since all tax revenues were collected at the republic level (or below) and
transferred upwards to the federal budget in agreed shares (the federal
government received all of the federal sales tax and all import duties). In
addition, fiscal “contributions” from the republics, over and above those
they made to the federal Fund for the Development of the Least Developed
Regions, were negotiated annually, to balance the federal government
budget.
The role of fiscal factors in the break-up

There were many reasons for the break-up of Yugoslavia, but it would be
incorrect to suggest that fiscal, quasi-fiscal, and redistributional factors
played no role. Redistribution policies of this scope were bound to nourish
political separatism, as prosperous republics became more and more reluc-
tant to release resources to poorer republics. The lack of transparency in
the system made it possible for each republic to argue that it bore the costs
of the intergovernmental system, while the benefits accrued to others, so
that each thought the others gained at its expense. Woodward (1995)
argues, compellingly, that the centralizing budgetary and financial policies
designed to give the central government more control over stabilization
policy, supported by the international community in the late 1980s, further

exacerbated such tensions, since they reduced the budgetary autonomy of
the better-off republics and empowered the central government, fiscally
and financially, with greater command over resources. The vulnerability
conferred by the design of the system itself finally surfaced, as the three
most prosperous republics – Slovenia, Croatia, and Serbia – withheld
federal tax revenues from the federal budget in 1991. It was with this
history that the partners at Dayton began their work.
Some lessons?
What lessons can be drawn from Yugoslavia’s break-up for the creation of
fiscal federalism in Bosnia? And what lessons might one hope to incorpo-
rate into the new structure that emerged from Dayton? There would seem
to be several.
First, because of the Yugoslav legacy, the designers at Dayton had to take
into account the inherited suspicion of a strong central state and its per-
ceived potential for abusing its power. This limited the role that could be
given to national-level taxing powers or centralized spending. Second, the
designers found that the notion of “ownership rights” over revenue col-
lected on one’s territory was hard to overcome – this was perhaps even
more so in the ethnic regions of Bosnia, recently emerging from war, than
in Yugoslavia, where territorial claims to revenues were very strong
indeed. Third, and related, the designers had to contend with a prima facie
need for, but apparent virtual intolerance of, fiscal transfers along the lines
of the Federal Fund or other cross-subsidies between areas, and sometimes
even within multi-ethnic municipalities.
But, even if no other fiscal lesson was learned from the demise of
Yugoslavia, the vulnerability of the federal fisc to transfers from below was
clear to the designers of the new structure. And yet, the state that emerged
had no revenue powers of its own. The vulnerability of the federal fisc to
a decentralized tax administration was similarly evident. And yet, the
bottom-up system of revenue administration remained. This reflects the
fact that the design of Bosnia’s new system was driven very strongly by
the interests of the entities, whose agreement had to be obtained to the new
structure.
The Dayton rules

Under the Constitution, major government budgetary responsibilities are
shared among the four levels of government – the state, the entities, the
cantons and the municipalities (World Bank, 1996). The state government
has exclusive responsibility for foreign and foreign trade policies, customs
policies, monetary policy, immigration, refugee and asylum policies, air
traffic control, payment of international financial obligations of Bosnia
incurred with the consent of both entities, regulation of inter-entity trans-
port, communications, and international and inter-entity law enforcement
(but not the raising of an army, which is an entity-level responsibility). All
government functions not explicitly assigned by the Constitution to the
state are assigned to the entities (though the entities may agree to relin-
quish some of their responsibilities to the state). The state government has
no independent tax sources under the Constitution (although the
Constitution provides that with the approval of the state parliament, the
state may levy its own taxes in the future), and is expected to finance its
activities almost entirely from transfers from the two entities. Two-thirds
of the state’s revenues are to come from the Federation and one-third from
the Serb Republic to cover the budget approved by the state parliament.
Under the Federation’s constitution (June 1996) the entity governments
have a wide range of expenditure responsibilities, including exclusive
responsibility in their territories for defense (there are currently three
armies: one in the Serb Republic and two in the Federation, one Bosniac
and one Croat), internal affairs, police, environmental policies, economic
and social sector policies, agriculture, industry, refugees and displaced
persons, reconstruction programs, and justice, tax, and customs
administration. This constitution provides that some of these functions
may be exercised jointly with the cantons, or separately, or by the cantons
with coordination by the Federation’s government. Thus, the Federation
and cantons exercise joint responsibility for health, environmental policy,
infrastructure for communication and transport, social welfare, and inter-
ior affairs, among others.
The cantons (in the case of the Federation) are responsible for all other
matters not granted expressly to the Federation. These include education,
culture, housing, public services, local land use, and social transfer expen-
ditures. Each canton is authorized by the Federation’s constitution to del-
egate its responsibilities to the municipalities in its territory.
The municipal governments in the Federation are granted “self-rule on
local matters,” including all responsibilities delegated to them by the
cantons. In the case of municipalities with a majority population that is
different from that of the canton as a whole, the canton must delegate all
responsibilities described in the preceding paragraph to the municipal
government.
Municipal governments in the Serb Republic are responsible for all
matters not explicitly the responsibility of the entity – and largely parallel
the combined responsibility of cantons and municipalities in the
Federation.
Can Dayton work?

More appropriately, is the Dayton structure the best for Bosnia, and if not,
how should the structure be altered? These are complex questions, particu-
larly since the equity implications associated with government structures,
and the particular importance linked to cross-subsidies in Bosnia, make it
very difficult to determine what is the best system.
The complicated, multi-level government structure arising from the
peace arrangements reflects an ardent aversion to a strong state govern-
ment that derives from the experience of being part of Yugoslavia. Better-
off minority groups are particularly concerned that their rights will not be
upheld, and as a result they envision being economically disadvantaged.
The fear is that a strong central government would continue the
Yugoslavian tradition of cross-subsidies providing greater benefits to the
less well-off majority group. The war has further diminished the willing-
ness of some nationalist groups to view Bosnia as a single country and to
work for the better of the whole. Creating a government with most expen-
diture responsibilities and revenue-raising powers at the subnational level
was expected to limit the opportunity for transfers to the majority. The
paranoia is sufficiently strong that favoritism of the majority is seen behind
every program. For example, apprehension exists within the Federation
that government-run economic development programs will be designed
to support industries that predominate in the majority areas.
The difficulty of creating governments in this environment can be seen
in the relationship between the Bosniacs and Croats in structuring the
Federation. Croat majority areas favor relatively strong canton govern-
ments, while the Bosniac areas favor a relatively strong Federation. The
Croats, as the better-off, but minority population, want to continue
decentralizing responsibilities to lower levels of government, where
ethnicity and incomes are relatively more homogeneous, and therefore
opportunities for cross-subsidies are lessened. The Bosniacs argue for
service delivery and revenue collection concentrated in the Federation
where they might better be able to effect cross-subsidies. Economies of
scale seem to indicate that many functions would be better performed at
the Federation level given the relatively small geographic area and popula-
tion size of the cantons. However, the discussions are heavily driven by
equity and not efficiency.
It is too early to judge fully the Dayton agreements on the basis of actual
events. Still, some general problems can be seen. First, the tax and expen-
diture assignments are imprecise, leaving the potential for disagreements
and requiring difficult political negotiations to reach accords. Even where
there is greater precision in the rules, some of the assignments are ineffi-
cient and need to be reconsidered. Second, little room is allowed for cross-
subsidies to exist anywhere in the structure. This means large inequities
can be expected to develop since economic conditions differ widely across
the country. Third, most of the government structures appear potentially
sustainable, if not efficient. However, the state’s roles are very limited, and
its funding sources are precarious. As a result, the state’s sustainability is
in question. Finally, government’s share in the economy after the many
transitions is still to be determined. The remainder of this section provides
a detailed examination of specific issues in the Dayton rules.
Imprecision of expenditure rules

Despite the appearance of clear guidelines on which government is
responsible for delivering which services, the Dayton rules are confused
and create overlapping responsibilities. Not surprisingly, very different
perspectives exist on what the final outcome should be. Overlap occurs
when the state is charged with one aspect of service delivery and the enti-
ties with another. For example, the state is to do “customs policy” and the
entities “customs administration.” These assignments leave in question
which level is to set tariff rates, which is to determine exemptions, which
gets control over the revenues, and a number of other practical issues.
Similar overlap occurs at the Federation and canton levels, which are
assigned joint responsibility for human rights, health, environmental
policy, transportation and communication infrastructure, social welfare
policy, immigration and asylum, tourism and natural resources.
Little is said about municipal responsibilities. The relatively centralized
structure already in place will remain in the Serb Republic. In the
Federation, the issue is which responsibilities are cantonal and which are
municipal. Municipalities are granted self-rule and are also permitted to
perform functions delegated by the cantons. The probable outcome is a
different relative set of duties in each canton, given their geographic and
demographic diversity. Differing patterns of responsibility will reflect
varying sizes, ethnic compositions, and political strengths in the cantons.
No problems should arise from assigning provision to different govern-
ment levels as long as governments are creative in achieving efficiency in
production and tax powers follow expenditure responsibilities.
A series of other specific problems with the expenditure rules can be

identified. In some cases there is no clear assignment of functions. Pensions
are such an example, though they might be categorized under one of the
general functions listed in the Dayton agreements. In other cases, assign-
ments are politically difficult, such as having defense delivered by the
Federation. In a third set of cases, the assignments may be economically
inefficient; higher education being set as a cantonal rather than a
Federation responsibility is such an example.
Expenditure responsibilities being imprecise or being assigned to the
wrong level of government does not necessarily mean the assignments will
be inefficient, since governments could negotiate efficient arrangements,
such as cooperative service delivery. However, the recent history of events
in Bosnia make such negotiations particularly difficult, implying that there
are likely to be economic inefficiencies over the short to medium term.
Expenditure policy
The war arrested the transition to a market economy and the correspond-
ing relative reduction in the size of the public sector. The war years led to
ad hoc and very distorted budgeting and expenditure practices in the
Bosniac, Croat, and Serb areas. Expenditures were heavily focused on mili-
tary purposes, with about 45 percent of each entity’s budgeted expendi-
tures going for military uses. Data are unavailable on the total extent of
military expenditures because large off-budget military purchases and
spending financed by foreign governments occurred as well. Budgets for
non-military purposes were only provided as necessary and depending on
available resources. Expenditures were made essentially on a cash basis.
Governments made payments when resources were available, and
suspended payments when they were not. Pensions and wages – for mili-
tary and non-military workers – often went unpaid. Factor payments bore
little relationship to market value.
Expenditure rationalization is an important area for reform. The size of
the public sector needs to be limited to make room for private sector
development, and the structure of public expenditure programs needs to
be improved to facilitate growth, while maintaining an acceptable level of
social protection. Public expenditures currently represent more than 40
percent of Bosnian GDP, and this is in the context of substantial public
sector arrears in wages, pensions, military, and debt service. Reduction of
overall public sector expenditures to sustainable levels as a percentage of
GDP will require major cuts in many areas of public expense; key areas
for reduction are defense, subsidization of the economy, and entitlement
programs such as pensions. Pension payments cannot be based on prewar

earnings and expectations, and must be scaled back to a level that is con-
sistent with reasonable wage tax rates. (Tax rates are currently in excess of
90 percent on net wages.) A rationalization of public expenditure programs
will be required to accompany this process of downsizing the government,
as well as to provide better targeting of public expenditure. Highest prior-
ity must be given to public expenditures which provide a framework that
will allow the private sector to flourish – appropriately selected infra-
structure is essential. Subsidization of outdated, unproductive public
enterprises must be given the lowest priority.
Great care must be taken to avoid building large administrative struc-
tures in the public sector – a potential problem that could be exacerbated
by the large number of overlapping governments. Early experience evi-
dences a propensity for cantons to overstaff their central offices with politi-
cal appointees. For example, Tuzla canton already has over 300 employees
performing administrative functions. These practices must be stopped at
the canton level and avoided in all governments. Efforts to rationalize
salaries for government employees resulted in large percentage salary
increases during 1996, expanding the size of the budget for administrative
staffs. The raises, mandated by central governments, have placed many
cantons and municipalities in the difficult position of having insufficient
resources to finance their expected salary payments. As a result, employ-
ees in some areas receive raises but do not in others.
Finance and tax administration

The Dayton agreements offer fewer specifics about financing arrange-
ments after 1996 than about expenditures. Three general guidelines are
given: the entities and their constituent governments are given
responsibility for finance and tax administration; the state is to receive
two-thirds of its funding from the Federation and one-third from the Serb
Republic; and there is to be a Federation tax administration. All three
create significant issues in practice.
Financing of the state

The Constitution indicates that the state is to adopt a budget for the expen-
ditures necessary to finance its responsibilities. No explicit own-source
revenues are provided for the state, but the state can generate revenues
through charges such as passport fees. Two-thirds of the financing that
remains is to be provided by the Federation and one-third by the Serb
Republic, a bottom-up arrangement reminiscent of the former Yugoslavia.
The state’s sustainability is a concern in this environment where a strong

degree of distrust exists. No limitation is placed on state spending, other
than that expenditures are to be the ones necessary to finance its
responsibilities. The financing structure differs from a conventional grant
system, since the recipient is telling the grantor how much to provide.
Disagreement between the state and the entities over what are necessary
expenditures should be expected, and the entities are likely to balk at times
about paying their share. Difficulties with funding of the United Nations
may be an appropriate parallel. Tax rates will need to be higher in the Serb
Republic than in the Federation to generate the respective revenue
contributions because of the greater aggregate population and income in
the Federation. As a result, pleas of unfairness can be expected.
The state may be less accountable for spending revenues raised by the
entities. Also, variability in revenue flows will be borne solely by the enti-
ties, which in principle are unable to reduce contributions to the state to
help lessen the effects of unexpectedly low revenues. There are two sides
to this coin, since entities do benefit if better than anticipated revenues
result.
Federation finance and tax administration

Tax structures, emanating from the Yugoslav tradition, are similar in the
two ethnic areas. The Bosnian structure as of November 1995 is given in
table 9.1. Ownership of tax revenues, control over tax rates and bases, and
tax administration are separable. Division of these, with, say, the
Federation administering taxes that are owned by both the Federation and
cantons, is normally based on conditions that may not be fully in place. A
level of trust must exist, so that the government owning revenues is certain
that the administering government will transfer the appropriate amounts
in a timely fashion.9 In some cases, local offices of the Russian State Tax
Service are passing resources to the regions and municipalities and giving
the Russian Federation government what remains, a pattern that is the
Bosnian concern. Cantons fear Federation collection will be to their detri-
ment and the Federation feels the same in reverse.
The owning government must also feel that the administering govern-
ment will make a faithful effort to collect revenues. Pressures and incen-
tives in the administering government may be such as to focus more of its
limited resources on collecting revenues that it can keep, than on taxes
going to other governments. Allowing the administering government to
retain part of revenues as a collection fee can help, but it still gets more
revenue benefits from collecting taxes that it totally retains.
In such a small country, administrative and compliance efficiencies
Table 9.1. Bosnia-Herzegovina: tax structure, November 1995
Tax Type of tax (description) Rate
Taxes on individuals and corporate profit

Taxes on individuals Schedular taxes on certain components of income are
supplemented by a further tax on gross personal
income exceeding a specified threshold
Payroll tax Withheld by employer 10%
Farming tax 5–10%
Self-employment tax 36%
Tax on income from property Progressive rates depending on the type of income
Tax on total personal income Progressive rates, scale from 5–50%
Republic reconstruction tax Base is net wages; withheld by employer 10%
Tax on corporate profit Tax is paid by enterprises and other legal entities 36%
realizing profit, whether or not domiciled in the
Republic
Tax on games of chance Tax paid by individuals 15%
Social security contributions

Employees
Old age insurance 17%
The gross wage is the base for payment of the
Health insurance ⎫ contribution ⎧ 14%
Unemployment insurance ⎬ ⎨ 2%
Employer ⎭ ⎩
Old age insurance 17%
The gross wage is the base for payment of the
Health insurance ⎫ contribution ⎧ 14%
Unemployment insurance ⎬ ⎨ 2%
⎭ ⎩
Tax on property
Tax on real estate transactions Tax paid by legal entities and individuals at the time of 15%
a real estate transaction
Gift and inheritance tax Paid on real estate and personal property over a 0–14.4%
certain value
Motor vehicle tax Tax paid by individuals on vehicles less than five years 1,500 dinars per year
old and engine size exceeding 1.6 liters
Boat tax Paid by individuals on boats exceeding 5 meters in 900 dinars per year
length
Sign tax Paid by legal entities and individuals on a sign DM 50 to DM 250 annually
displayed
Tax on weekend cottages Paid by individuals on the construction value of the From 0.10–1% on the value of construction
structure
Domestic taxes on goods and services

Sales tax on products Tax paid on final consumption, but on alcoholic and 5–20%
non-alcoholic beverages, tobacco products, coffee,
petroleum and petroleum products, paid on first sale
Sales tax on services Paid by legal entities and individuals on services 10%
subject to the tax
Sales tax surcharge A general surcharge on goods, levied on the sales-tax 10%
inclusive price
Excise taxes
Tax on petroleum products Paid by the producer or importer DM 0.08–0.4 per liter depending on the product
Tobacco tax Paid by the producer or importer DM 0.22–2.3 per pack
Beer tax Paid by the producer or importer DM 20–30 per hectoliter
Alcoholic beverage tax Paid by the producer or importer DM 8–10 per liter of pure alcohol content
Non-alcoholic beverage tax Paid by the producer or importer From DM 1–20 per hectoliter
Tax on imported automobiles Paid by the legal entity or individual on a new DM1 900–8 000
imported automobile with a power exceeding
75 kW
Tax on imported coffee Paid by the legal entity or individual who brings From DM 1–4 per kg
coffee into the Republic
Taxes on international trade

Import duties 0–21%
Source: International Monetary Fund.

should arise from Federation (or state) control over most tax rates and
bases, since uniform rates and bases will be more difficult to achieve with
devolved control. However, political incentives can cause Federation offi-
cials to make concessions on tax bases where the revenues are owned by
the cantons. Granting concessions against another level of government’s
tax bases offers all of the same political gains but fewer costs because the
concession-granting government does not have to face the budgetary con-
sequences of its actions. Tuzla canton has already complained that conces-
sions given by the Federation to reduce the electric company’s wage taxes
have significantly reduced Tuzla’s receipts. Again, willingness to give
control to another level of government requires a strong degree of trust,
despite the incentives and history creating a perception that the trust
would be misplaced.
Separate tax administrations, with offices located in each municipality,
are currently operating in the Bosniac and Croat areas. A significant next
step in development of the Federation must be creation of a Federation tax
administration, as called for in the Dayton agreement. Legislation was
passed in 1996 establishing most tax collection at the Federation level, but
the tax administration was not operating even in late 1997.
Structural conditions in Bosnia suggest that grants should be an impor-
tant financing source for cantons and municipalities. One reason is that
any assignment of revenue ownership is likely to lead to both vertical and
horizontal imbalances because data for estimating expected tax collections
are poor and the weak economies in some areas will leave them unable to
provide minimum service levels. Some mechanism, presumably operating
through a grant system, needs to be in place to smooth financing problems
during the interim as assignments are finalized and regional economies
rebound. Also, a system of grants could be used to exploit economies of
scale in tax collection, enhance equity in revenue distribution, and account
for geographic externalities in consumption. However, the very limited
tolerance for cross-subsidies between geographic areas suggests that
grants must play a small role in overall finance during the near term.
Unlike transfers made during the Yugoslavian years, any grants that are
introduced must be transparent and rely on objective indicators. A system
of grants based on negotiated formulas or ad hoc procedures would be par-
ticularly inappropriate for Bosnia, given the lingering distrust still pre-
vailing among different groups in society.
Debt policy
The IMF and World Bank require that their lending be to the internation-
ally recognized government, which is the state, but no agreement has been
reached on how external debt service is to be financed. The Constitution

says that “Each Entity shall provide all necessary assistance to the govern-
ment of Bosnia and Herzegovina in order to enable it to honor the interna-
tional obligations.”10 Like many Dayton rules, implementation of this
statement is unclear. One possible interpretation is that each entity is to
make contributions according to the two-thirds Federation, one-third Serb
Republic rule. However, blind adherence to this rule is likely to generate
controversy that could lead to one entity refusing to pay its share. Even
getting agreement from the entities to borrow through the state has been
very difficult. Determining a means for financing debt service is more com-
plicated.
The Dayton agreements allow little flexibility for macroeconomic policy,
but the governments may seek to operate at a deficit, both to ease financ-
ing problems and to stimulate the economy. Some examples are already
apparent. The original 1996 Serb Republic budget assumed a deficit equal
to one-third of spending. The 1996 state budget included authorization for
borrowing and for letters of guarantee. In practice, the ability to borrow is
extremely limited. The Republika Srpska budget did not pass when it was
recognized that there was no means of financing the deficit. Thus, with the
central bank operating as a currency board and a weak banking sector, con-
cerns about deficits may be unimportant in the short term.
Governance during 1996

The Dayton agreements were drafted with expectations that the complete,
anticipated government structure would not be operative in 1996. In any
event, a transitional period is necessary because time-lines in the agree-
ments were impractical, a difficulty exacerbated by political problems in
getting legislative approval for many facets of the agreements.
Revenue assignments in the Federation for 1996 were negotiated
between the IMF, the World Bank, and the government. The state was to
receive a share of customs tax revenues. The Federation was given access
to excise taxes and the share of customs taxes not passed to the state. Other
taxes in the Federation were to go to the cantons.
The state has performed minimal operations, in as much as broad-based
acceptance of the state’s role was hampered in the pre-election period. The
presidency was elected on September 14, 1996, but the government was
slow to be formed. The state budget was a relatively small DM 61.4 million
for 1996, and is mostly used for wages to operate ministries and for the set-
up costs of providing diplomatic functions and issuing documents. The
state received about 30 percent of customs tax revenues from the
Federation, but there has been considerable controversy over how much it
should get. The Serb Republic did not contribute to state finance in 1996.
The Serb Republic government structure was not altered by the agree-
ments, so no transition was necessary. However, the initial budget pro-
posed by the Ministry of Finance was 50 percent above expected revenues.
The parliament sent the budget back for adjustment, and the government
operated under an interim budget. Spending was limited to available
funding during the first three quarters of 1996, and revenues are running
about 15 percent below estimates. The Serb Republic had an explicit deficit
equal to about 1 percent of GDP in 1995, and a payment to cover some of
the 1995 shortfall was included in the 1996 budget. The majority of rev-
enues are collected at the entity level, and shares of the revenue are trans-
ferred to finance municipal functions. The municipality of Banja Luka has
argued strongly that for political reasons, it is receiving less than its legis-
lated share of taxes.
The Federation has not begun to operate as a fully fledged government.
Ministries have been formed and ministers and some staff appointed, but
only the Federation’s customs bureau is fully operative. A combined pay-
ments bureau just began to function at the end of 1996. Most functions are
still being performed by the former governments in the Bosniac and Croat
majority areas, or by cantons and municipalities. Tax revenues continue to
be collected by the former Bosniac and Croat governments, with some of
the revenues being transferred to the cantons and municipalities. The
Federation began receiving customs and excise tax revenues in May 1996.
By the end of September 1996, the Federation had received income of DM
183.9 million and had expended DM 173.4 million. Only DM 13.0 million
has gone toward salaries and materials.
All ten cantons had been formed by the time a law establishing their geo-
graphic boundaries was passed in June 1996. However, most of the cantons
have elected a legislature and president, and are establishing a working
framework that includes constitutions. Only the cantons of Tuzla, Zenica,
and Bihac can be described as functioning. Organization of these three was
well under way before peace, in part because these regions were cut off
from Sarajevo during the war and had, de facto, to set up their own institu-
tions.
Options for the future

It is difficult to be fully prescriptive about the best government structure
for a country where so much change is under way. The complexity of
making such prescriptions is exacerbated by Bosnia’s recent history, which
includes being part of Yugoslavia and participating in the war, and by the
various agreements that have already been made. Still, a series of
recommendations that would enhance the public sector’s operational effi-
ciency can be made.
Shifting functions to the state

A number of functions could be managed much more effectively by the
state than at the entity level. The entities would do well to identify where
efficiency gains are likely, and to seek agreement to transfer the functions
to the state. Examples are the administration of a value-added tax (should
one be legislated), enterprise profits taxes, and customs. The entities could
assign the functions to the state, but both entities would need to agree.
Three conditions need to be in place for delegation to occur: significant effi-
ciency gains must result; transfer must be Pareto-optimal; and there must
be reasonable security that the agreements will not be violated.
Nonetheless, political discord, fears that agreements will not be adhered
to, and envy could lead the entities to reject efficient delegation.
Coordination between the entities

Open borders that allow the free movement of goods and services and
permit the economy to expand are a requirement of the Dayton agree-
ments, but they oblige the entities to cooperate in setting tax administra-
tion and policy. For example, a value-added tax cannot be easily imposed
at a subnational level without internal borders, but such borders would
violate the Dayton agreements. Even a tax or boundary check between the
Serb Republic and the Federation would be inconsistent with rapid eco-
nomic growth, and would create ample opportunity for abuse. One form
of abuse would be imposing a recording fee at the checkpoint that could
quickly be escalated to become effectively a customs duty. The potential
for abuse is equally large in customs administration. Close coordination
must substitute for borders to ease administration of these taxes, unless
administration is delegated to the state.
Given the open economy (even at the borders with Croatia and
Yugoslavia), the tax policy of each entity must also be carefully coordi-
nated with that of the other.11 Differentials between areas in customs,
consumption, and capital tax rates12 provide significant opportunities for
tax avoidance and evasion and will result in concentration of imports, pro-
duction, and sales in the lowest tax areas.13 The best way to limit tax
competition and the resulting resource migration is for the entities to set
tax rates on mobile factors, consumption, and customs which are close to
each other. The coordination could be informal, based simply on observ-
ing the other entity’s tax policies, or more explicit, through negotiations.
More efficient tax policy is likely to result from explicit cooperation, but
there is currently no mechanism in place through which the dialog can
occur.
Size of government
Government must be reduced substantially to a size that is appropriate for
facilitating the private sector’s operation, providing a sustainable, reason-
able safety net, and ensuring delivery of services that are essential to an
acceptable quality of life. The multi-layer government structure raises the
challenge of attaining a smaller size while achieving public sector goals,
but emphasizes the importance of being vigilant in seeking efficiency in
government.
In recent years, budget control has been exercised by limiting spending
to available revenues, with no expectation that planned expenditures will
actually be financed. Budgetary practices need to be rationalized so that
expenditures are set to meet appropriate goals, with the expectation that
all expenditures, including wages and pensions, normally will be met.
Rationalization of spending will require that expenditures for defense, the
safety net, and subsidization of the economy be reduced.
In addition, the budget cannot be expanded to accommodate arrears. A
number of outstanding claims on Bosnia arise from frozen foreign
exchange deposits and unpaid pensions and wages. The total value of
these claims is more than DM 13 billion in the Federation, or about ten
years of current own-source revenues. Recently, foreign exchange bank
books worth nearly DM 5 billion were issued allowing former Republic of
Bosnia-Herzegovina soldiers to make explicit claims against the Republic.
The recognition of explicit claims on a case-by-case basis must be curtailed,
and previous decrees and laws repudiated. It is essential that any settle-
ment of these liabilities should occur using revenues from the sale of assets
(such as could arise from housing and enterprise privatization) and that no
mechanism for settling liabilities should have implications for the budgets
of the state, entities, cantons, or municipalities. Political and equity-based
concerns will be the key factors in deciding the relative compensation paid
for these claims.14
Budgetary savings can be obtained through more efficient operations.
First, better delivery mechanisms must be identified. Diseconomies of
small scale must be overcome through creative means of consolidating
certain functions. The state or inter-entity cooperation should be used

wherever significant efficiency gains are possible. Similarly, cantonal or
municipal cooperation could allow the scale necessary to deliver certain
services efficiently. Healthcare, higher and secondary education, trans-
portation, and environmental services are obvious areas for cooperation.
Privatization of service delivery is an option that permits both the benefits
of private sector incentives and the capacity to expand the geographic area
for service provision.
Second, public sector employment must be scaled down to acceptable
levels. In certain circumstances, public employment can be a stopgap for
job shortages, but it cannot serve as a long-term strategy for job creation or
as the safety net. Public sector employment should be determined by the
number necessary to deliver services efficiently. It is essential that current
administrative staffs be streamlined, and every service be examined for
areas of potential reduction.
Assigning expenditure functions in the Federation

Four criteria are used for recommending the optimal assignment of public
functions among the Federation, cantons, and municipalities:
• the potential for economies of scale
• the existence of interregional spillovers
• variation in preferences
• the desire for equalization and tolerance for cross-subsidies.
A recommended assignment of functions within the Federation (in some
cases including the state) based on the tradeoffs among these criteria is
shown in table 9.2. The Federation should be responsible for areas with
mainly Federation-wide impacts. The Federation’s constitution has
already assigned several functions to the Federation, including defense,
federal police, justice, and customs administration. It would be reasonable
to assign additional functions to the Federation, including university
education, some medical care, intercanton transport, environmental
control, and part of social welfare. Cantons and municipalities will have
responsibility for many functions including much of healthcare, educa-
tion, housing, fire protection, and utility services. Despite the substantial
responsibilities assigned to cantons and municipalities, Federation expen-
ditures are a significant share of total (see table 9.5).
Agreements should be reached between cantons and their constituent
municipalities regarding the specific sharing of responsibilities. The
Federation’s constitution requires that municipalities that are pre-
Table 9.2. Recommended assignment of public functions
Service category Type of service Level of government
Healthcare Primary Municipality

Secondary (hospitals, curative) Canton
Tertiary (infectious disease, research) Entity
Education Primary Municipality
Secondary Canton/municipality
University Entity
Transportation Roads/highways (intracity) Municipality
Roads/highways (intercity) Canton/entity/state
Airports Entity
Public transportation: intracity Municipality
Public transportation: intercity Entity
Private transportation, taxis Canton/municipality
International State
Environmental Air/water pollution Canton
Water/forestry Entity/state
Housing All Canton
Solid waste, water, sewer, fire All Municipality/canton
Land use/zoning All Municipality
Licensing/regulation All Canton
Cultural policy All Canton
Tourism All Canton
Social welfare All Canton/entity
Telecommunications All Entity/state
dominantly composed of a minority population in the canton be permitted

to deliver services locally. A key problem is determining the revenues, that
otherwise would go to the canton, that are to flow with the expenditures.
Minority municipalities must receive a revenue allocation that legitimately
allows them to meet their service responsibilities, without creating unrea-
sonable fiscal strain on local citizens. Careful design of the revenue
arrangements is important because ethnic problems often arise locally, yet
can frequently be solved if much of the power for decision-making is local
(Tishkov, 1993). Cantons could devise transfer schemes that would make
it very difficult for minority municipalities to finance service delivery,
which could lead to strong disagreements.
Tax assignment in the Federation

As a rule, canton and municipal revenues should be based on a combina-
tion of assigned revenue instruments and grants, with the goal of ensuring
vertical balance. This means that each level of government would have
sufficient revenue capacity to fund the services for which it is responsible.
Furthermore, tax and grant assignments must be made with an eye not
only on the funding of current service levels but also on future service
needs. That is, revenue buoyancy must be carefully considered when
assignments are made. Each level of government – the Federation, cantons,
and municipalities – must have sufficiently buoyant revenue sources to
provide the amounts needed to pay for the delivery of services over time.
As noted above, it is important to differentiate among the separate steps
of the taxation process when making tax assignments. These steps include
tax administration, definition of tax bases, setting of tax rates, and tax own-
ership. The level of government in charge of collecting a tax need not be
the one that defined the tax base, set the tax rate, or owns the tax revenues.
Locally owned taxes (cantonal or municipal taxes) need not be locally
administered, and federally imposed and collected taxes can be – and in
other countries often are – shared with other levels. Alternatively, taxes
could be collected at the canton level and shared with the Federation. In
fact, it is recommended that for the most important taxes, the definition of
the tax base and possibly the setting of the tax rate be done at the
Federation (and ultimately the state) level, while allowing cantons and
municipalities some flexibility over revenue instruments. Economic
integration, tax compliance, and tax administration would be greatly
enhanced by nationally uniform bases and rates for customs, excise, cor-
porate and personal income taxes, and sales or value-added taxes.
However, cantons must be protected from the Federation exempting tax-
payers or narrowing the base for taxes where the Federation has little or
no revenues at stake. A proposed set of assignments for the Federation is
in table 9.3.
Three criteria must be considered in determining the appropriate tax
ownership assignments. First, tax assignment must bear a strong corre-
spondence to the expenditure responsibilities of each government level
to ensure that adequate financing is available to deliver required services.
A perfect relationship between expenditure responsibilities and tax
revenue ownership is not required (and probably is not desirable)
because grants can be made from a government that owns revenues to
assist another government in financing its expenditure responsibilities.
Second, tax assignment must not violate the Constitution, the Dayton
agreements, and other international agreements. Third, tax ownership
must be consistent with the administrative ability to measure where the
taxable activity occurs, if revenues are to be distributed on a sites-of-
collection basis. For example, ownership of customs taxes based on
Table 9.3. Summary of suggested tax assignments
Tax type Tax administration Tax base Tax rate Revenue ownership
Customs Federal or state Federal or state Federal or state Federation and state
Excises Federal Federal legislation Federal legislation; surcharges set by Federation; cantons or
cantonal or municipal assemblies municipalities may impose
retail level taxes
Wage and personal income Federal or canton Federal legislation Basic rate set by the Federal Shared by the Federation,
legislation; surcharges set at the cantons, and municipalities
cantonal level
Corporate income Federal Federal legislation Federal legislation Federation
Sales tax (retail level) Federal; feasible but Federal legislation Rates to be set by Federal legislation; Federation, cantons and
complicated at the surcharges can be set by cantons municipalities
cantonal level
Property tax Municipal Federal legislative Municipal level Municipal level
framework
Real estate transfers Municipal Federal legislative Municipal assemblies Municipal level
background
Motor vehicles Municipal Federal legislative Cantonal level Cantons and municipalities
framework
Social security contributions Federal Federal legislation Federal legislation Funds
User fees, social services Service-providing Cantonal legislation Cantonal or municipal providers Service-providing agencies
agencies
User fees, utilities Service supply Regulatory Service-providing companies subject Service-providing
companies frameworks to regulatory oversight companies
VAT Federal Federal legislation Rates to be set by Federal legislation Shared between the
Federation and the cantons
Note: In this table, tax assignments are suggested between the state and the Federation’s government. Similar tax assignments between the state
and the Serb Republic entity could be suggested.
Table 9.4. Tax revenue assignments for the Federation, 1997 (percentages)
Revenues State Federation Cantons Total
Customs duties 7.4 92.6 0.0 100.0

Excise and special import duties 0.0 100.0 0.0 100.0
Wage and income taxes 0.0 50.0 50.0 100.0
Profits tax 0.0 50.0 50.0 100.0
Sales tax 0.0 0.0 100.0 100.0
Other taxes 0.0 0.0 100.0 100.0
Total revenues 1.3 40.0 58.7 100.0
Table 9.5. Tax revenue distribution for Bosnia, 1997 (million DM)
State Serb Republic Federation Cantons Total
Revenues
Customs duties 25.0 39.7 208.3 0.0 273.0
Excise and special import taxes 0.0 0.0 225.0 0.0 225.0
Wage and income taxes 0.0 60.0 20.0 20.0 100.0
Profits tax 0.0 0.0 70.0 70.0 140.0
Sales tax 0.0 78.0 0.0 500.0 578.0
Other taxes and fees 15.0 21.0 0.0 179.0 215.0
Total revenues 40.0 198.7 523.3 769.0 1 531.0
Expenditures
Expenditures (except army, 40.0 116.7 163.3 769.0 1 089.0
social programs and debt
service)
Expenditures: army, social 0.0 82.0 360.0 0.0 442.0
programs and debt service
Total expenditures 40.0 198.7 523.3 769.0 1 531.0
Note: The Federation and Serb Republic data do not include special funds for healthcare,
unemployment, and pensions. The Serb Republic data do not include municipal revenues.
where the imports are consumed is administratively impractical at the

municipal or canton level.
The Federation needs to take a number of important steps before tax
ownership can be determined precisely and implemented effectively.
Expenditure assignments must be structured before final tax assignments
can be made. Also, the full set of Federation tax laws need to be enacted
and enforced. A mix of laws from the Federation and the old Croat and
Bosniac areas is currently being enforced on the Federation side.
Tax assignments that would result in vertical balance, given expected
1997 expenditure assignments, were estimated (tables 9.4 and 9.5).15 The
assignments assume that no intergovernmental transfers occur, other than
those implicit in the assignments. The state needs to receive 7.4 percent of
Federation and 17.3 percent of Serb Republic customs tax revenues to
finance its expenditures. The difference in shares illustrates the effect of
having one-third of the state financed by the Serb Republic and the other
two-thirds financed by the Federation. No other detail was estimated for
the Serb Republic.
In the Federation, the remainder of customs taxes, and excise, special
import and enterprise profits taxes are assigned to the Federation. Wage,
income, and profits taxes are assumed to be split evenly between the
Federation and the cantons.16 The cantons and municipalities receive the
range of smaller taxes including the real estate, judicial and advertising
taxes.
The revenue assignments were proposed in order to create vertical
balance in the system, but several important cautions should be noted.
First, the assignments will allow horizontal imbalances at the canton and
municipal levels. Estimates suggest several cantons would have signifi-
cant deficits, arising from both the expenditure and revenue sides.
Revenues are constrained in some areas because of very weak economies
and expenditure needs are larger in some places than others. Second, the
expenditure assignments used here are similar to those in place for 1996,
except most spending for health, education, and police services is passed
to the cantons. Changes in the expenditure assignments will necessitate
changes in revenue assignments. Third, very minimal grants from the
Federation to the cantons are allowed for in the financing scheme. Own-
source revenues should be the major financing source because of account-
ability, concerns about cross-subsidies, and differences in service
demands, and for other reasons. Still, a greater role for grants exists, and
development of a transfer system would require additional tax assignment
to the Federation. Finally, the assignments fail to recognize differences in
revenue buoyancy, and will have to be altered over time if expenditures
grow at different rates from revenues. Subnational governments are often
assigned taxes with low buoyancy. However, more than 60 percent of the
revenues assigned here are from the relatively buoyant sales and wage
taxes.
Most taxes, including the corporate and personal income, excise,
customs, and value-added taxes would be best administered at the entity
level, and in several cases at the state level. A limited number of revenue
sources, such as the property tax and user fees, could be administered at
the subentity level. Taxes are administered at the entity level in the Serb
Republic and legislation has been passed for administration at the
Federation level as well. Advantages in terms of lower operational costs,
coordination of administration, and uniform practices all argue strongly in

favor of the Federation. The key is to design an administrative and revenue
distribution system that safeguards the cantons and municipalities so that
they will be willing to accept Federation administration.
Conclusion
Progress in developing the governments in Bosnia has been slower than
was envisioned when the Dayton agreements were penned. This is not sur-
prising since there is little or no world experience of developing govern-
ments in the short period of time assumed in the agreements. There are
positive signals in Bosnia, such as a number of government institutions
becoming operative and revenues being passed to both the state and the
Federation. Still, the challenge of achieving fully operational governments
remains.
Traditional economic models of federalism, like that on which this
chapter’s analysis is based, suggest a government structure assuming
there is an intent to achieve Pareto efficiency for the entire country. Current
attitudes in Bosnia suggest that many people in each ethnic group see
themselves as members of their group, rather than as Bosnians, and are not
broadly concerned about the welfare of the whole country or access to
public services outside their group. The motivation for the fiscal federal-
ism structure proposed in the Dayton agreements is better interpreted as
an effort to manage conflict between the ethnic groups. Federalism in a
conflict management sense does not require that each group be given its
own state, but does lead to the conclusion that institutions of power should
be brought closer to people so that decision-making can be more sensitive
to the different ethnic groups (Tishkov, 1993). Decentralization in this
context is a means to lessen the points where disagreement exists, rather
than a structure to obtain economic efficiency. Common institutions at the
state, entity or canton levels are maintained, but only for functions that
must be broader in scope. The fiscal (and other) interdependencies from
common institutions at the state, entity, and canton levels provide oppor-
tunities to build relationships and trust over time.
The government structure included in Dayton is workable. However,
unless governments negotiate other arrangements, the likely outcome in
the short to medium term is diseconomies of scale in provision of certain
services, as some that can be provided at lower unit costs at the state level
are provided at the entity level, and some with lower unit costs at the entity
level are delivered at the canton or municipal level. Services with geo-
graphic spillovers will be underprovided because governments will fail to
account adequately for the benefits received by members of other ethnic

groups. Further, little consideration will be afforded to equity among
groups, resulting in widely different access to services across the country.
Better service delivery mechanisms from a national, Pareto-efficient per-
spective will not be selected given the very strong distaste for cross-subsi-
dies and the fear by minority groups that they will be dominated by a
larger group. More efficient arrangements can be expected to evolve over
time as confidence in the government structure grows.
Horizontal imbalances will be large in the Federation, and grants will be
used to a very limited extent to offset these imbalances. The resulting prob-
lems will be lessened in the next several years to the extent that the former
Bosniac or Croat governments are able to provide some resources to the
more financially deprived areas, or the donors help fund infrastructure
and other needs. But donors will not finance operating expenditures, and
maintenance of new infrastructure investments will place greater
demands on limited budgets, so financial problems in the places least able
to pay could grow. Economic growth that increases revenue-raising capac-
ities across Bosnia is one approach to offsetting imbalances. Also, differ-
ences in access to services could be lessened with a system that allocates
more grants to places with the greatest expenditure needs and with the
least revenue capacity.
Notes
1 Bosnia’s prewar population was 4.5 million, of which some 3 million have had
to leave their towns and villages, with 1 million estimated to be refugees
abroad, mostly in Western Europe. Another 250,000 people are dead or missing.
2 It could be argued that the concentration of industrial conglomerates, made
non-viable because of transition shocks in Bosniac-majority areas, would have
led to some decline in the tax bases of this area even without the war.
3 In the short term, this is expected to be donor-financed, but as the economy
recovers, a domestic contribution to infrastructure finance is anticipated.
4 McLure (1994) has made these observations about the service provision differ-
entials in the Russian Federation and the issue of whether natural resource rev-
enues “belong” to all Russians, or to the territories (oblasts) in which they are
located. The same observations about preferences and tolerances for redistribu-
tion also apply to Bosnia.
5 There will be a one-time opportunity to issue currency when the currency board
is established, from which the state may accumulate seignorage revenue.
However, the revenue is expected to be limited, and cannot be relied upon as a
continuing revenue source. While the possibility of seignorage revenue is
removed by the establishment of a currency board, the state may earn some
revenue through interest earnings on foreign exchange reserves. In the interim
and until a new currency is in place, the seignorage from the circulation of the
Deutschmark in the country accrues to the Bundesbank in Germany (and of the
kuna to the Croatian Central Bank).
6 Kosovo along with Voivodina, which is largely Hungarian-populated, is now a
dependent province within Serbia-Montenegro.
7 78 percent of Federal Fund resources were mobilized from the three better-off
republics in the 1970s, falling to 75 percent in the 1980s (Federal Statistical Office
of Yugoslavia). Gross Social Product may be thought of, very roughly, as
equivalent to the more familiar GDP or GNP measures.
8 The financial channel was effected in two ways. First, commercial banks in each
republic could supplement their deposit resources by borrowing from their
republic branch of the National Bank of Yugoslavia. Empirical studies suggest
that the central bank did play a role in the regional allocation of credit through
this mechanism, although it is difficult to get a consistent picture of who were
the gainers and who were the losers (World Bank, 1989). Second, the commer-
cial banks in each republic met the financial losses of enterprises which
amounted to 5–6 percent of GSP in the latter part of the 1980s. The inflation this
process gave rise to was felt by all, but the “benefits” of the system (financing
of losses) were concentrated in Bosnia-Herzegovina, Croatia, and Serbia (World
Bank, 1989).
9 The peculiar Yugoslav institution, the SDK, (the Social Accounting Office),
which exists in a somewhat truncated form in Bosnia today, makes it possible to
pinpoint closely where consumption, employment, and production take place
and where taxes should be collected. So far this system has been inadequate for
generating confidence in the governments getting their appropriate shares.
10 Constitution of Bosnia and Herzegovina, Annex 4, Article III, Section 2(b).
11 Coordination between the Serb Republic and Yugoslavia and between the
Federation and Croatia is already well under way.
12 The Dayton agreements can be interpreted as requiring common customs tax
rates, but the entities are free to set the other tax rates.
13 Labor migration in response to tax differentials is not a significant issue in the
current environment, given the lack of freedom of movement between entities.
14 These claims do not include the international debt of Bosnia.
15 Revenues and expenditures were estimated separately. Expenditure require-
ments were based on budgeted and actual 1996 expenditures and assumed
growth rates for the state and Federation. Additional transfers to the state are
necessary if the state is to remit debt amortization payments. Canton and
municipal expenditures were estimated using actual 1995 expenditures for the
various services, and assumed growth rates. Revenue estimates for 1997 are
based on actual collections in 1995, the first nine months of 1996, and assumed
growth rates. No attempt is made to estimate the separate revenues of
municipalities and cantons in the Federation. Defense, social programs, and
debt service are used to balance expenditures and revenues and do not include
off-budget expenditures.
16 The split wage tax could be designed as a piggybacked wage tax, as has been
proposed by the IMF.
References
Bird, Richard M., Ebel, Robert D., and Wallich, Christine I., eds. (1995),
Decentralization of the Socialist State: Intergovernmental Finance in Transition
Economies, Washington, DC: World Bank.
Ding, Wei (1991), “Costs and benefits of union and interdependence of regional
economies,” Unpublished paper, World Bank, Washington, DC.
Farsad, Mansour (1991), “Yugoslavia: economic integration and viability of a
breakup,” draft Working Paper, Eastern Europe Department, World Bank,
Washington, DC.
Federal Statistics Office of Yugoslavia, Yugoslavia, annual.
McLure, Charles E., Jr. (1994), “The sharing of taxes on natural resources and the
future of the Russian Federation,” in Christine I. Wallich, ed., Russia and the
Challenge of Fiscal Federalism, Washington, DC: World Bank.
OECD (1992), “Regional developments and developments in the republics of
Yugoslavia,” March: 36.
Prud’homme, Rémy (1994), “On the dangers of decentralization,” Policy Research
Working Paper No. 1252, World Bank, Washington, DC.
ciency and macroeconomic aspects,” in Michael Bruno and Boris Pleskovic,
eds., Annual World Bank Conference on Development Economics, Washington, DC:
World Bank.
Tishkov, Valery A. (1993), “Nationalities and conflicting ethnicity in post-commu-
nist Russia,” Working Paper series, Conflict Management Group, Harvard
Law School, Cambridge, MA.
Vodopivec, Milan (1991), “Redistribution through the soft budget constraint,”
World Bank (October).
Woodward, Susan (1995), Balkan Tragedy, Washington DC: Brookings Institution.
World Bank (1989), “Yugoslavia: financial sector restructuring: policies and prior-
ities,” Report No. 7869-YU, World Bank, Washington, DC (November).
(1996), Bosnia and Herzegovina: Toward Economic Recovery, World Bank Country
Study, Washington, DC: World Bank.
Index
agency theory 23 Bahl, R. 2, 5, 16

Ahmad, J. 5, 14, 25, 34, 253, 256 Bird, R. 1, 4, 5, 8, 14, 17, 18, 25, 26, 33, 37, 72,
All India Financial Institutions 97 169, 170, 206, 207, 209, 276
apartheid 239, 239–46 Bosnia-Herzegovina 5, 21, 25, 27, 271–98
Black Local Authorities (BLAs) 240, break-up 276–77
242–43, 246, 253, 257, 260 centrifugal forces 271, 275
consequences 239–46 cross-subsidies 275, 279
homelands 241, 246 Dayton system 271, 272, 277–88
Johannesburg 244, 253, 257–63, 268 debt policy 286–87
local governments 241–42 delegation 289
national context 240–41 expenditure functions 291–92
regional governments 241 expenditure policy 281–82
Regional Service Councils (RSCs) 240, expenditure rules 280–81
245–46, 260 Federation: Bosniac-majority area 272,
Self-Governing Territories (SGTs) 240, 273, 279, 281, 286, 288; Croat-majority
241, 246, 247 area 272, 278, 279, 281, 288, 289
spatial policies 239, 241–46 finance 282–87
TBVC (Transkei, Bophutaswana, Venda, fiscal equalization 273–74
and Ciskei) 240, 241, 246, 247 government size 290–91
transfers 241 lessons 277
White Local Authorities (WLAs) 240, 242, Serbia-Montenegro 272
243–45, 253, 257, 260, 261 Serb Republic (Republika Srpska) 272, 278,
Argentina 5, 8, 21, 26, 31, 192, 206–32 279, 280, 281, 282, 288, 289
Buenos Aires 208, 220 tax administration 282–87
decentralization 214–21 tax assignment 292–97
delegation 225 vertical balance 293
devolution 232 Yugoslavia 272, 274–79, 283, 288
equalizing criteria 232 bottom-up approach, see demand-driven
federal fiscal arrangements 214–21 approach
fiscal decentralization, 206–08 borrowing ability 6
Fiscal Disequilibria Fund 223 Brazil 5, 18, 33, 192
fiscal gap 221–24 Britain 23
fiscal imbalance: horizontal 224–25; British rule, in Pakistan 116, 127, 145
vertical 221
Regional Disequilibria Fund 223 Canada 18, 249
revenue-sharing 213, 215 Chile 31
principal–agent model 230 China 5, 21, 22, 23, 25, 26, 27, 49–74, 134
public choice approach 230 central control 63
revenue-sharing system 231 centralization 72–74
setting 208–14 declining tax ratio 57–58
Stabilization Program 221 division of revenues 59
Australia 18, 21, 33, 34, 249 expenditure variations 68
301
302 Index
China (cont.) demand-driven (bottom-up) approach 3, 5,
earmarked grants 55–56 10
enterprise income tax 50–51 in Bosnia-Herzegovina 276
Finance Commission 82 in Indonesia 126
fiscal disparities 66 Denmark 31
indirect taxes 51–53 devolution 3, 12, 207
local government 63–65; autonomy of 65 in Argentina 206, 215, 232
macroeconomic policy 59–62 in Morocco 169
National Tax Service 53 in Tunisia 169
redistribution 68–69 see also public choice approach
reform 62–72; objective of 57–62
revenue capacity 66 equalization, horizontal 31
revenue-sharing 53–57 in India 95
revenue structure 49–52 equalizing criteria, in Argentina 232
shared taxes 54–55 European Union 5, 78, 207
tax effort 69–71
Colombia 5, 6, 8, 9, 10, 21, 25, 30, 34, 172–98, federal finance 23–28
265 “federal–provincial diplomacy,” in Canada
Antioquia (Medellin) 174 25
Bogotá 174 fiscal decentralization 1–40, 206–08
Commission on Intergovernmental conditions for success 12–14
Finance 173 definition and assessment 3–5
deficits 191–94 federal finance 23–28
earmarking and expenditure grants fiscal federalism 23–28
186–89, 195 international comparisons 16–22
education 181–83 lessons 34–40
fiscal effort 189–91 local capacity and taxation 8–12
fiscal flexibility 189–91 macroeconomic aspects 5–8
general purpose grant 184–86 patterns 22–28
health 181–83 transfers, 28–34
infrastructure 183–84 fiscal dependency 91
intergovernmental transfers 177–79 fiscal effort 189–91
lessons 195 “fiscal effort hypothesis” 190
participaciones municipales (PM) 177, 178, fiscal federalism 23–28
186 fiscal flexibility 189–91
policy objectives 179–86 fiscal imbalance
sistema nacional de cofinanciación (SNC) horizontal 11; in Argentina 224–25; in
177, 179 India 92–94, 109
situado fiscal (SF) 177, 178, 192 vertical 11, 28; in Argentina 221; in India
subnational governments 175–77 81, 89–92; in Pakistan 140–41
transfers, 191–94; design of 179–86 “fiscal laziness,” see substitution effect
Valle (Cali), 174 Fiszbein, A. 4, 14, 265
competitive equality 87 “flypaper effect” 190, 192
competitive federalism 81 France 23
“coparticipation” system (Colombia) 9 French model 152, 155
cost-benefit appropriability 87 Fox, W. 5, 25
Czechoslovakia 28
“gap-filling” approach 100–02
decentralization theorem (Oates’) 24, 80, Germany 18, 21, 25, 31
218, 220
deconcentration 3, 23, 207 hard budget constraint 13, 14
in Indonesia 126 in Colombia 196
in Pakistan 127 horizontal fiscal balance, see equalization,
delegation 3, 13, 23, 207 horizontal
in Argentina 206, 215, 225 horizontal fiscal imbalance, see under fiscal
in Bosnia-Herzegovina 289 imbalance
see also principal–agent model
Index 303
India 5, 18, 78–110 tax revenues 156–67
assignment question 80–86 transfers 157–60, 162–71
central sector 103–05
economic rationale for transfers 94 “need–revenue” gap 96
equalization 95 Netherlands: Dutch rule in Indonesia 116,
equalizing effects 105–08 145
expenditure assignments 80–86 New Zealand 255
federal fiscal arrangements 108–10
Finance Commission 97–108, 110; Oates’ theorem, see decentralization
transfers 100–02 theorem
fiscal disharmony 89–94 OECD 17
fiscal gap 95 “own-source” revenues 11
fiscal harmony 86–87
fiscal imbalances: horizontal 92–94; Pakistan 10, 18, 79, 115–16, 126–51
vertical 81, 89–92 bureaucratic factors 145–46
Indian National Congress 79 decision-making 131
intergovernmental competition 86–87 economic evaluation of transfers 139–40
intergovernmental transfers, 96–108 external participants 147
interstate competition 87–89 federal–provincial transfers 136–39
plan transfers 102–03 fiscal imbalance, vertical 140–41
Planning Commission 79, 97–108, 110 institutional factors 146–47
spillovers 96 Karachi Metropolitan Corporation 127
tax and expenditure overlapping 86–89 lessons 148–51
transfers 94–108 National Finance Commission 134, 137
Indonesia 9, 21, 23, 24, 115–26, 144–51 political factors 144–45
budgeting 126 private sector 129
bureaucratic factors 145–46 provincial structure 128–29
expenditure assignment 118–20 provincialization 126–44
external participants 147 public sector accountability 132
fiscal system 116–26 public sector strength 132
institutional factors 146–47 public service efficiency 131–32
Instruksi Presiden (INPRES) 123, 125, 126 public service equity 131–32
intergovernmental loan finance 123–25 reform 144–51
Jakarta 18 subnational government 143–44
lessons 148–61 taxing powers 132–36
local borrowing 125 transfers 136–43
local user charges 122–23 “post-welfare state” 1
monitoring 126 principal–agent model 25, 206, 207
planning 126 in Argentina 230
political factors 144–45 see also delegation
reform 144–51 public choice approach 208; see also
regional taxes 121–22 devolution
tax base assignment 120–21
transfers 123–26 Rao, G. 5, 14, 18, 86, 89
International Monetary Fund 53, 286, 287 rent-seeking 115
in Indonesia 115
Linn, J. 2, 16 in Pakistan 115
revenue-sharing system 231, 232
moral hazard problem 264, 266 Rezk, E. 18
Morocco 6, 7, 8, 21, 23, 25, 31, 152–71 Russia (see also Soviet Union) 5, 18, 21, 23,
communal system 155 28, 134, 283
funding mechanism 156–60
institutional setting 152–56 Shah, A. 4, 18, 25, 115, 116
intergovernmental finance 156–66 South Africa 5, 6, 8, 18, 21, 25, 26, 239–68
municipalities 154, 155–56 apartheid 239, 239–46
sociodemographic characteristics 152–53 Black Local Authorities (BLAs) 240,
subnational governments 153–55 242–43, 246, 253, 257, 260
304 Index
South Africa (cont.) Sweden 34
capacity and decentralized responsibility Switzerland 18, 21, 25
265
Cape Town 257–63 Tanzi, V. 207, 274
central fiscal transfers 260 tax
cooperative governance 263–64 principle of concurrence 85
efficiency considerations 254 principle of separation 82, 85
expenditure assignment 248 Tishkov, V. 292, 297
expenditure responsibilities 258–60 top-down approach, see supply-driven
Financial and Fiscal Commission 14, 249, approach
250–51, 263 transfers
fiscal balance: horizontal 251; vertical 251 in Colombia 177–86, 191–94
fiscal imbalance: horizontal 246; vertical in India 94–108
246 in Indonesia 123–26
fiscal stability 255 in Morocco 157–60, 162–71
homelands 241, 246 in Pakistan 136–43
horizontal inequality 249–50 role and design 28–34
income redistribution 254–55 in Tunisia 162–71
Johannesburg 244, 253, 257–63, 268 Tunisia 6, 18, 21, 23, 152–71
local governments 241–42 allocation formulas 165
KwaZulu 247 communal system 155
municipal services surcharges 261–63 direct taxes 160–61
Natal 247 envelope 164
national context 240–41 funding mechanism 160–62
payroll tax 260 institutional setting 152–56
political considerations 253–54 indirect taxes 161–62
provinces, new 246–52; and metropolitan intergovernental finance 156–66
governments 267–68 local governments 162–64, 169–70
property tax 260–61 local investment 166–68
public financial intermediary (PFI) 255 municipalities 154, 155–56
racial jurisdictions 246–48 own revenues 165–66
regional governments 241 sociodemographic characteristics 152–53
Regional Service Councils (RSCs) 240, subnational governments 153–55
245–46, 260 transfers 162–71
Self-Governing Territories (SGTs) 240,
241, 246, 247 United States 18, 249, 255
spatial policies 239, 241–46
TBVC (Transkei, Bophutaswana, Venda, Vaillancourt, F. 2, 87, 89, 170
and Ciskei) 240, 241, 246, 247 vertical fiscal imbalance, see under fiscal
tax assignment 260 imbalance
transfers 241 Vietnam 18, 22, 23, 26
urban governance 252–57
White Local Authorities (WLAs) 240, 242, Wallich, C. 1, 5, 26, 276
243–45, 253, 257, 260, 261 Wiesner, E. 191, 208
Soviet Union (see also Russia) 28, 78 Wildasin, D. 1, 266, 267
Spain 34 World Bank 8, 23, 25, 26, 52, 53, 64, 176, 178,
“stimulative effect” 192 182, 184, 185, 189, 191, 195, 273, 278,
“substitution effect” 192 286, 287
supply-driven (top-down) approach 4, 5,
10, 23 Yugoslavia 28, 272, 274–79, 283, 288
in Indonesia 126

Richard M. Bird, Francois Vaillancourt-Fiscal Decentralization in Developing Countries (Trade and Development) (1999) PDF

Uploaded by

Copyright:

Available Formats

Richard M. Bird, Francois Vaillancourt-Fiscal Decentralization in Developing Countries (Trade and Development) (1999) PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Richard M. Bird, Francois Vaillancourt-Fiscal Decentralization in Developing Countries (Trade and Development) (1999) PDF

Uploaded by

Copyright:

Available Formats

Fiscal Decentralization in Developing Countries

Recent years have seen a worldwide trend toward ﬁscal decentralization. In

Richard M. Bird is Professor of Economics, University of Toronto and a

François Vaillancourt is Professor of Economics, Université de

Other titles in the series

Cambridge University Press

© Cambridge University Press 1998

This publication is in copyright. Subject to statutory exception

First published 1998

Library of Congress Cataloguing in Publication data

Fiscal decentralization in developing countries / edited by Richard M. Bird, François

ISBN 978-0-521-64143-2 hardback

List of tables page viii

1 Fiscal decentralization in developing countries: an overview 1

1.1 Fiscal decentralization: the case-study countries page 19

4.1 A comparative impressonistic review of selected aspects of

7.11 Provincial public employment and spending on personnel,

Junaid K. Ahmad, World Bank

PARADI/ICTS* Conference held in Montreal, Canada on September

* The Programme d’analyses et de recherches économiques appliquées au développement

Fiscal decentralization in developing countries:

Recent years have seen worldwide interest in ﬁscal decentralization.

economic view is that the existence of different tax-spending packages in

Fiscal decentralization: what and why?

Deﬁnition and assessment

responsive and responsible government in general, it is not surprising that

decentralization. The bottom-up perspective may be particularly

Macroeconomic aspects of decentralization

may give rise to particular concern in developing countries where local

decentralizing public borrowing in the ﬁrst place – namely, risk diversiﬁca-

very limited taxing powers available to subnational governments that are

Local capacity and taxation

in other localities. Municipalities have started to share certain equipment.

sharing from even the poorest communities – usually their contribution

governments should be insulated from either the tax base consequences of

Conditions for success

the delegation of implementation responsibilities to local bodies may still

Just as accountability is the key to improved public sector per-

A quantitative overview of ﬁscal decentralization

Some international comparisons

average of 23 percent, was accounted for by local governments in the

and Slack, 1991). Countries clearly have considerable discretion in

Notes and sources:

Australia Brazil Canada Germany Russia Switzerland USA Vietnam

Notes and sources:

Broadly and with some overlapping of categories, ﬁve types of countries

Patterns of ﬁscal decentralization

Table 1.3. A taxonomy of ﬁscal decentralization

Fiscal federalism Federal ﬁnance

Developed countries France, Japan USA, Canada

Fiscal federalism and federal ﬁnance

Although ﬁnancial autonomy – the share of local expenditures ﬁnanced by

United States (at least in this century) or Switzerland to Germany (Bird,

to overcome the familiar agency problems of information asymmetry and

countries. Not only is it difﬁcult to restructure incentives constructively

The role and design of transfers

(3) Expenditure functions could be transferred up to the level with more

Many other countries (including most of the transitional countries) have

in the same gap-ﬁlling sense as vertical ﬁscal balance, an equalization