THE WORLD BANK S E C T O R S T U D I E S S E R I E S 21491 ON Co t-i o m. I r-1 p -o ~- I c: September 1 998 LOIJC A iON E ?o E' R -A . .$1 .N f.'G 8* - ) i! PA UN , C AT Ns Rethinking Decentralization in E Ni vO - N N Developing Countries Fij.ANWACIA- SECTOR D r V E L O P ME N N T E N ~) E N4 EA.tLT B. N 1T R; T i?.N aSiW, Po * . L AT ON!r 4 PR'I A r S DEOVELrJ?T-Y Tl PUBLIC SECTOR JENNIE LITVACK R u P i)E V f L O M-`EN t JUNAID AHMAD RICHARD BIRD S0CL. DEVELOP? rE;T FILE COPY _, P.15,RO E?:c-i-70N TrANS?POiRTATiON. WVAT-EPF? .k ,JiA D F ,E. L C, ?, FP D Poverty Reduction & Economic Management S E C T O R S T U D I E S S E R I E S Rethinking Decentralization in Developing Countries JENNIE LITVACK JUNAID AHMAD RICHARD BIRD THE WORLD BANK WASHINGTON, D.C. Copyright © 1998 The International Bank for Reconstruction and Development / THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing September 1998 The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility for any consequence of their use. The boundaries, colors, denominations, and other information shown on any map in this volume do not imply on the part of the World Bank Group any judgment on the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to the Office of the Publisher at the address shown in the copyright notice above. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to copy por- tions for classroom use is granted through the Copyright Clearance Center, Inc., Suite 910, 222 Rosewood Drive, Danvers, Massachusetts 01923, U.S.A. Jennie Litvack is senior economist, Public Sector Group, Poverty Reduction and Economic Management, The World Bank. Junaid Ahmad is the Bank's deputy resident representative, South Africa Resident Mission. Richard Bird is professor of economics, University of Toronto. ISBN 0-8213-4350-5 Library of Congress Cataloging-in-Publication Data (CIP) has been requested. Contents Foreword v Abstract vii 1. Why Should We Worry about Decentrahzation? 1 2. Decentralization: A Complex Phenomenon 4 Why Is Decentralization Happening? 4 What Is Decentralization? 4 What Are the Implications of Decentralization? 6 Successful Decentralization Depends on Institution-Specific Design 7 3. Designing Decentralization: The Building Blocks of Fiscal Federalism 10 Assigning Expenditure Responsibilities 10 Raising Revenue 11 Overseeing Subnational Borrowing 12 Designing Intergovernmental Transfers 12 Assembling the Building Blocks: A Systems Approach 13 4. Designing Decentralization: Incorporating Institutions 16 Adjusting the Regulatory Framework and Decentralizing Borrowing 16 Organizing Service Delivery 20 Establishing Information Systems and Competitive Governments 21 Adopting Asymmetric Decentralization 23 Synchronizing Policy 24 5. Rethinking Decentralization in Developing Countries 26 Incentives for Accountability 26 The Role of Capacity 27 Policy Dialogue and Project Design 28 More Case Studies, Data, and Research 29 Annex. World Bank Economic and Sector Work on Decentralization, 1988-Present 32 Notes 35 References 37 iii Foreword Central governments around the world are decentral- be able to identify more precisely what works and what izing fiscal, political, and administrative responsibili- doesn't in particular institutional settings. And by iden- ties to lower-level governments and to the private tifying what institutions are important for successful sector. Political pressure probably drives most de- decentralization in industrial countries and how these centralization efforts. But whatever the origins, de- institutions differ in developing countries, we can find centralization can have significant repercussions for ways to compensate for weak institutions in the short resource mobilization and allocation and ultimately for run and to build the basic elements of key institutions macroeconomic stability, service delivery, and equity in the long run. For these reasons, the World Bank is actively involved This paper, a product of the Public Sector Group and in understanding the design and impact of decentral- the Decentralization Thematic Group in the Poverty ization policy in many developing countries. Reduction and Economic Management (PREM) Net- Much of the literature on decentralization, norma- work, is intended to stimulate thinking and encourage tive and empirical, is based on industrial countries. a more nuanced approach to decentralization policy Developing nations, however, have very different insti- This approach recognizes the importance of institu- tutional frameworks. Given that the World Bank's work tions and policy design in determining the impact of is exclusively in developing countries, it must careful- decentralization on efficiency, equity, and macroeco- ly consider what unintended consequences these insti- nomic stability It implies the need to address decen- tutional differences might have on decentralization tralization within a broader institutional assessment. policies and what the implications are for project The paper draws on a wide body of literature and design and policy dialogue. recent experiences documented by the World Bank We do not know enough empirically to make defin- and proposes a framework for improving the design of itive recommendations about which types of decen- decentralization by incorporating information about tralization are best for which services in which institu- country-specific institutions. The paper is the product tional settings. But we do know that the best design will of a partnership between an academic researcher, oper- vary depending on circumstances and institutions and ational staff, and network staff, an exchange that has that this complexity has sometimes been overlooked in already enriched the policy dialogue in several coun- the haste to offer policy advice. If we can develop a tries. It is expected that PREM and other networks will common framework, we can begin to document cases continue to join in the policy dialogue through similar of decentralization, so that within a few years we may partnerships in the future. Masood Ahmed Vice President Poverty Reduction and Economic Management Network v Abstract This paper draws on the literature and growing expe- ernments, asymmetrical decentralization, and policy rience with decentralization in developing countries synchronization. to explore how a wide range of variables can affect The paper's starting point is the traditional fiscal fed- decentralization efforts and how policies and incen- eralism approach. But the primary measures for local tives can be designed to improve outcomes. The and central accountability assumed in most discus- paper highlights the fact that decentralization is nei- sions of decentralization may not hold or are different ther good nor bad for efficiency, equity, or macroeco- in many developing countries. Drawing on the evi- nomic stability; but rather that its effects depend on dence from the World Bank's operational work, there- institution-specific design. It discusses the building fore, the paper suggests the need for a stronger focus blocks of fiscal federalism (expenditure and revenue on institutions in designing decentralization policies. assignment, intergovermental transfers, and subna- This broader agenda suggests an enhanced focus on tional borrowing) and then discusses five means accountability, governance, and capacity in the context through which decentralization policy and institu- of designing policies for decentralization. This tions interact. These are the regulatory framework for approach has strong implications for the Bank's project subnational borrowing, the financing and delivery of design and policy dialogue and calls for a reinvigorat- services, information systems and competitive gov- ed research effort focused on developing countries. vii 1. Why Should We Worry about Decentralization? Central governments around the world are decentraliz- decentralization can greatly affect economic develop- ing fiscal, political, and administrative responsibilities ment and poverty reduction, it is no surprise that the to lower-level governments and to the private sector. World Bank is actively involved with decentralization Decentralization is particularly widespread in develop- policy in many developing countries. ing countries for a variety of reasons: the advent of mul- The Bank is involved in decentralization issues in a tiparty political systems in Africa; the deepening of variety of ways: democratization in Latin America; the transition from a * Since many countries are undergoing some form of command to a market economy in Eastern Europe and decentralization either by design or by default, a the former Soviet Union; the need to improve delivery growing number of Bank-funded projects are in of local services to large populations in the centralized effect supporting sectoral decentralization strategies. countries of East Asia; the challenge of ethnic and geo- Twelve percent of Bank projects completed between graphic diversity in South Asia, as well as ethnic ten- 1993 and 1997 involved decentralizing responsibil- sions in other countries (Bosnia and Herzegovina, ities to lower levels of government. A sectoral and Ethiopia, Russia) and the attempt to keep centrifugal regional breakdown is provided in Table 1. forces at bay by forging asymmetrical federations; and * The Bank is also supporting decentralization the plain and simple reality that central governments through loans to subnational governments. have often failed to provide effective public services. Although these loans are (necessarily) guaranteed Political pressure probably drives most decentral- by the central government, they are otherwise nego- ization efforts. But whatever its origins, decentraliza- tiated and undertaken by independent local author- tion can have significant repercussions for resource ities. Such loans are used for both specific projects mobilization and allocation, and ultimately macroeco- and state-level structural adjustment efforts (for nomic stability, service delivery, and equity Since example, in Andhra Pradesh state in India). A new Table 1: World Bank Projects with a Decentralization Component, by Sector and Region Sector Percentage Region Percentage Urban 43 Africa 19 Health 27 South Asia 14 Social funds 26 Latin America and the Caribbean 13 Environment 16 Middle East and North Africa 11 Water 13 Europe and Central Asia 6 Agriculture I 1 Transport 10 Education 9 Energy 3 Finance 1 Note: Given an average project cycle of five years, the database used for this analysis does not include projects that went to the Board after 1992. Thus the numbers presented here likely underreport current decentralization-related projects. Source: Portfolio Review, PRMPS, 1998. 1 2 Rethinking Decentralization in Developing Countries loan instrument was established in late 1997 to facil- unities for voice and exit are limited because of weak itate subnational adjustment lending. institutions. Democratic systems are often frail, ren- * Structural adjustment loans to central governments dering the electoral system a highly problematic have supported the design of intergovernmental fis- method of achieving accountability Strong, broadly cal relations and the process of decentralization (for based local participation can overcome weak formal example, in Kyrgyz Republic, Russia, and Vietnam). election systems, but powerful elites make this diffi- * Given the importance of decentralizaton in many cult in many places. Mobility is often constrained by countries and its potential impact on a range of poor information, infrastructure, and legal frame- development objectives, a number of recent Bank works, which result in weak markets for land, labor, reports have focused on decentralization and relat- and capital. Although all countries have at least some ed issues in specific regions and countries (see urban areas, in smaller municipalities and rural areas Annex 1 for a list of economic and sector work on it is often unrealistic to expect a family to sell their decentralization).1 land, learn of employment opportunities in other * Finally, a growing number of the Bank's country jurisdictions, physically move to the new area, and assistance strategies are giving greater prominence to borrow money in a new locality where they are decentralization (such as those for Hungary, Russia, unknown. The chances of overcoming these obstacles South Africa, Thailand, and Uganda, among others). are further reduced because low incomes and weak One reason decentralization has attracted so much social safety nets often make households risk averse in attention is that it is often a cross-cutting reform that developing countries. can relate to such important Bank concerns as the rela- A third channel for local accountability is through tion between fiscal and financial development; macro- hierarchical relationships within the public sector economic stability; poverty alleviation and the social (World Bank, 1997), including the formal rules and safety net; institutional capacity, corruption, and gover- oversight arrangements that exist within government. nance; investment in infrastructure; and the provision This is the primary channel for accountability in many of social services. Other international and national insti- developing countries, but it is often ineffective because tutions are currently studying various aspects of these of poor information and monitoring systems. In short, issues from their own perspectives.2 Given the impor- the primary measures for local accountability assumed tance of decentralization for economic management, in most discussions of decentralization may not hold the Bank also needs to invest in understanding it. in many developing countries. Much of the literature on decentralization, norma- Given that the World Bank's work is exclusively in tive and empirical, is based on industrial countries and developing countries, it must carefully consider what assumes the existence of institutions that are usually unintended consequences these institutional weak- very weak in developing countries. For example, much nesses might have on decentralization policies, and of the traditional public finance literature related to fis- their implications for project design and policy dia- cal decentralization relies on "voice" and "exit" to logue. This paper is intended to initiate this type of ensure local accountability and achieve the allocative analysis and help those working on decentralization efficiency gains expected from decentralization. in developing countries develop a more institutional- According to this view, if people are dissatisfied with ly sensitive perspective when analyzing decentraliza- decisions made by local leaders, they can vote them out tion in a particular country.3 We draw on the literature of power. If they do not like the package of local taxes and growing experience with decentralization in and public services offered, they can exit (or "vote with developing countries to explore how a wide range of their feet") by moving to a jurisdiction that better variables can affect decentralization efforts and how matches their preferences (Tiebout, 1956). policies and incentives can be designed to improve Yet governments in many developing countries are outcomes.4 often not responsive to their citizens, and decision- We do not know enough empirically to make defin- making is rarely transparent and predictable. Opport- itive recommendations about which types of decen- Why Should We Worry about Decentralization? 3 tralization are best for which services in which institu- compensate for the weak institutions in the short run, tional settings. Much of the discussion of decentraliza- and build key elements of important institutions in the tion reflects a curious combination of strong precon- long run. ceived beliefs and limited empirical evidence. But we The next section describes the complexity of the do know that the best design will vary depending on decentralization process. It concludes that the debate circumstances and institutions, and that this complex- on whether decentralization is "good" or "bad" is unpro- ity has sometimes been overlooked in the haste to offer ductive since decentralization is a political reality policy advice. If a common framework can be estab- worldwide-one that varies greatly in form within and lished, we can begin to assemble documented case among countries. The third and fourth sections focus studies of decentralization, so that within a few years on the lessons learned about the design of decentral- we may be able to identify with more precision patterns ization, with an emphasis on the building blocks of fis- of success and failure. Also, by identifying institutions cal federalism and on the importance of institutions in that are important for successful decentralization in decentralization efforts. The final section draws impli- industrial countries and that may be weak in develop- cations from this analysis and describes the research ing countries, we can also begin to identify ways to agenda that remains to be explored. 2. Decentralization: A Complex Phenomenon Decentralization-the assignment of fiscal, political, ernments to participate in the political and economic and administrative responsibilities to lower levels of process. government-is occurring worldwide for different rea- sons, at different paces, and through different means. The "why" of decentralization is as varied as the "how" What Is Decentralization? (Box 1). The complexity inherent in the decentraliza- tion process is further aggravated by its cross-cutting Decentralization is not easily defined. It takes many impact. For policymakers, who often have little control forms and has several dimensions. Indeed, a wide vari- over the political genesis or pace of decentralization, the ety of institutional restructurings are encompassed by challenge is to implement it in a way that ensures the this label, and several variants may be operating at the stability, efficiency, and equity of :he economic system. same time within a country, and even within a sector. Thus care must be used in labeling, and labels- including those used in this paper-must be inter- Why Is Decentralization Happening? preted with care. One widely used distinction is among deconcentra- In most countries decentralization reflects a broader tion, delegation, and devolution (Rondinelli, 1981 and process of political and economic reform (World 1989). Deconcentration occurs when the central govern- Bank, 1997). Political changes worldwide have given ment disperses responsibilities for certain services to its voice to local demands and the need to bring eco- regional branch offices. This does not involve any trans- nomic and political systems closer to local communi- fer of authority to lower levels of government and is ties. In some cases the very preservation of a national unlikely to lead to the potential benefits or pitfalls of political system has required the decentralization of decentralization. The "decentralization" that has oc- power. In addition, technological changes and global curred in many unitary countries is actually deconcen- integration of factor markets have changed the size of tration, since independent local governments (which are government needed to manage economic systems. On legally accountable to local constituents) do not exist one hand, economic mobility has led to the creation and local field offices of the central govemment are sim- of supranational bodies to manage the growing eco- ply used to improve the efficiency and effectiveness of nomic integration among nations. On the other, an service delivery. This is the case in many East Asian increasing number of public services can be efficient- countries and, until recently, was the rule in Eastern ly provided by decentralized (and often private) orga- European countries (Komai, 1992). Deconcentration nizations (World Bank, 1995a). For example, in met- can also exist for some functions in federal countries ropolitan Buenos Aires the entire water and sewerage when the central government maintains a strong inter- system is operated, maintained, and invested in by a est in ensuring delivery of a particular service. consortium of private companies (Triche, Mehia, and In contrast, the central issue for both delegation and Idelovitch, 1993). Finally, the collapse of central eco- devolution relates to the balancing of central and local nomic systems has encouraged regional and local gov- interests. Delegation refers to a situation in which the 4 Decentralization: A Complex Phenomenon 5 Box 1: Rationales for Decentralization Many rationales for decentralization may be discerned in arose mainly from design problems, including a soft bud- the literature and in practice. Not all will be relevant in get constraint between levels of government. any one country, nor are they all necessarily consistent or An interesting argument asserts that a primary econom- equally important to all relevant parties. As noted, most ic rationale for decentralization is to improve the "compet- are based on normative or empirical work in industrial itiveness" of governments-that is, decentralization will countries. make local govemments try to satisfy the wishes of citizens The most common theoretical rationale for decentral- (Breton, 1996; Salmon, 1987). Although suggestive, there ization is to attain allocative efficiency in the face of dif- has been little empirical examination of this idea (Kenyon, ferent local preferences for local public goods (Musgrave, 1991; Mueller, 1997). Some Bank work has occasionally 1983; Oates, 1972; Tiebout, 1956). Problems may arise recognized "horizontal" competition, particularly with with respect to coordination-which is itself costly respect to taxation (both at the national and local levels), (Breton and Scott, 1978)-where interjurisdictional although it has generally been viewed as a problem rather spillovers are important, including stabilization (Tanzi, than a potential benefit. Little attention seems to have been 1996; Wildasin, 1997) and distribution (Tresch, 1981). paid to the more troublesome question of "vertical" inter- Most World Bank work on decentralization has focused govemmental competition and the nature and enforcement on this rationale (World Bank 1988, 1994, and 1997). of the rules that may make intergovernmental competition Related issues, on which there is neither theoretical nor beneficial (Breton, 1996). However, Weingast (1995) pro- empirical agreement, concern the direction and impor- poses that competitive local govemments act as checks on tance of the relation between decentralization and the size the central govemment to prevent it from confiscating of the public sector (Mueller, 1996; Ehdaie, 1994) and wealth; thus federalism preserves markets. between decentralization and the rate of economic growth Another political rationale for decentralization is that (Martinez-Vazquez and McNab, 1997; Zhang and Zou, good governments are those closer to the people (Inman 1998). and Rubinfeld, 1997; World Bank, 1997). Since John Equity and distributional concerns about decentraliza- Stuart Mill, this argument has been closely related to ques- tion cut both ways. Some analysts argue that in some cir- tions of political participation and democracy, but it is still cumstances local governments achieve such goals more undeveloped in theoretical (Breton, Cassone, Fraschini, effectively than central governments (Pauly, 1973). Others 1998) or, especially, empirical terms (Martinez-Vazquez argue that central redistribution is needed both for effec- and McNab, 1997). Although there has been some indi- tiveness (Musgrave, 1983) and to overcome biases of local rect discussion of this proposition in relation to such gov- elites (Wilensky, 1974; Inman and Rubinfeld, 1997). ernance concerns as corruption (Prud'homme, 1995), Another view is that, regardless of what local governments little has yet been done to link decentralization, participa- may attempt to do in the way of redistribution, their tion, and good governance apart from an interesting gen- attempts will be frustrated by resource mobility and the eral empirical paper (Huther and Shah, 1998), one exam- openness of the local economy (Buchanan and Wagner, ining electoral accountability and economic policy choices 1971). All these views may be found in different Bank by U.S. govemors (Besley and Case, 1995), and a reveal- country studies in recent years. ing case study on Colombia (World Bank, 1995b). Decentralization's potentially destabilizing effect on A quite different political rationale for decentralization the macroeconomy has caused much concern to some is to accommodate pressure for regional autonomy and, (Prud'homme, 1995; Tanzi, 1996; Ter-Minassian, 1997). hence, perhaps increase the legitimacy and sustainability Others suggest, however, that such effects are more like- of heterogeneous national states. In a sense, there may ly to reflect inappropriate incentives than any problem sometimes be a tradeoff between political and economic inherent to decentralization (Spahn, 1997a and 1997b). stability (Treisman, 1998). Although touched on in a Bank In countries that have "decentralized" to offload fiscal study of Russia (Litvack, 1994), this idea has-perhaps imbalances from the center, it is of course not surprising understandably, given the nature of the Bank's clients- to see a strong association between decentralization and not been much developed in Bank literature. Still, the fiscal imbalance at lower levels (Wallich, 1994). At least emphasis in many studies on the role of "equalization" three World Bank studies-on Argentina (1996a), Brazil transfers can perhaps be attributed to political stability as (Dillinger, 1997), and Colombia (1996b)-have exam- well as a more magnanimous desire for interjurisdiction- ined this issue and suggested that destabilization effects al equity (Ahmad, 1996). 6 Rethinking Decentralization in Developing Countries central government transfers responsibility for deci- central government, for example, can only be translated sionmaking and administration of public functions to into actual powers being shifted if subnational govern- local governments or semiautonomous organizations ments have the fiscal, political, and administrative that are not wholly controlled by the central govern- capacity to manage this responsibility ment but are ultimately accountable to it. These orga- nizations usually have a great deal of discretion in deci- sionmaking. This form of decentralization can be What Are the Implications of Decentralization? characterized as a principal-agent relationship, with the central government as the principal and the local For policymakers and their advisers, the multidimen- government as the agent. From this perspective, the sional aspect of decentralization-the dispersion of fis- main design issue is to ensure that the self-interested cal, political, and administrative powers-suggests agent (the local government or semiautonomous orga- three implications that heavily influence the context for nization) faces incentives that induce it to act as close- thinking through decentralization. First, because ly as possible in accordance with the wishes of the prin- decentralization can change the mobilization and allo- cipal (the central government).5 cation of public resources, it can affect a wide range of Finally, devolution, a more extensive form of decen- issues from service delivery to poverty reduction to tralization, refers to a situation in which the central macroeconomic stability. (Thus we refer to it as a government transfers authority for decisionmaking, "cross-cutting" issue.) Second, the management of finance, and management to quasi-autonomous units decentralization requires intimate knowledge of local of local government. Devolution usually transfers institutions and a nuanced understanding of the responsibilities for services to municipalities that elect process of decentralization-that is, what is driving their own mayors and councils, raise their own rev- decentralization in a country (and sector) and which enues, and have independent authority to make stakeholders are involved. Third, limited empirical evi- investment decisions. In a devolved system, local gov- dence exists about what works and what does not. ernments have clear and legally recognized geograph- Together these three factors pose a daunting challenge ic boundaries over which they exercise authority and for those responsible for designing and managing within which they perform public functions (Ron- decentralization. dinelli, 1998). The shift in responsibility between tiers of govern- ment is underpinned by several fiscal, political, and administrative instruments. These define the extent to Decentralization is often implemented as a broadly which intergovernmental relations are deconcentrated, based reform that affects various sectors and levels of delegated, or devolved. Fiscal decentralization-who sets government. Outcomes reflect the interaction and and collects what taxes, who undertakes which expen- coordination of policies between different tiers of gov- ditures, and how any "vertical imbalance" is rectified- ernment. In many countries it is a considerable chal- has been especially prominent in recent discussions in lenge to coordinate sectoral reforms undertaken by a many countries, but as just indicated many of the more ministry of the central government with decentraliza- fundamental questions relate to political and adminis- tion of fiscal, political, and administrative responsibil- trative decentralization. Political decentralization refers at ities to local governments. one level to the extent to which political institutions map The cross-cutting influence of decentralization has the multiplicity of citizen interests onto policy decisions a clear implication for the World Bank. Understanding (Inman and Rubinfeld, 1997). Administrative decentral- the evolution and status of a nation's decentralization ization is concerned with how political institutions, once process is essential to understand and analyze policy determined, turn policy decisions into allocative (and issues across sectors. For example, the successful deliv- distributive) outcomes through fiscal and regulatory ery and financing of education or welfare systems, as actions. The political decision to devolve powers from in South Africa, depends on the public finances of mul- Decentralization: A Complex Phenomenon 7 titiered governments rather than the edicts of a central merits, its potential problems, and its effects on static line ministry. In general, major policy reform of indi- and dynamic allocative efficiency, income distribution, vidual sectors-such as health, education, or infra- macroeconomic stability, institutional demands on local structure-cannot be undertaken without an assess- capacity, potential for corruption, governance, and so ment of a country's intergovernmental system, or vice on. But the actual empirical evidence on these proposi- versa. This outlook needs to be embedded in the coun- tions is either nonexistent or conflicting. In some try team framework at the Bank. respects this is not surprising. Given the complexity and multiple dimensions of the concepts involved and the Country context context-specific nature of decentralization, it is to be expected that studies will show considerable uncer- Many factors affect both the ideal and the actual form tainty about the strength and even the direction of the of decentralization adopted in any country for any ser- relation between decentralization and, say, growth, gov- vice at any time: emance, distribution, and stability. Conflicting evi- * The number of subnational units as well as their dence and interpretations may be found with respect to absolute and relative sizes and wealth. decentralization and local tax efforts (Bird and Fiszbein, * The distribution of functions (relative to the "span" 1998), the relationship (if any) between increased local of public goods, externalities and jurisdictional democracy and allocative efficiency (Martinez-Vazquez spillovers, and so on). and McNab, 1997), the relationship between subna- * The nature of the "common" institutions (with par- tional expenditures and growth (Zhang and Zou, 1998; ticular attention to their effects on government com- Davoodi and Zou, 1998), and the ability to target the petition). poor under different forms of decentralization (Ra- * The role and status of the constitution (for example, vallion, 1998; Alderman, 1998; Inman and Rubinfeld, with respect to the independence of the judiciary 1997). A particular challenge in this respect is to under- and collective rights). stand how best to match fiscal, political, and adminis- * The technical characteristics and policy objectives of trative arrangements to achieve the potential benefits of specific public services. decentralization for any given service in any given * The current political situation. country. Much of the discussion of decentralization pre- sumes that it is a matter of choice or deliberate design, but in many countries it may equally well be either a Successful Decentralization Depends on political necessity (for example, to hold a fractious het- Institution-Specific Design erogeneous country together) or a default option (that is, a way to try to execute more satisfactorily some nec- Designing decentralization policy is difficult in any essary functions of the state in the face of central fiscal country because decentralization can affect many crisis and political weakness). Indeed, all these ratio- aspects of public sector performance and generate a nales may be at play at once, possibly to varying wide range of outcomes. But it is particularly difficult degrees, in a single country. For these and other rea- in developing countries because institutions, informa- sons, decentralization often encompasses an extreme- tion, and capacity are all very weak. The cross-cutting ly nuanced set of activities that can only be understood, nature of decentralization, the importance of local analyzed, and-to the extent possible-guided on the institutions in influencing the impact of decentraliza- basis of thorough knowledge of local institutions. tion, and the limited empirical evidence on what works and what does not make the design and implementa- tion of decentralization a considerable challenge (for Strong bel'efs but limited empirical evidence example, matching expenditures and revenues at each Many people hold strong beliefs about various aspects level of government, providing a regulatory framework of decentralization: its intrinsic political and economic that imposes a hard budget constraint on subnational 8 Rethinking Decentralization in Developing Countries governments, and incorporating local participation How decentralization affects equity also depends in and accountability in decentralization). Evidence sug- part on the extent of local accountability and local gests that the problems associated with decentraliza- political participation by the poor. Accountability can tion in developing countries reflect flaws in design and be enhanced when local leaders are elected and are implementation more than any inherent outcome of concerned about providing services to their con- decentralization. The following sections illustrate this stituents. When the poor participate in the political important point by discussing how design can alter the process, they can exert influence on leaders. The mere impact of decentralization on equity and macroeco- existence of a democratic political system is insufficient nomic stability unless there is meaningful political participation by all groups. For example, India has a long democratic tra- dition but since local participation depends on social caste, the poor often have little influence. In contrast, The impact of decentralization on interregional and in Cuba and Vietnam there is considerable participa- interpersonal equity can vary greatly depending on insti- tion at the grassroots level (World Bank, 1996c). In tutional arrangements and policy design details. If the Oaxaca, Mexico, the targeting of poverty expenditures central government makes no effort to redistribute depends heavily on how representative local govern- resources to poorer areas, fiscal decentralization will ments are (Fox and Aranda, 1996). result in growing disparities. Similarly, if provinces or Decentralization can also enhance access by the states do not redistribute within their jurisdictions, poor poor if it increases competition in the delivery of ser- people may lack access to public services. But decentral- vices (such as water and electricity) and this drives ization need not produce such outcomes. Depending on down prices. In addition, if policy reform separates the preferences of the national population, horizontal efficiency and equity objectives, decentralization can equity-that is, ensuring some level of comparability in enable service deliverers to charge user fees and focus ability to provide public services throughout the coun- on efficiency and then use the revenues to expand cov- try-can be achieved through intergovernmental trans- erage or improve quality, which can benefit the poor. fers that include equalization components. How inter- For example, the introduction of earmarked user fees governmental transfers are designed (block or specific for primary health care in Cameroon led to an improve- purpose, matching or nonmatching, flat rate matching or ment in quality and an increase in utilization-partic- progressive matching) and monitored largely determines ularly for the poor, who previously lacked access to the extent to which decentralization results in greater bal- alternative providers (Litvack and Bodart, 1993). ance or imbalance in service provision between regions, states, towns and villages, and households. Decentralization can affect interpersonal equity Decentralization and macroeconomic stability through public expenditure policy, tax policy, and the The stringent conditions for successful decentraliza- design of intergovernmental transfers. If decentraliza- tion have recently been emphasized with respect to tion of expenditure decisions leads to a greater share developing countries (Prud'homme, 1995; Tanzi, of public resources being spent on services used by the 1996). In particular, it has been argued that not only nonpoor, equity will suffer. Similarly, if decentraliza- can decentralization fail to improve local service deliv- tion leads to a higher tax burden for the poor (through ery, it may even risk national destabilization. Argentina higher user fees and local taxes and no offsetting in the 1980s is a commonly cited example, but others changes in other taxes), then this too can weaken equi- are not hard to find in the transition economies of ty Nonetheless, if the central government is con- Central and Eastern Europe (Bird, Ebel, and Wallich, cerned about preserving equity and protecting the 1995). Similar fears appear to have played a role in poor, it can do so in part-despite decentralized China's fiscal reforms (Bahl, 1998). expenditures and taxes-through the design of inter- Concerns about decentralization and macroeconom- governmental transfers. ic instability are fueled by instances where governments Decentralization: A Complex Phenomenon 9 have decentralized without adequate planning to con- current revenues and then attempt to close the gap trol the deficit behavior of local governments. For through borrowing. Others argue that since macro- example, international experience suggests that if coun- economic stabilization is properly a national govern- tries decentralize more expenditure responsibilities ment task, it is important that the national govern- than revenue resources, service levels will fall or local ment have full control over all the policy instruments governments will press-sometimes successfully-for it needs to carry out this task, including borrowing- more transfers, more loans, or both. One of the clearest particularly borrowing abroad. The key to addressing and most analyzed cases of this phenomenon is the this problem is to ensure that decentralization is Russian Federation (Wallich, 1994). On the other hand, undertaken in a way that increases rather than if more revenues than expenditures are decentralized, decreases accountability. This can be done by imple- local revenue mobilization may decline because of tax menting stronger rules and regulations (such as the competition, and again macroeconomic imbalances balanced budget rules in the United States) and by may emerge-as in Argentina, Brazil, and Colombia. changing the incentives facing various actors. This risk is greatest when revenues are decentralized If a national government wants to avoid macroeco- without adequate steps to ensure that local revenue nomic problems arising from subnational debt, it can mobilization is maintained and that local authorities are do so by not subsidizing such borrowing and by let- capable of carrying out the corresponding expenditure ting subnational governments that borrow too much responsibilities. Even if both sides of the budget are go bankrupt. One way to approach this is to make the decentralized in a balanced fashion, local governments rules of the game clearer and more transparent. This may not have adequate administrative or technical is exactly what was done in Morocco, where the gov- capacity to carry out their new functions. Such prob- ernment changed the subsidy scheme for local gov- lems may give rise to particular concem in developing ernments from one of budget-balancing grants, in countries where local governments are charged with which capital and interest payments on loans important social and economic infrastructure invest- increased transfer receipts, to a formula-based equal- ments (Bird, 1994). ization transfer that takes no account of borrowing Regardless of their empirical validity, concerns (Vaillancourt, 1998). In addition, lenders were explic- about macroeconomic disaster ensuing from fiscal itly told not to count on financial bailouts. As a gen- decentralization rank high in many countries. Thus eral rule, however, the difficulty of envisioning, let care must be taken to avoid unwanted outcomes in alone carrying out, bankruptcy in the public sector this respect. Some analysts fear that unchecked sub- provides good reason to require fairly stringent con- national governments may increase current expendi- ditions on subnational borrowing to ensure local gov- tures well above their capacity to finance them out of ernment accountability. 3. Designing Decentralization: The Building Blocks of Fiscal Federalism Much of the literature on decentralization, including congruent services at the local (street lighting, refuse that generated by the World Bank, reflects the "fiscal removal), regional (rural-urban roads, refuse disposal), federalism" orientation of the traditional public and national (intercity highways, environmental poli- finance approach to this issue (Walsh, 1992; Shah, cy) levels. 1994). The emphasis is on setting the appropriate Decentralized decisionmaking enlarges possibilities expenditure and tax assignment for each tier of gov- for local participation in development. However, ernment and on designing intergovernmental trans- accountability is often best promoted by establishing a fers. The framework is driven by the Musgravian clear and close link between the costs and benefits of principles of efficiency, equity, and stability public services, so that the overall amount of expendi- (Musgrave, 1983). Although this literature has con- ture responsibility assigned to a level of government sidered institutional issues, particularly in recent ideally will correspond to the amount of revenues that years (Huther and Shah, 1998), the main emphasis level has at its potential command. has been on economic policies. Less attention has Some national allocative objectives can be carried been paid to how these policies might be implement- out either directly by the central government or indi- ed in disparate institutional settings and what impact rectly by local governments responding to incentives they might have on the design of institutionally sen- created by national grants and regulations (as well as sitive policy. This section reviews the traditional interlocal or interregional agreements). Moreover, issues of public finance, while the next section national governments have obvious roles in formulat- explores the institutional issues further. ing stabilization and distribution policies, and atten- tion must be paid to possible local conflicts with these policies. In many instances it seems appropriate for Assigning Expenditure Responsibilities some functions to be shared in the sense that higher levels of government may exercise a regulatory or pol- Both theory and experience indicate that it is impor- icy role, while lower levels of government are respon- tant to state expenditure responsibilities as clearly as sible for delivering services. possible in order to enhance accountability and reduce Sectors need to be unbundled because some aspects unproductive overlap, duplication of authority, and of service delivery are appropriately undertaken by high- legal challenges. In theory, decisionmaking should er levels of government while other components are best occur at the lowest level of government consistent with done locally For example, there are four broad functions allocative efficiency (for example, the geographic area in water supply and sewerage (usually considered a local that internalizes the benefits and costs of decision- service), and each has several components: making for a particular public service). Thus the opti- * Policymaking (central and local, financial, econom- mal size of jurisdiction for each service and service ic, environmental, health-related, subsidies). component would likely differ, although in practice * Regulation (economic, environmental, water quali- economies of administration and transactions costs ty, service quality). would presumably lead to the grouping of roughly * Investments (planning, financing, execution). 10 Designing Decentralization: The Building Blocks of Fiscal Federalism 11 Operations (production and treatment, distribu- taxes on international trade should be assigned to tion, commercial operations and consumer rela- the central government. Similarly, local govern- tions, maintenance). ments should presumably, tax only immobile factors The characteristics of each function differ and thus (such as land and real estate) in addition to apply- have different implications for the role of the govern- ing user charges wherever feasible. ment and private sector, and the level of government * Because the central government has a large role to involved (Triche, 1993). play in achieving distribution and stabilization A key to designing good policy is a clear under- goals, it seems logical for it to be responsible for standing of what outcomes are important for the cen- progressive (redistributive) taxes (such as those on tral government, and what outcomes can be deter- wealth and personal income), as well as those taxes mined by local governments. The central government most sensitive to economic fluctuations (such as should retain control (either by directly providing ser- corporate income taxes and taxes on natural vices or by creating incentives for local governments to resources) and any specific taxes (for example, car- act according to central preferences) over functions for bon taxes) related to national objectives.6 On the which it desires certain outcomes (such as family plan- other hand, to ensure accountability to local resi- ning), and should relinquish control on issues where dents, local governments should to the extent fea- local divergence in priorities is not a big issue. Local sible be responsible for financing local services. governments must be held accountable to the center When local governments are responsible for such for nationally mandated services, while for other major expenditures as education, accountability devolved functions they must be held accountable to thus suggests more taxing power for local govern- their local constituents. ments than efficiency-or central policy-permits Finally, experience has shown that effective decen- in most countries. tralization requires complementary adaptations in a An important national policy consideration in coun- variety of institutional arrangements for intergovern- tries where subnational governments have indepen- mental coordination, planning, budgeting, financial dent taxing power is to ensure an adequate degree of reporting, and implementation. Such arrangements internal tax harmonization and coordination to pre- may encompass both specific rules (for example, in the serve the internal common market. This concern also design of fiscal transfers) and provisions for regular argues against giving lower-tier governments the intergovernmental meetings and periodic reviews of power to impose taxes on corporate income, value- intergovernmental arrangements. Detailed central con- added, or natural resources. For corporate taxes, such trol over local use of funds is seldom appropriate. governments may engage in inefficient tax competition instead, what is needed is better monitoring and over- or in equally undesirable (for accountability reasons) sight of local fiscal performance. interjurisdictional tax exporting. Tax exporting (from producer to consumer regions) and administrative complexity provide strong arguments against subna- Raising Revenue tional value-added taxes. Tax exporting and instability both argue against assigning resource revenues to local Two broad principles of tax assignment are the need for or state governments. The same arguments can be economic and administrative efficiency and the desir- made with respect to tax-sharing arrangements under ability (for accountability reasons) of having each gov- which local governments receive a fixed percentage of ernment finance its own expenditures out of its own certain national taxes collected in their jurisdiction. revenues. Unfortunately, these two principles general- Such arrangements create undesirable incentives for ly conflict. tax exporting and bias national tax policy (in the direc- * Economic and administrative efficiency suggests tion of raising taxes that do not have to be shared). that taxes on mobile factors (such as corporate and Moreover, tax rates in such systems are invariably set personal income taxes), value-added taxes, and by the central government, and the sharing rate is often 12 Rethinking Decentralization in Developing Countries applied uniformly throughout the country. Thus the it is sometimes argued that subnational governments accountability link is broken and subnational govern- should be restrained from accentuating cyclical pres- ments have no incentive to ensure that the amount and sures by borrowing-especially borrowing abroad and pattern of their spending is efficient. thus adding to the national obligation to service foreign As a rule, the allocation of revenues in most coun- debt (Ter-Minassian, 1997). On the other hand, alloca- tries provides most (and often all) lower-tier govem- tive efficiency and intergenerational equity often ments with insufficient own-source revenues to finance require that long-lived investment projects, especially the expenditures for which they are responsible. Apart those that will increase productive capacity, be from borrowing, there are two ways to bridge this financed by borrowing rather than relying solely on financing gap, either or both of which may be applic- current public savings or transfers from above. The able to varying extents in different countries. question, then, is how to distinguish good borrowing The first is to supplement local revenues with inter- from bad borrowing. governmental fiscal transfers-without undesirably In the past subnational borrowing was an issue of reducing local efforts to collect taxes (see below). The fiscal federalism because local borrowing was guaran- second is to permit subnational govemments to levy teed by central governments or secured through cen- their own broadly based taxes-so long as they burden tral intermediaries. This has changed significantly in local residents only In principle two such taxes are recent years with the rise of local and global capital possible: a retail sales tax and a personal income tax. markets, increasing political decentralization, and the Administrative difficulties rule out retail sales taxes in importance of the regulatory environments in which practice. Thus the only efficient broadly based subna- subnational borrowing occurs. For these reasons, sub- tional tax that seems both feasible and desirable is like- national borrowing is addressed in greater detail in ly to be a flat-rate surtax (or surcharge) on a national section 4. personal income tax-provided that such a tax exists and works moderately well.7 To ensure local accountability with respect to such Designing Intergovernmental Transfers a surtax, a local property tax, or local taxes in general, local governments must be restricted (to the extent Since transfers are the main source of revenue for sub- possible) from exporting taxes and allowed to set their national governments in most countries, the design of own tax rates. For efficiency it may be desirable to transfers is of critical importance to the success of assess the base of a tax centrally and even to have it col- decentralization. Transfers can be broadly divided into lected by the central government. But for accountabil- nonmatching (lump sum) and matching transfers. ity it is critical that the local authorities be responsible Nonmatching transfers can in turn be divided into (perhaps within limits) for setting the tax rate. selective (conditional) and general (unconditional). Conditionality ensures that the recipient government spends at least the amount received on the designated Overseeing Subnational Borrowing functions, though the fungibility of funds means that some money that would have been spent on such Subnational governments can be permitted to borrow, functions in any case may be diverted elsewhere. but a great deal of caution is required. Unchecked sub- Empirical studies of industrial countries suggest that national governments, particularly those highly depen- as a rule even unconditional grants are fully spent by dent on national transfers, may increase current expen- recipient governments rather than used to lower local ditures well above their capacity to finance them out of taxes (the so-called "flypaper effect," meaning that current revenues and then close the gap through bor- money sticks where it hits; Hines and Thaler, 1995). rowing (especially if, as in Argentina, local govern- Similar results have been noted in developing coun- ments control provincial banks), often pledging future tries, such as Colombia (World Bank, 1996b). revenue transfers to service the loans. More generally, Matching transfers require that the funds be spent on Designing Decentralization: The Building Blocks of Fiscal Federalism 13 specific purposes and that the recipient to some extent national stabilization objectives are not thwarted by matches the grant out of its own funds. Such transfers subnational finances (Box 2). One approach that may distort local priorities and be considered appears to achieve this dual objective is to set the inequitable in that richer jurisdictions can raise total level of transfers as a fixed proportion of total matching funds more easily But the latter problem can central revenues, subject to renegotiation periodi- be offset to some extent by varying matching rates cally (say, every three to five years). with jurisdictional wealth, and the former may be the * The transfer formula (or formulas) is transparent, desired outcome when the transfer is intended to, for based on credible factors, and as simple as possi- example, internalize spillovers or achieve overriding ble. Unduly complex formulas are unlikely to national policy objectives. prove feasible or credible in countries where, for When local governments are expected to play a example, there is serious dispute about regional major role in delivering social services, the design of population sizes. intergovernmental transfers is particularly important If several of the objectives discussed earlier are because the central government usually retains a strong applicable-for example, some degree of equalization interest in at least some of the outcomes. If the central is desired while at the same time there are clear nation- government is, in effect, using local governments as al policy objectives-it will generally assist both clari- agents in executing national policies-for example, to ty and effectiveness if separate transfers are targeted at provide primary education at a specified level through- each objective (see World Bank 1996b). out the country-then it is important to make the transfer conditional on the funds actually being spent on education or on the attainment of some level of edu- Assembling the Building Blocks: A Systems cational performance (Bird, 1993).8 Approach If key services are to be provided through decentral- ized governments, careful attention has to be paid to get- The assignment of expenditures, revenues, transfers, ting the prices facing service providers right (for exam- and subnational borrowing together compose the sys- ple, through a well-designed system of matching grants), tem of intergovernmental finance. Though there is setting up an information and inspection system capa- much to know about each of these components as well ble of ensuring that the desired services are delivered to as their sequencing, it is perhaps most important to the target groups, and devising some system (such as a know that these issues must be considered together, as national 'fail-safe" provision) for dealing with the non- part of a complete system. This system, combined with compliant without punishing the innocent. the institutional environment (see the next section), As Bahl and Linn (1992) show, the most appropri- determines impacts on efficiency, equity, and macro- ate form of a transfer largely depends on its objective. economic stability. For example, an intergovernmental Regardless of the particular design, however, good fiscal system might assign local governments the intergovernmental transfer programs share certain responsibility for financing primary school, collecting characteristics: local school fees, and administering central funds to Transfers are determined as objectively and openly target subsidies to poor schoolchildren. This set of as possible, ideally by some well-established formu- expenditure, revenue, and transfer policies can la. They are not subject to hidden political negotia- increase efficiency and equity. But if the three policies tion. The transfer system may be decided by the cen- are not designed with a common objective, the quali- tral government alone, by a quasi-independent ty of services could decline dramatically and the poor expert body (such as a grants commission), or by could lose access to schooling. some formal system of central-local committees. Although the ultimate impact of decentralizing any * Transfers are relatively stable from year to year to one fiscal instrument will depend on the design of all permit rational subnational budgeting. At the same of them, all components should not be designed simul- time they are sufficiently flexible to ensure that taneously. Expenditure functions should be assigned 14 Rethinking Decentralization in Developing Countries Box 2: Experience with Intergovernmental Transfers Any good system of intergovernmental grants must be sta- A number of developing countries distribute transfers ble. Yet it must also be flexible. How can both character- using a formula intended both to equalize public expendi- istics be achieved simultaneously? From the perspective tures in localities with different needs and capacities and to of a central government, the best system might be one in stimulate local fiscal efforts, although severe data problems which the total amount to be transferred (sometimes often constrain the parameters used in such formulas. Brazil called the "distributable pool" or the "primary distribu- and India, for example, allocate some transfers in accor- tion") is determined each year in accordance with budget dance with per capita incomes in the different states, but few priorities. But unless the amounts to be received are to other developing countries do so owing to data difficulties. some extent predetermined and predictable, recipient In the absence of such data, simpler approaches-like those govemments will not be able to budget properly and will used in Colombia and Morocco-based on, for example, not face an appropriately hard budget constraint. One way population and a simple categorization of localities (by size, to provide some degree of stability to local governments type, and perhaps region) are more likely to prove useful in and some degree of flexibility to the central govemment measuring general expenditure needs. Few developing without biasing central tax policy is to establish a fixed countries include explicit capacity measures in their for- percentage of all central taxes (or current revenues) to be mulas. India and Nigeria include a measure of tax effort in transferred-preferably in accordance with an objective the basic distributional formula to states, and Colombia has distributive formula (the "secondary distribution") that such an element in one of its transfer programs. Chile goes takes into account such factors as the needs and capacity further and actually "taxes" ncher localities to some extent of recipient govemments. If not too complex, such a sys- by reducing their transfers and raising those granted to tem will also be transparent. Moreover, interjurisdiction- poorer localities. Of course, this approach makes sense only al equalization can be built into the formula. if local govemments have the ability to vary local tax rates. Many transfer systems in developing countries possess The absence of much local autonomy with respect to local few or none of these characteristics. In Russia and other for- taxes, combined with data difficulties, probably explains the mer socialist countries transfers are negotiated every year in few examples of transfer programs incorporating explicit an obscure political process. In Bangladesh nominally pre- capacity measures in developing countries. determined transfers are not made available in a reliable In principle, a system requiring local governments to fashion. And in Argentina the factors in the formula are cover some portion of service costs out of their own funds complex, contradictory, and not optimally equalizing. is desirable for accountability and efficiency reasons. In Fortunately, better examples can be found around the addition, given that different localities have quite differ- world. The Philippines, for example, distributes 20 percent ent capacities to finance services, it may be appropriate to of national internal revenues to local governments. Similarly, require different degrees of local finance (matching Colombia distributes 25 percent of (non-earmarked) ratios). An example of such a system is in Zambia, where national current revenues to departmental (intermediate- local governments receive a transfer that equals the dif- level) govemments, in part in equal portions and in part on ference between the estimated cost of providing a speci- the basis of population. Colombia also distributes 30 per- fied level of local services and the expected revenues to be cent of value-added tax revenues to municipal govemments, raised locally by applying a standard set of local tax rates. largely on the basis of population. Although these systems The basic problem with this approach is that it is quite have their flaws, they are objective, stable, and transparent. demanding in terms of information. Even these systems, however, make no formal provision The design of intergovernmental transfers is particular- for periodic reexamination of the total amount to be dis- ly important for the social sectors, where central govem- tributed or the distribution formula (although Colombia's ments maintain a strong interest in certain expenditures 1991 constitutional reform provided for a one-time reeval- and outcomes (as in public health and primary education). uation of the municipal grant system after five years). In Rules and incentives set by the center through intergov- contrast, formally federal countries such as India and ernmental transfers are the key to ensuring that central Pakistan have established fiscal commissions to carry out mandates are realized. In Vietnam, despite central efforts such evaluations and to recommend changes every five to target more funds for primary education and basic years. India, for example, has had 10 such reports over the health care to poorer provinces, the funds do not neces- years and has made numerous changes in the distribution sarily reach the poorer districts and communes or even the amount and formula as a result. South Africa's Financial intended subsectors, since provinces are not provided with and Fiscal Commission may play a similar role. specific instructions on how to allocate the funds. Designing Decentralization: The Building Blocks of Fiscal Federalism 15 first. Assignment of expenditure functions can broad- ernmental transfers and local taxes. The transfers ly follow the subsidiarity principal, which addresses should address both vertical imbalances and horizon- interjurisdictional spillovers; however, fine-tuning of tal inequities, but assigning at least minimal tax instru- functional assignments within countries will lead to ments to local levels is an important part of account- different outcomes. Once expenditure functions are ability. The combination of transfers and taxes should determined, revenues should be assigned to different cover local recurrent expenditures. Subnational bor- levels of government to ensure that services can be rowing (under the conditions described in the follow- financed and there are no unfunded mandates. Local ing section) should serve as the last source of finance revenues will come from a combination of intergov- for the capital budget. 4. Designing Decentralization: Incorporating Institutions The impact of the intergovernmental fiscal system on * One common strategy is to restrain provincial and economic or service delivery outcomes depends on local borrowing by, for instance, limiting such bor- how that system is designed and on the institutional rowing to financing for capital expenditures, limit- setting in which it is implemented. Institutions can be ing debt service to a maximum percentage of cur- defined as the rules of the game in society (or the incen- rent revenues, or requiring prior approval of central tives and constraints that influence human behavior) government for borrowing. and the organizations and other means to enforce * A more fundamental approach is to remove the insti- them. Both the rules and the enforcement mechanisms tutional problems that give rise to unsustainable sub- influence the design of decentralization. national borrowing. This focus might include reas- Institutions have had an increasingly important signing revenues and expenditures to provide local influence on decentralization in recent years because of governments with some own sources of revenue with technological change, political demands for greater par- which to finance local expenditures, revising the ticipation, and the emergence of global and local capi- transfer system, introducing transparent, timely, and tal markets, among other factors (Box 3). This section reliable reporting systems, and establishing a stable, identifies five institutional factors that shape the design accepted periodic review process. These measures of decentralization and influence its economic impact: are related to the fiscal system that binds different the regulatory framework, the organization of service tiers of government. Equally important would be to delivery, information systems and competition, the ensure transparent reporting by and regulation of the potential for asymmetric decentralization, and the need financial sector-particularly as it relates to borrow- for policy synchronization. Taken together with the tra- ing by subnational governments. ditional public finance perspective, these factors reflect International experience suggests an important objec- an approach to decentralization that is based on the tive in the design of an intergovernmental fiscal system: design of institutional incentives and rules of gover- keeping the fiscal and financial systems separate (Box 4). nance as instruments for better economic management. Indeed, the lack of separation between fiscal and finan- cial systems-rather than the decentralization process itself-may have led to macroeconomic instability in Adjusting the Regulatory Framework and some countries, as the web of implicit obligations sud- Decentralizing Borrowing denly appeared as explicit budget commitments. In addition, government involvement in the financial sec- There are many sound justifications for allowing sub- tor may dampen the economic role of capital markets in national governments to finance their own investments allocating credit and signaling creditworthiness. Thus it (for example, the intertemporal nature of large capital could limit the financial sector's ability to promote eco- investments). But experience has shown that main- nomic efficiency and local accountability. taining a hard budget constraint for subnational gov- In Argentina, for example, provincial government ernments is both essential and extremely difficult. deficits have been financed in part by provincial banks, Several approaches are possible: many of which have gone bankrupt. As Wildasin 16 Designing Decentralization: Incorporating Institutions 17 Box 3: Decentralization in South Africa: Redesigning Intergovernmental Fiscal Relations and Creating Institutional Incentives South Africa has embarked on a comprehensive decen- ty and governance. In addition, the central government is tralization of political, fiscal, and financial responsibili- reforming budget systems. A uniform accounting system ties to provinces and local governments. At the political is being established for municipalities in hopes of setting level, the apartheid system was formally ended with the up clear budget rules to link all tiers of government to an elimination of racial jurisdictions, the holding of elec- integrated budget system. tions on a nonracial basis for all tiers of governments, and On the financial side, a regulatory framework is being the writing of a new constitution. Legislation is being developed to enable local governments to access local enacted to implement a new fiscal system for subnation- capital markets directly The framework will have clear al governments, to provide a framework for subnational rules for public sector bankruptcy This policy measure borrowing, and to provide institutional incentives for effi- complements the political and fiscal decentralization by cient service delivery. In addition, the central government decentralizing borrowing powers. Borrowing for long- has initiated a multipronged strategy for capacity build- term infrastructure development will be more appropri- ing to enable successful decentralization. The overall ate in large metropolitan areas, while smaller urban gov- framework does not, however, take a uniform view of ernments and local rural governments will still require decentralization. Guided by the demographic and eco- intergovernmental transfers to finance capital invest- nomic concentration of the urban sector (which accounts ment. A capital grants program from the center to the for 60 percent of the population and 80 percent of local level has been established to accommodate this GDP)-and perhaps by its political clout-decentraliza- need. But in the case of the provinces, access to capital tion is being implemented rapidly for local governments markets has been prohibited. This policy is consistent in urban areas, particularly in metropolitan centers, but with the policy of not granting provinces any significant with greater caution in the provinces. own-taxes because it avoids creating implicit liabilities A major objective of the fiscal system in the post- for the central government from provincial access to cap- apartheid era is to improve the distribution of income ital markets. To facilitate capital investment at the imposed by apartheid. The policies implied by this objec- provincial level, a program of capital grants is expected tive involve the efficient assignment of expenditures, cen- to be implemented. tral financing of redistributive measures, and the devel- In addition, legislation is enabling local governments opment of a stable and predictable intergovernmental to restructure the institutional organization of municipal grant system. Given their redistributive and spillover delivery systems. In particular, municipalities are legally impact, education and health have been designated as able to experiment with partnerships with the private sec- provincial responsibilities but are financed from the cen- tor (such as concessions) and nongovernmental organiza- ter. The central government uses its "divisible pool" of tions (such as community delivery systems). Thus the national revenues to support transfers to poorer jurisdic- decentralization framework not only assigns responsibili- tions through two channels: one for provinces and one ties to local authorities but also extends it to the private for local governments. Both systems are formula driven. sector and community groups. The institutional focus also For local governments the formula is based on household covers metropolitan areas, where a policy debate is under per capita income and for provinces on household per way about the appropriate structure of metropolitan gov- capita income augmented by a measure of the rural econ- ernments. Similarly, a debate is under way about the omy in a region. As a result the transfers have shifted appropriate structure of welfare and health systems, cur- resources into poorer jurisdictions. The system has also rently the responsibility of provinces, and its implication eliminated the ad hoc and inequitable fiscal transfers of for provincial governance. the apartheid era, which shifted resources from central to The central government has focused on capacity- subnational tiers for the purpose of balancing end-of-the- building measures to support the overall framework of year local deficits. decentralization. For example, rather than impose top- To complement the transfers, all local authorities have down, supply-driven systems of capacity measures, the been given access to user charges, property taxes, and central government provides grants to municipalities to some business taxes, and can set rates for user charges and buy capacity for projects involving the private sector in property taxes. But provinces do not have their own tax the delivery and financing of municipal services. instruments, creating potential problems of accountabili- Similarly, the central government (in partnership with 18 Rethinking Decentralization in Developing Countries donors and the private sector) has launched manage- ment is an important design issue in the formulation and ment training programs aimed at municipal authorities. implementation of intergovernmental systems. A similar program for provincial treasuries is being for- Decentralization is far from complete, and it is far too mulated. early to judge the outcome of the measures taken. But Finally, the central government is investigating institu- what is striking is the comprehensive vision of South tional settings for coordinating economic policies in a African policymakers in the design of the decentralization decentralized system. Currently, national policies (such as framework. Political, fiscal, financial, and institutional those for standards and personnel) are set by central line changes are being managed simultaneously and in differ- ministries, but financing and implementation are in the ent ways for different jurisdictions. Few countries have hands of subnational governments. This divergence is taken such a fundamental approach to reforming their very acute in the social sectors (such as education and intergovemmental systems. Given its potential to offer health) and, therefore, in the relationship between the other countries a host of lessons in the reform of inter- center and the provinces. How to coordinate national governmental systems, South Africa will be watched very policies with an emerging multitiered system of govern- closely by researchers and policymakers for years to come. (1997) points out, the central bank's policy of manag- evidence suggests that the de facto structure of inter- ing these banks and absorbing their losses provided governmental fiscal relations includes the use of state provincial governments with a circuitous mechanism banks, and their relationship to the central bank of inflationary finance. It also weakened incentives for through the financial regulatory system shifted implic- fiscal discipline at the provincial level. In Brazil too, it liabilities for state deficits to the central bank. In Box 4: Separating Fiscal and Financial Systems and the Evolution of Capital Markets Separating fiscal and financial systems helps the growth of As capital markets emerge with financial instruments implicit liabilities for the central government and so offering long-term finance, the need for discount facilities reduces the possible macroeconomic consequences of will disappear. And as long as the fiscal system is well decentralizing borrowing powers. It also raises several designed, subnational governments will be able to access issues on the nexus between the sequencing of policies for longer maturities. For fiscally weaker municipalities (such separating fiscal and financial systems and the institu- as rural local governments), however, transfers would tional development of capital markets. remain an important vehicle for access to funds. In countries without a domestic capital market, it may Finally, institutional changes in the organization of be preferable to have central fiscal transfers (from a cen- delivery systems provide mechanisms for accessing capital tral government borrowing from international sources) markets to expand the delivery of services (such as water provide additional resources to subnational govern- and electricity) while offering mechanisms to separate fis- ments, while focusing central government policy efforts cal and financial systems. For example, even where capital on developing a private commercial banking system. This markets are weak, the creation of privately managed deliv- approach avoids the creation of development finance ery systems (such as concessions for water and electricity) institutions, keeps fiscal and financial systems separate, may provide access to equity and debt from private inter- and potentially offers a better starting point for a munic- national markets. Similarly, smaller municipalities may ipal finance system. The reform of the financial sector in pool their service delivery systems by contracting out to a Eastern Europe is a case in point. As the private banking regional utility and take advantage of economies of scale system emerges and establishes itself, long-term finance on the production side. This option may alleviate the need can be provided through discount facilities funded by for the public sector to catalyze or manage a financial pool- central governments. Discount facilities take only the ing instrument to lower the cost of funds for smaller local maturity risk by stretching the term of commercial bank government units (for example, the Michigan Bond Bank lending to municipalities. Commercial banks, however, used such a financial pooling mechanism). What remains retain the credit risk for their retail lending, thus pre- important, however, is to develop a regulatory framework serving the separation of risks between the public and that ensures these delivery systems are not inherited by the private sectors. public sector in the case of bankruptcy. Designing Decentralization: Incorporating Institutions 19 commenting on the Brazilian story, Dillinger (1997) tral government's budget and a formal part of the cen- points out that the hard budget constraint imposed on tral govemment's liabilities. This measure will eliminate the fiscal side was circumvented because states have- a possible source of implicit liabilities and may provide wittingly or not-used the financial system to obtain a check through the budget system on the spending federal resources. Similarly, in China the weak revenue behavior of development finance institutions and base of the center has created political pressures on the municipal development funds.9 The credit allocation People's Bank of China to offer credit to lower-level mechanism must be made transparent and binding to governments. As Lall and Hofman (1995) suggest, reduce ad hoc and politically influenced allocation of such policy lending can have a negative impact on finance. In addition, the municipal development fund macroeconomic price stability and, presumably, fiscal must not provide subsidized funding since this will hin- discipline at the subnational level. der the development of local capital markets-which These examples suggest various mechanisms for ultimately will be the sustainable, sound source of local separating fiscal and financial systems. An important investment finance (Peterson, 1997). Finally, a credible issue is, where possible, getting the public sector out exit strategy for the institution would be needed, espe- of the financial system. This includes privatizing the cially as local capital markets develop. banking system. But by itself privatization is not suffi- Macroeconomic crises linked to subnational finance cient to enforce the separation. As the U.S. savings and have been caused more by a lack of institutional sepa- loan crisis and elements of the recent Asian financial ration between fiscal and financial systems than by crisis suggest, private banks must be supervised with a decentralization per se. In fact, decentralization may clear prudential, regulatory framework that monitors even provide incentives for a central government to the level and nature of their aggregate liabilities. institutionalize the separation of these systems. When The central government also must clarify the fiscal subnational governments have direct access to capital rules of the game and follow them. For example, it markets, the risk of inheriting liabilities may provide an should first clearly assign expenditure responsibilities impetus for the central govemment to monitor bor- to each level of govemment. After these responsibilities rowing, making the interaction between subnational have been assigned, tax instruments should be assigned governments and capital markets-whether through to subnational govemments to increase their access to public or private intermediation-more transparent. own-revenues, and arrangements should be fixed for In addition, it may be useful to develop a mecha- three to five years to improve predictability. Similarly, nism by which any emergency central support required ensuring the stability and predictability of the fiscal to work out such debt problems will carry with it the transfer system and implementing the legal framework obligation to introduce and make effective all these for pledging revenues as debt payments are also essen- reforms under the supervision of a central review board tial. Transfers and own-revenues provide the "collater- (World Bank, 1996a). To avoid any moral hazard al" necessary for access to capital markets. In fact, the behavior from this implicit insurance scheme, howev- lack of access to financial markets by subnational gov- er, it would be necessary to specify through legislation ernments often results in central govemment interven- the bankruptcy procedures. tion in financial markets when the real issue is design- Still, at least two basic guidelines for subnational ing and implementing an appropriate fiscal framework. borrowing seem to be needed. The first is a limit on On the other hand, if financial intermediation is borrowing solely for investment (which may, of course, undertaken by the public sector, as in the case of devel- be difficult to enforce in the absence of strictly segre- opment finance institutions or municipal development gated capital budgets). The second is a requirement for funds, the separation of fiscal and financial mechanisms explicit national approval for foreign borrowing. At a may need to be enforced in other ways. For example, it deeper level, given the heavy dependence on central would be important to ensure that the capital borrow- transfers of subnational governments in most develop- ing of public entities responsible for onlending to sub- ing and transition economies, stronger restrictions on national govemments is an explicit line item on the cen- borrowing may be warranted-particularly if some 20 Rethinking Decentralization in Developing Countries subnational units are large enough to affect national respect to such critical matters as the justification for, macroeconomic variables (Wildasin, 1997). Unless and appropriate level of, public subsidy and the role of subnational governments are able to save themselves different tiers of the public sector in financing, deliv- from fiscal crises by drawing on their taxing powers, ering, or regulating a service. Economic support ser- their only options are bankruptcy or bailouts. In some vices such as those involved in rural development may cases the only way to reduce the moral hazard implic- raise still other problems. it in this situation may be by imposing stricter limits In this connection, it is useful to distinguish on subnational borrowing than would be needed if between public delivery of services (production) and capital markets functioned well and the central gov- public financing of services (provision). It is also emnment's nonguarantee of local debt was credible. important to distinguish between production efficien- Finally it should be noted that there is almost no case cy (that is, the cost of delivering a given quantity and for central government lending for local investment in quality of service) and allocative efficiency (that is, the infrastructure-other than the possible argument that extent to which public expenditures reflect local such lending may help develop a private capital market demand). In principle, there is a wide-almost to finance such investment. Central finance may be unbounded-variety of forms in which particular pub- appropriate if the central government has an interest in lic services may be produced and provided, and there local infrastructure, but it should be provided through, is no necessary link between the two (Box 5). for example, competitive grants from a social investment Schools, for example, may be run by a local educa- fund (Glaessner and others, 1994) rather than through tion authority, a local government, a nonprofit body, or subsidized credit. As experience with municipal devel- a private company This is the production side, to which opment funds has shown, subsidized financing agencies issues of production efficiency (such as economies of almost inevitably become contaminated by political con- scale and scope) relate. In some instances-such as cerns and end up blocking rather than aiding the devel- schools-a case can be made that public production is opment of private capital markets. Thus they weaken a more efficient way of obtaining a desired output (say, rather than strengthen the potential regulatory role of uniform primary education) than is regulation and such markets with respect to subnational borrowing. monitoring of contracted production. In other cases public production may be warranted to achieve economies of scale-as in the technical design of super- Organizing Service Delivery highways in countries that do not afford sufficient scope for the development of competitive private producers The broader institutional approach to decentralization (and that do not want foreigners to enter the market). has several important implications. For example, to the It is important not to confuse the question of who extent that decentralization is intended to improve the should deliver a service with the question of how much delivery of services, it is essential to consider in detail should be provided and who should pay for it. The not only the nature of each service but also the struc- more fundamental question from an allocative point of ture of delivery. Not only does infrastructure invest- view is who pays, not who delivers. And, as noted, the ment raise quite different questions than do social ser- two can be quite different. It is also important to deter- vices, there are also important differences within each mine who decides who delivers services. It is still an category. As the World Bank's World Development Report open question whether local governments should be 1994 shows, for example, roads, power, telecommuni- allowed to choose to deliver services or contract out. cations, and water and sewerage each have specific Where procedures are transparent and basic rules of requirements with respect to the most appropriate procurement are in place, this may be a good idea. But institutional structure for efficient and equitable ser- without these preconditions, it may not. vice delivery (see also Kessides, 1993). Similarly, edu- In one sense pnvatization is the ultimate form of cation, health, and social assistance-or, more broad- decentralization. In recent years a wide variety of coun- ly, the "social safety net"-may be very different with tries have seen the private sector play an increasingly Designing Decentralization: Incorporating Institutions 21 Box 5: Alternative Ways to Produce or Provide Local Public Services Services can be produced (delivered) or provided * Build-operate-transfer (BOT) or build-own-operate (financed) by many different public institutional struc- (BOO) arrangements tures, including: * Other forms of public-private partnerships * Central government (department, decentralized * Development charges, exactions, and similar schemes agency, or enterprise) * Franchise arrangements * Regional government (department, decentralized * Service contracts agency, or enterprise) Finally, public services can also be both produced and * Local government (department, decentralized agency, provided by private agents, sometimes in response to or enterprise) coercive legal requirements and sometimes even volun- * Central-regional arrangement tarily: * Central-local arrangement * Compulsory provision by developers * Central-regional-local arrangement * Compulsory provision by individuals (through vouch- * Regional-local arrangement ers or self-financed) * Association of local governments * Voluntary provision (through formal or informal * Special-purpose local authority (encompassing more arrangements) than one local government, coterminous with a local * Provision by nongovernmental organizations (church- government, or covering less area than a local govern- es, enterprises) ment) Even this extensive list is incomplete. Many of the In each of these cases the service may be produced by arrangements listed have a number of possible variants, one structure (such as a regional enterprise) and provid- and there are ways to combine all these organizational ed by another (such as a specific local government structures. Moreover, different structures might apply for department). policymaking, regulation, financing, production, and so Services can also be provided in whole or part by the on. Finally, different structures may apply for different ser- public sector but produced by the private sector. Again, vices and for local governments with different character- there are many possible forms: istics (size, financial capacity, and so on). Source: Bird, 1995. large role in the delivery and to some extent the financ- vatized or fully funded through user charges without ing of many public services, depending on the extent breaching either efficiency or equity concerns. Although of externalities involved. New forms of service delivery regulation of private services can help deal with some incorporating private, public-private, and horizontal issues-for example, content of educational curricu- and vertical combinations of public and private actors lums-it cannot deal with others-for example, equal already exist, and more are being created all the time. access for the poor-without continuing an explicit or Private delivery of services is important in developing implicit subsidy. Consequently, although privatization countries because it loosens the traditional monopoly may often be one way to deal with many problems of of public service providers, which often provide ineffi- public service delivery, in other ways it may make the cient, poor-quality services. task of the institutional policy designer even more com- Market solutions to service delivery offer many plex, since the number of agents has expanded and the advantages in terms of efficiency and may even be equi- degree of control over agents has diminished. table (for example, by enabling public subsidies to be targeted to the poor, or by reducing the travel costs for the poor to obtain quality services). Most public ser- Establishing Information Systems and vices, however, embody some degree of "publicness," Competitive Governments some relevant externalities, some broader political or social purpose (often redistribution), or some combina- One of the theoretical arguments supporting decen- tion of these features. Such services cannot be fully pri- tralization is that local governments have better 22 Rethinking Decentralization in Developing Countries information about local preferences; thus decentral- for competition is, on one hand, the ability of citizens ization should enhance allocative efficiency to compare governments in terms of the services they Information asymmetries make it difficult for central provide and the taxes they levy ("exit") and, on the governments to monitor local government agents. other, the ability of citizens to affect and alter the deci- The decentralization literature developing this argu- sions of government ("voice"). It is important to ask ment has implications for the design of intergovern- how relevant this mechanism is for developing coun- mental grants and the earmarking of transfers and tries, where poor information, weak democracies and revenues (Cremer, Estache, and Seabright, 1995; traditions of participatory decisionmaking, and often Boadway, Marceau, and Sato, 1997). But no serious weak rural labor markets constrain both popular par- research has been done in developing or transition ticipation and interjurisdictional mobility (voice and economies to test such basic propositions as the pre- exit). Moreover, poor information and weak capacity sumed greater knowledge of local preferences by local make hierarchical accountability particularly difficult rather than central governments. There is, however, in developing countries. anecdotal evidence that decentralization increases the One key lesson relevant everywhere is thus that the availability of local information (Crook and Manor, more that is known, and the more publicly it is 1994). Recent work on Colombia is particularly inter- known, the better the outcome of decentralization esting in this respect (World Bank, 1995b). efforts is likely to be, whatever their rationale and Nonetheless, it is important to keep informational whatever the circumstances in which they occur.12 constraints in mind when analyzing and prescribing Because the feasible and desirable implementation of decentralization strategies. this generalization may vary widely from place to However the decentralization literature is inter- place and for different services, World Bank contribu- preted, it seems clear that the efficiency of any com- tions to supplying and supporting information rele- petition engendered by decentralization will largely vant to understanding and evaluating the impacts of depend on the rules of the game that are set and decentralization can play a crucial role in improving enforced by the central government or perhaps by an outcomes. As with the information-theoretic ap- independent judiciary. As Inman and Rubinfeld proach to decentralization, the competitive approach (1996) show, electoral competition alone is unlikely to signals both the importance of improving knowl- produce efficiency-although as Breton (1996) edge-and public awareness of knowledge-in order shows, government competition can take many other to improve outputs and the importance of considering relevant forms in addition to counting votes. Very lit- in detail the institutional structure of particular decen- tle research has been done-even in industrial coun- tralization arrangements. It is not enough to discuss, tries-on the effect alternative institutional forms for example, tax assignment or the case for or against have on outcomes. This issue is particularly important equalization transfers. Close attention must also be in developing countries, where institutional settings paid to the details of how such fiscal institutions actu- vary dramatically and can strongly influence the out- ally work in the context of particular countries in comes of decentralization. Thus developing a better order to understand and analyze the incentives facing understanding of this issue should be a top priority for various decisionmakers. Indeed, assigning revenue the World Bank. 1° sources to localities that lack channels for public par- One conclusion that can be emphasized, however, ticipation may be counterproductive. is that the key to usefully competitive governments is An unfortunate side effect of decentralization in to make relevant decisionmakers accountable for their some countries has been the virtual disappearance decisions, and that the key to effective accountability from the central government's cognitive horizon of reli- is to make relevant comparative information publicly able information on the provision of such services. The available.11 Accountability and public information are role of the central government in ensuring and moni- easier to achieve in industrial countries than in devel- toring effective and efficient decentralization is espe- oping countries. At its base, the ultimate mechanism cially critical when the main goal is to enhance service Designing Decentralization: Incorporating Institutions 23 delivery, particularly for services (such as health and Adopting Asymmetric Decentralization education) that are important not only for national development but also for poverty alleviation and gen- Economic, demographic, and social diversity is often eral welfare. Decentralization does not mean that the reflected in a multitude of government and delivery central government no longer has any responsibility in structures even within a single country. For example, these areas. What it means is that the nature of this South Africa has provincial and regional governments, responsibility has changed from delivering services rural districts, one and two-tiered metropolitan gov- directly to regulating and monitoring the efficiency and ernments and smaller urban municipalities, and spe- equity of services delivered by others-usually local cial service delivery units such as water boards and governments. The essential tool for this task is an accu- health districts. Given such diversity, both in the rate information system-generated, for example, by nature of political jurisdiction and the characteristics requiring local governments to file uniform and infor- of households, "one size fits all" is definitely not true mative budgets and financial reports. Uniform infor- for decentralization. Different instruments may have mation is important so that comparisons can be made very different effects in different circumstances, and across communities and yardstick competition used. very different approaches may be needed to achieve Fiscal transparency is fundamental to sound decen- similar (or acceptable) results. For example, privatiz- tralization policy and, indeed, public policy in gener- ing water services may achieve efficiency and equity al. All government accounts should be comprehensive, objectives in a dense, urban setting, but it may fail to comprehensible, and widely available. Budgets should reach similar goals in a sparsely populated rural be drawn up to display the real status of public region. Private sector delivery and financing of water finances, ideally including the long-run implications services may then have to be complemented with pub- for taxation of current spending decisions. This acute lic sector and community delivery systems for specif- need for reliable information has been neglected in ic areas. To accommodate the need for such diverse most decentralizing countries (Box 6). approaches, asymmetrical central policies-treating Some of the needed incentive for compliance with different units differently-may be required to achieve information demands can be created for local gov- similar outcomes. ernments by making timely submission of such An important element of such an approach is the reports a condition of receiving fiscal transfers. This principle of asymmetrical decentralization. For exam- requirement may imply significant setup costs for sys- ple, in many countries it may be feasible to decen- tem design, training, and implementation. To spend tralize political, economic, and administrative re- money wisely in a decentralized system may thus sponsibilities to large urban areas. Similarly, at the require some initial investment. While such concerns regional level, fiscal and administrative capacity may are less important when full responsibility for the effi- make it easier to decentralize responsibilities only to ciency and equity of services are devolved (as in a for- some provinces or states. In other cases it may be fea- mal federal structure) rather than simply delegated to sible to decentralize responsibilities directly from subnational governments, the need for an adequate central government to the private sector rather than information system remains critical to macroeco- to local governments. nomic management even in this instance, given the But asymmetrical decentralization may raise a fun- importance of such services as education and health damental political problem: the perceived need to in total public expenditure. Regardless of the form of have a law that treats all units similarly, in the face of decentralization, an important institutional problem the reality that there are wide and relevant differences is thus how to ensure that the relevant central agen- between them. In South Africa, for example, there is cies have adequate incentives to monitor subnational growing political debate about the capacity of many activity And, as noted earlier, such information is newly formed provinces-especially those that have even more necessary when democratic institutions inherited significant administrative problems from the are important. apartheid system-to manage the delivery of social 24 Rethinking Decentralization in Developing Countries Box 6: Why Do We Know So Little about Local Government Finance? Good data on local finance are hard to find in developing Incidence and other problems of economic analysis are countries. In some countries careful examination of critical in this connection, but surprisingly little empiri- national data sources may disclose some relevant infor- cal progress has been made toward resolving such prob- mation, but data are often old, incomplete, budgeted lems even in the most developed, and statistically sophis- rather than actual, aggregated (say, by region), or in some ticated, countries. Finally, and in the end perhaps most other way not easily usable for analytical purposes. important, local finance data are poor because no one is Moreover, information is often hard to interpret. Few really asking for improvement: other things seem much developing countries compile local fiscal data in a way more important, and resources are scarce, so this omis- that is comparable over time, let alone across countries. sion may seem understandable. What information exists is often the result of special and Such neglect is often a mistake, however. Local gov- occasional studies; the main international example is the ernments are among the most important service deliver- massive work of Bahl and Linn (1992). A common reac- ers in most countries; unless they do better, the public sec- tion to this work by those with little experience in the field tor is not likely to do much better, and how can their is, "Why do so many of the numbers refer to the early performance be assessed in the absence of information? 1980s or even the 1970s?" But those who have carried out Moreover, governments everywhere are decentralizing; such work can only applaud the massive data collection again, how can decentralization policies be properly effort reported in this volume. No similar effort seems to designed, or adequately implemented, in the absence of have been carried out anywhere since then. the information needed to assess them? Finally, in the Local finance data are poor for a number of reasons. absence of good information, policies and assessments are First, few countries attach much importance to its sys- too often based on what central officials or visiting experts tematic collection. Second, collecting such data is difficult happen to know, or think they know, about what is going in countries with low administrative capacity and poor on in local governments. Given the variability of local real- communications systems. Third, local government struc- ity in most countries, such biased judgments may be very tures are often complex and differentiated, which makes misleading. It is difficult to read accounts of local finance it difficult to design adequate data collection procedures in China, for example, without being reminded of the tale and to ensure that the data gathered is reliable. Fourth, of the seven blind men describing an elephant, each inap- local government finance is complicated: tracing fiscal propriately generalizing from the bit they happen to know flows among and within the array of local governance All in all, paying more attention to getting the facts straight bodies found in most countries is not easy. Fifth, even if about local (or, more broadly, subnational) finance would information is obtained, how should it be interpreted? seem to be a worthwhile investment in most countries. Source: Bird, 1995. services such as health and education. An option may decentralization of political and economic powers and be to decentralize responsibilities only to those does not leave it to the discretion of the center but, provinces that show the capacity to manage such ser- rather, to the actions of elected representatives of sub- vices. But this option may be politically difficult to national governments. implement because it will require some elected repre- sentatives to agree to receive less autonomy than their counterparts in other provinces. In such cases, and Synchronizing Policy depending on the country, solutions may be sought through such means as making the application of a As noted earlier, a difficult but important design issue particular decentralization provision conditional on is how to match fiscal, political, and administrative the satisfaction of a number of preconditions, on the arrangements to achieve the potential benefits of decen- voluntary assumption of certain financing or other tralization. Most countries do not get the mix right- obligations, or on the signing of a contract that per- they devolve decisionmaking to local levels without mits individuation of both the timing and extent of providing budgets to make meaningful decisions, or decentralization. Such legislation assures the ultimate they decentralize finance without ensuring adequate Designing Decentralization: Incorporating Institutions 25 local accountability through political decentralization. tion but must be responsible to the central government Even within the fiscal realm, coordination is extremely for achieving nationally specified outcomes. The whole- difficult. Countries sometimes decentralize expenditure sale decentralization of the health sector led to the fail- responsibility without providing adequate revenues (as ure of one state to undertake a family planning program in Russia and other countries of the former Soviet deemed important by the central government in the Union), or decentralize revenue but mandate virtually Philippines, and to a drop in vaccination coverage in all expenditures (as in Ecuador). Or the central gov- Papua New Guinea (Kolehmainen-Aitken and emnment devolves a service fiscally and politically, yet Newbrander, 1997). determines all wage and employment policies, or issues The issue of providing flexibility to local levels in the locality with a long list of service delivery mandates determining wages and job requirements-decentraliz- that provides very little flexibility in local decisionmak- ing the design of labor market conditions in the public ing. Decentralization has been likened to a souffle sector-provides an important example of the need for where all ingredients must be present in the right policy synchronization in the design of decentralization. amounts and prepared in the right way to achieve suc- In the Philippines, for example, decentralization of the cess (Parker, 1995). Moreover, like a souffle, the best health system resulted in health care workers doing the method of preparation will depend on the environment samejob being paid at three different rates based on their and the best mix of ingredients is a matter of taste. previous employment (central, provincial, or local). This Policy synchronization is difficult not only because gave rise to morale and budgetary problems. Inadequate of the many fiscal, political, and administrative issues attention to the financing of pension obligations when that require careful consideration, but also because each transferring employees from one level of govemment to service and even each function within a service will dif- another has complicated decentralization efforts in fer with regard to the appropriate form of decentraliza- Colombia and many other countries. The failure to give tion. Depending on the nature of the service (external- local governments any leeway in staff complements or ities and interjurisdictional spillovers), the political salaries largely vitiated initial decentralization efforts in landscape, and possibly the administrative capacity, the a number of Eastem European countries (such as amount of autonomy given to a local government will Romania). Such issues are especially important with differ (Bossert, forthcoming).13 For example, certain respect to large, labor-intensive activities such as health public health programs (family planning, immuniza- and education, where matters are further complicated tions) are of national concern, and thus local govern- by externalities (for example, with respect to public ments may be given some autonomy for implementa- health) and basic welfare (accessibility) concerns. 5. Rethinking Decentralization in Developing Countries To debate whether decentralization is good or bad is attained. Many developing countries, though, have unproductive and misleading since the impact of weak representative decisionmaking processes and decentralization depends on design. Yet it is essential local elites are often deeply entrenched. Local elites to consider the wide range of issues that influence often take leadership roles, and although not necessar- decentralization. The traditional fiscal federalism ily bad, this can result in the hijacking of resources approach is a starting point, but the need for a stronger unless transparency and accountability are somehow focus on institutions-both the rules that influence the enforced (Narayan, 1998). One way to improve local behavior of actors at different levels of government, in participation and accountability is through transparent the private sector and in civil society, and the organi- budgeting processes and public procurement proce- zations that implement those rules-is increasingly dures. Experience in Brazil and Mexico has shown that evident in the World Bank's operational work. This participatory budgeting (that is, where all citizens are broader agenda has led to an enhanced focus on invited to a public meeting to discuss budgetary prior- accountability and capacity, has strong implications for ities) can provide a critical link between communities the Bank's project design and policy dialogue, and calls and government. 14 Governments should actively seek for a reinvigorated research effort focused on develop- community participation and use performance-based ing countries. budgeting so that their constituents know not only what inputs were used but what outcomes were achieved. Similarly, publicizing procurement bids, Incentives for Accountability proposals, and selection enhances community aware- ness and ultimately accountability Stressing the Decentralization is dispersing fiscal, political, and importance of community participation in local deci- administrative responsibilities across different tiers of sionmaking should be an important component of the government and between the public and private sec- dialogue on decentralization. tors. Responsibility for delivering services, for exam- Second, by diffusing responsibilities across different ple, may lie with all or some tiers of government, with entities, decentralization provides a basis for compari- community groups, or with the private sector. The son and competition (even if indirect). In developing challenge is to design decentralization so that it creates countries, where interjurisdictional mobility may be incentives that hold each entity accountable for its constrained, competition among service providers of responsibilities and makes explicit the institutional goods that can be privately consumed within a jurisdic- relations between each entity tion increases options for residents. 15 (Decentralization How can decentralization provide such incentives? and private sector development have an interdependent First, in democratic settings political decentraliza- relationship whereby the former enables the latter, and tion and elections provide direct political accountabil- the latter strengthens the former.16) ity More broadly, as Breton (1996) demonstrates, in all Third, in distributing fiscal instruments to alllevelsof functioning political structures there are a variety of government, with the right to set rates, decentralization means by which indirect political accountability is creates incentives for fiscal accountability Being forced 26 Rethinking Decentralization in Developing Countties 27 to tax one's constituency-either to deliver additional and respond to individual preferences? Does the cen- services or to pay for policy mistakes-is an important tral government have the administrative capacity to restraint on political decisionmaking. But in developing provide technical and financial support where appro- countries, where it may not be easy for people to move priate? Do local governments have the capacity to to the jurisdiction where the tax and service bundle deliver promised services at low cost? Political capaci- matches their preferences or to vote the incumbent deci- ty has to do with the channels for local participation sionmaker out of power, maximizing accountability and administrative capacity with the processes re- means selecting tax instruments that closely match taxes quired of central and local administrators and with and services. In this respect, user charges are particular- their training. The evidence on these issues is sparse. ly important for accountability because they create a Local administrative capacity is sometimes consid- close link between the delivery agent and the client. 17 ered inadequate because bureaucratic requirements Finally, access to well-functioning markets by gov- imposed by the center are inappropriate for local deci- emments and households may provide additional sionmakers. In reality, if the appropriate requirements check on politicians and administrators. For example, were assigned to each level of government according to decentralizing borrowing powers offers a mechanism the information required for them to perform their for using capital markets to provide accountability by functions, local capacity would probably not be as sending market signals on the performance of govern- problematic. In other cases local administrative capac- ments and private firms. Well-functioning land mar- ity is identified as a problem by the central government kets make "voting with one's feet" a credible threat when in fact the central government may also lack the because policy decisions get capitalized in land values. capacity to manage local affairs. In yet other cases the In both cases information helps ensure that markets design of intergovernmental fiscal relations does not function efficiently. Strengthening the regulatory provide guidelines, resources, and incentives that framework to improve the functioning of markets is would lead to strong local capacity All these factors essential for establishing the self-correcting methods of may be in play at the same time. For example, in South accountability that will lead to successful and sustain- Africa limited administrative capacity at the provincial able decentralized decisionmaking. level has constrained the decentralization of education Properly designed and sequenced decentralization and welfare payments. There has been less debate on may deliver each of these forms of accountability. The whether the central government has the capacity to framework of incentives and accountability is especial- manage the delivery process or whether the failure is ly important given the uncertainty that generally per- driven by the design of the intergovernmental system vades the decentralization process. We may not be able (Ahmad, 1995). Local capacity is a complicated issue, to say exactly what the "correct" form of decentraliza- and the appropriate way to improve it may not simply tion is for a particular country, but we do know that cor- be through increased training of local officials. rect institutional incentives are essential both to reveal Colombia provides an example of how local gov- mistakes and to provide a self-regulatory mechanism. ernments have relied on demand-driven processes for both infrastructure investment and capacity building. This approach has increased allocative efficiency The Role of Capacity (matching investments to local preferences) and direct- ed resources to the poor (Box 7; World Bank, 1996b). Increasing fiscal, political, and administrative respon- The experience with developing capacity for local sibilities and setting clear rules holding each tier of gov- decisionmaking and administration is encouraging for emment accountable for those responsibilities lead to believers in the potential allocative and democratic an important set of questions: Does the public sector- virtues of decentralization. As in such well-known that is, central and local governments-have the Asian cases as the Orangi project in Karachi, Pakistan capacity to support and manage decentralized respon- (Bird, 1995), local participation appears both to reveal sibilities? Does it have the political capacity to identify preferences and to keep costs down. Depending on the 28 Rethinking Decentralization in Developing Countries Box 7: Developing Capacity in Colombia Under Colombia's "coparticipation" system, local com- has totally privatized road maintenance; another has put munities provide labor and local materials for new pro- private developers in charge of constructing urban roads. jects, and municipal governments contribute a portion In other localities computers have been introduced to of the cost. This approach not only fosters community monitor water and sanitation services. In addition, munic- involvement in identifying needs and choosing projects, ipalities have started to share equipment and they have it also promotes community participation in the execu- improved their technological capability for internal orga- tion, operation, and maintenance of the works. nization, planning, and monitoring to ensure better pro- Municipalities have to prepare projects that are then ject management. appraised using technical and environmental criteria. Underlying these improvements is a more basic Project beneficiaries should be low-income rural fami- change: Colombian municipalities have been moving lies, and projects can be carried out by any contractor from a supply-driven (top down) to a demand-driven (private contractors, nongovernmental organizations, (bottom up) approach to public services. Increasingly, state agencies, universities) that competes to supply the reflecting the new liveliness of local politics and (with sub- works and services. stantial variations from area to area) more extensive com- Preliminary evidence is surprisingly encouraging on munity participation, people are getting what they want local capacity to carry out such functions. A recent study rather than what someone in the capital thinks they of 16 municipalities found that decentralization has should want. Emphasis has been put on roads, education, enhanced local capacity in terms of labor, capital, and and water works; those are the needs that people perceive, technology (World Bank, 1995b). Colombian municipal- and those are the needs that newly empowered and ities are, for example, increasing the skills of local bureau- responsive local governments are attempting to satisfy cracies through such means as competitive hiring, sharing Opinion surveys suggest that the resulting sectoral alloca- the services of professionals among municipalities, train- tion of resources is consistent with community prefer- ing municipal employees, and rotating personnel across ences, with most respondents indicating that they trust different departments in the same municipality. Capital the local government more than the national government spending capacity has also increased. One municipality to deliver goods and services. Source: World Bank, 1995b. nature of the project, community involvement may private sector and nongovernmental organizations but also enhance targeting of the poor, since participants intergovernmental contracts and agreements as well. who self-select communal work projects are willing to In addition, the discussion of capacity constraints volunteer their labor without remuneration. may incorrectly assume that all subnational govern- In assessing the link between decentralization and ments are similar. In reality, urban governments (par- capacity, therefore, several issues and competing ticularly in large cities) have more capacity to manage hypotheses need to be raised. The initial impression and finance service delivery than do rural govern- that capacity constraints may inhibit decentralization ments. Similarly, different regional governments- seems to be an offshoot of a paradigm that looks at provinces and states-have different fiscal and man- capacity building in a top-down, supply-driven frame- agement capacity In this context the principle of work. The policy implication of this view is that capac- asymmetric decentralization offers an approach to ity building should precede decentralization. A com- decentralizing responsibilities where feasible rather peting hypothesis suggests a more dynamic and than relying on an all or nothing approach. demand-driven relationship between decentralization and capacity building: shifting responsibilities to lower-tier governments may provide the incentive for Policy Dialogue and Project Design public officials to invest in capacity building or seek creative ways to tap into existing sources of capacity The institutional perspective makes clear the impor- The latter could include not only outsourcing to the tance of understanding the concrete and specific cir- Rethinking Decentralization in Developing Countnes 29 cumstances of the country concerned. What can be done imposed on local decisionmakers by citizens ("voice"), in terns of decentralization largely depends on where, by competition ("exit"), and by the public sector ("'hier- when, and how it is done. Initial conditions determine archy"; World Bank, 1997). Governments are general- the level of trust, the reputation of the various actors, the ly not as responsive, transparent, or predictable as they existence (and rigidity) of constraints to institutional are in industrial countries. Thus the same policies change, and so on. The interests of the main stakehold- implemented in an industrial country may have very ers in the existing system must be understood to ascer- different consequences in a developing country (and tain the existing or potential political support for change this outcome will vary between developing countries and the resources available, and to determine the best depending on their institutions). strategy (objectives) and tactics (timing, sequence, dura- Since the outcome of decentralization depends heav- tion) for change. Decentralization is political, and ade- ily on the institutional arrangements with which decen- quate political and institutional analysis is needed to tralization policy interacts, successful decentralization undertake it successfully and appraise its success-or to (that is, a process that achieves efficiency objectives assist or intervene in the process effectively. without jeopardizing equity or macroeconomic stabili- As noted, decentralization has strong implications ty) requires a careful examination of each country's for resource mobilization and use. Because these issues institutions. A framework for assessing institutional affect all of the World Bank's main policy concerns- weaknesses and factoring them into the design of growth, macroeconomic stability, service delivery, and, decentralization policy is shown in Table 2. This list is most important, poverty reduction-the Bank must meant only as a starting point. We must identify which engage countries in a dialogue on the design of decen- institutional structures are most important to achieve tralization. The potential benefits of decentralization the benefits of decentralization and avoid the pitfalls, can only be achieved-and the potential pitfalls can encourage their development, and compensate for only be avoided-if policy design focuses on creating institutional weaknesses (at least in the short run) when the appropriate institutional arrangements in which advising decentralizing governments on policy design. decentralization can occur. In the longer run, institutional strengthening is essen- Even if decentralization efforts are far from optimally tial. But given that decentralization is happening now, designed, the Bank may still be able to improve their and institution building takes time, our advice should allocative and distributive outcomes-for example, by take these institutional weaknesses into account. enhancing local capacity and working directly with local In countries that seem to lack most forms of officials (as in Colombia) or by using adjustment lend- accountability, aspirations for decentralization should ing to support a productive decentralization agenda (as be modest. In these cases decentralization can do more in the Kyrgyz Republic). Useful dialogues on decentral- harm than good, and the best technical advice the Bank ization have been held in countries as diverse as can provide is how to go forward with some benign Vietnam-where focusing on rural service delivery forms of decentralization (such as local trash collec- rather than central-provincial relations proved helpful- tion) while encouraging the institutional development and South Africa-where a low-key but potentially high necessary for further advances. payoff dialogue on decentralization has been sustained through a pivotal period in the country's history. Tools of public policy exist in all countries to bring More Case Studies, Data, and Research about some desired consequences of decentralization. Yet institutions differ dramatically between industrial Because decentralization is an inherently country-spe- countries and developing countries, and these differ- cific (and sometimes activity-specific) process, consid- ences have implications for the extent and type of erable contextual information is required to analyze decentralization that may be appropriate. In develop- and assess, let alone assist. Detailed, information- ing countries weak democracies, factor markets, and intensive case studies are the essential building blocks regulatory frameworks affect the accountability to knowledge in this field. 30 Rethinhing Decentralization in Developing Countries Table 2: Moving toward Better Design of Decentralization in Developing Countries Institutional Weakness Impact Design Feature to Compensate Weak democratic institutions * Less local accountability through "voice" mechanisms * Create channels for community and processes * Greater chance of local elites capturing benefits participation * Decisionmaking less transparent, predictable, * Initiate process of participatory budgeting responsive * Legislate open, public procurement procedures Weak legal and regulatory * Less interurisdictional mobility and less * Diversify local service providers to give systems accountability through "exit" mechanisms more choice (provide technical assistance Weak markets for land, * Decisionmaking less transparent, predictable, to facilitate public-private partnerships) labor, and capital responsive * Consider earmarking user fees to improve accountability Weak information systems * Less "hierarchical" accountability * Create more local incentives for compliance Weak regulatory systems * Decisionmaking less transparent, predictable, with central objectives responsive * Improve informaticn systems Prioritize key regulations Weak information systems * Moral hazard * Establish transparent and internally Weak regulatory systems * Soft budget constraint between levels of govemment consistent multitiered budgeting systems Weak financial systems * Potential central govemment bailout * Establish subnational debt reporting, Nontransparent fiscal monitoring, and rules for central systems government intervention L Legislate independence of central bank All of the above * All of the above . Encourage only modest decentralization of certain local services Detailed data should be collected from country case Given the importance of accurate, timely, and reli- studies and a cross-country dataset developed to pro- able information with respect to what is going on and vide a context against which to assess country experi- its effects on various policy outputs, the Bank should ences. Only those who have tried to do cross-country consider fostering and supporting not only the devel- comparisons of decentralization can appreciate how opment of information systems but also, and perhaps bad, inadequate, and incomparable is the existing data- more important, the creation of governmental and base on almost anything one might wish to examine. It especially nongovernmental forums and centers of is not much of an exaggeration to say that one can expertise (such as the Center for Local Government in prove, or disprove, almost any proposition about Hungary or the official but still independent Financial decentralization by throwing together some set of cases and Fiscal Commission in South Africa). Unless and or data. More comparable and meaningful cross-coun- until the citizens of decentralizing countries have bet- try data are needed on various measures of decentral- ter access to information and analysis, they, the gov- ization (Huther and Shah, 1998), effects of alternative ernments involved, and outside agencies (such as the transfer designs (Ahmad, 1996), and so on. Though in Bank) will continue to stumble along with respect to recent years much work has been done on this issue by many of the issues that are most critical to designing academics and international agencies, much more and assessing decentralization policies. remains to be done. The World Bank's Decentralization More generally, decentralization is an issue about Thematic Group has undertaken a project to merge which it may truly be said that "the devil is in the several cross-country data collection efforts from dif- details." Much research needs to be done-primarily ferent parts of the Bank and from several external part- in case studies-on a variety of important issues, par- ners in order to form a comprehensive database that ticularly with respect to the outcome of alternative includes fiscal and institutional country variables. An institutional changes. These include but are by no effort will be made to regularize the collection of sub- means limited to the effect of intergovernmental trans- national data in all client countries. fers on local tax efforts and service expenditures, the Rethinking Decentralization in Developing Countries 31 effect of fiscal decentralization on macroeconomic extent to which different decentralization initiatives indicators and capital market development, alternative improve (or worsen) governance, and potential uses forms and methods of strengthening local institution- and limitations of user charges as a means of financing al and service delivery capacity, assessments of the decentralized services. Annex. World Bank Economic and Sector Work on Decentralization, 1988-Present Report Report Number Type Date Country Title General 13399 ER 10/26/94 - Stability in a decentralized economy 11620 SR 10/1/94 - Health priorities and options in the World Bank's Pacific member countries 8634 ER 5/1/90 - Public sector decentralization: economic policy reform and sector investment programs Sub-Saharan Africa 16593 ER 11/26/97 Ethiopia Public expenditure review 13011 SR 11/30/94 Madagascar Decentralization and local government reform 12930 SR 11/10/94 Eritrea Options and strategies forgrowth 9643 SR 3/1/93 Malawi Public sector management review: selected issues 8997 SR 4/23/92 Kenya Local government finance study 9101 ER 6/1/91 Madagascar Beyond stabilization to sustainable growth 8995 SR 6/1/91 Zaire A review of govemment expenditure 8974 SR 2/1/91 Nigeria Urban transport in crisis 7844 SR 1/1/91 Nigeria Road sector strategy paper 8221 SR 12/1/89 Ghana Fiscal decentralization 17963 SR 10/1/89 Mozambique Food security study 7495 SR 10/1/89 Tanzania Population, health and nutrition sector review 6930 SR 8/1/89 Zaire Urban sector development in Zaire: a new approach to poverty alleviation East Asia and the Pacific 16737 SR 6/1/97 Papua New Guinea Accelerating agricultural growth: an action plan 16047 SR 2/25/97 Lao PDR Priorities for rural infrastructure development 15745 ER 10/31/96 Vietnam Fiscal decentralization and the delivery of rural services: an economic report 15573 SR 6/27/96 China Higher education reform 15898 SR 6/11/96 Philippines Education financing and social equity: a reform agenda 14933 SR 11/13/95 Philippines A strategy to fight poverty 14540 ER 10/18/95 China Public investment and finance 14680 SR 9/5/95 Philippines Public expenditure management for sustained and equitable growth 13847 SR 4/3/95 Korea, Rep. of Transport sector: resource mobilization challenges and opportunities 13313 SR 11/30/94 Philippines Power sector study: structural framework for the power sector 13143 ER 9/26/94 Vietnam Public sector management and private sector incentives 12929 SR 9/15/94 China Power sector reform: toward competition and improved performance 12343 SR 5/23/94 Philippines Devolution and health services: managing risks and opportunities 11764 ER 11/1/93 Malaysia Managing costs of urban pollution: country economic report 12407 ER 10/1/93 Indonesia Fiscal decentralization: towards a new partnership for progress 11094 SR 7/28/93 China Budgetary policy and intergovernmental fiscal relations 10716 SR 1/1/93 Philippines Fiscal decentralization study 10050 ER 11/27/91 China Reforming intergovernmental fiscal relations 9354 ER 6/1/91 China Economic development inJiangsu Province 8532 ER 8/1/90 Lao PDR Issues in public economics 8074 ER 6/1/90 China Reforming social security in a socialist economy 7927 SR 12/1/89 Indonesia Power sector institutional development review 32 Annex. World Bank Economic and Sector Work on Decentralization, 1988-Present 33 Report Report Number Type Date Country Title 7605 SR 6/1/89 China Revenue mobilization and tax policy 7483 ER 6/1/89 China Macroeconomic stability and industrial growth under decentralized socialism 7098 SR 3/31/88 Philippines Transport sector review 7007 ER 3/1/88 Indonesia Selected issues of public resource management Europe and Central Asia 16420 ER 7/3/97 Estonia Public expenditure review update 16112 SR 6/25/97 Ukraine Public expenditure review: restructuring govemment expenditures 15601 ER 2/24/97 Azerbaijan Poverty assessment 15353 ER 7/15/96 Kazakhstan Transition of the state 14546 SR 6/28/96 Bulgaria Private sector assessment 14925 SR 12/19/95 Estonia Financing local governments 14862 SR 11/29/95 Russian Federation Fiscal management in the Russian Federation 14929 SR 8/1/95 Russian Federation Housing reform and privatization: strategy and transition issues 14470 ER 7/20/95 Latvia Local government expenditures and resource transfers 14110 SR 6/13/95 Russian Federation Poverty in Russia: an assessment 13187 SR 12/12/94 Russian Federation Restructuring the coal industry: putting people first 13039 ER 9/28/94 Poland Growth with equity: policies for the 1990s 12273 ER 2/23/94 Bulgaria Public finance reforms in the transition 11302 ER 12/1/92 Russian Federation Intergovernmental fiscal relations in the Russian Federation 11190 SR 9/1/92 Romania Decentralization and local government reform 10446 ER 7/1/92 Poland Decentralization and reform of the state 10061 ER 5/1/92 Hungary Reform and decentralization of the public sector Latin America and the Caribbean 15543 SR 5/30/96 Belize Environmental report 15298 ER 4/29/96 Colombia Reforming the decentralization law: incentives for an effective delivery of services 15044 ER 4/26/96 Paraguay The role of the state 15272 ER 2/22/96 Bolivia Poverty, equity and income: selected policies for expanding eaming opportunities for the poor 14085 SR 7/7/95 Colombia Local government capacity: beyond technical assistance 13304 SR 1/20/95 Guatemala Basic education strategy: equity and efficiency in education 12673 SR 8/8/94 Colombia Poverty assessment report 12293 SR 6/29/94 Paraguay Poverty and the social sectors in Paraguay: a poverty assessment 11933 SR 3/2/94 Colombia Toward increased efficiency and equity in the health sector: can decentralization help? 11130 SR 12/23/93 Venezuela Venezuela 2000: education for growth and social equity 11160 ER 12/23/92 Venezuela Decentralization and fiscal issues 8972 ER 1/21/92 Venezuela Public administration study 8924 SR 7/1/91 Mexico Decentralization and urban management: urban sector study 8918 ER 1/1/91 Ecuador Public sector finances: reforms for growth in the era of declining oil output 9093 SR 11/1/90 Honduras Social sector programs 8994 SR 8/1/90 Colombia Decentralization reform: a review of political and administrative aspects 7870 SR 10/1/89 Colombia Decentralizing revenues and the provision of services: a review of recent expenence Middle East and North Africa 16522 SR 5/8/97 Tunisia Higher education: challenges and opportunities 13233 SR 7/26/96 Iran Education, training and the labor markets 14043 SR 5/1/95 Islamic Intergovemmental fiscal relations and municipal finance Rep. of Jordan management sector study 10157 SR 6/1/92 Morocco Issues and prospects in the public sector 8782 ER 8/1/90 Morocco Decentralization in local government expenditures and management 7150 SR 3/1/88 Tunisia Municipal finance and management 7115 SR 2/1/88 Morocco Municipal finance sector report 34 Rethinking Decentralization in Developing Countries Report Report Number Type Date Countrv Title South Asia 15677 SR 7/31/97 India The Indian oilseed complex: capturing market opportunities 16506 ER 5/30/97 India 1997 economic update: sustaining rapid growLh 16558 SR 5/1/97 Bangladesh Municipal finance management sector study 16113 ER 12/20/96 Bhutan Country economic memorandum 15905 ER 7/31/96 Bangladesh Public expenditure review 14960 SR 6/6/96 Pakistan Improving basic education: community participation, system accountability, and efficiency 15092 SR 11/20/95 Pakistan Supporting fiscal decentralization in Pakistan 9518 SR 12/20/91 India Irrigation sector review 7946 ER 1/1/90 Bangladesh Poverty and public expenditures: an evaluation of the impact of selected government programs Note: ER is Economic Report; SR is Sector Report. Notes The authors are grateful for helpful comments provided by citizens or simply as the superior jurisdiction-should domi- Thomas Bossert, Shanta Devarajan, William Dillinger, Cheryl nate, and the design problem is as just stated (Seabright, 1996). Gray, Robert Inman, Christine Wallich, and Dana Weist. 6. The tax and expenditure decisions made by local govern- 1. It is interesting to note that Africa, the region with the high- ments can also have a major impact on redistribution. est proportion of decentralization projects, has completed the 7. Bird and Gendron (1998) suggest that surcharges on a least formal economic and sector work on the broader topic of value-added tax may also be feasible in some circumstances, decentralization or intergovemmental fiscal relations. By con- provided administrative capabilities are high enough. trast, Europe and Central Asia, the region that has focused its 8. An altemative view is the "federalist approach," where the analytical work most on intergovemmental fiscal relations, has primary objective is to ensure that all regions of the country are the lowest proportion of decentralization projects. The experi- able to provide such services at acceptable minimum standards ence from these regions suggests that there may be inadequate and have adequate resources to do so. In this case simple lump- connection between sectoral and project issues on one hand and sum transfers, with no conditionality other than the usual the macroeconomic and broader policy dialogue on the other. requirements for financial auditing, are called for (Shah, 1994). 2. See, for example, recent studies by the Intemational In this approach it is essentially assumed that the fact that the Monetary Fund (Ter-Minassian, 1997), Inter-American funds flow to locally responsible political bodies will ensure suf- Development Bank (1994, 1997; Lopez-Murphy, 1996), and ficient accountability and that it is neither necessary nor desirable CEPAUGTZ (Aghon and Krause-Junk, 1996), as well as the for the central govemment to attempt to interfere with, or influ- recent conferences sponsored by the U.S. Agency for Inter- ence, local expenditure choices. national Development (Washington D.C., October 1997) and the 9. The success of this option would depend, of course, on Food and Agriculture Organization (Rome, December 1997). the rules that guide the budget process. 3. The traditional methods used by the Bank to promote 10. For example, recent World Bank research on decentral- transparency and accountability-procurement rules and ization of education has indicated that the most important ele- financial audits-are inadequate for improving the quality of ment in improving outcomes (as part of education reform) is govemment at the local level because of the lack of basic insti- parent involvement in teacher supervision (King and Ozler, tutions to ensure political and economic accountability. 1998). Knowing which aspects of decentralization are most 4. Dillinger explored the importance of institutional issues important can help in creating the appropriate institutional for Bank work in urban areas in his discussion of urban finance structures to benefit from decentralization. in 1994. This paper furthers that analysis. 11. Democracy without good information is not enough; 5. Alternatively, as in the traditional economic theory of "fis- nor, of course, is information without democracy Nonetheless, cal federalism" (Oates, 1972), both levels of govemment may even in countries without truly democratic institutions in which be seen as agents of different groups of citizens, with the cen- decentralization is simply another instrument of the central tral government encompassing the subsets included within the government, good information is essential to improving service purview of the different local governments. So long as local gov- outcomes. ernments are hierarchically subordinated to the central govem- 12. Toward this end, the Economic Development Institute is ment, both approaches lead to the same conclusion: the central working in Africa (Zimbabwe) and Latin America (Venezuela) government's preferences-whether seen as representative of all to develop pilot experiences with municipalities to increase 35 36 Rethinking Decentralization in Developing Countries transparency through participatory budgeting and public infor- 15. Pure public goods will be underprovided and under- mation campaigns. consumed if left to the market and thus must still be provided 13. Since some functions are of national importance, the by government. Thus accountability for local public goods must principal-agent framework (whereby the center provides man- still rely on electoral processes or interjunsdictional mobility dates and incentives so that local govemments will achieve 16. Technical assistance can be provided to local govern- desired results) is appropriate. Yet decentralization provides ments to facilitate public-private partnerships. For example, risk myriad opportunities for independent local decisionmaking management templates for solid waste have been designed and that affects efficiency, equity, and quality of services. How much used in the Philippines to help local governments analyze the local discretion (or "decision space") is provided to local gov- structure of financing and the risks faced for different aspects of emments will differ depending on the sector and the particular service delivery under different contractual agreements. function. (For more discussion, see Bossert, forthcoming.) 17. If indeed mobility is constrained, then general local 14. Porto Allegro, Brazil, offers a fascinating case of trans- taxes-such as a sales tax or a piggybacked income tax-are parent budgeting and community participation in local govern- appealing from an efficiency standpoint. (The problem of ment fiscal decisionmaking. Indeed, Porto Allegro and a dozen interjurisdictional tax competition would not exist.) Yet in municipalities in Brazil publish their budget on the Internet (see most developing countries the administrative capacity is weak http://www.prefpoa.com.br/opart/default.htm). Regrettably, for such tax collections. If local democracy is also weak, then most participatory budgeting in Latin America has focused only it may be more likely that funds raised through earmarked on capital works, not on recurrent revenue and expenditures user fees will be spent more efficiently to better reflect local (see also Paul, 1995, on India). preferences. References Aghon, Gabriel, and Gerold Krause-Junk. 1996. Bird, Richard M., and Francois Vaillancourt, eds. 1998. Fiscal Descentralizaci6n Fiscal en America Latina: Balance, Pricipios, Decentralization in Developing Countries. Cambridge: Desafi6s. Santiago, Chile: Proyecto Regional de Cambridge University Press. Descentralizaci6n Fiscal CEPAUGTZ. Bird, Richard M., Robert D. Ebel, and Christine 1. Wallich, eds. Ahmad, Ehtisham. 1996. Financing Decentralized Expenditures: 1995. Decentralization of the Socialist State. A Regional and An International Comparison of Grants. Cheltenham, U.K.: Sectoral Study Washington, D.C.: World Bank. Edward Elgar. Boadway, Robin. 1997. "Public Economics and the Theory of Ahmad, Junaid. 1995. "Funding the Metropolitan Areas of Public Policy" Canadian Journal of Economics 30 (4a): South Africa." Finance and Development 32 (September): 753-72. 50-53. Boadway, Robin, Nicolas Marceau, and Motohiro Sato. 1997. Alderman, H. 1998. "Decentralization and Targeted Transfers: "Subsidies versus Public Provision of Private Goods as Social Assistance in Albania." living Standards Measurement Instruments for Redistribution." Discussion Paper 942. Study Working Paper 134. World Bank, Washington, D.C. Queen's University, Institute for Economic Research, Canada. Bahl, Roy W 1998. "China: Evaluating the Impact of Bossert, Thomas. Forthcoming. "Decentralization of Health Intergovernmental Fiscal Reform." In Richard Bird and Systems: Decision Space, Innovation and Performance." Francois Vaillancourt, eds., Fiscal Decentralization in Social Science and Medicine. Developing Countries. Cambridge: Cambridge University Brennan, Geoffrey, andJames M. Buchanan. 1980. The Power to Press. Tax. Cambridge: Cambridge University Press. Bahl, Roy W, and Johannes Linn. 1992. Urban Public Finance in Breton, Albert. 1996. Competitive Governments. Cambridge: Developing Countries. New York: Oxford University Press. Cambridge University Press. Besley, Timothy, and Anne Case. 1995. "Incumbent Behavior: Breton, Albert, and Anthony Scott. 1978. The Economic Vote-Seeking, Tax-Setting, and Yardstick Competition." Constitution of Federal States. Toronto: University of Toronto American Economic Review 85 (March): 25-45. Press. Bird, Richard M. 1993. "Threading the Fiscal Labyrinth: Some Breton, Albert, Alberto Cassone, and Angela Fraschini. 1998. Issues in Fiscal Decentralization." National TaxJournal 46 (2): "Decentralization and Subsidiarity: Toward a Theoretical 207-27. Reconciliation." University of Pennsylvania Journal of .1994. "Decentralizing Infrastructure: For Good or For International Economic Law 19 (1): 21-51. Ill?" Policy Research Working Paper 1258. World Bank, Buchanan, James, and Richard Wagner. 1971. "An Efficiency Washington, D.C. Basis for Federal Fiscal Equalization." In J. Margolis, ed., - 1995. "Financing Local Services: Patterns, Problems, Analysis of Public Output. Princeton, NJ.: Princeton and Possibilities." Major Report 31. University of Toronto, University Press for National Bureau for Economic Research. Centre for Urban and Community Services. Cremer, Jacques, Antonio Estache, and Paul Seabright. 1995. Bird, Richard M., and Ariel Fiszbein. 1998. "Colombia: The "The Decentralization of Public Services: Lessons from the Central Role of the Central Governrnent in Fiscal Theory of the Firm." In Antonio Estache, ed., Decentralizing Decentralization." In Richard Bird and Francois Vaillancourt, Infrastructure: Advantages and Limitations. Washington, D.C.: eds., Fiscal Decentralization in Developing Countries. World Bank. Cambridge: Cambridge University Press. Crook, Richard, and James Manor. 1994. Enhancing Bird, Richard M., and Pierre-Pascal Gendron. 1998. "Dual VATs Participation and Institutional Performance: Democratic and Cross-Border Trade: Two Problems, One Solution?" Decentralisation in South Asia and West Africa. A Report to International Tax and Public Finance 5 (3): 429-42. ESCOR, the Overseas Development Administration. 37 38 Rethin king Decentralization in Developing Countries Davoodi. Hamid, and Heng-fu Zou. 1998. "Fiscal Kenyon, Daphne. 1991. Interjurisdictional Tax and Policy Decentralization and Economic Growth: A Cross-Country Competition: Good or Badfor the Federal System? Washington, Case Study"Journal of Urban Economics 43: 244-57. D.C.: Advisory Commission on Intergovernmental Relations. Dillinger, William. 1994. Decentralization and Its Implicationsfor Kessides, Christine. 1993. Institutional Optionsfor the Provision UrbcmSurviceDelivery. Urban Management Programme Series of Infrastructure. World Bank Discussion Paper 212. 16. WVorld Bank, Washington, D.C. Washington, D.C. -. 1997. "Brazil's State Debt Crisis: Lessons Learned." King, Elizabeth M., and Berk Ozler. 1998. "What's Economic Notes 14. World Bank, Country Department 1, Decentralization Got to Do with Learning? The Case of Latin America and the Caribbean Region, Washington, Nicaragua's School Autonomy Reform." Impact Evaluation of D. C. Education Working Paper Series 9. World Bank, Washington, Ehdaie, Jaber. 1994. "Fiscal Decentralization and the Size of D.C. Government," Policy Research Working Paper 1387. World Kolehmainen-Aitken, Riitta-Liisa, and William Newbrander. Bank, Washington, D.C. 1997. Decentralizing the Management of Health and Family Fiszbein, Ariel. 1997. "Local Government Capacity in Planning Programs. Boston, Mass.: Management Sciences for Colombia." World Development 25 UJuly): 1029-43. Health. Fox, Jonathan, and josefina Aranda. 1996. "Decentralization Komai, Janos. 1992. The Socialist System. Princeton, NJ.: and Rural Development in Mexico: Community Participation Princeton University Press. in Oaxacas Municipal Funds Program." Monograph Series. Lall, Rajiv, and Bert Hofman. 1995. "Decentralization and the University of California, Center for U.S.-Mexican Studies, Government Deficit in China." In Macroeconomic San Diego. Management and Fiscal Decentralization. EDI Seminar Series. Glaessner. Philip J., Lee Kye Woo, Anna Marian Sant'anna, World Bank, Washington, D.C. Jean-Jacques de St. Antoine, and others. 1994. Poverty Litvack, Jennie. 1994. "Regional Demands and Fiscal Alleviation and Social Investment Funds: The Latin American Federalism." In Christine Wallich, ed., Russia and the Fxperience. World Bank Discussion Paper 261. Washington, Challenge of Fiscal Federalism. A Regional and Sectoral Study D.C. Washington, D.C.: World Bank. Grossman, David. 1996. "Strengthening the Institutional Litvack, Jennie, and Claude Bodart. 1993. "User Fees Plus Framework." In "Colombia: Reforming the Decentralization Quality Equals Improved Access to Health Care: Results of a Law: Incentives for an Effective Delivery of Services." World Field Experiment in Cameroon." Social Science and Medicine Bank, WNashington, D.C. 37 (3): 369-83. Hines,J., and R. Thaler. 1995. "The Flypaper Effect."Journal of Lopez-Murphy, Ricardo, ed. 1996. Fiscal Decentralization in Economic Perspectives 9 (4): 217-26. Latin America. Washington, D.C.: Center for Research in Huther, Jeff, and Anwar Shah. 1998. "Applying a Simple Applied Economics, Inter-American Development Bank. Measure of Good Governance to the Debate on Fiscal Martinez-Vazquez, Jorge, and Robert McNab. 1997. "Fiscal Decentralization." Policy Research Working Paper 1894. Decentralization, Economic Growth, and Democratic World Bank, Washington, D.C. Governance." Paper prepared for USAID Conference on Inman, Robert P, and Daniel L. Rubinfeld. 1996. "The Political Economic Growth and Democratic Governance, Economy of Federalism." In Dennis C. Mueller, ed., Washington, D.C., October 9-10. Perspectives on Public Choice. Cambridge: Cambridge May, RonaldJames. 1971. Financing the Small States in Australian University Press. Federalism. New York: Oxford University Press. - 1997. "Rethinking Federalism." Journal of Economic McLure, Charles E., Jr. 1995. "Comment on Prud'homme. ' The Perspectives 11 ( 4): 43-64. World Banh Research Observer 10 (2): 221-26. Inter-American Development Bank. 1994. "Fiscal De- Mueller, Dennis C., ed. 1996. Perspectives on Public Choice. centralization: The Search for Equity and Efficiency" In Cambridge: Cambridge University Press. Economic and Social Progress in Latin America: 1994 Report. . 1997. "Breton Competitive Governments: An Baltimore, Md.: Johns Hopkins University Press for the Inter- Economic Theory of Politics and Public Finance." A Book American Development Bank. Review Canadian Journal of Economics 30 (4a): 997-1003. - 1997. 'Fiscal Decision-making in Decentralized Musgrave, Richard A. 1983. "Who Should Tax, Where, and Democracies." In Economic and Social Progress in Latin What?" In Charles McLure, ed., Tax Assignment in Federal America: 1997 Report. Baltimore, Md.: Johns Hopkins Countries. Canberra: Centre for Research on Federal Financial University Press for the Inter-American Development Bank. Relations, Australian National University References 39 Narayan, Deepa. 1998. "Participatory Rural Development." In Fiscal Decentralization in Developing Countries, Ernst Lutz, ed., Agriculture and the Environment: Perspectives on Washington, D.C., May 7. Sustainable Rural Development. Washington, D.C.: World Bank. . 1997b. "Decentralized Government and Macro- Oates, Wallace. 1972. Fiscal Federalism. New York: Harcourt economic Control." Paper prepared for the International Brace Jovanovich. Institute of Public Finance 53rd Congress, Kyoto, Japan. Parker, Andrew N. 1995. "Decentralization: The Way Forward Tanzi, Vito. 1996. "Fiscal Federalism and Decentralization: A for Rural Development?" Policy Research Working Paper Review of Some Efficiency and Macroeconomic Aspects." In 1475. World Bank, Washington, D.C. Michael Bruno and Boris Pleskovic, eds., Annual World Bank Paul, Samuel. 1995. A Report Card on Public Services in Indian Conference on Development Economics 1995. Washington, Cities: A View from Below Bangalore, India: Public Affairs D.C.: World Bank. Centre. Ter-Minassian, Teresa, ed. 1997. Fiscal Federalism in Theory and Pauly, Mark. 1973. "Income Distribution as a Local Public Practice. Washington, D.C.: International Monetary Fund. Good," Journal of Public Economics (2): 35-58. Tiebout, Charles. 1956. "A Pure Theory of Local Expenditures." Peterson, George E. 1997. "Building Local Credit Systems." Journal of Political Economy 64 (5): 416-24. Urban Management Program Policy Paper. World Bank, Treisman, Daniel. 1998. After the Deluge: Regional Crisis and Washington, D.C. Political Consolidation in Russia. Ann Arbor: University of Prud'homme, Remy. 1995. "The Dangers of Decentralization." Michigan Press. The World Bank Research Observer 10 (2): 201-20. Tresch, Richard. 1981. Public Finance. Plano, Texas: Business Ravallion, Martin. 1993. "Poverty Alleviation through Regional Publications. Targeting: A Case Study for Indonesia." In Karla Hoff, Avi Triche, Thelma. 1993. "The Institutional and Regulatory Braverman, and Joseph Stiglitz, eds., The Economics of Rural Framework for Water Supply and Sewerage: Public and Private Organization. Oxford: Oxford University Press. Roles." TWURD, WS-9. World Bank, Washington, D.C. - 1998. "Reaching Poor Areas in a Federal System." Triche, Thelma, Abel Mejia, and Emanueal Idelovitch. 1993. World Bank, Development Research Group, Washington, "Arranging Concessions for Water Supply and Sewerage D.C. Services: Lessons from Buenos Aires and Caracas." TWURD, Rondinelli, Dennis. 1981. "Government Decentralization in WS-10. World Bank, Washington, D.C. Comparative Perspective: Theory and Practice in Developing Vaillancourt, Francois. 1998. "Morocco and Tunisia: Financing Countries." International Review of Administrative Sciences 47 Local Governments-the Impact of Infrastructure Finance." (2): 133-45. In Richard Bird and Francois Vaillancourt, eds., Fiscal - 1989. "Decentralizing Urban Development Programs: Decentralization in Developing Countries. Cambridge: A Framework for Analyzing Policy." U.S. Agency for Cambridge University Press. International Development, Office of Housing and Urban Van de Walle, Dominique, and Kimberly Nead, eds. 1995. Programs, Washington, D.C. Public Spending and the Poor: Theory and Evidence. Baltimore, .1998. "What Is Decentralization?" World Bank Poverty Md.: Johns Hopkins University Press. Reduction and Economic Management Knowledge Wallich, Christine I., ed. 1994. Russia and the Challenge of Fiscal Management System, Washington, D.C. Federalism. A Regional and Sectoral Study. Washington, D.C.: Salmon, Pierre. 1987. "Decentralization as an Incentive World Bank. Scheme." Oxford Review of Economic Policy 3 (2): 24-43. Walsh, Cliff. 1992. Fiscal Federalism: An Overview of Issues and Seabright, Paul. 1996. "Accountability and Decentralization in a Discussion of their Relevance to the European Community. Government: An Incomplete Contracts Model." European Canberra: Federalism Research Centre, Australian National Economic Review 40 (1): 61-89. University. Shah, Anwar. 1994. "The Reform of Intergovernmental Fiscal Watts, Ronald. 1996. Comparing Federal Systems in the 1990s. Relations in Developing Countries and Emerging Market Kingston, Canada: Institute of Intergovernmental Relations, Economies." Policy and Research Series 23. World Bank, Queen's University Washington, D.C. Weingast, Barry. 1995. "The Economic Role of Political Shah, Anwar, and Zia Qureshi. 1994. Intergovernmental Fiscal Institutions: Market-Preserving Federalism and Economic Relations in Indonesia. World Bank Discussion Paper 239. Development." Journal of Law, Economics, and Organization 11 Washington, D.C. (1): 1-31. Spahn, P Bernd. 1997a. "Decentralization in Transition Wheare, Kenneth C. 1969. Federal Government. 4th ed. Economies." Paper prepared for the World Bank Seminar on London: Oxford University Press. 40 Rethinking Decentralization in Developing Countries Wildasin, David. 1997. "Externalities and Bailouts: Hard and . 1995b. Local Government Capacity in Colombia. A Soft Budget Constraints in Intergovernmental Fiscal World Bank Country Study Washington, D.C. Relations." Policy Research Working Paper 1843. World . 1996a. "Argentina: Provincial Finances Study" Report Bank, Washington, D.C. 15487-AR. Washington, D.C. Wilensky, Harold. 1974. The Welfare State and Equality: . 1996b. "Colombia: Reforming the Decentralization Structural and Ideological Roots of Public Expenditures. Law: Incentives for an Effective Delivery of Services." Report Berkeley: University of California Press. 15298-CO. Washington, D.C. World Bank. 1988. World Development Report 1988: Op- - . 1996c. "Vietnam: Fiscal Decentralization and the portunites and Risks in Managing the World Economy. New Delivery of Rural Services." Report 15745-VN. Washington, York: Oxford University Press for the World Bank. D.C. .1994. World Development Report 1994: Infrastructurefor . 1997. World Development Report: The State in a Changing Development. New York: Oxford University Press for the World. New York: Oxford University Press for the World Bank. World Bank. Zhang, Tao, and Heng-fu Zou. 1998. "Fiscal Decentralization, .1995a. Better Urban Services: Finding the Right Incentives. Public Spending and Economic Growth in China."Journal of A Development in Practice book. Washington, D.C. Public Economics 67: 221-40. THE WORLD BANK 1818 11 Strcet. N.W. Wiashingtovi, D).C. 20433 lISA Iclephone: 202-477-1234 Fflacsimile: 202-4;77-6391 TIclcx: MCI 64145 WOR 11)IANK MCI 248423 WORIA.)lIANK W\orld Wide \Vch http://!vww.worldhanlk.org/ I.-mtiail: hooks(alwvorldhank.org 9 7rS1111180 Si182111:1N 14151 0 9 780821 343500 I ISBN 0-8213-4350-5