FISCAL EQUALIZATION
FISCAL EQUALIZATION Challenges in the Design of Intergovernmental Transfers
edited by Jorge Martinez-Vazquez Department of Economics, Andrew Young School of Policy Studies, Georgia State University, Atlanta, USA Bob Searle Consultant on Fiscal Decentralization, Formerly the Commonwealth Grants Commission, Australian Government
a- Springer
Library of Congress Control Number: 2006935554
ISBN-I 0: 0-387-48987-8 (Printed on acid-free paper) ISBN-13: 978-0387-48987-2
O 2007 Springer Science+Business Media, LLC All rights reserved. This work may not be translated or copied in whole or in part without the written permission of the publisher (Springer Science+Business Media, Inc., 233 Spring Street, New York, NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafier developed is forbidden. The use in this publication of trade names, trademarks, service marks and similar terms, even if they are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights.
Contents
Contributing Authors PART 1
INTRODUCTION
Chapter 1
Challenges in the Design of Fiscal Equalization and Intergovernmental Transfers Jorge Martinez- Vazquez and Bob Searle
PART 2
THE NATURE OF EQUALIZATION OBJECTIVES AND CONSEQUENCES
Chapter 2
Fiscal Capacity Equalization and Economic Efficiency: The Case of Australia Jeffrey D. Petchey and Sophia Levtchenkova
Chapter 3
Ensuring Inter-Regional Equity and Poverty Reduction Bert Hofman and Susana Cordeiro Guerra
Chapter 4
The Impact of Equalization on Service Delivery Catherine Hull and Bob Searle
61
Chapter 5
Harmonizing Objectives and Outcomes at the National and Sub-national Levels through Citizen Engagement and Capacity Building (with special references to the Philippines) Alex B. Brillantes, Jr. and Jose 0. Tiu Sonco I .
95
Discussant comments: David Pkloquin PART 3
THE INSTITUTIONAL SETTING
Chapter 6
A Framework for Evaluating Alternative Institutional Arrangements for Fiscal Equalization Transfers Anwar Shah
Chapter 7
Intergovernmental Transfers: The Funding Rule and Mechanisms Paul Bernd Spahn
Chapter 8
Intergovernmental Transfers: The Vertical Sharing Dimension Roy Bahl and Sally Wallace
163
Discussant comments: Giorgio Brosio PART 4
CHALLENGES I N IMPLEMENTING EQUALIZATION
Chapter 9
Expenditure-Based Equalization Transfers Richard Bird and Francois Vaillancourt
Chapter 10
Designing Intergovernmental Equalization Transfers with Imperfect Data: Concepts, Practices, and Lessons Jameson Boex and Jorge Martinez- Vazquez
Chapter 11 A Model for Public Infrastructure Equalization in Transitional Economies Sophia Levtchenkova and Jeffrey D. Petchey Discussant comments: Serdar Yilmaz
345
vii
PART 5
THE RELATIONSHIP OF EQUALIZATION TO OTHER POLICIES
369
Chapter 12
Revenue Sharing, Natural Resources and Fiscal Equalization Bob Searle
371
Chapter 13
The Nature and Functions of Tied Grants Bob Searle and Jorge Martinez- Vazquez
403
Chapter 14
Intergovernmental Loans: Their Fit into a Transfer System Dana Weist
435
Discussant comments: Rimy Prud'homme
453
PART 6
THE BIGGER PICTURE
46 1
Chapter 15
The Political Economy of Equalization Transfers Stuti Khemani
463
Discussant Comments: Musharraf R. Cyan
485
INDEX
497
Contributing Authors
Roy Bahl Department of Economics, Andrew Young School of Policy Studies, Georgia State University Richard M. Bird University of Toronto, Canada Jameson Boex Department of Economics, Andrew Young School of Policy Studies, Georgia State University Alex B. Brillantes, Jr. National College of Public Administration and Governance, University of the Philippines
Giorgio Brosio Department of Economics, University of Torino, Italy Susana Cordeiro Guerra The World Bank Musharraf R. Cyan Asian Development Bank Bert Hofman The World Bank Catherine Hull The Commonwealth Grants Commission, Australian Government Stuti Khemani The World Bank Sophia Levtchenkova School of Economics and Finance, Curtin University, Australia
Jorge Martinez-Vazquez Department of Economics, Andrew Young School of Policy Studies, Georgia State University Jeffrey D. Petchey School of Economics and Finance, Curtin University, Australia David PCloquin Department of Finance, Government of Canada R6my Prud'homme Professor Emeritus, University of Paris XII, France Bob Searle Consultant on Fiscal Decentralization, Formerly the Commonwealth Grants Commission, Australian Government Anwar Shah The World Bank Paul Bernd Spahn Goethe University, Frankfurt Jose 0. Tiu Sonco I1 National College of Public Administration and Governance, University of the Philippines Fran~oisVaillancourt Economics Department, UniversitC de Montrkal Sally Wallace Department of Economics, Andrew Young School of Policy Studies, Georgia State University Dana Weist The World Bank Serdar Yilmaz World Bank Institute
1 INTRODUCTION
Chapter 1
CHALLENGES IN THE DESIGN OF FISCAL EQUALIZATION AND INTERGOVERNMENTAL TRANSFERS JORGE MARTINEZ-VAZQUEZ* AND BOB SEARLE** *Department of Economics, Andrew Young School of Policy Studies, Georgia State University,** Consultant on Fiscal Decentralization, Formerly the Commonwealth Grants Commission, Australian Government
Responding to a variety of forces over the past decade, many developing and transitional countries have embarked upon serious fiscal decentralization reform initiatives. In recent years, many developed economies have also deepened their decentralized government structures. Very often the allocation of expenditure and revenue assignments in a fiscally decentralized nation leads to horizontal fiscal imbalances because of the different fiscal capacities and expenditure needs of sub-national governments. Decentralization design also often leads to vertical imbalances in favor of the central government because tax revenue sources are seldom as decentralized as expenditure responsibilities. In overcoming these imbalances and attempting to achieve other policy objectives, equalization grants and other intergovernmental transfers have become key elements of intergovernmental finance reform around the world. Policymakers in both developed and developing countries face significant challenges with the introduction and reform of equalization grants and other intergovernmental transfers. One of these challenges is a lack of any clear framework, in either the decentralization literature or the recorded details of international practice, from which to consider numerous hard issues. For example, should capital expenditure needs of sub-national governments be considered part of equalization grants? Are independent grants commissions a preferred institutional set up for implementing equalization grants? What is the appropriate relationship between the untied and tied grants elements of the system? Should we try to equalize differences in fiscal capacity or
4
Martinez- Vazquez and Searle
expenditure needs, or both? And how do we measure these differences with limited data? This volume is a collection of invited original essays that highlight the state of knowledge in intergovernmental transfer design, especially for developing and transitional countries where data and administrative capabilities may be weak. It presents new thinking about a list of challenging policy issues and provides useful options for policyrnakers. The essays were presented at a conference held in Atlanta in October 2004 entitled "Challenges in the Design of Fiscal Equalization and Intergovernmental Transfers." It was sponsored by the Andrew Young School of Policy Studies at Georgia State University. It is our hope that the volume will deepen our understanding and offer useful solutions to the challenges of intergovernmental fiscal relations reform. In particular, we aimed at providing options for consideration in the design of equalization grants and other intergovernmental transfers, especially in the large number of developing and transitional countries embarking upon decentralization strategies. There are five specific themes in this collection of essays:
1. The fundamental nature and objectives of equalization grants and their consequences on efficiency and equity; 2. The appropriate institutional setting for the design and implementation of equalization grant systems (whether by independent grants comrnissions or central government ministries); 3. The challenges in the design of formulas with limited data availability for recurrent and capital purposes; 4. The coordination of equalization grants with other related policies, including specific purpose grants, revenue sharing and borrowing; and 5. The political economy behind equalization transfers. Each of these themes is represented by a section in the volume. The first set of essays in part 2 examines the objectives and impacts of equalization policies on the efficiency of allocation of resources and interregional equity. The paper by Jeffrey D. Petchey and Sophia Levtchenkova examines how fiscal capacity equalization, when not done correctly, can create an incentive for regions to act strategically in order to influence the size of their grant, leading to inefficiencies in the provision of local public goods. In addition, they show how the interregional transfer from equalization policy may lead to inefficiency in the spatial allocation of mobile factors of production. The result may be an overall lower level of social welfare. The policy implication of this chapter is that the standards chosen for the design of equalization transfers need to be exogenously set. Petchey and Levtchenkova conclude that choosing standards that are a
Challenges in the Design of Fiscal Equalization
5
function of what sub national governments do on average is not safe. On the other hand, they are less optimistic about the possibility of removing the inefficiencies resulting from the spatial allocation of mobile factors caused by fiscal capacity equalization, unless we are willing to shift Gom equalization based on fiscal capacity to equalization based on economic efficiency. Bert Hofman and Susana Cordeiro Guerra review the causes and consequences of interregional fiscal inequities among sub national governments and empirically examine the case of East Asian countries. A fundamental reason for differences in fiscal disparities is the large difference in economic endowments across regions. In the case of East Asian countries, the current transfer systems are only mildly equalizing so that, even after equalization, the remaining fiscal disparities are still large. Hofman and Guerra discuss the possible consequences of these remaining fiscal disparities for service delivery inputs, outputs, and outcomes. The authors provide some guidelines for how to make progress in increasing transparency of inequalities at the sub-national level, and how transfer systems can be made more equalizing without undermining the incentives for local governments to raise their own revenues and to spend their resources efficiently. Catherine Hull and Bob Searle focus on the impact of fiscal equalization on service delivery, using data for Australia. There are many factors capable of influencing service delivery levels. In the Australian case, the system of equalization provides for all States to have the capacity to provide the same standard of services. However, within the autonomy provided by a federal system and the unconditional nature of the equalization grants, it means that States may choose not to provide the same standard of services. Equality in the standard of services would follow only if that satisfied the preferences of the people who live in each State jurisdiction. In short, in Australia the fiscal equalization system is not designed to achieve uniformity but to give flexibility or "to deliver equality in diversity." Thus, there is no reason to expect that sub national service levels will converge under equalization. In reality, service levels in Australia have moved over time within a stable band around the average level of provision. Alex B. Brillantes and Jose 0. Tiu Sonco I1 take a comprehensive view of the role of transfers in the fiscal architecture of a country and conclude that the concepts, rationale, justifications and objectives of intergovernmental fiscal transfers boil down to how sub national governments can finance the delivery of public goods and services in a decentralized government environment where there are vertical and horizontal imbalances. Being a financial input, intergovernmental fiscal transfers do not guarantee achievement of objectives and outcomes set by the central authorities. The list of desirable objectives to be pursued in the design of transfers, according
6
Martinez-Vazquez and Searle
to Brillantes and Tiu Sonco, includes the improvement of service delivery, equitable growth, poverty reduction, and regional and national development and contributing to the attainment of the Millennium Development Goals. In reality, even the most fundamental objective of a more equitable regional distribution of fiscal resources may not be achieved. Because of inferior system design, this is currently the case in the Philippines. Perhaps one reason is that too many objectives are being pursued with a single or with few instruments. David PCloquin's comments, which complete the first section, integrate and add to the insights of the four essays. His contribution is centered on three themes - the equity, efficiency and "good governance" issues surrounding equalization. He concludes with possible lessons for the design of equalization policies and the key governance arrangements that underpin them. Part 3 addresses the issues surrounding the optimal institutional set up for the implementation of equalization transfers. Anwar Shah develops a framework for evaluating alternate institutional arrangements for fiscal equalization transfers. This is an issue that has received little attention in the literature despite its importance in creating a stable fiscal transfer regime. Institutional arrangements vary significantly across countries but can be broadly classified into five stylized models: (1) central government ministqdagency; (2) independent agency (grants commission) reporting either to the executive or the legislature on a permanent or periodic basis; (3) intergovernmental forums including intergovernmental cum civil society forums; (4) national legislature; and (5) sub-national government forums. Comparing the performance of different institutional arrangements is a complex task because relative success depends upon a multitude of factors. The Shah chapter focuses on the incentives and interactions facilitated by the different arrangements and draws implications for their impacts on the transaction costs involved in achieving society's objectives. Shah concludes that, both in theory and practice, the advantages provided by an independent grants commission are greatly exaggerated. Paul Bernd Spahn examines the institutions surrounding the funding rules and distribution mechanisms used for equalization grants. The analysis of the performance of these institutions can be conducted in the context of a specific country, which risks some type of path-dependence in the conclusions, or can be done in a more controlled fashion by looking at a stylized fictional country. Even though all decentralized and federal governments have to respect their historical and political conditions in their reforms, a more abstract analysis can provide sharper insights and serve as a benchmark for potential reforms in the real world. Be that as it may, Spahn uses an abstract approach to find that the h d i n g of transfers has to respect the overall constraint on public resources, so a general principle should be
Challenges in the Design of Fiscal Equalization
7
the closed funding of transfers (achieved either through the assignment of a fixed amount or proportion of the donor government's budget or through tax sharing). If the pool is formed on a single or narrow resource base, there will be excessive volatility or structural shifts that would impact negatively on sub national budgets. On the other hand, Spahn argues that there are no uniform rules for designing the distribution machinery of a transfer system. However, some basic principles are always desirable, including transparency, some degree of inter-jurisdictional fairness and the use of objective criteria that cannot be manipulated by recipient governments, Roy Bahl and Sally Wallace complement the focus of most of the other essays in the volume. Their concentration is on the vertical dimension of the intergovernmental transfer system rather than on horizontal disparities. The vertical share component of an intergovernmental transfer is the claim of the sub national governments on the revenues raised by the higher level government that makes the transfer. In this essay, Bahl and Wallace offer criteria for the evaluation of the vertical share and examine the range of practices and the variation of vertical shares over time. Available data suggests that the vertical share is, on average, about 19 percent in industrialized countries and between 13 and 14 percent in developing and transition countries. It has been stable in the industrialized countries over the past two decades, increasing in the developing countries and declining in transitional countries. Giorgio Brosio discusses these three essays and identifies the unifying theme of the key role played by the explicit institutions and financial arrangements that frame the implementation of equalization grants. The next three papers, in part 4, focus on the challenges of implementing equalization grants in the real world. Richard Bird and Francois Vaillancourt ask whether there is a role for expenditure needs equalization or whether equalization efforts should concentrate exclusively on equalizing differences in revenue capacity. They find that, on the whole, expenditure-based equalization is more likely to play a meaningfbl role in unitary than in (truly) federal countries. But even in unitary countries, Bird and Vaillancourt argue, many of the purported objectives pursued by the inclusion of expenditure needs in equalization formulas could be more simply achieved by implementing special 'regional' transfers. In fact, this is the approach followed in some federal countries, like Canada, where the equalization system only addresses disparities in fiscal capacity. At times, those special transfers come in the shape of special arrangements for border regions or those inhabited by minority groups. Bird and Vaillancourt argue that these explicit 'asymmetric' arrangements are more transparent and would seem superior to attempting to incorporate all the especial characteristics of every region in the country in one overall equalization formula.
8
Martinez- Vazquez and Searle
Jameson Boex and Jorge Martinez-Vazquez focus on the challenge of designing intergovernmental equalization transfers with imperfect data, in particular how to put together distribution formulas when data for fiscal capacity and expenditure needs are not readily available. Often it appears that the only data available to measure fiscal capacity and expenditure needs in many developing and transitional countries are actual revenue collections and actual expenditures. But, of course, using actual data, as they are directly affected by government behavior, would introduce negative and perverse incentive effects on sub national fiscal performance. This chapter seeks to achieve two objectives. The first is to provide a comprehensive overview of the alternative approaches that can be (or have been) taken to measure local expenditure needs and fiscal capacity when the "first best" data are not available. The second is to analyze the implications of designing an equalization transfer formula with less-than-perfect data. The paper aims to provide comparative evidence on what "second-best" approaches might perform best in the design of intergovernmental equalization transfers. Boex and Martinez-Vazquez find that simple formulas with very few proxies for capacity and need perform best in that they get closest to the results obtained when formulas can be built without data limitations. Sophia Levtchenkova and Jeffrey D. Petchey re-examine the issue of how to account for capital expenditure needs in the context of an equalization policy. The authors build a simulation model to enable policy-makers to estimate regionally disparate capital needs and direct capital expenditures so as to improve standards of public capital for services such as education and health. Levtchenkova and Petchey calibrate the model with data for South Africa and conclude that that country would need to commit about 2 percent of GDP to supplement public capital expenditures if it were to make substantial inroads into attaining some form of public infrastructure equalization over time. The model they have developed has several important features, including its ability to take into account the intertemporal nature of capital arising from its durability and an ability to provide a flow of services over many years. This enables it to account for past capital expenditures and hence the current capital stock as well as what capital services are to be provided in the future (the transition path), while still allowing for the fact that capital depreciates over time. Serdar Yilmaz reviews these three essays, and provides further insights on the issues of estimating the gap between fiscal capacity and expenditure need, data needs and availability, and estimation tools. Part 5 examines the relationship of equalization policy with other closely related policies, including revenue sharing, conditional or tied grants, and credit lending to sub national governments. Bob Searle looks at the relationships between policies associated with reducing vertical fiscal imbalances and the achievement of horizontal fiscal equalization. As
Challenges in the Design of Fiscal Equalization
9
possible ways of reducing vertical fiscal imbalances, this essay discusses public sector revenue sharing and, in particular, the revenues raised from natural resources. Because of differences in the 'own-source' fiscal capacity of local governments, any injection of additional revenues in the form of revenue sharing is most likely to increase the levels of horizontal fiscal imbalance, thus increasing the pool of funds necessary to achieve equalization objectives. This is even more likely when natural resource revenues are being devolved to or shared with local government. Unfortunately, decisions taken to address vertical imbalances are too often taken in isolation of the decisions involving the desirable level of equalization. Bob Searle and Jorge Martinez-Vazquez analyze the relationship between equalization grants and tied (or conditional) grants. Tied grants are typically a key element of intergovernmental fiscal transfer systems. They are an effective tool for upper-level governments to pursue a variety of objectives, including addressing externalities and stimulating sub national expenditures in areas of national priority. This essay discusses many aspects of the design and management of tied grants and concludes that they need far more careful consideration than is often given to them. Searle and Martinez-Vazquez pay special attention to how tied grants and equalization transfers should interact. Essentially, if we assume that equalization is about the equalization of service provision by sub-national governments, the issue for inclusion or exclusion of tied grants in the equalization formula is whether or not the services we are equalizing include or exclude those funded from the tied grants. Dana Weist's essay underlines the importance in developing and transitional countries of integrating intergovernmental capital fmancing systems (loans and transfers) with intergovernmental transfer systems aimed at recurrent service provision. When these two systems are considered separately, as is often the case, there can be many unintended outcomes. In most countries, the predictability of transfers from higher levels of government is key to enabling sub national governments to finance their basic service delivery or infrastructure investments. Other links between the two systems affect other areas as well. To pursue equalization objectives, promote an efficient use of resources and develop municipal credit systems, intergovernmental loans and transfers should be integrated within a consistent financing framework. Without such a framework, poorer local governments are unlikely to invest in needed infrastructure, and intergovernmental transfers may impede the development of a local credit market, especially when grants are available for the most credit-worthy borrowers for infrastructure investments that could be financed through borrowing. The essay illustrates many of these issues in reference to the recent experiences of several East Asian countries.
10
Martinez- Vazquez and Searle
RCmy Prud'hornme comments on the topics of the three papers in this section, concluding that too narrow a view of fiscal equalization grants can be misleading or dangerous. It is necessary to look at fiscal equalization in a broader setting and to integrate a number of additional mechanisms, objectives and instruments in the design of an equalization grant scheme. The political economy dimensions of intergovernmental transfers design is the subject of part 6. Stuti Khemani finds that there is consistent and robust international evidence that politics impacts on the distribution of transfers across sub-national jurisdictions, with particular forms of political decision-making leading to particular forms of political distortions in the distribution of resources across regions. The impact of political considerations on distribution may not necessarily be regressive, especially when low-income regions are informed and active participants in the political process. By reference to the circumstances in India, Khemani also finds that fiscal institutions such as formula-based transfers or delegation of distribution decisions to grants commissions have limited success in curbing political influence. These institutions do change the terms of political bargaining but they do not cancel completely the impact of political considerations on the distribution of resources across regions. Thus, the objective of equalization of resources across local jurisdictions might be best served through equalization of political representation; that is, through political institutions. Political institutions are also key to ensuring the accountability of sub national governments. Without adequate accountability, these governments are likely to be able to circumvent conditionalities imbedded in transfers from upper-level governments to pursue their own political objectives. Musharraf Cyan's comment on Khemani's paper agrees on the importance of politics as a determinant of the final outcomes of transfers. However, Cyan believes there are two ways to look at the issue. One looks at politics as a spoiler. The second sees politics as a determinant that is just as significant as technical considerations, maybe even legitimately so. Recognition of the limitation of technical solutions can lead to appropriate actions to strengthen equalization schemes. In summary, it is our hope that the essays in this volume will provide guidance to policy makers charged with the reform of intergovernmental transfer systems, and will deepen our understanding of the relevant problems and issues surrounding equalization transfers and other grants design issues
s!
THE NATURE OF E UALIZATION OBJECTIVES AND ONSEQUENCES
Chapter 2
FISCAL CAPACITY EQUALIZATION AND ECONOMIC EFFICIENCY: THE CASE OF AUSTRALIA1 JEFFREY D. PETCHEY AND SOPHIA LEVTCHENKOVA School of Economics and Finance, Curtin University,Australia
1.
INTRODUCTION
Central governments in many countries including Canada, Germany, Switzerland, Japan, India, the UK and Australia use fiscal equalization models when distributing grants to sub-national governments. The models used vary somewhat in their precise structure but also share general features. Some, for example the Australian model, estimate the revenue and expenditure needs of sub-national governments in generating the distribution of grant funds while others, such as that used in Canada, estimate only the revenue needs of regions. Cost disabilities may also be incorporated into schemes that take account of expenditure needs. The logic here is that a high cost region needs more grant funds than a lower cost region simply to provide the same level of service. The additional funds are required in order to compensate for the relatively higher costs. An instance here is Switzerland where differences in the cost of providing services in mountainous areas are taken into account in calculating expenditure needs for equalization - one disability proxy measures the relative significance of agricultural land above 800 meters. The Swiss also have proxies for population density in their allocation model (based on the idea that it is relatively more expensive to provide services to a geographically dispersed population). In Japan too, cost disabilities are included in the scheme that allocates equalization grants to local governments. The types of disabilities accounted for include population density, population growth, climate, area and geography, degree of urbanization, and industrial diversification. The UK also takes account of cost disabilities when allocating grants from the national government to local
14
Petchey and Levtchenkova
jurisdictions. These disabilities are constructed using a cross-section regression analysis. Equalization in practice is almost always motivated by equity concerns with the basic idea being to ensure equality of access to public services regardless of where a citizen lives. This is usually attempted by designing a system that attempts to equalize "fiscal capacities" across regions. The idea is illustrated from the following official statement of the aims of Australian equalization: "States should receive hnding from the Commonwealth such that, if each made the same effort to raise revenue from its own tax bases and operated at the same level of efficiency, each would have the capacity to provide services at the same standard" (Commonwealth Budget Paper No. 3)2. By equalizing fiscal capacities through inter-State transfers, it is thought that citizens of a federation with the same preferences and incomes can enjoy the same standard of State-provided public services with identical tax burdens, regardless of where they live. A federation with equalized State fiscal capacities is one that, in principle, replicates the equity of a unitary system while at the same time providing the benefits of decentralization, namely, the ability to have different packages of local public goods and taxes in accordance with local preferences. Fiscal capacity equalization results in inter-regional income transfers. Such transfers can be quite substantial, as Table 1 shows for Australia. Column 1 in Table 1 shows how the pool of grant funds directed to the States is allocated using fiscal capacity equalization. Column 2 shows how the funds would be distributed on a simple equal per capita basis, while the last column shows the difference between the two allocation methods. One can see that New South Wales, Victoria and Western Australia are "losers' while the remaining States and Territories are "winners". Economic efficiency arguments for such inter-regional transfers have appeared in the fiscal federalism literature. One, the 'efficiency in migration case', argues that local public goods create region-specific fiscal externalities, and that fixed factors of production create region-specific economic rents. The location decisions of mobile capital, and labor, are affected by these externalities and rents. The result is that, in equilibrium, these factors of production will be located inefficiently across regions. In short, 'too many' mobile factors locate in regions relatively rich in fiscal externalities and rents (e.g. resource rich regions). It is argued that there is an optimal inter-State transfer that corrects for the distorting effects of externalities and rents, and establishes an optimal spatial distribution of
15
Fiscal Capacity Equalization and Economic EfJiciency
Table I. Equalization, Australia, 2002-03. Distribution Using CGC Model, Equal Per Capita Distribution $m
YO
$m
YO
Difference
29.0 20.8 19.6 9.2 9.9 4.2 1.9 5.4 100.0
(2) 8967.4 6650.7 5029.5 2633.2 2045.0 637.8 429.3 272.1 26664.9
33.6 24.9 18.9 9.9 7.7 2.4 1.6 1.O 100.0
(1) - (2) -1243.6 -1097.8 207.4 -185.3 596.7 480.8 90.3 1151.5 0.0
(1)
New South Wales Victoria Queensland Western Australia South Australia Tasmania ACT NT
Total (Pool)
7723.9 5552.9 5236.9 2447.8 2641.6 1118.6 519.6 1423.6 6664.9
Source: Commonwealth Budget Paper No. 3, "Federal Financial Relations, 20022003".
mobile factors. Of course, the existence of an optimal transfer does not justify any of the schemes that we see in operation since none actually implement the optimal transfer3. Fiscal capacity equalization has also been criticized on efficiency grounds. Swan and Garvey (1991) attempt to show that equalization grants induce strategic behavior by regions, or gaming against the distribution formula. The result is sub-optimal provision of local public goods by regions. It has also been argued that equalization creates 'transfer dependency' (an issue for the Atlantic Provinces in canada4) making regions reluctant to pursue economic development objectives. Equalization may also slow down economic adjustments that would otherwise take place to facilitate inter-State income convergence, such as changes in relative wages between regions, or migration from low to high-income regions. Therefore, though often justified by policy-makers on equity grounds, equalization and inter-State transfers have been both supported and criticized on economic efficiency grounds. There seems to be no consensus on the efficiency debate, with conclusions depending on particular lines of argument taken, assumptions made and modeling structures adopted. This paper attempts to provide some cohesion to the debate. It does this by bringing together, within one model, the major efficiency rationale for inter-regional transfers - the presence of regional externalities and rents together with one of the major efficiency arguments against equalization: the potential for strategic behavior by regions and sub-optimal provision of local public goods. The advantage of this approach is that it enables us to
16
Petchey and Levtchenkova
examine exactly how equalization affects efficiency in a world in which regions act strategically, and where externalities and rents affect location decisions. We are also able to draw conclusions about the efficiency (and welfare) implications of equalization in such a world. The model adopted is novel in the sense that it integrates a real world fiscal equalization scheme into a standard model of a decentralized (e.g. federal) economy, with optimizing regional governments and factor mobility. Each region chooses the provision of a local public good to maximize the utility of a representative resident, while taking account of the migration responses to its decisions (regions are non-myopic with regard to migration responses). The central government collects taxes in each region and distributes the revenue raised to the regions through an equalization formula that estimates regional revenue and expenditure needs. The equalization formula used in the model is the Australian one. This is because Australian equalization is, arguably, the most comprehensive in the world: it equalizes for revenue and expenditure needs, and incorporates cost disabilities. As will be shown, many schemes used in other countries are special cases of the Australian approach. The efficiency properties of a Nash equilibrium are then explored and the results are as follows. The first is that equalization distorts State decisions over public policy, mainly because (at least in Australia's scheme) the standards used to assess whether a State has expenditure andlor revenue needs are dependent on state policy choices. Therefore, States, through their choice of public policy, are able to influence these standards, and hence the grant they receive. This gives them an incentive to distort their public policies away from what might otherwise be optimal. The second result relates to the efficiency of location choices made in a decentralized economy with equalization. As noted, some have argued that in economies with free mobility, location choices are inefficient because of region-specific rents and fiscal externalities. This necessitates an optimal transfer. What we show is that in the equilibrium in our model, there will be an inter-regional transfer, but it is not the one required for efficiency in the spatial allocation of mobile factors. The paper outline is as follows. Section 2 develops the model of a decentralized economy with optimizing sub-national governments, factor mobility and a fiscal equalization system implemented by a central government. Section 3 examines public policy choices made by the subnational governments, while Section 4 examines the efficiency implications of those decisions in the presence of equalization. Section 5 concludes with suggested policy implications.
Fiscal Capacity Equalization and Economic Efficiency
MODEL Suppose that we have a decentralized economy with N citizens who have identical incomes and preferences. For convenience we will think of this as a federal economy with i = 1,2 States. State i has ni residents who each supply one unit of labor. The national population (labor supply) is therefore
The production process in each State is simple. There are two inputs, the first, immobile and in fixed supply, can be thought of as land, fixed physical capital, or natural resources. We denote the supply of this factor in State i as Li . The second factor is labor. Since each citizen supplies one unit of labor, ni is State i's labor supply. As shown below, labor is perfectly mobile between States and its supply can vary from the perspective of each State. The two factors are combined using a production technology based on constant returns to scale to produce a numeraire good whose price is set at one. The value yi of a State's production of the numeraire (the value of aggregate State output) is represented by the production function
Since the immobile factor is in fixed supply in each State, from now on we define the aggregate output of State i as yi = f,(ni) and suppose the following: fie(ni)> 0, fim(ni) < 0 . Though States have the same production technologies, we allow them to have different endowments of the fixed factor. Competitive factor markets are also assumed, implying that each person in a State receives a wage, w i , equal to their marginal product. Since citizens of a State are identical, each receives the same wage, but because State specific supplies of land may differ, inter-State wage rates may not be the same. The residents of a State own equal portions of that State's fixed factor5and each receives an equal per capita share of the State's fixed factor income, or economic rent. Since we have assumed constant returns to scale, and hence that output is exhausted by factor payments, the income of a representative citizen in State i is the State's average product
Petchey and Levtchenkova
18
Part of the numeraire output in each State is transformed, by a State government, into a pure local public good denoted as qi , with no inter-State spillovers, and the rest is consumed directly by State citizens. Per capita consumption of the numeraire is denoted as x i . From now on we think of this as private good consumption. There is implicitly a transformation frontier defined between private and public good consumption that is assumed to be linear. The (constant) slope of the frontier is the marginal rate of transformation between the two goods that is equal to the marginal cost of xi over the marginal cost of q , . Under the assumption of perfect competition it is also equal to the price of the numeraire (one) over the price of the public good6. Each citizen has a quasi-concave, continuous and differentiable utility function,
As noted, citizens are also assumed to be perfectly mobile across States so that, in equilibrium u(x,, q,) = u(x,,q,) . (Equal utility condition)
(5)
The equal utility condition can be thought of as a social welfare function; W = u, = u, . Since citizens of a State receive location-specific fixed factor rents, and because of the presence of local public goods which generate fiscal externalities, in this model labor will, in general, be allocated inefficiently between States. This is a well-known feature of federalism models with the underlying regional structure developed here.
Fiscal Capacity Equalization In practice, the size of the pool of funds to be distributed among the States is determined by tax and spending assignment between the national and subnational governments, which commonly leads to a fiscal gap (an excess of revenue relative to expenditure) at the federal level. This gap is then distributed back to the States on the basis of various distribution formulas. We abstract from the complexities of how the pool is created and concentrate on the gaming behavior of States over the distribution of the pool. It is the efficiency effects of how the pool is distributed, rather than any distortions related to the creation of the pool, which are of primary focus here.
Fiscal Capacity Equalization and Economic Efficiency
19
Therefore, it is supposed that some pool G is created by a federal government using a per capita lump sum tax on citizens, denoted as s. In addition, we do not model central government provision of public goods (national public goods). Rather, the only role given to the central government is one of creating a revenue pool that is then distributed to the States using an equalization methodology. This is clearly a major abstraction and simplification of central government behavior, but again, it is one that allows us to focus on the issue at hand: the distortions created by gaming over the allocation of a given pool. For simplicity we also assume that s is given7. It represents a quantity of the numeraire that is surrendered by a citizen to the national government. Since the numeraire produced in each State is the same, the quantity collected by the national government can be aggregated to create a single 'pool' of the numeraire, denoted as G = sN. Since s and N are fixed, G is a parameter.
2.2
The State - Specific Grant
As noted, we have chosen to model equalization using the Australian approach. The grant pool in Australia is allocated between the States using the Commonwealth Grants Commission (CGC) equalization formula8. We integrate the key components of this formula into the federal model, and in doing so, abstract from the inessential parts. Though the formula applies in a federation of multiple States, we also suppose there are only two States, i = 1,2, consistent with our model. The CGC's formula defines the per capita grant, g, ,to State i as:
G and N are the total grant pool and national population (as previously defined) and so G/N = s is the per capita amount of funds available for distribution to the States. The variable E is defined by the CGC as total expenditure by the States on the services included in the model (for example, education, health, transport, welfare). Here, there is only one service, a local public good, so that the following holds: E = p,q, p,q, . Therefore, E/N is the per capita average expenditure on the local public good across both States. The CGC calls this 'standard expenditure'. As will be shown below, it is the expenditure that is used as a benchmark to assess a State's 'expenditure need'. The variable T is defined by the CGC as the total (own-source) revenue raised by all States to fund their public expenditures. Assuming balanced State budgets, this is equal to total expenditure by all the States, less what
+
20
Petchey and Levtchenkova
they receive as grants from the federal government, G. Therefore, ownsource revenue can be defined as T = E - G, or alternatively, T = plql + p,q, -G . Furthermore, TIN is the per capita average own-source revenue raised by all States, known as 'standard' revenue in CGC terminology. Again, this name is given to the term T/N because, as will be seen, it is the benchmark used to assess whether a State has a 'revenue need.'
2.3
Cost and Revenue Disabilities
Another part of (6) is the cost disability, y i . It captures the cost of providing each service in State i, relative to the average cost for all States. The CGC calculates a set of cost disabilities for each service provided by the States. The calculations are complex and since we have only one service we have only one cost disability for each State (for the local public good). A State may have a cost disability in the provision of a particular service for a variety of reasons. For example, it may have a geographically dispersed population and have to provide schools in remote locations. This means that a unit of education service may have a higher cost than the average across all States. Other factors contributing to cost disabilities include the agelsex profile of the population, ethnicity and the presence of groups with special health/educational requirements, and economies of scale. Australian equalization is unique in the sense that it puts a great deal of effort into estimating such cost disabilities and then allowing them to determine the distribution of the grant pool, G. We adopt a simplified definition of a cost disability that, of necessity, abstracts from this complexity but captures the essence of the idea. Namely, we define the cost disability for State i as
So defined, if yi > 1 , then State i is a relatively high cost provider of the public good (has a cost disability) and if yi < I , it is a relatively low cost provider. Thus, the cost disability variable is normalized around the number one. The prices of the public good are exogenous so the disability is treated as exogenously given by the States. The CGC also estimates a State specific 'revenue disability' for each State tax base. In Australia's case, such disabilities are estimated individually for each of the State taxes, including payroll tax, the major State tax, and mineral royalties. Again, we abstract from this complexity and suppose that a State has only one revenue disability, pi, which is greater than one if the
Fiscal Capacity Equalization and Economic Efficiency
21
State has a relatively strong tax base, and less than one if the State has a relatively weak tax base. As with the cost disability, we assume that States take the revenue disability as determined exogenously by the CGC.
Expenditure and Revenue Needs The term (E / ~ ) ( y-, 1) in equation (6) is the expenditure need of State i. If we multiply E/N through the brackets we can see that the need has two parts. The first, (E / N) - yi , is the standardized expenditure of State i. This is the expenditure that State i would have to undertake, taking account of its cost disability, to achieve the per capita standard, E/N. State i's standardized expenditure is greater than or less than the standard, depending on the magnitude of its cost disability. The second term, E 1N , is just the standard expenditure of all States. Thus, the expenditure need of the State is equal to its standardized expenditure less standard expenditure. If the State's cost disability is greater than one, the State has a positive expenditure need. Otherwise, it is negative. Similarly, (T/N)(l-pi) is the revenue need of State i. Multiplying through the brackets one can see that it also has two parts. The first is just T/N, or standard own-source revenue. The second term, (TIN) - pi, is the standardized own-source revenue of State i. This is the revenue that State i would raise if it applied the average tax effort to its own tax base. If the State's revenue disability is greater than one, then its standardized revenue will be higher than the standard, and its revenue need will be negative. Alternatively, if the State's revenue disability is less than one, its standardized revenue will be less than the standard, and the State will have a positive revenue need. Thus, under Australian equalization, a State receives an equal per capita share of the pool, G/N, adjusted by the expenditure and revenue need terms. A State will receive more than its equal per capita share of the grant pool if the sum of its needs is positive; and less than its equal per capita share if the sum of its needs is negative. Finally, if the expenditure and revenue needs cancel each other exactly, the State will simply receive its equal per capita share, G/N. Note also here that, due to the differences in their definition, the disabilities are applied differently in equation (6). Namely, for the expenditure disability, standard expenditure is multiplied by (yi - 1) where yi > 1 implies that the State has relatively high costs, while for the revenue disability standard revenue is multiplied by (1 -pi) where pi > 1 denotes a State with a relatively rich tax base9.
Petchey and Levtchenkova
2.5
Summary
Each endogenous variable in the model is a function of the exogenous variables and parameters. For later discussion, it is useful to explore this in more detail for one of the endogenous variables, for example the grant to State i. In this regard, one can define from (6) the per capita grant to a State as
where
F = [s N] is a vector of variables determined by the federal
government, P = [p, p,] is a vector of the local public good prices,
CGC = [I, pi c] is a vector of variables determined by the CGC and S=
lq,q,] is the strategy set of the two States. Within F, the variable s is
determined by the federal government. The total federal population N is determined by things such as the birth and death rate, but also by international migration and hence, to some extent, the population policy of the federal government. Within the vector CGC, the variables yi ,pi,c are all determined by the CGC, while the public good provision levels within S are determined by the States. As discussed below, we assume that each State perceives s, N, public good prices and the CGC variables (except the adjustment term c) to be exogenously given. This is reasonable since in practice the States have no impact on s and only a marginal impact on the CGC variables. It is true that the States rent-seek over the cost and revenue disabilities but, in reality, this meets with limited success. As also discussed below, we assume that each State adopts Nash conjectures with regard to the level of provision of the public good in the other State. For example, when optimizing, State 1 perceives q2 to be given, and similarly for State 2. Thus, each State perceives its grant to be a fbnction of its own public good provision. The adjustment variable c is also among the variables contained in the vector CGC. Given the previous discussion, we know that States also view c as a function of their joint policy choices. The general function (1 l), which links State policies and equalization variables to the State specific grant, will hold in any federation with equalization, though the specific variables to be included will vary depending on the particular structure of the equalization formula used. Thus, one can think of (1 1) as a general fhction defining the State specific grant, and equation (6) as the specific function for the Australian model.
Fiscal Capacity Equalization and Economic Efficiency
STATE POLICY CHOICES Taking into account private good consumption, provision of the public good, the national government tax and the equalization grant, the aggregate budget constraint in State i is:
Suppose that the government of each State is benevolent and perfectly represents the preferences of its citizens. The implication is that State and citizen interests are synonymous: a State will choose its level of provision of the public good to maximize per capita utility within its jurisdiction (recall that all residents within a State have the same income and preferences). In making its choice, a State is assumed to take account of the equal utility condition and the equation defining the total supply of labor. Thus, States are non-myopic with regard to the migration effects of their public policy choices. It is also assumed that States take account of the grant response to any changes in public good provision, through the equalization formula. This implies that each State's choice of public good provision will affect the welfare of citizens in the neighboring State, both through the migration condition and the equalization formula. Such policy interdependence means that States can act strategically. One can, therefore, think of the problem as a two-player simultaneous move game in continuous pure strategies, in which the States are players and the payoffs are the per capita utilities in each State. The strategy set for the game, defrned previously, is s = (q, ,q,). Nash conjectures are assumed so each State chooses its public good provision taking provision in the other State as given. However, it is supposed that States have sufficient foresight to take account of the impact of their choice on migration and the equalization grant. Given this, each State will choose its provision of the public good to maximize within-State per capita utility subject to the Statespecific budget constraint, the national labor supply condition, the equal utility condition and the equalization formulas. For convenience, from now on we conduct the analysis from the perspective of State 1. Rewriting the aggregate budget constraint for State 1 in terms of per capita private good consumption and substituting the result into the per capita utility function, the problem of State 1 can then be written as
Petchey and Levtchenkova
Subject to: Migration and Labor Supply Constraints
(ii) n, +n, = N Fiscal Capacity Equalization Constraints
G E T -l)+-(l-p,)+c N N N G E T (iv) g, =-+-(y, -1)+-(1-~2)+c N N N T G E (v) c=---(n I ~ +I % ~ 2+) E ( n ~ ~n 2 ~~ 2 ) N N (vi) E = Plq, + p2q2 (iii) g, =-+-(y,
(vii) T = E - G . From State 1's perspective, the exogenous variables are s, N, p,,p2, G (defined as G =sN), y,,y2 (defined by (7)) and p,, p,. These are the variables determined by the federal government and the CGC, and perceived to be exogenous by the States. With Nash conjectures, State 1 will also treat q, as given. The endogenous variables are, therefore, q, , n,,n, , g,,g,, E,T and c, which is an endogenous function of both exogenous and endogenous variables. The constraint set has seven equations. There are 8 unknowns and hence one free dimension to maximize over. Also, we do not use G = sN in constraints (iii) and (iv) because we want these constraints to reflect the way that the CGC formulates its model. Of course, if we do use G = sN in constraints (iii) and (iv) then G/N simply becomes s, the per capita tax levied by the federal government10. State 2 solves an analogous problem with identical exogenous variables and with Nash conjectures it takes q, as given when it choosesq, . The necessary condition for provision of the local public good is found by differentiating the objective function in (13) with respect to q,, for given q, under Nash conjectures, and fixed values of s, N, the CGC variables (except adjustment factor c) and public good prices. This yields
Fiscal Capacity Equalization and Economic EfJiciency
where MRS:, = u,, l u x , is the marginal rate of substitution between the local public good and the private good in State 1 (or the marginal benefit of extra units of the public good). The remaining parts of (14) are explained below. The term b, is the net benefit of an additional migrant in State 1 and is defined as
where (w, - x,) is the difference between a migrant's marginal product (wage) and their per capita consumption, and (s-g,) is the difference between the federal tax they pay (output foregone by State 1) and the per capita equalization grant the migrant attracts to the State. The net benefit of an extra migrant for a State is more complex than in traditional federal models because we have to account for the grant and equalization consequences of migration into a state". The term dg, / dq, in (14) is the change in State 1's equalization grant in response to a small increase in the State's public good provision. The expression for this is found by differentiating constraint (iii) with respect to q, ,for given q, ,and fixed values of the exogenous variables. This yields
The first term on the right side, (p, IN)(?, -l), is the change in the expenditure need of State 1 resulting from a small change in its provision of the local public good. The expenditure need changes because a variation in the provision of the local public good changes standard expenditure, which is in turn used by the CGC in its per capita grant formula to determine the standardized expenditure of each State, and hence the State specific expenditure needs. Similarly, the term (p, 1N)(1 - p,) is the change in the revenue need of State 1 resulting from a small change in its revenue raising. Again, the revenue need changes because, as discussed previously, standard revenue, T/N, is a hnction of the revenue choices of the States. The third term, dc/ dq, , is the change in the balanced budget condition adjustment c.
26
Petchey and Levtchenkova
The total change in the grant is just the sum of the changes in the expenditure and revenue needs, and the balanced budget adjustment. The sign and magnitude of the changes in the two types of need, and the adjustment, will determine the sign and magnitude of the own-grant response term. The sum of these changes can also be expressed simply as(p, / W Y ,-P,) +dcl%1 The term dn,ldq, in the public good necessary condition is the change in State 1's population resulting from a small increase in its provision of the local public good. An expression for this has been derived but at this level of generality it cannot be signed and is available from the authors on request. Similarly, an expression for the change in the adjustment term in response to a small increase in State 1's public good expenditure has been derived, cannot be signed, and is available on request.
4.
EFFICIENCY
The inefficiency of provision of the local public good in State 1 is where the marginal cost illustrated in Figure 1. Efficient provision is at of the public good (mc, = p,) is equal to the marginal rate of substitution between q and x (marginal benefit of the public good). Provision with equalization is at a point such as q, where the marginal cost, adjusted by the migration, grant and adjustment factor responses, is equal to the marginal benefit. Figure 1 illustrates a case of over-provision of the local public good, but as noted above, we might also have under-provision depending on the sign of the own-migration and grant responses (which are in general indeterminate) 12. The other interesting question from an efficiency perspective is whether the mobile population will be distributed optimally across States in equilibrium. This issue can be assessed by examining the first-best Pareto optimal outcome in a federation that has the same regional structure as our model. The Pareto optimal outcome is characterized by supposing that it is governed by a benevolent central planner who chooses the private and public good provision in either of the States to maximize per capita utility in that State. At the same time, the utility of a representative citizen in the other State is held fixed at some predetermined level. The central planner also takes into account the national feasibility and labor supply constraint^'^. The solution (not presented here but available on request) highlights that there are two conditions that must be satisfied in a Pareto optimum. One is that the Samuelson rule must hold in each State (as shown above, this is not the case in our model). The second is that in equilibrium the mobile population must
qr
Fiscal Capacity Equalization and Economic Eficiency
Figure 1. Sub-optimal Public Good Provision be distributed between States such that the following 'equating at the margin rule' is satisfied, (w, - x,) = (w, - x,) . It is well known that this condition will, in general, not be satisfied because of the presence of fiscal externalities and location specific economic rents that distort migration decisions. Moreover, there is an efficient transfer from State 1 to State 2 that corrects for this distortion and establishes an optimal distribution of the mobile population consistent with the equating at the margin rule. This transfer is found by solving for t from (w, - x, - t / n,) = (w, - x, + t 1n, ) to yield14
For the mobile population to be allocated efficiently across States in our model, the inter-State transfer that results from the equalization process would have to be consistent with this eff~cienttransfer. We know that gini = s x ni must hold in equilibrium because the CGC is assumed to be
x 1
i
setting the adjustment parameter in each State's per capita grant formula to
28
Petchey and Levtchenkova
ensure that the condition is satisfied. This implies that the equilibrium interState transfer under equalization, denoted by t E, is t E = n, (s - g , ) = n, (s - g, ) . (Transfer with Equalization) (18) Since g, ,g,, n, and n, are functions of State policies, for given values of the federal government variables, public good prices and CGC variables, t E is a function of collective State policies. Thus, through the equalization formula, the inter-State transfer is determined by the collective policies of the States. But there is no reason why the transfer under equalization should be the same as the transfer required for efficiency: hence, t E # top' and the distribution of labor is inefficient.
5.
CONCLUSION AND POLICY IMPLICATIONS
We develop a model that integrates a real world fiscal capacity equalization scheme, namely the Australian one, into a standard model of a decentralized (e.g. federal) economy with optimizing regional governments and factor mobility. This allows us to examine the efficiency of an equilibrium where States are linked together through capacity equalization and where factors of production make location choices while taking into account region-specific rents and externalities. The key result is that the modeled equilibrium with fiscal capacity equalization is inefficient for two reasons. The first is that it distorts State decisions over expenditure on local public goods. Second, the inter-State transfer that results under fiscal capacity equalization is not consistent with the inter-State transfer required for global efficiency. Thus, in the outcome with fiscal capacity equalization, the spatial allocation of mobile factors is inefficient. What are the implications for the design of systems of fiscal equalization, given that Australia is seen as having a benchmark model? The first is that standards need to be exogenously set. In the Australian model they are endogenous (see equation (6) where E - a h c t i o n of States' policies - is an element in the grant function for State i), and this is the source of the inefficiency due to strategic behavior. This raises the question of where standards should come from if they are not to be made a function of what States do on average. Second, capacity equalization yields an inefficient spatial allocation of mobile factors because it is based on factors such as expenditure and revenue needs rather than fiscal externalities and rent sharing (important for efficiency). Regardless of what design changes are made, this source of inefficiency will remain.
Fiscal Capacity Equalization and Economic EfJiency
29
An option is to shift from equalization based on fiscal capacity to equalization based on economic efficiency. This would require that countries implement an inter-regional transfer scheme consistent with (17) where the transfer is determined by variables that matter from an efficiency rather than equity point of view. In a world of perfect mobility, such a reform would be unambiguously welfare enhancing as it would make the residents of all States better-off.
Notes The background research for this paper was undertaken under the auspices of an Australian Research Council (ARC) Large Grant. The authors would like to thank the ARC for its support. The "Commonwealth" here refers to the Australian national government. Papers that analyze this idea include Boadway and Flatters (1982), Myers (1990) and Petchey and Shapiro (2000,2002). See Courchene (1 984). This rules out the possibility that people may be 'absentee landlords' (i.e. live in one State and own some of the fixed factor in another). The assumption also implies that as a person migrates from, say, State 1 to State 2, they immediately forfeit their right to fixed factor ownership in State 1, and gain a share of the fixed factor in State 2 upon enhy to that State. Defining
Pi
as the price of the public good, the marginal rate of transformation is
simply l / p i . With a linear frontier this ratio is unchanged as we adopt different combinations of Xi and the public good. Therefore,
pi is treated as a parameter.
Since
it can differ across States, the ratio l i p i can also vary across States. Thus, generally, States will have differently sloped frontiers. The determination of s could be made endogenous by explicitly modeling national government optimizing behavior. If States took account of the impact of their policy decisions on s, then additional distortions, not directly related to equalization, would be introduced. We abstract from these considerations by supposing that States treat s as given. See CGC ( 1999). There is no reason, a priori, for the sum of the aggregate grants across all States to exactly exhaust the available grant pool, G. Therefore, one must introduce an adjustment to the per capita needs-based grant estimated for each State to ensure that a 'balanced grant pool condition' is satisfied, i.e. that the sum of the aggregate grants exactly exhausts the pool. The last term, C ,in (6) is included to do just that. An explicit expression for c is derived in Annex A. A State also perceives its population size to be a function of its own policy choice. To see this, note that for given q,, q,,p,, p,, s, g, andg,, constraints (i) and (ii) determine n , andn,. We know that g, and g2 are hnctions of s, N, public good prices, CGC variables and joint State policies. Therefore, for State 1, n,(F, P,CGC,S) where F, P, CGC and S are as previously defined. State 1, when optimizing according to (13) with Nash conjectures, will perceive everything to be fixed, except its own level of provision
30
1 1. 12. 13. 14.
Petchey and Levtchenkova of the public good. Hence, it perceives its population to be a function of its own policy choices. In standard models where States are not linked through an equalization model, the net benefit of a migrant is just b, = (w, - x,) . In general, the curves in Fig 1 would be non-linear. They are drawn as linear for convenience. For example, see Flatters, Henderson, and Mieszkowski (1974) and Myers (1990) who solve analogous central planner problems. See also Boadway and Flatters (1982) and Boadway, Cuff and Marchand (2003). We can write the optimal transfer alternatively as a function of State specific rents and fiscal externalities (see Petchey (1993)).
References Boadway, R.W. 2004. The theory and practice of fiscal equalization. CESifo Economic Studies, 50(1): 2 11-54. Boadway, R.W., Cuff, K., and M. Marchand. 2003. Equalization and the decentralization of revenue-raising in a federation. Journal of Public Economic Theory, 5(2): 201-28 Boadway, R.W., and F. Flatters. 1982. Efficiency and equalization payments in a federal system of government: A synthesis and extension of recent results. Canadian Journal of Economics, 15: 613-33. Buchanan, J.M and C.J. Goetz. 1972. Efficiency limits of fiscal mobility: An assessment of the Tiebout model. Journal of Public Economics, I: 25-43. Commonwealth Grants Commission. 1999. Report on general revenue grant relativities, Vol. 11: Methods, assessments and analysis. Canberra: Common wealth of Australia. Courchene, T. 1984. Equalization payments; past, present and future. Ontario Economic Council, Toronto. Flatters, F., Henderson, V, and P. Mieszkowski. 1974. Public goods, efficiency and regional fiscal equalization. Journal of Public Economics, 3: 99-1 12. Mansoorian, A., and G.M. Myers. 1993. Attachment to home and efficient purchases of population in a fiscal externality economy. Journal of Public Economics, 52: 117-132. Myers, G.M. 1990. Optimality, free mobility and the regional authority in a federation. Journal of Public Economics, 43: 107-21. Petchey, J.D. 1993. Equalization in a federal economy with inter-state migration. Australian Economic Papers December: 336-54. Petchey, J.D. and P. Shapiro. 2000. The efficiency of state taxes on mobile labor income The Economic Record, 76:234; 285-96. Petchey, J.D. and P. Shapiro. 2002. State tax and policy competition for mobile capital. The Economic Record, 78:241; 2 18-28. Swan, P. and G. Garvey. 1991. The equity and efficiency implications of fiscal equalization. Mimeograph. Swan Consultants Pty Ltd. Tiebout, C.M. 1956. A pure theory of local expenditures. Journal of Political Economy, 64(5): 4 16-24.
Chapter 3
ENSURING INTER-REGIONAL EQUITY AND POVERTY REDUCTION BERT HOFMAN AND SUSANA CORDEIRO GUERRA' The World Bank
1.
INTRODUCTION
The decentralization of revenue sources and expenditure responsibilities to subnational levels of government can increase the efficiency of spending, and increase participation in decision making by local constituents. It is also true, however, that decentralization can also result in inequities in service delivery among citizens of the same country depending on where they live. While decentralization need not cause these inequities, devolution of revenue sources combined with disparities in endowments of regions is likely to lead to disparities in fiscal resources at the subnational level. If left by themselves, these disparities could lead either to lower levels of services in fiscally poor regions, andlor lead to higher tax rates for similar levels of government services in those regions. In turn, disparities in service delivery could thwart poverty alleviation efforts, as public services that are usually provided at the subnational level such as primary health and education are critical in empowering the poor. Disparities in the level and quality of these services could therefore eventually perpetuate inequalities in income levels, or at least delay conversion of incomes across regions. Moreover, large fiscal disparities could induce migration to regions that have higher revenues and better services, even though the migrant could have been more productively employed elsewhere.*Finally, large disparities in public service delivery may cause social unrest in regions that are left behind, and could undermine the sense of unity in a country. Such diversity coupled with increasing disparities across regions may constitute the basis for regional insurrections such as, for example, the Muslim regions of Mindanao in the Philippines and West Papua in Indonesia (Hill, 2000). Most governments take an interest in the level and distribution of public services provided to their citizens, even in a decentralized system. For
32
Hofinan and Guerra
some, this is reason enough not to decentralize the financing or the provision of services deemed critical to national goals. Indeed, income redistribution is seen to be primarily a central function because central governments are seen to be better able to manage one of the key instruments for this goal, a progressive income tax.3 Yet, at the same time, services that are likely to affect income distribution or poverty alleviation, or similar national goals are decentralized. Moreover, many governments have entered into commitments on the outcomes of services that are routinely at least partially decentralized to subnational levels of government.4 So they should be concerned about the level and distribution of fiscal resources among levels of governments, as those resources enable subnational governments to deliver services. Even if governments care about the distribution of services, they may not take policy action to correct the distribution of fiscal resources among subnational governments--or horizontal fiscal imbalances. This could occur for two reasons. First, policymakers may count on market adjustments. Thus, for some countries migration is considered to be countervailing equalizing force, driving people to the constituency that delivers the most beneficial level of public services at tax rates deemed appropriate by the constituents. Second, there are legitimate policy tradeoffs that need to be weighed: an aggressive transfer policy may be seen to dampen needed incentive for increasing own-revenue mobilization by sub-national governments-a short-term trade-off that de-emphasizes equalization in order to provide for an own-revenue foundation for a future of reduced fiscal disparities and transfer dependency.' Similarly, an equalization-only policy (perhaps only in the early stages of decentralization) may compete with broader considerations of efficiency andlor growth that are on the nation's agenda. In China, for instance, the coastal development strategy of the 1980s and 1990s deliberately left more resources in the regions with stronger growth prospects. Using panel data covering the 1985 to 1998 period of fiscal decentralization in China, Qiao et a1 find that inequality in the distribution of fiscal resources across provinces was positively related to higher economic growth, and higher growth led in turn to increased inequality (Qiao et al, 2002). Likewise, Yan also finds a similar tradeoff between growth and equity in China, and further shows that the fiscal reforms, in 1985 and 1994 respectively, did not contribute towards increased equalization (Yan, processed). In other countries governments compensate for horizontal fiscal imbalances not through redistribution, but by centrally providing certain services in poor regions, while leaving rich regions to fend for themselves. This approach may indeed be beneficial also from an efficiency point of view, as there is some evidence suggesting that the center is better at some
Ensuring Inter-Regional Equity and Poverty Reduction
33
services critical for poverty alleviation, including the targeting of a social safety net (Ravallion, 1998). Finally, redistribution of fiscal resources may remain limited for political reasons: rich regions also tend to be powerful regions, and taking resources away from them to give to poorer regions may simply be politically unfeasible. Most East Asian governments care about equitable services to its people, and consequently take an interest in issues of the distribution of fiscal resources among subnational governments that deliver much of these services. Countries such as Indonesia have included sub-national fiscal equity as an explicit goal in the c~nstitution.~ Other countries such as the Philippines or China include strong commitments on equal access to services in their constitution, whereas the delivery of many of these services is devolved to sub-national governments.7 Similarly, international commitments to public service outcomes such as the Millennium Development Goals often relate to services delivered at the subnational level. This would imply that the State should care about whether subnational governments are capable of delivering such services, and this implies a concern about subnational fiscal capacity and its distribution over subnational entities. In fact, most East Asian countries show they care to some extent by having some form of fiscal equalization mechanism in place. This paper reviews causes and consequences of fiscal inequities among subnational governments in East Asia. It shows that endowments and fiscal capacities among subnational entities are large, and these inequities translate into large fiscal inequities before equalization. The transfer systems currently in place in East Asian countries are only mildly equalizing, so that even after equalization the remaining fiscal disparities are still large. The paper discusses the possible consequences for service delivery inputs, outputs, and outcomes of these disparities. It concludes by some guidelines on how to make progress in increasing transparency on inequalities at the sub-national level, and how transfer systems can be made more equalizing without undennining the incentives for local governments to raise their own revenues and to spend their resources efficiently.
2.
FISCAL DISPARITIES IN EAST ASIA
2.1
How Large are Sub-national Disparities?
East Asian countries show a large variety in natural endowments, economic opportunities, level of development, and poverty (Figure 1). Indeed, the richest province in Indonesia has 17 times the GDP per capita as the poorest one, whereas in China that figure is 11 times, and in Viet Nam
34
Hojinan and Guerra
9.5.8 Disparities in income per capita have been relatively stable over time, and little convergence seems to have taken place (Hill 2000, Garcia-Garcia and Soehstianingsih 1998). In such an environment, devolving expenditure responsibilities and revenue sources could lead to undesirable inequities because of disparities in revenue potential among regions as well as disparities in the cost of delivering public services. Indeed, sub-national disparities in revenues generating capacity are large within East Asian co~ntries.~ The disparities in own revenues generated are larger than the disparities in income per capita (Table l).1Ā° In China, for instance, own revenues per capita of the richest province is fifteen times that of the poorest region, whereas in Indonesia, the Philippines, and Vietnam the numbers are even significantly higher." In the last three countries natural resource revenues collected or shared by the center with the regions exacerbates apparent inequalities. Below the provincial level, inequalities are even larger.12 In Indonesia, for instance, the richest local government had 50 times the own revenues of the poorest one in 2001 (World Bank, 2003b), whereas the richest county in Gansu province had 82 times the revenues per capita of the poorest one (World Bank, 2002a).13 This large magnitude of inter-regional fiscal disparities across East Asian countries is not due to recent trends. Data for China (Table 2) show that fiscal disparities have been persistently large across time.14 While expenditure disparities have slightly increased, with the inter-provincial
Table I. Disparities in Province-level Revenues before Grants, per Capita, (US$, Latest Available Year)
Maximum Minimum Average MadMin Standard Deviation Coefficient of Variation
China
Indonesia
Philippines
283.2 18.1 55.7 15.7
59.5 3.8 12.1 15.7
7.57 0.21 1.46 35.43
Thailand
Vietnam
343.4 6.5 36.9 53.0
54.8
11.6
1O .
1O .
0.92
1.65
Source: SABER Database, Indonesia Regional Fiscal Information System. Notes: The table presents consolidated province-level data. Because of the different
nature of the grant systems (see text), the absolute dollar amounts are not comparable. The revenue numbers include own and shared revenues for Indonesia. Since the IRS in the Philippines is an equalizing transfer, it is excluded from its revenues before grants.
35
Ensuring Inter-Regional Equity and Poverty Reduction
Figure I. Provincial Poverty Map of East Asia Source: World Bank EAP Data.
Table 2. Disparities in China's Provincial Per Capita Fiscal Outcomes, (1979-2002), (RMB yuan, per capita)
Maximum Minimum Average Max/Min Standard Deviation Coefficient of Variation
Revenues 1979 2002 1525 4362.8 -12 273.7 148.2 805.8 -127.1 15.9 293.5 939.2 2.0 1.2
Expenditures 1979 2002 274 5307 34 654.5 97.9 1620.5 8.1 8.1 67.4 1218.4 .69 .83
Source: Qiao et al, 2003.
Notes: The table presents consolidated province-level data. The negative minimum value for China is due to the definition of revenues, which includes subsidies to compensate state enterprise losses due to policy as negative revenue.
Hojnan and Guerra
36
coefficient of variation rising from .69 to -83, revenue disparities have fallen from 2.0 to 1.2. Yet, although revenue disparities have fallen, they continue to be large with Shanghai having almost 16 times the revenue per capita of Tibet. Similarly, in terms of inter-provincial expenditure disparities, Tibet's expenditure per capita was 8 times that of Hainan in 1979, while Shanghai's is currently 8 times that of Henna. In Indonesia, disparity in total revenues after grants is by and large as big as it was in 1994 (Table 3). Regarding own revenues, however, disparities were already on the rise prior to the beginning of decentralization in 1999. Ultimately, own revenues plus shared revenues show an increase in disparity post decentralization, especially due to natural resource revenue sharing.
Table 3. Cumulative Coefficients of Variation in Indonesia, (1994-2002) Own Resources Shared Taxes Shared Non-Taxes SDOIPWB Inpres Total Revenue Disparity
1994
1999
2000
2002
0.866 0.727 0.742 0.792 0.685 0.682
0.876 0.695 0.74 0.506 0.508 0.512
0.982 1.282 1.244 0.641 0.68 0.679
0.9 0.859 1.182 0.614 0.635 0.661
Source: SABER Database Notes: SDO (subsidi daerah otonom) or subsidy for autonomous region. After 2001, IMPRES (instruksi presiden) was made equal to the DAU.
2.2
Equalization Mechanisms in East Asian Countries
Indonesia, China, Vietnam, Thailand, and the Philippines have all taken measures to address fiscal disparities though an explicit equalization grant system (Table 4). Many of these grant systems have desirable features: all have a formula-based distribution of resources and three of the five countries have a formula-based determination of the pool of resources to be distributed. In three of these countries, the equalization system takes into account both revenue capacity and some form of expenditure needs, whereas that in Thailand and the Philippines only considers the expenditure needs side. The size of the distribution pool differs greatly from country to country: whereas for Indonesia and the Philippines the equalization grants system makes up the largest share of central-local grants, this is very different for Thailand and China, where earmarked grants dominate. In
Ensuring Inter-Regional Equity and Poverty Reduction
37
addition to the equalization grant, the distribution of earmarked grants sometimes includes equalizing elements, but in other cases these grants are not targeted to the poor regions, and may even have a counter-equalizing effect. Finally, although numbers are hard to come by, central spending also has a regional dimension, and if such spending is done on functions that are considered as subnational ones, such spending could be considered as part of the equalization system. This section reviews the current equalization grant systems in Indonesia, China, Vietnam, Thailand and the Philippines. In Indonesia, the equalization grant, Dana Alokasi Umum or DAU, is the mainstay of the intergovernmental fiscal system. The funding for the DAU consists of 25 percent of central government revenues after tax sharing with the regions." Ten percent of the DAU goes to the provincial level, which plays a relatively minor role in public services, and 90 percent goes to the local government level. On aggregate, this grant finances some 70 percent of local government spending, and some 50 percent of provincial government spending. The DAU is distributed according to a formula that takes both revenue capacity and expenditure needs into account. Revenue capacity is defined as potential own-source revenues plus shared tax revenues plus 75 percent of shared natural resource revenues.16 Expenditure needs are defined as a function of population, poverty rate, land area, and construction cost index as an indicator of "geographical circumstances." In addition to the formula allocation, part of the DAU is distributed based on past spending patterns-by and large to accommodate transitory effects that occurred in the 2001 decentralization. Finally, a lump sum per region plays a role in the allocation. The new earmarked grants system (DAK) is still small compared to the general grants system (about 3 percent of total grants), but also includes an element of equalization through required counterpart funding: regions with low fiscal capacity pay the minimum of 10 percent counterpart funds whereas those with high fiscal capacity pay up to 50 percent in counterpart funds. In China, an ad-hoc amount is dedicated to transfers to the 16 poorest provinces, and distributed on an equalizing basis. Although the 1994 Tax Sharing System (TSS) reform introduced a formula-based equalization scheme, almost a decade later it is still in a "transitional" status with limited funds." The formula-based scheme relies on variables like provincial GDP, student-teacher ratios, number of civil servants, and population density.18 The scheme remains small, and each beneficiary province receives only a fraction of its fiscal needs as determined by the transfer-allocation formula. In 2001, the equalization scheme accounted for only 3 percent of total central transfers.19 Although the TSS reform marked the first time that the equalization grant was explicitly budgeted, until the Government allocates sufficient funds to equalization, this formula-based scheme may only have a
Table 4. Equalization in intergovernmental transfer systems, selected countries Indonesia All regions receive an equalizing general grant
China 16 regions receive an equalizing grant. (Very small, in 2001 only 3 percent of all transfers.)
Philippines All regions receive a fixed share of central government tax revenues (IRA).
Formula-based source?
Yes, 25 percent of actual central government revenues after revenue sharing.
No; level of hnding decided by the annual budget based on ad hoc principles.
Yes; LGC sets IRA of LGUs at 40 percent of average internal revenue tax collections three years prior to the current year.
Main features of Formula
Expenditure needs and revenue capacity based, but 50 percent determined by transitional elements. Expenditure needs
Expenditure needs and revenue capacity based formula derived fiom regression analysis on "standard budget." Relies on variables
IRA divided among LGUs as follows: provinces (23 percent), cities (23 percent), municipalities (34 percent), and barangays
Equalization Grant Principles
Viet Nam Allocated to those jurisdictions where approved expenditure budgets exceed the sum of own revenues and the 100 percent retention of all shared revenues. Partially. Formula based on a calculated budget transfer between the center and the provinces. Negotiated expenditure needs. SNG expenditure needs minus total revenue from taxes shared 100 percent with SNG and total budget revenues shared between the central
Thailand Discretionary allocated to PAOs, Municipalities and TAOs and then further allocated to individual localities on a formula base.
No; total amount of different types of grants differs on a yearly basis.
5 percent reserved for unfounded devolving functions; 95 percent local authorities; of this amount, PAOs (7 percent),
Table 4. Equalization in intergovernmental transfer systems, selected countries (Continued) Main features of Formula (Continued)
Equalizing properties (weak, medium, strong)
Indonesia Expenditure needs as a function of population, poverty rate, land area, and construction cost index. Revenue capacity estimated as standardized own revenues (based on average efforts) plus shared tax revenues plus 75 percent of natural resource revenues.
China like provincial GDP, studentteacher ratios, number of civil servants, and population density.
Philippines (20 percent). IRA allocated on the basis of population (50 percent), land area (25 percent) and equal sharing (25 percent).
Viet Nam and SNG budgets
Medium; due to transitional elements and imperfections in formula.
Weak; due to limited resources.
Medium; IRA equalizing effect is not enough to counteract large disparities in tax base because and revenue capacity
Weak, but improved from ad hoc negotiated transfers used in the past; introduced clear objectives and stability by fixing formulas, thereby
Thailand Municipalities (52 percent), and TAOs (4 1 percent). The allocation across local authorities is based on the following criteria: equal share (25 percent), population (30 percent), area (5 percent), invert to local revenues, excluded grants (20 percent) and invert to specific grants received (20 percent). Weak; lack of transparency in allocation leads to self-interested politics; delays in allocation decisions have presented
Table 4. Equalization in i n t e r g o v e m e n t a l transfer systems, selected countries (Continued)
properties (weak, medium, strong) (~on&ued) Equalization through specific grants (i.e. ad hocl conditional)?
Indonesia
China
Philippines it weakly compares expenditure needs.
Viet Nam decreasing role of bargaining.
Yes. Special allocations dependent on fiscal capacity, but are small
No. Special grants de-equalizing, such as tax rebates state enterprise support. Lacks monitoring mechanisms to ensure effective use.
No; matching grants from central government agencies (augmentation funds) are proportionally small and usually subject to political interests.
No; absence of conditional grants in the conventional use of the term but targeted national programs which function in a similar way (see below).
Thailand difficulties for LG planning and managing finances throughout fiscal year. Yes; other general grants include: tax effort promotion, local good governance promotion, devolution of compensatory functions; train ticket compensation; local development and education. Specific grants include educational (capital projects) and development projects that are allocated on a project base.
Table 4. Equalization in intergovernmental transfer systems, selected countries (Continued) Other equalizing elements of the fiscal system
China Indonesia Distribution of natural [ Transitory period grants determined resources to regions on an ad hoc basis bordering producing considering annual regions. budgetary demands.
I
Source: Various PERs, Provincial expenditure reviews China, Indonesia.
Philippines
Viet Nam
I Targeted "national programs" channeled through provincial budgets to the final recipients, in support of poorest communes, reforestation, or the national health program.
Thailand
42
Hofinan and Guerra
symbolic significance (World Bank, 2002a). Earmarked grants in China, which make up over 95 percent of grants, have no equalizing element Two thirds of those grants are the so-called "tax return grants" which can perhaps be better understood as revenue sharing on a derivation basis. In Vietnam, the tax sharing cum contracting system de facto equalizes. The revised State Budget Law (2002) defines the equalization transfer amount as that given to those jurisdictions where approved expenditure needs exceed the sum of own revenues and the 100 percent retention of all shared revenue^.^' According to the equalization formula, the local tax administration branch determines revenue potential based on the actual revenue collections from previous years, factoring in any changes in tax policy as well as expected annual economic growth (Martinez-Vazquez, 2003). Expenditure needs are based on the prevailing system of expenditure norms, specified as money amounts per capita, rather than physical standards as prior to 1996, which were unaffordable. Regions are also allowed to keep revenues collected in excess of agreed shares. Beyond the equalization transfer, there are no conditional grants in Vietnam, rather a series of "national programs" for the poorest communes, reforestation goals andfor the national health objectives. These "national programs" are targeted, thus channeled through provincial budgets to the final recipients. In Thailand, the fiscal equalization grant is among the general type of grants, which for the most part have a formula-based al~ocation.~'Fiscal equalization grants have a strict formula with the following allocation criteria to local authorities: equal share (25 percent), population (30 percent), area (5 percent), invert to local revenues, excluded grants (20 percent), and invert to specific grants received (20 percent). The second type of grants specific grants - include educational (capital investment projects) and development projects which are allocated on a project based criteria. Although grants were the greatest source of intergovernmental transfers in 2003 (i.e. 38 percent of local revenues), the amount allocated varies on a yearly basis, and actual allocations are not known until well after the fiscal year begins. This leads to a lack of transparency across tiers of government and an extremely politicized system. In addition, the unpredictable nature of grants creates planning and budgeting difficulties for LGs. Ultimately, grants are unable to reflect or alleviate the broader intergovernmental framework and the vertical fiscal imbalance between the Central and subnational governments. Under the Local Government Code (LGC) in the Philippines, local governments (LGUs) receive a fixed share of central government tax revenues called the Internal Revenue Allotment (IRA). The Local Government Code sets the aggregate IRA share of LGUs to be 40 percent of internal revenue collections three years prior to the current year.22 Of this share, provinces and cities each receive 23 percent, municipalities receive 34
Ensuring Inter-Regional Equity and Poverty Reduction
43
percent, and barangays 20 percent.23Within each tier of government, the share is allocated according to three basic criteria: population (50 percent), land area (25 percent), and equal sharing across provinces (25 percent). Prior to the LGC, although allocation criteria were similar, their relative weights differed: population (70 percent), land area (20 percent), equal sharing (10 percent). The increased emphasis on equal sharing (i.e. rise fiom 10 to 20 percent) coupled with the fall in population weight (i.e. fall from 70 to 50 percent) shows a potential shift in equity concerns.
2.3
Do Grant Systems in East Asian Countries Equalize?
To answer this question, one first needs to define what is meant by "equalization." A commonly used one defines fiscal equality as "the capability of sub-national governments to deliver similar levels of services at similar levels of taxes," (Searle, 2002). While such a definition is important since it points to how an equalization grant might be best designed, it proves too difficult to be operationalized as it requires information on differences in the cost of services (which can be large) among subnational levels of government, and this data is just not yet available.24Here we therefore use a simpler approach, and ask whether the disparities among subnational government's revenues decrease as a result of the intergovernmental grants system describes above. The answer to that question is that indeed, for all the countries reviewed, the distribution of revenues per capita becomes more equal after transfers. The coefficient of variation of revenues per capita drops in each of the cases, although the equalizing effect of the transfer system varies significantly among the countries, and is strongest in Vietnam, and weakest in the Philippines (Figure 2 and Table 1& 5). Further evidence of the equalization properties of the transfer systems comes from cross-section analysis of revenues and expenditures per capita against income per capita. If the income elasticity of expenditures is lower than the income elasticity of revenues, it can be argued that the transfer system is equalizing.25This is indeed the case for the East Asian countries for which sufficient data are available (Table 6), suggesting that grants systems equalize. Even after transfers, though, the disparities in revenues per capita remain large. The richest province in the Philippines has 28 times more revenues per capita than the poorest one, while the same numbers for China, Indonesia, and Viet Nam are 8, 10, and 22 respectively. For comparison, in the US, the poorest state has about 65 percent of the revenues of the average state, and in Germany, any state falling below 95 percent of the average level gets subsidized through the LLFinanzausgleich"(and any receiving more than 110 percent gets taxed). In Brazil, the richest state has 2.3 times the revenues per capita of the poorest state (World Bank, 2002b).
44
Hofman and Guerra
Table 5. Disparities in Province-level Revenues After Grants, per Capita, (US$, Latest Available Year)
Maximum Minimum Average MadMin Standard Deviation Coefficient of Variation
China 444.4 42.8 100.7 10.4 83.1
Indonesia 431.4 39.8 106.3 10.8 78.9
Philippines 117.48 4.18 14.76 28.13 13.25
0.8
.74
.90
Thailand
Vietnam 393.1 25.1 65.9 15.7 65.2
.99
Source: SABER Database, China Statistical Yearbook, Finance Yearbook of China, Philippine Statistical Yearbook, Vietnam Statistical Yearbook, Ministry of Finance (Vietnam), Indonesia Regional Fiscal Information System, PERs, RPERs, Staff estimates. Notes: The table presents consolidated province-level data. The revenue numbers include own and shared revenues for Indonesia.
China
Indonesia
Philippines
Vietnam
Figure 2. Disparity in Provincial per Capita Revenues Before and After Transfers (Coefficient of Variation) Source: See Notes in Table 1 and 5.
Ensuring Inter-Regional Equity and Poverty Reduction
Table 6. Evidence of Equalization
Indonesia China Viet Nam
Income elasticity of revenue
Income elasticity of expenditures
0.8 1 1 .03
0.57 0.6 1
1.26
0.72
Note: Reported are the estimated parameter for the log of income per capita in a regression that regresses the log of revenues (expenditures) per capita against the log of income per capita and a constant. All the reported elasticities are significant at the 1 percent level. In Russia, disparities are larger: the richest of the 89 regions has revenues per capita some 40 times higher than the poorest (Martinez-Vazquez et al, 1998). However, Russia's regions are smaller than the average province in most East Asian countries, and indications are that the smaller the sub-national entity in a country, the larger the measured inequality. It is not surprising, therefore, that disparities are larger at the local level than at the provincial level: in Indonesia, the richest local government has thirty times the revenue per capita of the poorest one in 2002 while the comparable number for 2001 was over 50 times (Hofman et al, 2003) whereas within China's Gansu province, the revenues per capita of the richest county were 37 times as large as the poorest in 1999 (World Bank, 2002a). Thus, intra-provincial disparities appear to be larger than inter-provincial ones. A number of studies on individual countries c o n f m the modest performance on equalization of the intergovernmental fiscal system. A recent empirical study on the Philippines suggests that the IRA equalizing effect has not been sufficient to counter the disparities in tax base across the LGUs (Manasan, processed). The same study found that the IRA has a counter-equalizing effect in provinces between 1995-2000 and in municipalities in 1999-2000. In Indonesia, Lewis (2003) found the DAU distribution to be equalizing, but less than would be warranted on the basis of revenue capacity and expenditure needs alone whereas Hofman, Kajatmiko and Kaiser (2003) elaborate on the weak equalization performance of the DAU. In Vietnam, equalization performance of the fiscal system seems more pronounced. The distribution of revenue collections and allocations shows that a substantial proportion of revenues collected in the wealthier provinces is redistributed to the poorer provinces for spending. While revenue collections were significantly higher in wealthier provinces, the revenues ultimately allocated to them were significantly lower than those allocated to poorer regions and vice versa (World Bank, 2000a). Rao (2001)
Hofman and Guerra
46
presents cross-section estimates of the revenue and expenditure elasticity of income, showing that the latter is much lower than the former. This implies a significant degree of equalization, in line with the steep drop in coefficient of variation presented in this paper. For China, Yan (2003) shows through a simple regression analysis that the two fiscal reforms (in 1985 and 1994) were not able to exert meaningful effects on interregional equalization. Evidence on the distribution of earmarked grants indicates that they are predominantly flowing to the richer provinces (Achmad et al, 2000, and World Bank 1994). The fiscal system has been unable to alleviate the deequalizing effects of market reform and appears to have exacerbated regional disparities (Yan, 2003). Disparities in revenues per capita after grants translate into disparities in expenditures per capita, although not one-to-one. The difference may be access to borrowing, or use of built up reserves.
2.4
Do Fiscal Disparities Matter?
Do the observed fiscal disparities matter? Policy makers may be concerned about these disparities for a number of reasons, but from a poverty alleviation perspective they matter if they translate into large disparities in service delivery and service outcomes among subnational governments. Indicators of service delivery at the sub-national level are as scarce as fiscal data, but Human Development Indicators ( H D I ) , ~persons ~ per hospital bed:' literacy, 28 and life expectancy:9 are available for most countries under review. Persons per hospital bed can be considered as a service input indicator, whereas the other indicators are examples of outcome indicators. These indicators show a large variation across and within countries, but variation is lower than the fiscal indicators analyzed before.30 When considering persons per hospital beds, Indonesia's poorest ranking province is 7.6 times worst than the better ranking one, while the same numbers for Vietnam, China, Philippines, and Thailand are - if not as bad, yet still discouraging - 3.3, 3.4, 4.4, and 6.8 respectively.3' As in literacy rates, China's highest-ranking province is almost 3 times as literate as the lowest ranking one, while Indonesia (1.34) and Vietnam (1.9) are close behind. HDI and life expectancy indicators also confirm the same trend of disparity with Vietnam's highest province ranking nearly twice as high as the lowest one in both indicators, while China and Indonesia also rank slightly below. To assess whether there is a systemic relation between fiscal disparities and disparities in service delivery indicators, we use a simple regression approach. We use two specifications (Table 7): specification I is a simple regression of social outcome indicators (i.e. HDI, persons per hospital beds,
Ensuring Inter-Regional Equity and Poverty Reduction
Table 7. Disparities in Province-level HDI Indices, 1997-2002 Maximum Minimum Average MaxIMin Standard Deviation Coefficient of Variation
China
Indonesia
Philippines
Thailand
Vietnam
0.9
0.7
0.6
0.8
0.5
0.5
0.3
0.5
0.7
0.6
0.5
0.7
1.6 0.1
1.3 0.04
1.7 0.1
1.7 0.1
.10
0.1
0.2
0.08
Source: UNDP Human Development Indicators.
Notes: Data is not available across all provinces for Thailand. Data for the Philippines is at the regional level.
Table 8. Disparities in Province-level Persons per hospital bed, 1997-2002
Maximum Minimum Average Max/Min Standard Deviation Coefficient of Variation
China
Indonesia
Philippines
Thailand
Vietnam
650
3969
3503
1114
867.9
190
524
797
165
26 1.2
405.3
1862
2125.3
579.7
524.8
3.4 105.3
7.6 904
4.4 1537
6.8 494.7
3.3 152.9
0.3
0.5
0.7
0.9
0.3
Source: UNDP Human Development Indicators. Statistical Yearbooks
Notes: Data is not available across all provinces for Thailand. Data for the Philippines is at the regional level.
Hofman and Guerra
48
Table 9. Disparities in Province-level Life Expectancy, 1996-2002 Maximum Minimum Average Max/Min Standard Deviation Coefficient of Variation
China 78.1 64.3 71.2 1.2 3.2
Indonesia 72.2 59.3 66.4 1.2 3.1
Philippines 73.7 60.32 70.7 3.2
Thailand 79.8 61.6 72.3 1.2 35.7
Vietnam 74.5 54.6 70.4 1.4 4.0
.04
.05
.05
.49
.1
1.2
Source: UNDP Human Development Indicators, Statistical Yearbooks
Notes: Data is not available across all provinces for Thailand. Data for the Philippines is at the regional level and for males only. Data for Vietnam are for males only
Table 10. Disparities in Province-level Literacy Rate, 1994-2002 Maximum Minimum Average Max/Min Standard Deviation Coefficient of Variation
China 93.6 33.8 83.2 2.8 11.0
Indonesia 98.8 74.4 90.9 1.33 6.27
Philippines 98.8 73.5 92.2 1.3 5.9
.13
.07
0.1
Thailand
Vietnam 96.9 51.3 88.2 1.9 8.3
.I
Source: UNDP Human Development Indicators, Statistical Yearbooks
Notes: Data is not available across all provinces for Thailand. Data for the Philippines is at the regional level.
life expectancy, and literacy rates) as a hnction of a constant and expenditures per capita. In specification 11, we also include GPP per capita as one of the explanatory variables to control for potential correlations between provincial (regional) income per capita and expenditures - and their respective implications in terms of service delivery outcomes.32 Specification I suggests a modestly significant correlation between social outcomes indicators and subnational expenditures. For the most part in
Ensuring Inter-Regional Equity and Poverty Reduction
49
China and Vietnam, the four indicators are significantly correlated with expenditures. In China, the only exception is literacy rates, which although positive, is not significant; moreover, the reason may be that education financing is still partly under the auspices of the central government given recent efforts to re-centralize education expenditures. In Vietnam, the only exception is persons per hospital beds and the reason is that the government's role in the financing of health expenditures is relatively modest; "In 1998, the public expenditure share in aggregate health spending was only about 20 percent, with households accounting for the remaining 80 percent" (World Bank, 2000a). Accordingly, given the predominance of private fimding in health services, government subnational revenues then would not make a tangible difference in outcomes. In this case, then, infant mortality, for example, might be a better health outcome measure. Indeed, differing from life expectancy, which may imply other variables such as food source, for example, infant mortality is more directly related to health services. In Indonesia, however, the results are only significant for persons per hospital bed and life expectancy, and the reason for such relatively modest results may possibly be due to the very nascent decentralization process in the country. In the second specification, expenditures per capita are not found to be correlated with service output and outcome indicators. A possible explanation for these ambiguous results may be multicollinearity, which reduces the efficiency of the estimation-although the estimator remains unbiased3' Literature on public spending and outcomes shows that although public spending can be critical for improvements in health and education outcomes, there is large variation in the effectiveness of the use of finds. For example, although Thailand and Peru both increased public spending on primary schooling, primary schooling completion rates fell in Thailand while they increased in Peru. Conversely, Mexico and Jordan health spending diverged, while child mortality fell in both countries.34 Another possible explanation for finding only weak correlations between fiscal disparities and service delivery indicators is that the latter only change slowly over time. For all indicators, one would expect that only persistent low spending in a region would result in a deterioration of the indicators compared to regions that spend more. Only China has long enough time series to allow for testing the hypothesis that persistent fiscal inequities result in disparities in service delivery indicators. In this third set of regressions, we use a longer time series for provincial per capita expenditures (1985-2000) as the explanatory variable for service delivery indicators at the provincial level. We also add a fifth social indicator-
50
Hofman and Guerra
Table 11. Expenditure Disparities and Public Services Outputs and Outcomes. Results of regression analysis China
Indonesia
.102* -.O 1 -.48* -.39* .04* -.02* .03 -.03
.008 .002 -0.29* -.I5 .02* .O 1 .02 .O 1
Vietnam
Dependent Variable and Specification HDI I
HDI 11' Persons per Hospital Beds I Persons per Hospital Beds I1 Life Expectancy I Life Expectancy I1 Literacy I Literacy I1
.096* -.OO 1
.04 -.I5 .04* -.02 .06** -.02
Source: Staff estimates Note: The table presents the results of regression analysis with the output or outcome indicator as a dependent variable. Dependent variables are a constant and expenditures per capita in specification I (y-a + P*expenditure per capita), and a constant, expenditures per capita, and provincial GDP per capita in specification I1 (y=a + P*expenditure per capita + y*GPP per capita). A '*' ('**') indicates significance at the 5 (10) percent level. Average number of observations for each specification: : China (28), Indonesia (27), Vietnam (59). China results exclude Chongqing, Tibet and Qinghai. Indonesia results exclude Aceh, Papua and Maluku. Vietnam results exclude Ba Ria Vung Tau and Long An.
combined student enrollment for primary, secondary and tertiary schools to further test our hypothesis.35 In this case, excluding life expectancy, both specification I and I1 suggest a significant correlation between the four social outcome indicators and subnational expenditures across time, thereby suggesting that persistent fiscal inequities do matter in social outcomes.36 In sum, although the variation in outcome indicators is much less than the variation in revenues, there is some evidence that outcomes are affected by different levels of sub-national revenues across provinces. In the short run, effects seem dominated by differences in income per capita, but evidence on China suggests that persistent fiscal inequalities result in inequalities in social indicators beyond those that can be explained by income per capita.
Ensuring Inter-Regional Equity and Poverty Reduction
CONCLUSIONS This paper has shown that fiscal disparities among sub-national governments in East Asian countries are considerable. Equalization mechanisms diminish subnational fiscal disparities, yet even after equalization, disparities remain large. The paper finds disparities among service delivery outputs and outcomes to be considerable as well, but data and methodological issues prevent us from establishing a strong link with the large fiscal disparities. For the one country for which sufficient data are available (China), the paper finds that persistent fiscal disparities do seem to matter in health and education outcomes. Nonetheless, there are numerous reasons why inequalities may persist. The first is that there may be significant variation in expenditure needs. This could be for two reasons: (i) a large variety in costs among the sub-national governments and (ii) an asymmetry in decentralization, or in other words, some regions do more than others. A second is the central government's emphasis on revenue mobilization. Too much equalization, if not well designed, could reduce the incentive for own revenue mobilization, to the detriment of general government's tax take in the economy. A third could be that inequalities among regions could induce a desirable migration to regions with better economic prospects. A fourth reason is that poor regions may be less capable in handling money than rich ones, or less concerned with poverty alleviation than the center. A fifth is a more political one: rich regions are also powerful regions, and they do not like to lose out against the poorer regions. Particular reasons in the East Asian context include the coastal development strategy in China, which allowed "some provinces to get rich first" and Indonesia, where natural resource regions with separatist tendencies were bought with a larger share of the natural resource revenues. A fifth reason is simply that rich regions are often also powerful regions, and it is hard for the center to tax away and redistribute resources from the rich regions. Based on these preliminary findings and on the broader implications of inter-regional equity, we have a set of recommendations in four areas: (i) information management, (ii) transfer systems, (iii) central government role (iv) future research and analysis. The policy debate on fiscal inequalities, and on intergovernmental fiscal relations more general, requires more data. Without more and sound information on how large inequalities are and how they have evolved over time, a policy debate on what inequalities are acceptable or desirable is impossible. In each of the countries in East Asia there is a dearth of sub-national fiscal-related data. The data for the countries analyzed in this paper are far from perfect, but it has to be emphasized that many other country cases had to be dropped because there were no fiscal data at all available at the subnational level. The lack of data hold even
52
HoJinan and Guerra
more true for levels of government below the first tier sub-national government. In addition, data on service delivery cost differentials is also critical to assess inter-regional disparities in access to services. Better data requires setting up monitoring systems in govemment, an undertaking that requires significant resources. Indonesia, for example, has managed to maintain a database on sub-national fiscal information at the center, which, supported by the legal requirement of regions to report, has information on most of its 410 local governments. In China, the data are there at the originating level, but the aggregation of information at each level of government implies that the central government has little relevant information on the fiscal situation at the sub-national level. For some countries in order to get better data, adjustments in accounting systems and budget classification is needed as well. For instance, in China, significant fiscal resources are classified as "extrabudgetary" and the information on those funds is even scarcer than that on budgetary funds, and they are large enough to hdamentally alter conclusions on fiscal disparities and other policy issues. Apart from more data, more policy analysis should be done on the data to inform the policy debate. Ultimately, countries should aim to regularly review the results and progress of their respective intergovernmental fiscal systems, including issues of fiscal disparities and service delivery disparities. For example, following the highly successful example of South Africa, Indonesia has published (March 2004) the first of what is to be a regular intergovernmentalfiscal review. Such a report would allow policy makers to evaluate their intergovernmental fiscal system on a regular basis. Irrespective of whether more or less fiscal equalization is desirable, there is a significant scope for improvement in the design of intergovernmental systems. In most countries, there is a need to identify a more comprehensive objective for the equalization system as a whole. The center must determine its equalization objectives and set of priorities (i.e. income levels, fiscal capacity, expenditure needs, per person revenues available) within a politically viable scenario, thereby defining the h d i n g sources (i.e. central govemment revenues versus fraternal contributions) (Bahl, 2000). Objectives pursued by equalization grants are frequently unclear, wherein some grants embody the features of earmarked grants - usually based on temporary considerations. A more comprehensive objective, then, would be to aim for each local government to at least be able to deliver a minimum level of public goods and services. Country-specific recommendations include: China's transfer system could aim to reflect local revenue capacity and expenditures needs in a rules-based manner, rather than through an ad hoc distribution and derivation. In addition to a more transparent system,
Ensuring Inter-Regional Equity and Poverty Reduction
53
China should also aim for simpler equalization mechanisms such as potentially consolidating the various equalization grants into one overarching grant. To further enhance both transparency and simplicity, it would be helpful to systematically identify the total pool of equalization fbnds - to be decided based on potential tradeoffs between equalization, growth and overall incentives for local revenue mobilization and expenditures. In Vietnam, negotiation and discretion remain primarily for "surplus" provinces and local governments. Moreover, it should consider the introduction of conditional grants and/or other transfer instruments with or without matching provisions - to create incentives and provide financing for expenditures in national priority areas such as health and education. This fhding can also be used for social assistance programs at the local level along with grants for capital infrastructure among other options. Furthermore, Martinez-Vazquez argues for the need to structure transfers at the sub-provincial level (Martinez-Vazquez, 2004b). Although Indonesia's transfer system introduces the notion of expenditure needs and revenue capacity through its equalization grant, it should move towards defining a more equalizing DAU by phasing out the transitional elements of base amount and "hold harmless" in the allocation. In terms of transparency and simplicity, Indonesia should primarily decide on a more consistent treatment of natural resource revenues in revenue sharing and in the equalization formula. In addition, it should introduce a selective system of specific grants combined with an (on) lending window to promote the financing of national priorities at the local level. As for future research agenda, and analysis, the data we collected for the most recent year available shows that although there are huge fiscal disparities across subnational governments, they are relatively modest for health and education outcomes. Why? What are the countervailing forces within each country or groups of subnational provinces and/or municipalities? Are there migration tendencies? Is there a push for asymmetrical decentralization? What are government preferences and what role do this play in actual policy-making? Moreover, discussion about the center's role and the impact of its central programs are critical in the context of decentralizing inter-governmental relations. These and many other questions are a fruitful line of research requiring fiuther empirical analysis.
54
Hojinan and Guerra
Notes The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not represent the views of the World Bank, its Executive Directors, or the countries they represent. The authors are very grateful to Dana Weist, Roland White, Paul Smoke, Bob Ebel, Jorge Martinez-Vazquez, Roy Bahl, Tamar Manuelyan Atinc, Richard Bird, David Rosenblatt, Roy van der Weide, Sjamsu Rahardjs, Kai Kaiser, Blane Lewis, Arvind Gupta, Shan Gooptu, Minh Van Nguyen and Jessica Tisch for helpful comments, and wish to express thanks to colleagues who provided subnational fiscal data (Ed Mountfield, Bambang Suharnoko, Bastian Zaini, Joven Balbosa, Amit Mukherjee, Khuankaew Varakornkarn, Xiaofan Liu, Christine Wong). Migration can be considered excessive in an economic sense, if the marginal productivity of a worker would be higher in his place of origin, or if the congestion costs in the region of destination are larger than the private benefits obtained from the better services; see Ahmad and Craig in Ter-Minassian, eds. (1997). For an extensive discussion of this point, see Tanzi (1995) and Prud'homme (1 995). A review, assessing constitutional rights to education and health care in 187 countries, concludes that of the 165 counties with written constitutions available, 116 made reference to a right to education and 73 to a right to health care (Gauri 2003). In the Philippines, for example, IRA allocations have somewhat detracted from LGUs incentive to improve the collection of own source revenues (World Bank 2003a). Art. XVIII (a) sub 2 of the Indonesian Constitution states "The relationship in finances, public services, utilization of natural resources and other resources .....shall be regulated and executed fairly and equitably based on the law." Art I1 section 9 and 10 of the Philippine Constitution states "The State shall promote a just and dynamic social order that will ensure the prosperity and independence of the nation and free the people from poverty through policies that provide adequate social services, promote full employment, a rising standard of living, and an improved quality of life for all. The State shall promote social justice in all phases of national development." Excludes the oil-producing region of Ba-Ria Vung Tao (Vietnam). Data limitations force us to restrict our sample to 5 countries: China, Indonesia, Philippines, Thailand, and Vietnam. SABER Database refers to the authors' original data collection process. Fiscal disparities can differ depending on the indicator of inequality chosen. In addition to the method chosen, a future draft of this paper will also show fiscal disparity calculations using the population weighted Theil indicator, aiming to have a better measure of the impact of fiscal disparities from the individual's perspective. This is a systemic point: the smaller the entities analyzed, the larger the disparities found. The level of fiscal disparities is sensitive to cost differentials across provinces; however, data on provincial cost of living is not readily available for the countries in the sample. Data limitations, along with concomitant changes in respective countries' decentralization processes, make similar type of analysis very difficult for other countries in our sample. In principle, the law prescribes this to be 25 percent of actual revenues, but for now the practice has been to take budgeted revenues as the basis for the aggregate amount of DAU. 16. For detail, see Hofman, Kajatmiko and Kaiser (2003).
Ensuring Inter-Regional Equity and Poverty Reduction
55
As a result of the 1994 TSS reform revenue sharing arrangements were changed from a negotiated, mildly equalizing system to one based on strict tax assignments (Wong, processed). For detail, see Ahmad, Ehtisham, Li Keping, Thomas Richardson (2000). World Bank, 2004e; see China Country Review. World Bank, 2004e; see Vietnam Country Review. In addition to fiscal equalization, general grants include tax effort promotion, local good governance promotion, devolution of compulsory functions, train ticket compensation, local develovment. and education. Excmtions to the formula-based allocation criteria are: local good governance, train ticket compensation and local development. Although the aggregate IRA share is fixed by law, subsequent to the 1998 Asian financial crisis, the central government withheld - without any consultation processes 5 percent of the IRA. Despite the Supreme Court ruling in favor of LGUs, there still remains disagreement whether the formula should be subjected to appropriation or whether it should be decreased as part of the appropriation process in Congress; see Manasan (200X). Prior to the LGC, inter-tier allocation differed greatly: provinces (27 percent), cities (22 percent), municipalities (41 percent) and barangays (10 percent). There has been some success in utilizing this data for purposes of grant design in the United States. See Robert Rafuse, Representative Expenditures, US Advisory Commission on Intergovernmental Relations (1992). This ignores the effect of borrowing. However, since it is likely that richer regions have more access to borrowing, the conclusions on equalization are likely to hold. This report adopts the UNDP procedure for computing HDI. The computation of provincial HDI is based on three indicators: longevity, as measured by life-expectancy at birth; education attainment, as measured by a combination of adult literacy (two-thirds weight) and combined primary, secondary and tertiary enrolment ratios (one-third weight); and standard of living, as measured by real GDP per capita in terms of PPP$. (UNDP, 1999). This paper broadly defines persons per hospital beds as the number of people per each inpatient bed available in public hospitals, unless specified. In Indonesia, the indicators includes special and general hospitals as well. This paper broadly defines literacy rate as the percentage of people ages 15 and above who cannot, with understanding, read and write a short, simple statement about their everyday life (World Bank, 2003~). This paper broadly defines life expectancy as the number of years a newborn infant would live if prevailing patterns of mortality at the time of its birth were to stay the same throughout its life (World Bank, 2003~). Variation is measured through the coefficient of variation, which is the standard deviation divided by the average. Although within countries max-to-min ratios are able to pick up the same trends as more sophisticated measures (e.g. gini coefficient), across countries they are more responsive than other measures to the tail-end of the distribution. Specification I1 aims to address the problem of reverse causality between expenditures (=revenues) and outcomes. Otherwise, reverse causality would suggest the possibility that our regression estimates may be upwardly biased. The possible problem of multicollinearity in specification 11, then, will later be addressed in the third set of regressions in the paper, when examining a longer time series for China. Indeed, expenditures are found to be highly correlated with GPP for China (.67), Indonesia (.42), and Vietnam (.67) with significance at the 1 percent level. In this case, subject to the availability of data, it would be interesting to see if the same results hold at even lower levels of governments. The only exception to the rule is China. (World Bank, 2004a).
56
Hofinan and Guerra
35. As demonstrated in the previous set of regressions, life expectancy yields ambiguous results since it is affected by a broad range of factors that are not controlled for in our specifications. Consequently, infant or child mortality would be a better indicator. In the absence of this kind of data, we have to chosen to use combined gross enrollment as our fifth outcome indicator in this third and final set of regressions. 36. The only exception to the results is HDI (in specification 11), yet this is understandable since GDP is one of its components.
References Ahmad, Ehtisham, and John Craig. 1997. Intergovernmental transfers, In Fiscal Federalism in Theory and Practice, Teresa Ter-Minassian, ed., Washington DC.: International Monetary Fund. , Li Keping and Thomas Richardson. 2000. Recentralization in China. Paper presented at Conference on Fiscal Decentralization, November. Bahl, Roy. 2000. Intergovernmental transfers in developing and transition countries: Principles and practice, Urban and Local Government: Municipal Finance. Washington DC: World Bank. ,and Jorge Martinez-Vazquez. Fiscal federalism and economic reform in China. International Studies Program, Georgia State University. Processed. Balisacan, Arsenio M., Ernesto M. Pernia, Abuzar Asra. 2003. Revisiting growth and poverty reduction in Indonesia: What do subnational data show?. Bulletin of Indonesian Economic Studies, 39(3): 329-5 1. Birdsall, Nancy, Estelle James. 1990. Eficiency and equity in social spending: How and why governments misbehave. Policy, Research, and External Affairs, Population and Human Resources Operations. Washington DC: World Bank. Bourguignon, Franqois. 2003. Transition of China's Northeast: The need for combining regional and national policies. Paper for development strategy for Northeast China Seminar, Shenyang, Liaoning Province, China. December 3-4. Brodjonegoro, Bambang. 2000. Regional autonomy and fiscal decentralization in democratic Indonesia. Hitotsubashi Journal ofEconomics, 4 1. Buchanan, James M. 1950. Federalism and fiscal equity. The American Economic Review, 40(4). Ebel, Bob. 2003. Intergovernmental transfers. Paper present at Decentralization and Intergovernmental Fiscal Reform Workshop. Washington DC:World Bank. March 24-26. . 2003. Subnational revenues and intergovernmental relations. Paper present at Decentralization and Intergovernmental Fiscal Reform Workshop. Washington DC: World Bank. March 24-26. Filmer, Deon. 2003. The incidence of public expenditures on health and Education. Background note for World Development Report 2004: Making Services Work for Poor People. Washington DC: World Bank. Garcia-Garcia, Jorge and Lana Soehstianingsih. 1998. Why do differences in Provincial Income Persist in Indonesia. Bulletin of Indonesian Economic Studies, 34 (1). Gauri, Vamn. 2003. Social Rights and Economics : Claims to health care and education in developing countries. Development Research Group. Washington DC: World Bank.
Ensuring Inter-Regional Equity and Poverty Reduction
57
Gill, Nicholas, and Andres Rodriguez-Pose. 2003. Is there a global link between regional disparities and devolution?. Research Papers in Environmental and Spatial Analysis No. 79. Department of Geography & Environment, London School of Economics and Political Science, London. Grosh, Margaret E. 1995. Toward quantifying the trade-off: administrative costs and incidence in targeted programs in Latin America. In Public Spending and the Poor: Theory and Evidence, Dominique Van de Walle and Kimberly Nead, eds., Washington DC: World Bank. Hegedus, Jozsef. 2001. Creating change networks for local governance. Paper present at WBI/LGI/UNDP Initiative for Central Asia, Bratislava. July30-August 3. Hill, Hal. 2002. Spatial disparities in developing East Asia: A survey. Asian-Pacific Economic Literature, 16(1). . 2000. Intra-country regional disparities. Paper present at the Second Asian Development Forum. Singapore. June 6-8. Hofman, Bert, Kadjatmiko, Kai Kaiser. 2003. Fiscal equalization in Indonesia's 2001 "Big Bang" decentralization. Forthcoming. Jimenez, Emmanuel, Donald Cox. 1995. Private transfers and the effectiveness of public income redistribution in the Philippines. In Public Spending and the Poor: Theory and Evidence, Dominique Van de Walle and Kimberly Nead, eds., Washington DC: World Bank. Lewis, Blane. 2003. Local government borrowing and repayment in Indonesia: Does fiscal capacity matter?. World Development, 3 l(6): 1047-1063. Manasan, Rosario. Fiscal decentralization: The case of the Philippines. Philippine Institute for Development Studies. Forthcoming. Martinez-Vazquez, Jorge, and Jameson Boex. 1998. Fiscal decentralization in the Russian Federation: Main trends and issues. Paper prepared for EDIIWorld Bank. Martinez-Vazquez, Jorge, and Jameson Boex. The design of equalization grants: Theory and applications. Course on Intergovernmental Relations. World Bank Institute, World Bank: Washington DC. Retrieved from http://www. w o r l d b a n k . o r g / w b i / p u b l i c f i n a n c e / d e c e n t r a Martinez-Vazquez. Jorge, and Robert M. McNab. 2003. Fiscal decentralization and economic growth. World Development, Forthcoming. . 2003. Improving the design of fiscal decentralization in Vietnam. Paper presentation. February. . 2004a. The design of fiscal decentralization: Principles and best international practice. Paper presentation at PER-IFA Mission. Hanoi. April. . 2004b. Improving the design of fiscal decentralization in Vietnam. Paper prepared for World Bank Vietnam Public Expenditure Review, Forthcoming. McLure, Charles E. Jr. 1999. The tax assignment problem: Conceptual and administrative considerations in achieving subnational fiscal autonomy. Washington DC: World Bank. Milanovic, Branko. 2001. Conflict between horizontal equity and maximum poverty reduction, how best to allocatefunds to regions: An empirical analysis. Washington DC : World Bank. Pritchett, Lant, Jonah Gelbach. 2000. Indicator targeting in a political economy: Leakier can be better. Washington DC: World Bank. Pritchett, Lant, Deon Filmer, Jeffrey Hammer. 1999. Health policy in poor countries: weak links in the chain. Washington DC: World Bank. Prud'homme, R h y . 1995. The dangers of decentralization. World Bank Research Observer, 1O(2).
58
Hofman and Guerra
Qiao, Baoyun, Jorge Martinez-Vazquez and Yongsheng Xu. 2003. Growth and equity tradeoff in decentralization policy: China's experience. International Studies Program Working Paper No. 02-16, Georgia State University, Atlanta, GA. Rahse, Robert. 1992. Representative Expenditure. US Advisory Commission on Intergovernmental Relations. Rao, M. Govinda. Fiscal assignment and service delivery. Paper present at Institute for Social and Economic Change. Bangalore, India. . Poverty alleviation under fiscal decentralization. Institute for Social and Economic Change. Bangalore, India. Forthcoming. . 2001. Challenges in fiscal decentralization: An Asian perspective. Paper presented at the conference Public Finance in Developing and Transition Countries. Atlanta, Georgia. , Richard M. Bird, Jennie Litvack 1998. Fiscal decentralization and poverty alleviation in a transitional economy: The case of Viet Nam. Asian Economic Journal, 12 (4): 353-378. Ravallion, Martin. 1999. Are poorer states worse at targeting their poor?. Economic Letters, 65. . 1990. Gaurav Datt. Regional disparities, targeting, and poverty in India. Policy, Research, and Extemal Affairs, Population and Human Resources Operations. Washington DC.: World Bank. . 1998. Reaching poor areas in a federal system. Policy Research Working Paper, Development Research Group. World Bank, Washington DC. . 1999. Is More Targeting Consistent with Less Spending. Washington DC : World Bank. . 1993. Poverty alleviation through regional targeting: A case study for Indonesia. In The Economics of Rural Organization: Theory, Practice and Policy, Karla Hoff, Avishay Braverman, and Joseph Stiglitz, eds., Washington DC and New York : World Bank and Oxford University Press. Searle, Bob. 2002. Federal fiscal relations in Australia. Paper present at International Centre for Economic Research. Torino, Italy. Smoke, Paul. Forthcoming. Fiscal decentralization in developing countries: A review of current concepts and practice. United Nations Research Institute for Social Development, Palais des Nations. Tanzi, Vito. and L. Shuknecht. 1995. The growth of government and the reform of the state in industrial countries. IMF Working Paper. December. Ter-Minassian, Teresa 1997. Fiscal Federalism in Theoty and Practice. Washington DC: International Monetary Fund. United Nations Development Program, China Human Development Report 1999, United Nations: New York. Van de Walle, Dominique. 1995. Incidence and targeting: An overview of implications for research and policy. In Public Spending and the Poor: Theoty and Evidence, Dominique Van de Walle and Kimberly Nead, eds., Washington DC: World Bank. Wong, Christine. 2002. Issues of equalization in China. Paper present at Intergovernmental Fiscal Relations in East Asia Workshop. World Bank: Washington DC, January. World Bank. 1993. China: Budgetary policy and intergovernmental fiscal relations. Report no. I 1094-CHA, Washington DC. World Bank. 2000a. Managing Public Resources Better. Vietnam Public Expenditure Review. Washington DC.
Ensuring Inter-Regional Equity and Poverty Reduction
59
World Bank. 2002a. China: National Development and Subnational Finance: A Review of Provincial Expenditures. Washington DC. World Bank. 2002b. Brazil: Issues in Brazilian Federalism. Brazil Country Management Unit, Poverty Reduction Economic Management Sector Unit, Latin America and the Caribbean Region. Washington DC. World Bank. 2003a. Philippines: Improving Government Performance: Discipline, Efficiency and Equity in Managing Public Resources. Poverty Reduction Economic Management, East Asia and the Pacific Region. Washington DC. World Bank. 2003b. Decentralizing Indonesia. Regional Public Expenditure Review, Poverty Reduction Economic Management, East Asia and the Pacific Region. Washington DC. World Bank. 2003c. World Development Indicators. World Bank: Washington DC. World Bank. 2004a. World Development Report: Making Services Workfor Poor People. Washington DC: World Bank and Oxford University Press. World Bank. 2004b. Chinese 11" Five Year Plan. Concept Note, Development Economic Research Group. Washington DC. World Bank. 2004c. 2004 Public Expenditure Review & Integrated Fiduciary Assessment. Aide Memoire, Multi-Donor Mission. World Bank. 2004d. Inequality in Latin America and the Caribbean: Breaking with History?. World Bank Latin American and Caribbean Studies. Washington DC. World Bank. 2004e. Managing Decentralization in East Asia Poverty Reduction Economic Management, East Asia and Pacific Region. Washington DC: World Bank. Forthcoming. Yan. 2003. Regional Equalization in China's Public Finance. Forthcoming. Zhang, Zhihua, and Jorge Martinez-Vazquez. 2003. The system of equalization transfers in China. International Studies Program Working Paper No. 03-13, Andrew Young School of Policy Studies, Georgia State University, Atlanta, GA.
Chapter 4
THE IMPACT OF EQUALIZATION ON SERVICE DELIVERY CATHERINE HULL* AND BOB SEARLE**' *The Commonwealth Grants Commission, Australian Government. **Consultant on Fiscal Decentralization, Formerly the Commonwealth Grants Commission, Australian Government
1.
THE AUSTRALIAN EXPERIENCE
The founders of the Australian federation2 recognized the need for transfer payments from the national to state level governments and made provision in the Constitution to achieve that end. Transfers are required to overcome the vertical fiscal imbalance (VFI) between the levels of government and also to provide greater financial assistance to some States than to others. The arrangements for providing extra assistance to particular States were first formalized in 1933 with the creation of the Commonwealth Grants Commission (the Commission). In 1936, the Commission (and the Australian Government) took a concept developed earlier by one of its members3 and decided that grants to the States would be provided on the basis of horizontal fiscal equalization (HFE). However, it was not until the early 1980s that a comprehensive system of HFE was introduced. As a result of this comprehensive system, the Commission has 26 years of data, from 1977-78 to 2002-03, on: i) The revenues and expenses of the States on a largely comparable accounting and classification basis ii) Australian average per capita State revenues and expenses iii) Standardized revenues and expenses of the states4 iv) The relative revenue raising capacities of the States, and their relative costs of providing Australia average levels of services; and v) The relative revenue raising efforts made by the States, and the relative levels of services they provide.
Hull and Searle
62
Using this database, it is possible to consider whether the move to a comprehensive HFE system post-1982 has had an influence on the levels of services provided by the States and their revenue raising efforts. Since fiscal capacity equalization is practiced in Australia, it is unlikely that service standards or revenue efforts will ever converge. However, it is possible that the differences have decreased since comprehensive equalization was introduced. Under the comprehensive equalization now practiced, it might be expected that service levels will fluctuate in a band around an average as jurisdictions ensure their residents do not 'miss out7on appropriate services. Of course, electoral events and other central government policies, expressed through conditional or tied grants, will also have an influence. This paper provides information on the design of the Australian fiscal transfer system and uses the Commission's data to consider whether the move to comprehensive equalization has had any impact on relative service delivery outcomes.
1 .
The Fiscal Transfer System in Australia
In Australia, there are three levels of government: national, state and local. The national government is responsible for Australia wide services such as defense, trade, foreign affairs, communications and social security. The State governments are responsible for the major services such as health, education, and law and order. Local governments are minor players, with responsibility for services such as garbage collection, local roads and other municipal amenities. All levels have their own taxing powers, although the national government has the major taxes such as income and company tax, customs and excise, while State governments have access to a range of indirect taxes such as payroll tax, land and mining revenues, and stamp duties. Local governments raise their revenues from land rates and user charges. The amount of revenue raised by the national government is much larger than its own purpose outlays require. By contrast, the States7own purpose outlays greatly exceed the revenues they raise from their own sources there is a vertical fiscal imbalance. Similarly there is an imbalance at the local level. All nations with more than one level of government have some VFI, although it is not often as large as in Australia. Fiscal transfers fi-om the Australian Government to the States and local governments are used to redress VFI. This paper focuses on the impact of these transfers to the States. Currently, about 60 per cent of the transfers to the States are in the form of untied funds (funds the States can spend as they choose). The other 40 per cent are tied funds, or specific purpose payments
The Impact of Equalization on Sewice Delivery
63
(SPPs), which are grants that must be spent on the functions for which they are given. The absolute amount of the untied funds provided to the States is the revenue raised from a Goods and Services Tax (GST) imposed by the Australian Government (less the costs of tax administration), and a number of other payments. The GST pool5 is estimated to be A$42 billion6 in 200405' and represents the largest single intergovernmental transfer in Australia. It represents about 35 per cent of the gross operating expenses of the States in total8 and is growing. Over the five years to 2002-03, it varied fi-om 30 per cent in Western Australia to 62 per cent in the Northern Territory. The Australian Government legislation titled A New Tax System (Commonwealth-State Financial Arrangements) Act 1999 requires that the GST pool be distributed among the States according to HFE principles. The Commission is charged with the responsibility of recommending appropriate shares for the States. The Act also makes it clear that the GST revenue allocated to each State may be spent according to each State's own priorities - that is, the Australian Government has no control over how States choose to spend the funds. To achieve national policy objectives in areas of State constitutional and traditional functional responsibilities (for example, in the health and education areas), the Australian Government provides SPPs. SPPs are used to influence State priorities and to support specific policy objectives of the Australian Government to which the States have agreed. The Australian Government does not seek to take over responsibility for State functions. There are around 90 SPPs for education, health, housing, social welfare, transport and local government functions. Some cover capital as well as recurrent programs. Distribution arrangements differ for each SPP. Block grants for government schools are largely distributed on a per student basis, with additional payments for students with special characteristics, such as Indigeneity or remoteness. Home and Community Care grants are funded on a matching basis and other payments, such as that for the Natural Heritage Trust of Australia, are made on the basis of funding applications.
1.2
Horizontal Fiscal Equalization in Australia
The objective of HFE as applied in Australia has remained much the same since the very early days of the Commission, though its expression and how it has been implemented have varied over time. The objective is presently stated as: State governments should receive funding from the pool of goods and services tax revenue and health care grants such that, if each made the
Hull and Searle
same effort to raise revenue from its own sources and operated at the same level of efficiency, each would have the capacity to provide services at the same ~tandard.~ The Commission uses this principle to make recommendations to the Australian Government on the share of the GST pool that each State should receive to enable it to provide services at the average standard, without having to make different tax efforts. The deftnition makes it clear that States' fiscal capacities, not their performances, are being equalized. Fiscal equalization is directed towards giving States the capacity to provide the same standard of service, but does not require them to do so. As noted above, it is an essential feature of intergovernmental transfers in Australia that the GST revenue distributed among the States under equalization arrangements is untied in the hands of the States. States have their own role in the federation and may follow their own policies on both sides of their budgets. The untied nature of these finds has important implications for service delivery outcomes. The Australian scheme is based on a concept similar to that of net fiscal benefit, or the difference between taxes paid and public services received1'. It allows for different levels of taxation and public services across the States, according to the preferences of residents. However, net fiscal benefits in Australia are not necessarily the same for similar individuals in all States. State government budgets are equalized, but this may or may not result in equity for similar individuals, depending on government decisions. Vaillancourt and ~ i r d "conclude that the basic rationale for equalization schemes is to enable sub-national governments to 'replicate the outcomes of a unitary State - if they choose to do SO'. In a general sense, this is the purpose of the Australian scheme. It aims to ensure that service delivery and taxation outcomes could be the same, if jurisdictions wished them to be. The implementation of HFE in Australia changed in 198212. Prior to this, financial assistance grants and special grants (both untied) were paid to States under a set of arrangements which could best be described as ad hoc. In 1982, the first assessment of relative needs of the six States was undertaken by the Commission and used to distribute a pool of revenue provided by the Commonwealth. Since then, the system has become more comprehensive over time. The Northern Territory was included in 1985 and the Australian Capital Territory (ACT) in 1993 and, since 1999, all State services (based on full operating costs -recurrent and capital) and revenues have been considered in developing the GST revenue sharing formula.
The Impact of Equalization on Service Delivery
1.3
65
The Commonwealth Grants Commission's Model
To implement WE, the Commission uses a closed model. Historical financial data are used to construct a picture of 'what States did' in terms of their revenue raising and service provision over the previous five years. National per capita averages, or standards, are calculated for each function13 of government and head of State revenue. This information is derived from State budgets, Government Finance Statistics (GFS) published by the Australian Bureau of Statistics and other sources. It is these standards, not the actual expenses or revenues of each individual State, which are the starting point. That is, the States are equalized to what, on average, they actually do. Next, the Commission measures differences in States' per capita costs of providing services and per capita capacities to raise revenue. Only influences outside the control of State governments are recognized as 'disabilities' or 'needs714and taken into account in the equalization model. This requires data on State populations and their demographic and geographic characteristics, on resource endowments and on economic conditions. It also requires judgment to ensure that only those differences between the States that cannot be influenced by State policies are recognized. For example, if State governments choose to pay higher wages to their staff, the Commission would not recognize this as a valid reason for an adjustment to the standard. If, however, States need to pay higher wages to attract staff to very remote or highly urbanized areas, this would be considered a valid reason for an adjustment to the average per capita spend on a function. Such influences on the costs of providing services can be either positive or negative. The standardized per capita expenses of a State on a function are calculated when these per capita needs are added to the standard, or subtracted from it when the estimated per capita costs are below standard. The standardized expense per capita represents the amount a State is considered by the Commission as needing to spend to provide the Australian average level of service. Similarly, if one State has a smaller (larger) revenue base than another for reasons beyond its control, and is not able to raise as much (more) revenue by applying the average rates of tax, then it has a greater (lesser) need for GST revenue. The standardized per capita revenue of a State is the amount the Commission calculates it would raise if it applied the Australian average effort to its tax base.
Hull and Searle
The Commission's model can be presented mathematically in two ways. i) A State's per capita share of GST revenue equals its per capita share plus its assessed needs per capita (expense and revenue), less per capita needs due to SPPs; or ii) A State's per capita share of GST revenue equals its standardized per capita expenses less its standardized per capita revenues less its per capita SPP receipts, less a standard per capita budget result. A State's share of the GST revenue reflects the amount it would spend to provide the standard level of services, after taking account of the revenue it would raise from its own sources by making a standard revenue raising effort, its receipts from SPPs and allowing for a standard per capita budget deficit (or surplus). The second presentation recognizes that the States can, and do, operate with a surplus or a deficit each year and provides for an average outcome to be reflected in the 'equalization result'. There are no 'needs' associated with the budget result as it is the same per capita amount for all States. A full mathematical exposition of the modelI5 and an explanation of how the Commission implements HFE can be found in its most recent report16.
WHAT THE AUSTRALIAN DATA SHOW As a by-product of the Commission's work, the data collected and generated can be used to calculate ratios to compare the level of services States provide and the effort they put into raising revenue. A level of service provision ratio (or expense effort) is the ratio of actual expenses incurred by a State on a service to its standardized per capita expenses - what it actually spends compared to what the Commission considers it would spend if it were to provide the standard (average) level of services. An index above 100 indicates the extent to which the cost of providing the service is above the Australian average due to policy choices about service levels andlor relative inefficiencies. A revenue raising effort ratio is the ratio of actual per capita revenue raised by a State to standardized per capita revenue -what it actually raises compared to what the Commission considers it would raise if it applied the average tax imposts to its revenue bases. An index above 100 indicates effort above the Australian average. Table 1 shows the level of service provision ratios for total State services and some specific services for 2002-03, together with the States' overall revenue raising effort ratios. The service provision ratios suggest that five
68
Hull and Searle
States provided services above the Australian average and three provided below average services. For the specific services included in the table: i) Victoria, Western Australia and the ACT provided above average levels of schools education, hospitals and housing, and below average levels of per capita roads expenses; ii) the Northern Territory provided services overall at above average levels but the specific services at below average levels17; iii) New South Wales and Queensland provided all but roads at below average levels; and iv) other States chose a mix of above and below average services. In terms of total revenue raising effort, the table shows that two States New South Wales and Victoria - operated at close to the Australian average effort, three States - Queensland, the ACT and the Northern Territory - made below average effort and the rest made above average efforts. Queensland alone was a low taxing, low service provision State in 2002-03. It is possible to compare the ratios across States as in Table 1, or over time for particular States. Attachment A summarizes the level of service provision ratios for the period 1977-78 to 2002-03 for all States for all services, Government Primary and Secondary Schools Education, and Hospitals services. The ratios for Housing and Roads are also included but cover only the later part of the period18. Government Primary and Secondary Schools Education, Hospitals, Housing and Roads account for about 40 per cent of State expenses. Attachment B includes the total revenue raising effort ratios on an annual basis for the complete period. In using these data, it is essential that their limitations be recognized. The ratios and what they tell us are only as good as the data used in their calculation. There can be errors in both the numerator -the actual fmancial data - and the denominator - the standardized figures arrived at by the Commission. As noted above, the financial data are compiled using a range of sources. The States are largely responsible for reporting the information and it may not always be filly consistent across States or over time. In estimating the standardized figures, the Commission uses a range of data of varying quality and reliability to quantify differences between the States in the non-policy influences on their costs. Generally, data measuring differences in the average use of services are more readily available and of better quality than data on differences in unit cost. The data can be as good as those available on student numbers to as poor as information available on what it costs to educate a non-English speaking student compared to an English speaking student. The Commission also uses its judgment to decide
The Impact of Equalization on Sewice Delivery
69
the range of influences it recognizes. It could miss an important influence or simply not be able to obtain the data to measure the impact of an influence on State costs. For example, the Commission has found it impossible to quantify the impact of different climatic conditions on State depreciation rates (and therefore expenses). Over time, the nature of services and the scope of transactions used to define a service have changed. There has also been a change in GFS from a cash to an accrual basis, commencing in 1998-99. However, these changes should have affected State actual expenses and Australian standard expenses in the same way and should not cause major breaks in the series. If State expenses are affected differently from the average, then discontinuities in the time series ratios for particular States will occur. The changes that the Commission makes to the way it calculates relativities in each reviewI9 can cause a noticeable break in the series. As well as changes to the methods used to calculate standardized revenues and expenses, the scope of the system has changed over time. As noted earlier, the Territories were not included in the equalization system until the 1985 and 1993 reviews. The years when a break in the series might be expected because of a review of assessment methods have been noted in the attachments.
2.1
Total Services
Figure 1 shows that service levels in the Australian States fall within a bandwidth - generally no more than 20 per cent above or below the average. The HFE system has provided States with fimding to enable this degree of consistency. However, there is no evidence of convergence2'. Figure 1 also shows a number of interesting trends. i) Queensland has been a low service level State for the entire period, although its service levels have been increasing since the early 1990s. In 1989-90, Queensland elected its first Labor government after twenty years. ii) Victorian service levels have generally been above average, particularly in the period 1984-85 to 1993-94. In 1993-94, they fell following the election of the Liberal (conservative) government in 1992. Since the return to Labor governments, they have been increasing again. iii) The ACT and the Northern Territory have tended to provide above standard services. These two Territories were administered directly by the Australian Government before they gained self-government and were included in the equalization process. This may have conditioned service
The Impact of Equalization on Service Delivery
71
expectations. The ACT experienced a dip in service levels in the early to mid 1990s, coinciding with the election of a Liberal government. The apparent fall in the Northern Territory level of service provision from 1997-98 may be due to data problems because of the change to accrual based GFS collections in that year because, at that stage, the Northern Territory Government had not adopted accrual budgeting and its accrual statistics were estimated. iv) The service levels of the other States have moved around the average.
2.2
Government Primary and Secondary Schools Education
As for total service provision, the level of service provision ratios for government schools education suggest that States tend to provide services within 20 per cent of the Australian average. Some States traditionally provide services below the Australian average and others provide services well above that average. New South Wales, Queensland and Western Australia have generally had ratios below the Australian average, although Queensland and Western Australia are closer to the average now than they were at the start of the period. South Australia has remained well above average for the period, as has Victoria, although its ratios have been trending closer to the average. This suggests that there are some States that value spending on education more than others. Again, the increase in education spending in Queensland came with a change in the party in government. Figure 2 illustrates the trends. For most States, Figure 2 suggests that relative service levels converged towards the average in the mid to late 1990s but diverged again more recently. However, comparing the average range in the early years with the range in the later years showed no evidence of a convergence in the ratios over the full period. An unexpected trend is the apparent fall in the level of government education services provided by the Northern Territory. It is possible that the observed fall to below 80 after 1998-99 is due to data anomalies (the change from cash to accrual GFS occurred in that year and the Northern Territory's actual expenses also showed a decline in that year). However, the downward trend in the Northern Territory's level of service provision ratios commenced in 1984-85, although it was not until the mid-1990s that service levels fell below the Australian average. This trend is of concern, given that twenty nine per cent of the population of the Northern Territory is Indigenous and Indigenous Australians have much lower educational outcomes than non-Indigenous ~ustralians~'.
Figure 2. Level of Service Provision Ratios(")- Government Schools Education
Source: Commonwealth Grants Commission, Reports, 198 1 to 2004. Notes: (a) A ratio of 100 indicates that the level of services is at the Australian average.
The Impact of Equalization on Service Delivery
73
The observed differences in service levels have been shown to be the result of differences in State policies. In 1994, the Commission published a paper22 which found that high service provision States (Victoria, SouthAustralia and the ACT) paid their teachers more, provided more teachers per student and had more students in the non-compulsory years of schooling (over 15 years). State government policies can be influenced by lobby groups. Nongovernment school groups in some States have used the Commission's data to argue that the s e ~ t o ?was ~ receiving a poor deal compared to the sector in other States. Lobby groups have also used Commission data to argue for increased funding in State budgets. New areas of funding, or increased funding, can be observed in State budgets in response to community demands, particularly where those groups can show that services are below Australian standards. The provision of government school services within 20 per cent of the Australian average and the convergence in the mid-1990s could be due to the influence of the Australian Government which, together with the States, is seeking to raise educational standards (increases in literacy and numeracy levels are a clear objective) and to ensure that the education services made available to students are more uniform. As a result, SPPs for school purposes have been substantial and have been conditional on States adopting the same starting age, reporting standards and testing procedures to evaluate literacy and numeracy.
2.3
Hospital Services
The picture painted by the Hospital services ratios (Figure 3) shows a greater range in the level of services provided in different States. Again there has been no convergence. In the early years, there was some movement towards the average, but that is not where service levels stayed. Figure 3 shows that, again, Queensland is the State providing below average levels of service for almost the entire period. At the start of the period, many States, but especially Western Australia and Tasmania, provided well above average services but these have been gradually reduced towards the average. In recent times, Victoria, Western Australia and the ACT have been providing well above the average level of services. The five other States are below the average. It is possible to speculate that the reductions that occurred in the early 1980s in most States came about because the Commission's first report in 1981 provided information on the States' relative spending levels compared to those of the other States. However, as the service levels reduced and the populations of some States did not support this policy, governments were forced to invest more in hospital services. That is certainly what has
The Impact of Equalization on Service Delivery
75
happened in Victoria in recent times. After the cutbacks (and efficiencies) in the 1980s and 90s, the Victorian population is now demanding a higher level of services in its hospitals. Funding has been increased to support this. In New South Wales, hospital service levels fell in the early 1990s. This was the result of Commission of Audit inquiries24that advised the State government that there was 'over servicing'25occurring in some areas. These inquiries used Commission data to observe levels of service provision above the Australian standard. They concluded that the higher costs were not due to better quality services (desired by the community) but to inefficiencies that needed to be removed. Similarly, a decline in service levels occurred in South Australia in the mid-1990s as a result of the findings of a Commission of ~ u d i t ~ ~ . Interestingly, service levels have increased again in recent times, suggesting that the people of South Australia did not agree with the Commission of Audit findings. Some State governments27 have told Commission staff that the Commission's data are used in formulation of their budgets and by other Government inquiries2' to check whether appropriate service levels are being delivered. Where a service appears to be above or below the Australian average level, departments and agencies are asked to justify the level in terms of community preferences. Although the largest tied grants in the Australian fiscal transfer system, the central government's Health Care Grants for hospitals do not appear to be causing any move towards uniformity in service standards. This may be because clear national outcomes have not been agreed. Conditional arrangements within the HCG program are at the margin, such as reporting requirements, or requirements to reduce waiting lists. Funding does not appear to be tied to improving health outcomes. Not surprisingly, the States and the Australian Government blame each other for poor performance of the hospital sector which they often attribute to cost shifting - as one level of government increases funding, the other moves it out.
2.4
Housing
Housing services were included in the equalization budget for the first time in the 1993 Review, but on a net cost basis. In the following review, the Commission decided to undertake separate assessments of expenses and revenue raising capacity associated with public housing services. It did this because it believed that the needs relating to each were not offsetting. As a result, the data set for housing gross expenses commences in 1991-92. Looking at Figure 4, the ratios suggest very large differences in levels of public housing services. Again, Queensland's services are below average and the ACT considerably above. Historically, the ACT and the Northern
The Impact of Equalization on Service Delivery
77
Territory had high levels of public housing dwellings per capita because the Australian Government provided them to attract public servants to live and work there. These levels have decreased over time, although for the ACT, they are still about twice the average. South Australian housing services are well above average because of the long-term effects of a public housing policy applied in the 1950s and 1960s to provide high levels of public housing to encourage a migrant workforce for the State's rapid industrialization. Services in most of the other States appear to be converging towards the average. The fall in the Northern Territory's service level in the period since 1998-1999 could be due to the misallocation of depreciation expenses.
2.5
Roads
The level of service provision ratios for roads shown in Figure 5 suggest widely fluctuating levels of service - from 60 per cent below to 60 per cent above average Australian standards. Again there is no convergence of service standards over the period. Anecdotally, travelers in Australia say that they can tell when they are crossing State borders, not because of the signs, but because of the quality of the roads. The widely fluctuating service levels can probably be understood in terms of how governments in Australia manage road maintenance and depreciation expenses. There are widely varying standards of roads constructed in the different States, and different maintenance regimes are applied. In part this is a response to traffic conditions, but States also make different decisions relating to the trade-off between construction standards and maintenance, and on what standards they consider appropriate for their different regions. Timing of roadworks programs is also flexible. As a result, actual road expenses fluctuate in most States from year to year, although the Commission's assessments of needs are relatively stable. Figure 5 is still surprising in that Queensland, a low service provision State and a State recognized for below standard roads, appears to be providing a high standard of service. Queensland argues that this is because the Commission fails to recognize all road lengths for which the Queensland government is responsible. The Commission is investigating this concern. The graph is also interesting in that the ACT and Victoria, with reputations for good quality road systems, have below Australian average service level ratios. These observations suggest that the Commission's assessments may not adequately be identifying State disabilities, or that road construction and maintenance cannot be assessed as simply as the Commission has done in the past. High levels of expenses in the past can give an appearance of high standards of roads that may be misleading, particularly if they are not adequately maintained thereafter (note Victoria's declining ratios).
Figure 5. Level of Service Provision Ratios (" -Roads
Source: Commonwealth Grants Commission, Reports, 1981 to 2004.
(a) A ratio of 100 indicates that the level of services is at the Australian average.
Notes:
The Impact of Equalization on Service Delivery
79
As with Housing, there is a possible misclassification of depreciation expenses for the Northern Territory from 1998-99.
2.6
Revenue Raising Effort
An examination of State revenue raising effort over the period (Figure 6) is revealing. It suggests that while convergence has not occurred in relation to State services, there has been convergence in terms of revenue raising effort. The range in the early period was significantly greater than in the later period. In fact, the large increase in effort in 1981-82 in Queensland, South Australia and Tasmania suggests that they chose to adjust their revenue efforts rather than service levels when their grants from the Commonwealth were to be reduced because of the introduction of equalization for the six Western Australia began to increase its revenue raising effort earlier, but after the Commission had been asked to consider the needs of all States and possibly because it recognized that its service provision levels were high, a fact substantiated by the Commission's inquiry. Figure 6 and Figure 1 together suggest that States' revenue raising efforts tend to be above average if States are providing above average services and below average if the State is supporting below average services. Queensland is the classic example. Its revenue raising efforts were well below average in the earlier years - less than 60 per cent of the Australian average, rising to 90 per cent of the Australian average as its service provision levels have crept up. Similarly, South Australia's revenue raising efforts have increased from below average to above average to support its increased service standards. New South Wales and Victoria have maintained revenue raising efforts generally slightly above the Australian average to support above average services. Figure 7 compares the revenue raising efforts and service provision levels of Queensland, South Australia, and Victoria. It suggests that, under comprehensive W E , the choice of overall service standards must be supported by a State's willingness or otherwise to make a revenue raising effort. Figure 7 shows that States' revenue raising efforts and their levels of service provision track well, although that was not the case in the years prior to the first Commission review (1981). It has largely been the case since, although there have been some divergences. Where a State chooses a lower revenue raising effort than its service levels, it needs to borrow above average amounts to support the difference. Similarly, where a State imposes a greater revenue effort than its service level requires, it can repay an above average amount of borrowing or operate at a larger than average surplus.
Figure 6. Revenue Raising Effort (" - Total Revenue
Source: Commonwealth Grants Commission, Reports, 1981 to 2004. Notes: (a) A ratio of 100 indicates that the level of services is at the Australian average.
81
The Impact of Equalization on Service Delivery
F i ~ u r e7. Comparison of Revenue Raising Efforts and Service Levels, Three
Victoria 140.00 120.00 -80.00 60.00
-Revenue Raising Effort - - - - - - Level of Service Provision ~
s
~
,
~
,
,
,
~
~
~
~
~
~
~
~
~
Queensland 120.00 100.00
-
Revenue Raising Effort
-;
- - - - - - Level of Service Provision
South Australia 140.00
----~-
60.00 40.00 m
t
-
o
m
w
t - o m t - m m m m m m m m m m m m m
m
m
o m
* o
~
Revenue Raisin Effort Level of Servicl Provision
~
~
1
~
*
Hull and Searle
CONCLUSION We note that: i) HFE in Australia has not resulted in all States providing the same level of services, but that it is not designed to do so because it is a capacity equalization system. ii) On average, and over time, service levels in each State are within 20 per cent of Australian average levels. iii) Service levels for particular hnctions can vary widely, reflecting State government policies. These policies in turn reflect community choices relating to service requirements and revenue raising preferences. They are expressed in the types of governments that are elected and through lobby groups and other community pressure. iv) Service levels have not converged since the introduction of comprehensive equalization, but revenue efforts have. v) Commission data have been used to inform decision makers about service delivery. It has been used to ask questions about the effectiveness and efficiency of service delivery in particular States. Levels have changed as a result. There are many influences on service delivery levels in Australia. Australia is a federation and it has an HFE system that provides certainty that all States have the capacity to provide the same standard of service. The federal system and the untied nature of the grants provided also means that State governments can choose not to do so. They can choose instead to satisfy the preferences of the people who live in their jurisdiction. These preferences are expressed at elections through changes in government and through lobby groups that tell govements about community expectations. The HFE system is not designed to achieve uniformity but to give flexibility - to deliver equality in diversity3'. We see that in the service levels and revenue efforts observed in this paper. There is no reason to expect that State service levels will converge under capacity equalization. However, over time, service levels appear to move within a band (so much above or below the average). This is likely to be the result of the State and Australian Governments (through tied grants) seeking to achieve similar goals or objectives in service provision. Australian residents in any State do not want 'to miss out on something others have'. This seems to place restrictions on, or boundaries about what States can do differently from the average. Other State government policies, such as those relating to revenue raising effort, also have an influence on service levels. Higher service levels must
The Impact of Equalization on Service Delivery
83
be supported through higher revenue efforts. This puts a ceiling on service levels. In addition, the information generated as a by-product of the HFE system, despite its imperfections, allows governments and other groups to ask questions about why standardized per capita expenses on services differ between States. If they are higher for reasons that relate to the preferences of the population, this is good government decision making. If they are the result of inefficiency or government inertia, they are changed through the electoral or other political processes. All the policy instruments used by the Australian Government and the States play their role in helping to achieve appropriate service standards for Australian residents, regardless of where they live. Both equality and diversity (choice) are important goals in Australian culture and our systems work to achieve these goals. The fiscal transfer system and the HFE objective within that system are key components of Australian public finances.
Annex Tables Annex Table A-I. Service Provision Level Ratios (a), Australian States, 1978-88 to 2002-03 - All Services Year
NSW
Vic
1977-78 1978-79 1979-80 1980-81 1981-82 1982-83 1983-84 1984-85 1985-86 1986-87 1987-88 1988-89 1989-90 1990-91 1991-92 1992-93 1993-94 1994-95 1995-96 1996-97 1997-98 1998-99
99.93 101.70 101.70 102.80 104.15 99.30 98.82 98.75 98.83 100.11 97.58 98.10 96.56 99.54 100.17 96.82 94.84 99.62 100.78 98.12 102.14 98.62
101.95 102.98 102.23 102.12 102.80 108.38 107.02 98.94 108.98 110.28 114.26 114.20 114.16 111.13 110.50 114.50 115.18 105.16 100.82 99.65 96.13 98.49
Qld 87.33 85.78 85.33 85.10 84.17 85.51 85.70 85.82 84.92 8 1A2 80.17 78.11 81.48 82.79 85.36 85.98 85.02 87.69 90.55 95.76 93.49 95.35
WA 107.07 104.53 106.90 105.35 96.19 96.54 99.36 97.37 98.31 97.54 101.82 104.09 103.32 97.04 93.47 93.07 99.33 98.34 99.67 103.93 103.71 101.77
SA 104.93 102.15 101.80 101.46 101.50 103.82 104.01 102.03 102.90 104.57 105.88 103.89 103.66 105.09 106.54 105.99 108.76 109.82 110.69 109.79 111.93 107.13
Tas 101.93 103.02 106.33 102.06 107.47 102.32 103.12 102.27 101.77 98.26 97.08 99.77 104.81 103.12 103.13 105.20 97.42 98.52 99.88 104.79 98.09 101.60
ACT na na na na na na na na na
na 105.76 112.28 111.27 118.81 104.51 112.17 103.77 107.38 103.23 104.61 101.83 127.91
NT na na na na 100.84 97.93 98.64 103.43 99.42 99.3 1 102.66 102.64 101.39 100.38 100.12 99.70 107.95 118.12 115.41 102.99 101.66 118.77
CGC Report 1981 Review 1982 Review
1985 Review
1988 Review
1993 Review
1998 Update 1999 Review
2004 Review
-
Annex Table A-1. Service Provision Level Ratios (a), Australian States, 1978-88 to 2002-03 All Services (Continued) Year 1999-2000 2000-0 1 2001-02 2002-03
NSW 98.93 95.51 95.96 94.73
Vic 99.26 100.23 98.72 105.33
Qld 91.03 97.92 102.73 94.01
WA 104.19 102.82 100.75 107.75
SA 112.17 109.75 107.16 106.70
Tas 104.10 102.49 102.90 97.89
ACT 119.24 126.89 119.13 119.10
NT 116.23 112.57 105.30 105.05
CGC Report
Source: Commonwealth Grants Commission, Reports, 1981 to 2004. Notes: a) A ratio of 100 indicates that the cost of providing services is at the Australian average. b) na indicates data not available.
Annex Table A-2. Service Provision Level Ratios (a),Australian States, 1978-88 to 2002-03 - Government Primary and Secondary Schools Education Year 1977-78 1978-79 1979-80 1980-81 1981-82 1982-83 1983-84 1984-85
NSW 94.78 96.96 98.22 100.13 103.00 98.56 97.93 95.09
Vic 115.58 113.02 112.51 108.51 111.24 118.76 117.53 117.62
Qld 84.41 8 1.62 79.59 80.89 78.16 76.37 77.78 78.61
WA 86.78 87.99 89.42 88.93 85.58 85.89 87.43 88.93
SA 108.41 111.21 109.79 113.66 101.41 102.15 103.87 105.91
Tas 111.08 115.59 117.86 115.36 116.12 107.39 110.35 109.47
ACT na na na na na na na na
NT na na na na 100.00 95.69 97.63 116.92
CGC Report 1981 Review 1982 Review 1985 Review 1988 Review
Annex Table A-2. Service Provision Level Ratios (a), Australian States, 1978-88 to 2002-03 - Government Primary and Secondary Schools Education (Continued) Year 1985-86
NSW 96.26
Vic 117.27
Qld 79.44
WA 89.58
SA 107.17
Tas 106.58
Source: Commonwealth Grants Commission, Reports, 1981 to 2004. Notes: a) A ratio of 100 indicates that the cost of providing services is at the Australian average. b) na indicates data not available.
ACT na
NT 112.85 112.12 112.78 114.22 113.70 109.81 108.29 103.97 99.58 104.94 95.20 93.60 92.61 78.20 75.03 80.22 73.34
CGC Report
1993 Review
1998 Update 1999 Review
2004 Review
Annex Table A-3. Service Provision Level Ratios (a), Australian States, 1978-88 to 2002-03 - Hospitals Year
NSW
Vic
Qld
WA
SA
Tas
ACT
NT na na na na 100.00 89.54 92.68 121.11 117.42 109.05 134.30 128.06 135.12 135.51 136.68 134.57 106.26 100.83 106.25 82.93 96.28 91.52
CGC Report 1981 Review 1982 Review
1985 Review 1988 Review 1993 Review
1998 Update 1999 Review
2004 Review
Annex Table A-3. Service Provision Level Ratios (a), Australian States, 1978-88 to 2002-03 - Hospitals(Continued) Year 2000-0 1
NSW 92.04
Vic 125.05
Qld 79.44
WA 117.41
SA 92.60
Tas 83.75
ACT 117.97
NT 71.83
CGC Report
Source: Commonwealth Grants Commission, Reports, 198 1 to 2004. Notes: a) A ratio of 100 indicates that the cost of providing services is at the Australian average. b) na indicates data not available.
Annex Table A-4. Service Provision Level Ratios (a), Australian States, 1978-88 to 2002-03 - Housing Year 1991-92
NSW 93.79
Vic 62.71
Qld 80.74
WA 115.26
SA 148.26
Tas 119.86
ACT 457.57
NT 239.25 201.53 275.11 301.50 324.90 227.10 283.86 80.42 55.70 56.24
CGC Report
1998 Update 1999 Review
2004 Review
Annex Table A-4. Service Provision Level Ratios (", Australian States, 1978-88 to 2002-03 -Housing (Continued) Year 2001-02 2002-03
NSW 95.18 93.45
Vic 106.38 114.72
Qld 73.60 69.74
WA 96.89 101.67
SA 174.97 167.55
Tas 100.21 94.06
ACT 203.20 196.48
NT 55.48 54.90
CGC Report
Source: Commonwealth Grants Commission, Reports, 1981 to 2004. Notes: a) A ratio of 100 indicates that the cost of providing services is at the Australian average. b) na indicates data not available.
Annex Table A-5. Service Provision Level Ratios (a),Australian States, 1978-88 to 2002-03 - Roads Year 1987-88
NSW 72.88
Vic 151.58
Qld 111.20
WA 66.70
SA 92.84
Tas 88.19
ACT 52.98
NT 132.04
CGC Report 1993 Review
1998 Update 1999 Review
Annex Table A-5. Service Provision Level Ratios (", Australian States, 1978-88 to 2002-03 - Roads (Continued) Year 1998-99
NSW 102.26
Vic 61.1 1
Qld 165.78
WA 67.34
SA 79.64
Tas 99.68
ACT 159.66
NT 75.78
CGC Report 2004 Review
Source: Commonwealth Grants Commission, Reports, 1981 to 2004. Notes: a) A ratio of 100 indicates that the cost of providing services is at the Australian average. b) na indicates data not available.
Annex Table & I . Revenue Raising Effort ~atios'~), Australian States, 1978-88 to 2002-03 - Total State Own Source
Revenue Year 1977-78 1978-79 1979-80 1980-81 1981-82 1982-83 1983-84 1984-85
NSW 95.55 102.25 103.92 106.51 102.78 104.49 101.04 102.77
Vic 102.08 99.12 95.76 91.10 106.49 108.42 110.84 112.65
Qld 56.83 54.88 56.44 61.94 86.17 86.03 80.39 84.38
WA 82.61 73.67 74.06 82.89 91.41 97.81 102.93 101.39
SA 66.09 64.3 1 58.67 52.09 98.84 9 1.49 94.78 99.03
Tas 53.14 54.89 54.84 51.82 99.91 90.04 89.41 89.35
ACT na na na na na na na na
NT na na na na 89.20 89.53 83.26 80.55
CGC Report 1981 Review 1982 Review 1985 Review 1988 Review
A n n a Table B-I. Revenue Raising Effort ~ a t i o s ~Australian ), States, 1978-88 to 2002-03 - Total State Own Source Revenue (Continued) Year 1985-86
NSW 104.42
Vic 109.17
Qld 86.29
WA 101.16
SA 104.48
Tas 90.17
Source: Commonwealth Grants Commission, Reports, 1981 to 2004. Notes: a) A ratio of 100 indicates that the cost of providing services is at the Australian average. b) na indicates data not available.
ACT na
NT 96.26 86.13 70.84 76.54 86.19 78.94 86.50 98.24 99.03 120.69 131.23 111.22 107.87 77.10 86.88 84.42 84.36
CGC Report
1993 Review
1998 Update 1999 Review
2004 Review
Hull and Searle
92
Notes Catherine Hull is an Assistant Secretary at the Commonwealth Grants Commission. Bob Searle is a former Secretary of the Commission. The views expressed in this paper are those of the authors and not the Commission. We would like to thank Rosa Benedictos and Linda Pure for help with the data analysis. The Australian Federation comprises six States - New South Wales, Victoria, Queensland, Western Australia, South Australia and Tasmania - and two Territories the Australian Capital Tenitory (ACT) and the Northern Territory. All participate in the equalization system. In this paper, the word State(s) should be read to include the ACT and the Northern Territory unless the context indicates otherwise. Professor L F Giblin, a member of the first Grants Commission, discussed the concept in a 1925 Tasmanian submission prepared for the proposed Royal Commission on Tasmanian Finances and in an appendix to the 1930 Tasmanian Submission to the Australian Parliament's Joint Committee of Public Accounts. The standardized revenue of a State is what it would raise if it applied the Australian average revenue imposts to its revenue bases. Its standardized expenses are what it would have to spend to provide the Australian average level of services, taking into account differences between the average experience and the use and unit costs of services it experiences. This comprises the GST revenue and most of the Health Care Grants paid to the States by the Australian Government. In US dollars, this equates to $29.4 billion. One Australian dollar is worth about 75 cents US. The $29.4b would be the equivalent of a grant scheme of about $US350 trillion in the US context. Australian Government (2004), p 10. Commonwealth Grants Commission (2004a), pp4-5, available at www.cgc.gov.au. Commonwealth Grants Commission (2004a), p4. Boadway (200 1). Vaillancourt and Bird (2006), this volume. Commonwealth Grants Commission (1995). Because they contribute to the overall standards of services provided, expense standards include any funding provided by the Australian Government through a tied grant. Disabilities or needs are defined as differences in the economic, demographic and physical circumstances of States that lead to differences in their relative costs of providing services and their relative revenue raising capacity. They are beyond the control of an individual State government. Commonwealth Grants Commission (2004b), Attachment B: The Distributional Model - A Mathematical Representation, pp341-350. Commonwealth Grants Commission (2004a), Chapter 2, p4. According to Commission figures, the ~ o r t h e h~er&toryis spending well above average amounts on providing preventive health services, higher education, services to industry, general public services, depreciation and debt charges. These services were included in the equalization budget for the first time in 1993. Comparable data are available from 1987-88 for Roads but only from 1991-92 for Housing. Prior to 1991-92, Housing needs were calculated on a net basis and standardized gross expenses are not available. The Commission first calculated State relativities in 1981 and undertook reviews of its methods in 1982, 1985, 1988, 1993, 1999 and 2004. A system of five yearly reviews was adopted in 1988 and the 1999 Review was a 'one off delay to better align reviews with Census data availability.
T7ze Impact of Equalization on Service Delivery
93
Statistical testing for differences in the average ranges for different early and later periods suggested no significant difference at the 5 per cent level. Commonwealth Grants Commission (200 I), p 195. Commonwealth Grants Commission (1994). The Commission also published similar papers that identified policy differences affecting Urban Transit and Police services. Catholic Education Office, Archdiocese of Canberra and Goulburn. New South Wales Commission of Audit (1988). , , The Grants Commission never uses its data to talk of over servicing, but the levels of service provision ratios can be used to compare service levels with the Australian average level. South Australian Commission of Audit (1994) Personal communications between State Treasuries and Commission staff. For example, the Forde Inquiry into the Abuse of Children in Queensland Institutions (1999). The Commission was asked to calculate relativities for the States in 1978 and brought down its first report in 1981. Commonwealth Grants Commission (1995).
References Australian Government. 2004. Federal financial relations 2004-05, Budget Paper No 3. Canberra: CanPrint Communications Pty Limited. Boadway, R. 2001. Inter-governmental fiscal relations: The facilitator of fiscal decentralization. Constitutional Political Economy, 12. Commonwealth Grants Commission. 1994. Information paper no I , Education services, analysis and discussion of state policies, 1994. Canberra: Australia Government Publishing Service. Commonwealth Grants Commission. 1995. Equality in diversity: History of the commonwealth grants commission, Second Edition. Canberra: Australian Government Printer. Commonwealth Grants Commission. 200 1. Report on indigenous funding, 2001. Canberra: CanPrint Communications Pty Limited. Commonwealth Grants Commission. 2004a. Report on state revenue sharing relativities, 2004 Review. Canberra: CanPrint Communications Pty Limited. Commonwealth Grants Commission. 2004b. Report on state revenue sharing relativities, 2004 Review supporting information. Canberra: CanPrint Communications Pty Limited. Forde, M. 1999. Inquiy into the abuse of children in Queensland institutions. Brisbane: Queensland Government Printer. New South Wales Commission of Audit. 1988. Focus on reform, report on the state's finances. Sydney: Government Printer New South Wales. South Australian Commission of Audit. 1994. Charting the way forward, improving public sectorperjormance. Adelaide: Gillingham Printers Pty Ltd. Vaillancourt, F. and Bird, R 2006. Expenditure-based equalization transfers, this Volume.
Chapter 5
HARMONIZING OBJECTIVES AND OUTCOMES AT THE NATIONAL AND SUBNATIONAL LEVELS THROUGH CITIZEN ENGAGEMENT AND CAPACITY BUILDING (WITH SPECIAL REFERENCES TO THE PHILIPPINES) ALEX B. BRILLANTES, Jr. AND JOSE 0.TIU SONCO 11' National College of Public Administration and Governance, University of the Philippines
1.
INTRODUCTION
In general terms, this paper explores the idea that good governance is the key to harmonizing the objectives and outcomes of intergovernmental transfers. In this context, good governance is the active participation and engagement of citizens in the planning and budgetary processes at the national, but mostly at the local levels, together with an accurate determination of the capacities of such levels of government. The paper has eight sections. The first discusses the context within which intergovernmental transfers occur. The second discusses the framework of the paper and tries to look at the bigger picture, suggesting that the system of intergovernmental transfers occurs within the policy framework of decentralization in general and fiscal decentralization in particular. It also suggests that the principles of good governance impact on the implementation of intergovernmental transfers. For instance, the predictability of transfers to subnational levels of government enables rational and realistic planning. The third section discusses the rationale, objectives and outcomes of intergovernmental transfers. It suggests that human development outcomes - operationalized through the delivery of appropriate basic services - have to be "harmonized" with the objectives of the transfers which are essentially to provide adequate financial resources to fund public service delivery at subnational government levels.
96
Brillantes and Tiu Sonco II
Section four discusses the general concepts of intergovernmental fiscal transfers, their rationale, objectives and links to development outcomes. Section five discusses the objectives, goals and outcomes of fiscal transfers as well as some issues and concerns. It presents the intergovernmental fiscal transfers experience of the Philippines and relates the amount of transfers to local revenue-raising capacities and human development indicators. Section six discusses the engagement of civil society and citizen groups in planning and public finance at the local level, an innovation in governance that improves both fiscal management and public service delivery. Section seven highlights the importance of capacity building not only for local government officials and staff but also national agencies in operationalizing decentralization, thus contributing to the harmonization of objectives and outcomes. Section eight provides some summary notes and indicative action points leading to the Harmonizing of Objectives and Outcomes at the National and Subnational Levels through Citizen Engagement and Capacity Building. The paper provides examples of countries where citizen participation and capacity building may be contributing to harmonizing objectives of fiscal transfers and human development outcomes. Referring to a well designed system of intergovernmental transfers, Bahl, Boex and Martinez-Vazquez (2001) point out the critical importance of matching expenditure needs and revenues at the subnational or local government level. This information is then critical in determining the amount of the transfers from the national level. According to them, "in order to know how much transfer is necessary, one must estimate the difference between the revenues available to subnational governments as a whole, and the expenditure needs of each level of government" (2001:3). The expenditure needs at each level can only be determined if the responsibilities and assigned h c t i o n s at the subnational levels are clearly defined and recognized. This is very much in accordance to the "finance follows function" principle advocated as Rule Two in Bahl's Rules of Fiscal ~ecentralization'. In deciding the appropriate functions that are to be delivered by the subnational levels of government, it is imperative that the citizens' demands and needs be correctly assessed and determined within the context of responsive local governance. This is where active people participation and citizen engagement plays a key role, but it must be calibrated with the capacities of the subnational levels of government to perform the functions. The paper does not attempt to romanticize participation. It is recognized that there are certain goods and services ("merit goods" according to Bahl et a1 2001) such as education, which will be underprovided if the consumption decision is left to the consumers (or lower level governments) because the consumer does not recognize the true value or benefit of that good. Indeed,
Harmonizing Objectives and Outcomes
97
there are cases where the central government is in a better position to determine the socially optimal level for this good than either individual citizens or local governments (Bahl et al, 2001, p4). This however does not obviate the need to seriously consider the participation of citizens, this time at higher levels of government where collective demands upon the politicoadministrative system may be better aggregated and articulated, and hence factored into the design of the intergovernmental transfer system. It must be stated though that other equally important factors such as the imperative to improve both vertical and horizontal fiscal balances should continue to be included in the design of intergovernmental fiscal transfer system.
2.
INTERGOVERNMENTALTRANSFERS: FRAMEWORK FOR HARMONIZING OBJECTIVES AND OUTCOMES
Intergovernmental transfers have been at the center of debates in many countries that have adopted decentralization as a strategy for development and good governance. They have been adopted as mechanisms to concurrently fund national development efforts and public service delivery at the local government level. This is a key feature of decentralized governments that can be seen as a response to the overall framework of decentralization. Financial resources transferred from national to subnational levels of government have been the major sources of revenues for many local governments in developing countries (Litvack 2003; World Bank 2003). Furthermore, appropriate local government finance is imperative if the policy goals and outcomes of decentralization are to be achieved. Intergovernmental fiscal transfers refer to the transfer of money from central governments to lower level, subnational or local government units (states, provinces, cities/municipalities, etc.). The term 'transfers' is often used to refer to a number of public financing instruments between central and subnational or local governments and includes, among others, sharing of tax revenues, intergovernmental grants, subsidies, and subventions. (Bahl, Boex and Martinez-Vazquez, 2001; Bird and Smart 200 1) Among the elements of good governance are accountability, transparency, participation and predictability. Figure 1 presents a schematic diagram that shows the harmonization of these objectives with outcomes of intergovernmental transfers from national to subnational governments. Decentralization provides the policy framework at the national level. More specifically, fiscal decentralization from the central to subnational
98
Brillantes and Tiu Sonco II
governments provides the policy context for the operationalization of intergovernmental fiscal transfers. Such transfers attempt to vertically and horizontally balance fiscal needs and resources between governments. Vertical fiscal imbalance is the disparity between revenue sources and expenditure needs at the central and subnational government levels. It occurs when the expenditure responsibilities of subnational governments do not match their revenue-raising powers. Meanwhile, horizontal imbalance occurs when own fiscal capacities to carry out the same functions differ between jurisdictions within a level of subnational government. The capacity of local government units (LGUs) to deliver public goods and services depends on the capacity of their staff and elected officials, and their fiscal soundness. Local development and services are prioritized through local development planning, budgeting and implementation. This requires better capacities of local governments - both staff and elected officials - to ensure that plans are prepared in accordance with the national goals as well as the needs and demands of the local people. Implementation, monitoring, and review of local plans require capabilities of local leaders and staff to ascertain the efficiency, effectiveness and responsiveness of policies and programs. The fiscal capacity of subnational governments to generate revenue, through their own-source revenue sources (tax and nontax) and transfers, enables them to deliver basic services as well as pursue local development efforts. Figure one also suggests that citizens can be engaged in the process, starting with the design of intergovernmental transfers. It has been noted that decentralization is a process that enables citizens to participate in the process of governance. It is based on the "belief that participation of a variety of key local stakeholders, including subnational governments and local communities, is critical to realize development and poverty reduction goals. There has also been a growing recognition of the need for broader partnerships among the various levels of the public sector, the private sector, civil society and international development agencies in promoting more equitable and sustainable development" (ADB, 2003; TA 5902). It is within this context that we suggest that intergovernmental transfers would be more predictable and transparent if the citizens, as well as the various subnational governments and other stakeholders, were involved in the design of the intergovernmental transfer system and the overall framework of public service delivery. Different subnational governments have different requirements and objectives for intergovernmental transfers. Hence, their participation in the effective design of intergovernmental transfers is important. It should also be mentioned, however, that there is a school of thought (with which we agree) that suggests citizen participation in subnational planning and budgeting results in better local governance.
Participation IGT Objectives
Framework
tT
Decentralization
I/
Fiscal Decentralization
I
Transparency
/
Intergovernmental Fiscal Transfers Vertical Equalization Horizontal Equalization Minimize interjurisdictional spillovers (externalities) Correct major administrative weakness and streamline bureaucracy
I Capacity Building /
Figure I. Harmonizing Objectives and Outcomes Framework
Outcomes
Goals and
Equitable growth Regional and National Development Improved living conditions Poverty reduction Contribution to the Millennium Development Goals
I
Participation
/
100
Brillantes and Tiu Sonco II
Citizen participation improves prioritization of policies; promotes transparency and accountability in the use of limited financial resources; improves service delivery; and responds to local needs. As such, subnational governments have a better opportunity not only to achieve local goals but also to contribute to national outcomes. Outcomes, after all, are the ultimate goals of decentralization and national policies. These outcomes include improved service delivery, equitable growth, improvement in socio-economic conditions, better quality of life of local residents, contribution to regional and national development, poverty reduction and the attainment of the Millennium Development Goals (MDGs).
3.
DECENTRALIZATION: THE CONTEXT FOR FISCAL DECENTRALIZATION AND INTERGOVERNMENTALTRANSFERS
This section draws on "Decentralization Imperatives: Lessons from Some Asian Countries" by Brillantes 2004. It discusses the broad context within which intergovernmental transfers occur and suggests that decentralization involves the transfer of functions, powers, authorities and responsibilities from national to local levels, as well as fiscal transfers. Depending on one's paradigms and motivation, decentralization can have different meanings and dimensions. It is a management approach that is claimed to deliver public services more efficiently by improving both allocative and productive efficiencg (Kakhonen 2001). It decongests top management, hastens decision-making processes and minimizes unnecessary delays and red-tape. It can be a means of "load shedding" wherein central authorities transfer functions and responsibilities to subnational institutions because of the inability or unwillingness of the former to continue funding or providing such functions. It can broaden the reach of national government and enable the "penetration" (Cheema and Rondinelli 1983:15) of national government policies into the remote rural areas of the polity. It can be a means of recognizing the special status of certain regions that differ markedly from the rest of the nation due to ethnic composition and makeup, or the availability of natural resources (Bahl, 2002). Decentralization suggests democratization by broadening the base of participation and providing a voice to marginalized and non-mainstream sectors of society, such as cultural and ethnic minorities. It contributes to operationalizing democracy at the local level by providing citizens with access to structures and processes of governance. Decentralization can mean building the capacities of subnational institutions to enable them to respond
Harmonizing Objectives and Outcomes
101
to local needs: it can lead to more autonomous local authorities that are less dependent on central institutions. Decentralization can also mean more innovations and flexibility at the local level: it allows local governments to design and implement programs customized to the unique needs of their locality. It encourages creativeness and provides the opportunity to depart from standard and formula-based (one size fits all) approaches to development challenges. Indeed, decentralization may mean any or all the above and more.3 Overall, "decentralization entails the transfer of functions, powers, responsibilities and accountability to lower level institutions for better governance4. This definition is consistent with the classic definition developed by Cheema and Rondinelli that has become some kind of "industry standards. According to them, decentralization is the "transfer of planning, decision-making or administrative authority from the central government to its field organizations, local administrative units, semiautonomous and parastatal organizations, local governments or nongovernmental organizations" (1983:p18). Our defmition suggests that given the recognition in contemporary development analysis of the imperative to reduce poverty, and given that the lack of effective governance has been identified as the "missing link" in failed poverty reduction efforts the discourse should illustrate that decentralization, when correctly implemented and with an appropriate policy and capacity mix at the national and subnational levels, has the potential to be a very powerfbl tool to effect good governance.6
Forms of Decentralization7 Building on various theoretical constructs in local government and decentralization literature, this paper suggests that decentralization may be classified into: Deconcentration; Devolution; and Debureaucratization
Deconcentration may also be referred to as "administrative decentralization." It entails the transfer of functions fi-om central authorities to lower level administrative institutions such as field offices. It embraces the notion of "delegation" since central authorities decide and identify what functions can be "delegated" to subnational institutions. Such institutions, however, still retain their "national character" since they are simply authorized by their principals at the central government to act administratively on routine matters that need not go to the center and clog
102
Brillantes and Tiu Sonco N
operations there. It also includes transfer of authorities over some fiscal matters, such as a determination of the value of transactions for which the field offices can make decisions without having to obtain authority from the central office. It is referred to as "deconcentration7'because it lessens the concentration of the load at the center. The extent of delegated authority is determined and may be revoked by the center. A distinguishing feature of deconcentration is that final authority still rests with the center. It is within this context that deconcentration has been referred to as "pseudo-devolution." It is therefore important to consider administrative decentralization within the context of the overall decentralization scheme. To a certain extent, it represents a weak form of decentralization. According to Dalton (2003:9) "administrative decentralization often distracts attention from building towards devolution and in some cases misses the point entirely. " Dalton also suggests that we should not lose sight of the ultimate objective of decentralization - to move towards good governance based on local democracy.
Devolution is also referred to as "political decentralization." This entails the transfer of powers to lower level political institutions - specifically local governments. Local governments participate in a political nature when they: a) have elected officials, e.g., elected local chief executive such as the mayor and/or the local legislative body; b) have jurisdiction over a specifically defined geographical area; c ) have clear responsibility for the performance and delivery (and financing) of specified basic services and are held accountable for such; and; d) have the power to generate revenues and levy taxes. * Local governments are thus clothed with a degree of autonomy that enables them to decide on local matters without interference by the center. The imposition of taxes is authorized by the local elected authority, not the central government.9 Debureaucratization ("getting out of the bureaucracy") refers to the process of transferring public functions, powers and authorities to the private sector, business organizations, voluntary and non-governmental organizations, people's organizations and to civil society in general. It is essentially enabling non-government and extra-governmental structures to deliver services and perform functions that traditionally belonged to government. Debureaucratization is recognition of the limitations of government's ability to deliver some services due to such things as lack of resources, graft and corruption. More importantly, it recognizes that there
103
Harmonizing Objectives and Outcomes
are some services and functions that may be more efficiently delivered by the private sector. There can be various modalities of debureaucratization. These can include government-non-government partnerships, joint ventures, cofinancing between government and the private sector, contracting out and privatization. As suggested earlier, the relationship between intergovernmental transfers and decentralization may be explained within the context of "fiscal decentralization." One of the primary concerns of intergovernmental transfers is to be able to achieve the desired results and outcomes based on objectives set by national and subnational governments. It concerns the overall impact of local development initiatives through local governments, with the objectives of equitable growth, poverty reduction and better human development outcomes. This necessitates a good fiscal transfer system that equalizes the vertical and horizontal imbalances. In this way, fiscal transfers are able to effect change and produce better outcomes at the subnational government level. Figure Two represents these various modalities of decentralization.
CentraVNational Government
Deconcentration
Field Offices
I
Debureaucratization
Local Government
Private Sector / NGOs Civil Society
Figure 2. Types of Decentralization
104
Brillantes and Tiu Sonco II
Box I. Experience of Intergovernmental Transfers in some Asian Countries
Cambodia heavily targets resources to poorer provinces through its revenue-sharing formula but the formula is being phased-in as part of a decentralization process thal is beginning with some of the better-endowed areas. Pakistan's main provincial revenue-sharing formula (population-based) that provides special allocations to backward provinces has led to higher per capita expenditures in those provinces, but they still have enormous backlogs in access and higher costs of service provision. In the Philippines, the Internal Revenue Allotment is found to substantially increase the aggregate resources at the local level and possibly has a modest effect on development status, but it worsens the horizontal fiscal imbalance across the various subnational levels of government. India and Indonesia make more sophisticated attempts to deal with fiscal imbalances by considering expenditure needs relative to revenue capacity, but imbalances remain significant, partly because of deficiencies in the design and management of the transfer programs and partly due to the offsetting effects of other activities.
Source: ADB, Intergovernmental Fiscal Transfers in Asia.
4.
INTERGOVERNMENTAL FISCAL TRANSFERS: RATIONALE, OBJECTIVES AND OUTCOMES
There are three principal reasons why intergovernmental fiscal transfers are important in decentralized nations - whether they are industrialized or developing countries (Bahl, Boex and Martinez-Vazquez, 2001; ADB, 2003). First, central governments have advantages over subnational governments in raising revenue from many particularly productive sources, while subnational governments have advantages in providing many public services. Second, there are often substantial disparities in revenue-raising capacity across subnational levels of government. Some subnational governments
Harmonizing Objectives and Outcomes
105
are inherently more capable of raising revenues than others because of their jurisdictional characteristics, e.g., natural resources, population, tax base, geographical location, availability of infrastructure, and quality of political leadership and attitude of people. Some can raise sufficient revenue, while others cannot. Third, resources from the central government can be used to ensure that national priorities will be met in all subnational government jurisdictions. As discussed above, devolution transfers responsibility for the delivery of services from the national to the subnational level of government to ensure that they are delivered. There remains a need, however, for the national government to be able to see that national objectives are met. Subnational governments' fiscal condition concerns the available local revenue resources and their expenditure requirements in service delivery, public infrastructure development, promotion of economic sustainability, and contribution to national development, economic growth and stability. More often than not, there is an imbalance between the expenditure requirements and the revenue-raising capacity of subnational governments or a vertical fiscal imbalance. Revenue and expenditure assignments in a nation give rise to both vertical and horizontal fiscal imbalances within the nation's government structure (Yilmaz and Bindebir, 2003). The fiscal gaps between revenue and expenditure and the disparities among subnational governments in their capacities to earn enough revenue must therefore be filled by either giving them more revenue raising powers or by transfers kom the central government. The latter is better because subnational governments have different characteristics that can be more easily overcome through a transfer system. Intergovernmental fiscal transfers thus have four principal objectives: 1. 2. 3. 4.
to equalize vertically (improve revenue adequacy); to equalize horizontally (interjurisdictional redistribution); to minimize interjurisdictional spillovers (externalities); and to correct major administrative weaknesses and streamline bureaucracy
Table 1 presents these four objectives, together with indicators of needs and the key dimensions in intergovernmental fiscal transfers. Table 2 presents the links between the rationales, objectives and outcomes of intergovernmental fiscal transfers.
Brillantes and Tiu Sonco II
106
Table I. Objectives and Key Dimensions o f Intergovernmental Fiscal Transfers Objectives
Fiscal Gap, Fiscal Need and Fiscal Performance Areas
Key Dimensions
To equalize vertically (improve revenue adequacy)
Vertical imbalance refers to insufficiency of local revenue resources to meet the costs of providing public services Horizontal imbalance relates to the disparities across subnational governments' fiscal capacities to raise revenue from local sources
Fiscal gap between expenditure and capacity to raise revenue from local sources
To equalize horizontally (interjurisdictiona 1redistribution)
To minimize interjurisdictional spillovers (externalities)
To correct major administrative weaknesses and streamline bureaucracy
There is need to maximize the use of central fiscal transfers in better performing subnational governments There is need to address negative externalities from nearby subnational governments r Need for administrative efficiency in tax collection and revenue management Central governments have greater capacity than subnational governments to manage revenue collection Equitable redistribution of income from central government to subnational governments
Unequal revenue or tax bases, natural resources or wealth across subnational governments Variation in socioeconomic characteristics of population Differences in geographical location and climate conditions Benefits from wellperforming subnational governments spillover to nearby jurisdictions Costs of poor-performing subnational governments extend to neighboring local governments
Weak administrative processes and redistributive function of national resources
Harmonizing Objectives and Outcomes
107
Table 2. Rationale, Objectives, and Outcomes of Intergovernmental Fiscal Transfers Rationale 1. Central governments have natural advantages in raising revenues from some sources, while subnational governments have natural advantages in providing many types of public services.
Objectives To equalize vertically (improve revenue adequacy)
2. Substantial disparities in revenue-raising capacity across decentralized levels of government
To equalize horizontally (intejurisdictional redistribution)
3. Use national resources to meet basic national priorities at the subnational jurisdictions
To minimize interjurisdictional spillovers (externalities)
4. Administrative inefficiency of some local governments
To correct major administrative weaknesses and streamline bureaucracv
Source: ADB, 2003; T A 5902
Outcomes Match cost of expenditure and available resources at the subnational governments Deliver public services and infrastructure development needs at the subnational governments Promote regional development and local economic stability Contribute to national economic growth, sustainability, and development Improve living conditions and quality of life of people Raise standard of public service delivery in lower-income subnational jurisdictions Promote equitable redistribution and allocation efficiency Meet requirements to support local economic development, e.g., provision of key infrastructure and services Subnational constituents adhere to national priority goals by providing basic services such as health, education, water and sanitation at prescribed levels Constituents of poor-performing subnational governments enjoy services from better-performing local governments nearby (spillover/ externalities) Better administrative efficiency in the redistributive function of national resources
Brillantes and Tiu Sonco II
HARMONIZING OBJECTIVES, GOALS AND OUTCOMES OF INTERGOVERNMENTAL TRANSFERS: SOME ISSUES AND CONCERNS The delivery of public services by subnational governments requires financial resources. Local governments are confronted with two fiscal challenges in realizing the policy goals and programs prescribed in their local development plans. These are revenue mobilization and revenue utilization. Revenue mobilization refers to the generation of local revenues such as: (i) local taxes, fees and charges, business permits, etc; (ii) proceeds from government services and business operations; (iii) income from assets; and (iv) funds from grants and aid. Revenue utilization concerns expenditure management and involves the allocation of resources in the provision of goods and services (Tomas, 1999). Many subnational governments suffer from financial constraints to fund the provision of basic services and local inf?astructure development. In the Philippines, for instance, most local government units (LGUs) have inadequate resources to finance local development and implement national policies due to the imbalance in resources across levels of governments, inequitable sharing of the Internal Revenue Allotment (IRA),undefined tax bases and poor local revenue-raising capacity (for both tax and non-tax revenues). Harmonizing the objectives and outcomes of intergovernmental fiscal transfers is not a simple matter. It does not only refer to achieving vertical or horizontal objectives or equalization, but more importantly: achieving desired results and outcomes at the subnational levels; reducing poverty; and contributing to regional and national economic development, sustainability and growth. The intergovernmental fiscal transfer design has to link national and subnational governments in harmonizing objectives and outcomes. "As such, well-designed intergovernmental fiscal transfers can play a critical role in promoting regional development, developing local infrastructures, delivering essential services and reducing poverty in a more efficient and equitable The equalization of development outcomes across local governments represents one of the main challenges in intergovernmental transfer design. Actually, in some cases the design turns out to be counter-equalizing.
Harmonizing Objectives and Outcomes
The Philippine Intergovernmental Transfer Experience Much can be learned from the experience of the Philippines when it implemented a fiscal transfer system through the enactment of the Local Government Code of 1991 (LGC of 1991). Experience has since shown that there is a need to closely examine the purpose and ideology of the transfers. To what extent have the transfers brought about vertical and horizontal equalization? The current intergovernmental transfer (IGT) design in the Philippines follows a fixed, formula-based computation which determines the share of the different levels of LGUs - the provinces, citieslmunicipalities and barangays. It is otherwise known as the internal revenue allotment (IRA) and is the revenue sharing mechanism through which all LGUs receive a fixed or predetermined share from the national government income collected by the Bureau of Internal Revenue (BIR). It is based on a sharing of resources set by the LGC of 1991 and comprises 40 percent of the national internal revenue tax collections." The formula determining the share of each LGU within each level of LGUs is as follows.
a) Population Population of LGU Total of Population of all LGUs at that Level
X (Total IRA - Cost of Devolution Fund) X 50%
b) Land Area
Land Area of LGU Total of Land Area X [(Total IRA - Cost of Devolution Fund) X 23%] of all LGUs at that X 25% Level c) Equal Sharing
[Total IRA - Cost of Devolution Fund) X 25% x 25% Total LGU at that Level Over the years, the IRA has been subject to very close examination. During the initial stages of devolution under the LGC of 1991, the LGUs, as well the academe and other stakeholders, were apprehensive and concerned about: (i) the sufficiency of the IRA for the devolved functions and the payment of salaries for devolved personnel; (ii) uncertainty of LGUs
110
Brillantes and Tiu Sonco I1
receiving their estimated IRA shares and the equality of sharing among levels of LGUs; and (iii) the responsiveness of the criteria (i.e. population, land area and equal sharing) to the funding needs of poorer LGUS.'~There were also questions pertaining to the use of population as a basis for computing local fiscal and social needs. Did population result in equitable distribution based on need? Was population size really a good indicator of local fiscal need? Those involved in the design of the IRA formula noted that the purpose of fiscal transfers was to cover the "cost of devolution." However, the cost of devolved functions referred only to the cost of personnel and facilities that were actually transferred to LGUs in the devolution program of 1992-03 (Manasan) and did not include the cost of actual service delivery of devolved functions. Hence, it has not been based on the "true cost" of devolution. The experience of many local governments, especially at the provincial and municipal levels, has shown that the IRA has been unable to comprehensively cover the costs of devolution. Many local governments have been unable to afford the salaries of national government personnel devolved to them, much less adequately perform their sectoral responsibilities. The IRA has been insufficient to provide services to the people, e.g. no funding for health services, environmental protection, agricultural modernization and public infrastructure.13 Fourteen years after the introduction of the LGC, it can be observed that the intergovernmental fiscal transfers system in the Philippines has been unable to correct andlor equalize the vertical and horizontal fiscal imbalances, not only in financing the cost of devolved functions - involving some " u n h d e d mandates" - among LGUs, but also in effecting better local development outcomes. In 1992, Manasan observed that the IRA formula was counter-equalizing as the grant shares and per capita income of LGUs were positively correlated. In a related study in 1996, Manasan et al. observed that higher per capita IRA tended to be associated with higher per capita social service expenditures. This relationship was found to be statistically significant in the case of total social service expenditure, human priority expenditure and health expenditure. Moreover, Figure Three shows the negative correlation between IRA shares and poverty incidence across regions. Box 2 provides a simple analysis and comparison across regions in regard to the relationship between human development outcomes and their IRA share, local revenue earning capacity from local taxes and non-tax sources, and poverty incidence.
Harmonizing Objectives and Outcomes
111
Regions
ERA Share----8- Poverty Incidence-1 Figure 3. Regional Comparison of IRA Shares and Poverty Incidence in the Philippines, By Rank, 2000
In summary, the current IRA formula has been unable to promote equitable sharing of financial resources of the national government among LGUs in the country. It has resulted in a widening of the gap between the richer and poorer LGUs. It has aggravated not only the vertical and horizontal imbalances, but also differences in local and regional development outcomes. The richer LGUs get greater hnding and are able to bring about change, economic growth and development, and improvement in the living conditions of their constituents. On the other hand, poorer LGUs are the areas in which local development outcomes are relatively low.I4 It is within this context that proposals have recently been made to radically modify the formula to determine the transfers to subnational governments. These include: (1) the inclusion of a poverty index that would determine the IRA shares of the local governments - the poorer the LGU, and the more the poor among its population, the higher will be its IRA shares; and (2) the inclusion of performance criteria - the share received by the LGU will depend on its tax collection performance. A matching grant scheme may even be devised. For instance, every peso collected by the LGU will be matched by the national government in terms of the IRA allotment
112
Brillantes and Tiu Sonco II
Box 2. IRA Share vis-ti-vis Poverty Incidence, 2000 The 16 regions were ranked based on the: (i) IRA share received by LGUs within the region; (ii) Local Income Taxes earned within the region; and (iii) poverty incidence of LGUs within the region.
A. IRA Share Regions IV, VI, 111, and NCR receive the highest IRA shares; while CAR, Region XII, CARAGA and ARMM receive the lowest shares from the IRA B. Local Tax Revenue NCR, Regions IV, 111 and VI are the high earning regions from local income taxes; while, ARMM, CARAGA, Regions XI1 and I1 are the least earning regions. C. Poverty Incidence ARMM, Regions V, XII, and CARAGA are among the regions with the highest poverty incidence; while NCR, Regions 111, IV and I1 have a relatively lower poverty incidence
IRA Share
Regions IV, Better Performing VI, 111, and Regions NCR
HDI
Local Revenue
NCR, Regions NCR, Regions 111, IV and I1 IV, VI, and 111
Poor CAR, Region ARMM, ARMM, Performing XII, Regions V, XII, CARAGA, Regions CARAGA and CARAGA Region XII, and and ARMM I1
IRA Share vis9-vis Poverty Incidence NCR, Region 111, Region IV ARMM, Region XII, CARAGA
Source: Brillantes and Tiu Sonco. Policy Paper on Strengthening Devolution through Meaningful Financial Decentralization: Improving Fiscal Transfers to LGUs.
Harmonizing Objectives and Outcomes
Box 3. Challenges in Designing and Implementing Intergovernmental Transfers Based on the studies of 5 Asian countries, the ADB has enumerated several challenges in designing and implementing intergovernmental fiscal transfers. They are: National government officials recognize the importance of stable sources of revenue for subnational government to meet their increasing fimctions and responsibilities; however, they are worried about the macroeconomic implications of institutionalizingmajor intergovernmental transfers. Intergovernmental transfers are often intended to meet a variety of difficult, and sometimes conflicting, objectives. Hence, programs have to be prioritized to meet different objectives. Devising mechanisms to allocate intergovernmental transfers can be very challenging. Selecting the appropriate criteria and measuring performance are difficult and problematic. The objectives of intergovernmental fiscal transfers are often compromised by political and institutional interference. The overall effect of intergovernmental fiscal transfers and other national policies related to subnational governments on broader development goals are difficult to determine. Source: ADB, Intergovernmental Fiscal Transfers in Asia.
6.
INTERGOVERNMENTAL FISCAL TRANSFERS: CITIZEN PARTICIPATION IN LOCAL PLANNING AND BUDGETING"
Participatory planning and budgeting refer to the involvement of citizens in identifying local priorities, policies, programs and projects that require allocation of resources. Participatory planning and budgeting provide the opportunity for peoples' participation in the allocation of resources to priority social policies, and for them to monitor public spending and policy performance. As such, local constituents gain ownership of the policies/programs/projects for local development, thus increasing their commitment to support social policies and the LGU development initiatives. Engaging civil society and citizens' groups in public management have enabled improved service delivery and accountability of the public sector. It has given the people greater opportunities to influence policy-making processes and the implementation of policies and programs. The idea of
114
Brillantes and Tiu Sonco II
engaging the people in the management of public affairs is a key dimension of good governance. People participation - involving the constituents in the politico-administrative processes - means that their needs and aspirations are heard and mainstreamed. Empowerment of citizens and their involvement in the decision-making process is vital for supporting pro-poor policies, improved service delivery, poverty reduction and the attainment of the Millennium Development Goals (MDGs). The collegial and partnering role of civil society groups with government has evolved over the recent years. Civil society organizations (CSOs) and non-governmental organizations (NGOs) have become active participants in national and local development planning activities and the implementation, monitoring and evaluation of policies, programs and projects. Public finance is another entry point for civil society participation. Participatory governance underscores the involvement of the local constituents in development planning processes. People participation usually starts at the grassroots or the subnational government level and has worked well in fostering needs-based local planning activities. However, there has been insufficient linking of development plans to budgets. On the one hand, good policies and plans are ignored in favor of politically strategic projects, creating a disconnect between what is planned and what is budgeted. On the other hand, good programs and projects are often expensive and local revenues simply cannot accommodate the large budgetary requirements. Participation in subnational planning and budgeting is a key tool in making local constituents understand and appreciate local development initiatives. One innovation could be citizens "deciding who to tax and how much, and how to spend these revenue^."'^ Budgeting means developing a plan - a periodiclannual budget based on a local development plan - in monetary terms. Participatory budgeting uses participatory governance and results in progradproject interventions that represent the people's perceptions of needs. Programs/projects financed by the local government have direct and significant impact on the lives of the local constituents. Participatory budgeting could be viewed as a mechanism whereby the citizens are directly involved in actual budgeting processes budget formulation and execution - thereby ensuring transparency and accountability in the management of local funds. It is anchored on pro-poor service delivery. "Participatory approaches to expenditure management refer to the range of methods, tools and choices that introduce/involve ordinary citizens and civil society in general into the process of resource allocation, tracking disbursement and monitoring the use of public resource (See Figure Four.) They contribute to transparency where people have access to public information; efficient service delivery and needs fulfillment.
115
Harmonizing Objectives and Outcomes and
Public Information
Implementation
Figure 4. Framework for Citizen Participation in Subnational Budgeting Source: World Bank Thindwa, Jeff. Entry Points for Civil Society to Influence
Budget Processes.
In the Naga City (Philippines), case study presented at the end of this section, both formal and informal participatory mechanisms have been used to enable people participation in local development planning and budgeting processes. Local planning processes have been enhanced through multisectoral consultation and active participation of the citizens and civil society groups. Hence, local policies, programs and projects have been prioritized and decided by the people. Moreover, the representation of a formal community-based body in the local legislative council provides the people an opportunity to ensure that the policies put forward during the planning processes are allocated enough fiscal resources during the budgeting process. Community or civil society groups can be seen as important partners of local governments in four areas: (1) local development planning and budgeting processes; (2) making local governments accountable in the allocation of local resources; (3) increasing local revenues; and (4) monitoring utilization of resources and the impact of local policies and programs. These have a large impact on aligning the objectives and outcomes of fiscal transfers by ensuring the effective, efficient and responsive utilization of fiscal resources.
116
Brillantes and Tiu Sonco I1
Box 4.Participatory Planning and Budgeting: Successhl Redistribution Resulting in Better Outcomes Studies show that the successful redistribution and utilization of resources in decentralizing states result in better outcomes. A case in point is Porto Alegre, Brazil, where the government was able to balance expenditures and earnings, and the equitable redistribution of resources to regions, by adopting participatory budgeting processes. The redistribution of resources to lagging regions is an important function of intergovernmentaltransfers. This is not to say that transfers are not important: rather, there has to be an interface between intergovernmental transfers and other stages of financial management and resource utilization at the subnational governments. Source: World Bank. Thindwa, Jeff. Entry Points for Civil Society to Influence Budget Processes
Box 5. Some International Experience on Participatory Budgeting Process Porto Alegre in Brazil has become a model for participatory budgeting. Gijurat in India has been noted for the successful engagement of civil society in budget review and analysis. The Public Expenditure Tracking Surveys in Uganda are a model for participatory expenditure tracking, and the Philippines is noted for its performance monitoring: citizen's report card.
Source: World Bank. Thindwa, Jeff. Entry Points for Civil Society to Influence Budget Processes
Among the indicative action points that enable participatory budgeting in subnational governments are: a) Identifying and designing clear points of entry for civil society participation in local governance; b) Developing a policy to provide the framework for participatory budgeting; c) Investment in capacity building - not only of local governments and civil society but also of national government agencies; d) A realization that participatory budgeting cannot be fast-tracked;
Harmonizing Objectives and Outcomes
117
e) A management that can overcome long established systems and procedures; f) A formal system of accountabilities being in place; and g) Efforts to institutionalize participatory governance through participatory planning and budgeting that tap institutions already working in the area. Obviously, citizen participation in the local government budgeting processes has not yet been fully operationalized. There are, however, many cases showing how local governments' improved policy choices and program implementation can be attributed to engaging civil society in the budget process. A policy framework for a participatory budgeting process would provide the necessary entry points of citizen participation in subnational budgeting process. This would advance any mechanisms for participatory framework for civic participation in local planning that are already in place. The institutionalization of participatory budgeting in the infrastructure of local development planning activities would be the foundation for the practice of participatory local budgeting. Both the local government units and civil society would have critical interest in such a mechanism, thus engaging in it. Participatory budgeting cannot be fast-tracked. It is a continuing effort that eventually leads to a successful engagement of civil society in the formal budgeting processes. It should be noted, however, that in the Philippines, this is just one of many reform imperatives not only in local governance but also in the administrative system in general. More importantly, there has to be efforts to ensure clear and strategic activities directed towards the attainment of the policy goals of decentralization, enhancement of living conditions of local constituencies, improving local service delivery and contributing to national development.
A Case Study: Naga City, Philippines - Engaging Civil Society in Planning and Budgeting for Local Development In Naga City, local governance places high hopes in good leadership; performance of local officials and workers; and the active participation of groups, community members and citizens in local decision-making processes. There are two components working towards good governance and improvement of service delivery: (1) the City Hall factor; and (2) the people factor. The City Hall factor comprises the elected Mayor and local officials, the appointed city hall officers, and other employees who work for Naga City government. The people factor refers to the community -
118
Brillantes and Tiu Sonco 11
especially the urban poor, senior citizens, women, youth, disabled, children, religious, academe and the media. Engaging civil society in pursuit of local development has had a big impact in Naga City. The city works within a conceptual framework for good urban governance which has three elements: (i) progressive leadership, which shows consideration for the public rather than self-interest; (ii) partnerships, which increase resources and maximize potentials by persistent tapping of partners of the LGU; and (iii) participation, which refers to innovations and mechanisms to involve groups and individuals, thus involving all citizens. Both partnerships and participation enable the sustainability mechanism for good local governance through the presence and active participation of citizens and community organizations. Naga City has adopted both formal and informal participatory mechanisms for individuals and organizations. These include: (i) continuing NGOs accreditation, (ii) multi-level consultation mechanisms, (iii) referenda on development issues, and (iv) an empowerment ordinance. Naga City was among the few LGUs that accredited NGOs after the enactment of the Local Government Code of 1991. Through the NGOs, it has conducted multi-level consultations that provide valuable inputs into identifying development priorities and the citizens' stand on policy issues. The city was also the first to implement city-wide referenda on development issues, wherein the result of each referendum serves as the basis for policy decisions of the LGU. Matters that have been decided by referenda include: (i) stopping the development of a country club golf course; (ii) a shelter program; (iii) establishment of a central bus terminal; (iv) the floatation of bonds to finance local development; and (v) color-coding of tri-mobiles (a popular mode of intra-city transport). In 1997, the Naga City People's Council (NCPC) was created under the Empowerment Ordinance of Naga City. It comprises duly accredited businesses, people's organization (POs) and NGOs within the city and has 40 to 45 members. In addition to its eight elected members of the local legislation council, the NCPC has a representative on the local decisionmaking board. Among other things, the NCPC: 1. Appoints representative to local special bodies; 2. Observes, votes, and participates in deliberations and planning; 3. Proposes legislation and participates up to the committee level of Sanggunian; and 4. Acts as the people's representative in the exercise of their constitutional rights to information on matters of public concern and access to official records and documents.
Harmonizing Objectives and Outcomes
Box 6. Naga City Special Bodies and Membership Compositio Special Body I. Naga City Investment Board 2. Naga City Urban Development and Housing Board 3. Solid Waste Management Board 4. Traffic Management Task Force 5. Task Force Tubig 6. Naga City Cooperative Development Council 7. Naga City Disaster Mitigation Council 8. City Agriculture and Fisheries Council 9. Bids and Awards Committee (government procurement) 10. People's Law Enforcement Board 11. Local School Board 12. Local Health Board 13. Peace and Order Council 14. Naga City Tourism Council
Membership 10 member body, 5 come from NCPC, one sits as vice chair of Board 20 member body, 10 from NCPC 10 member body, 2 from NCPC 10 member body, 2 fkom NCPC 5 member body, 1 from NCPC 12 member body, 1 from NCPC 7 member body, 1 from NCPC 12 member body, 1 from NCPC 7 member body, 1 observer from NCPC, 2 observers from engineering and accountancy professions 5 member body, 1 from NCPC 5 member body, 1 from NCPC 15 member body, 1 from NCPC 27 member body, 1 from NCPC
Source: ADB. Empowering the Poor: Key to EfSective Pro-Poor Services
Aside from the formal representation of the NCPC in the local legislative council, various groups and organizations are involved in other city policymaking bodies. Members of the urban poor, for instance, are part of the Housing Board. NGOs and community organizations are active members in the Integrated Livelihood Manage Council (ILMC) which oversees the livelihood program of the city. The private sector comprises 50 percent of the membership of the Naga City Investment Board (NCIB), the Naga City Tourism Council (NCTC), and the Environment Management Council (EMC). Box 6 lists the various special bodies in Naga City and their membership composition, indicating the NCPC representation.
Brillantes and Tiu Sonco II
CAPACITY BUILDING FOR EFFECTIVE UTILIZATION OF TRANSFERS, RESOURCE MOBILIZATION AND SUSTAINABILITY Some of the experiences of decentralization in Asian countries suggest that local governments everywhere might face similar problems and challenges. For instance, the lack of capacities - financial and human resources - continues to be a concern and the "absorptive capacities" of local governments under decentralization have to be carefully studied. Capacity building should always be a priority and is a continuing process. It must be emphasized however that "lack of capacities" is not the exclusive domain of subnational institutions. Central government agencies and civil society organizations are also faced with the same problem as they adjust to a decentralized governance structure. Capacity building is much more than capability enhancement and training of personnel. It looks into three levels of capacity: (i) the systemic level; (ii) the institutional level; and (iii) the individual level. The systemic level refers to the legal framework and supporting policies that provide the environment for an effective intergovernmental fiscal transfers system, utilization of local revenue resources (transfers and local revenues) and the attainment of better human development outcomes. The institutional level looks into the elaboration and establishment of appropriate management systems, structures, processes and procedures, resources, links, performance measures and systems of accountability. It also highlights the management of relationships between the different levels of governments and across local governments. The individual level focuses on the process of equipping LGU staff and elected officials, local constituencies and other stakeholders with the understanding, knowledge and access to information, skills and training, and the appropriate attitude; thus, providing them with the capability to effectively carry out their roles and responsibilities. Figure 6 shows the conceptual framework for capacity building and assessing capacity building needs. As mentioned earlier, the challenges for effective local governance include the lack of capacities of staff and elected officials, and the LGU fiscal capacities. One of the primary concerns is to enhance the competencies of LGU staff and elected officials, and to continue to refine the formula and conditions for transfers. Civil society participation in the local budget process and capacity building presents challenges as well as opportunities for better local governance. In view of the imperative for good governance and local governance reform, there should be a continuing effort to improve the enabling environment for participatory governance and capacity building. Participatory planning has become a key feature of local development
Legal Framework Supporting Policies
r-l Systemic Level
I
Goals, objectives, strategies Structures, processes, and procedures Resources (human, physical, financial) Linkages Communications and information systems Performance measures System of accountabilities
Institutional Level
m Individual Level
I
Knowledge Skills Attitudes
Figure 6. Critical Components for Capacity Building Source: Astillero and Mangahas, 2003.
Capacity to implement effective intergovernmental fiscal transfers and fiscal sustainability
122
Brillantes and Tiu Sonco I .
Box 6. Capacity Building for Local Financial Sustainability in Cambodia Local councils in Cambodia rely heavily on fiscal transfers to cover both their administration and development expenditures. The design of fiscal transfers in Cambodia recognizes local capacity building as a primary objective. It recognizes that transfers play a critical role in developing and sustaining an autonomous local-level development planning, budgeting and service implementation process. The objectives of intergovernmental fiscal transfers are not limited to correcting the vertical and horizontal imbalances; there is an additional aim of developing a system and incentives for mobilization and management of local public resources. There are mechanisms whereby the local councils have the power and capacity to raise their own revenue, thus providing the long term solution to financial constraints and fostering local accountability. Source: ADB, 2003.
planning. Mechanisms for civil society participation in the local government budgeting processes must be institutionalized. This would further "democratize" local development processes and must be supported through capacity building. Building capacities should not only focus on local governments but also on national governments where efforts can concentrate on strategies to streamline the structures and processes of national government agencies. In the Philippines for instance, the capacities of the Department of Budget and Management have to be strengthened to support innovations in local governance. Capacity building efforts should also focus on civil society groups. This can include designing strategies for active civil society engagement in local governance in general and in local budgeting processes in particular; thus increasing the capacity of local officials and civil society/NGOs in areas within the budgeting process where they can be involved. The capacity of local governments to increase local revenue collection has to be strengthened. Fiscal autonomy should be seen not only as the transfer of authority to disburse resources to the local government level but also the greater responsibility to increase revenue collection from local sources. This should address LGU dependency on fiscal transfers and allow them to enjoy a sustainable fiscal revenue-raising capacity. It should also mean building the capabilities of local government officials.
Harmonizing Objectives and Outcomes
SUMMARY AND CONCLUSION The concepts, rationale, justifications and objectives of intergovernmental fiscal transfers boil down to how subnational governments can finance the delivery of public goods and services in a decentralized government environment. Intergovernmental fiscal transfers should be viewed as a part of fiscal decentralization and the larger system of decentralization. Intergovernmental fiscal transfers do not guarantee achievement of objectives and outcomes. The case studies on intergovernmental fiscal transfers in Asia offer a valid observation that intergovernmental transfers are not the only factor that influences better outcomes at the subnational government level. Indeed, intergovernmental transfers must not be situated or considered outside the overall framework of fiscal decentralization, or decentralization. Rather, fiscal transfers should be viewed as part of the system and their design should be tailored to work within the system of local governance. The design of transfers has to be placed within the overall systems and processes of subnational governance, i.e. the legal and regulatory framework; local development planning; budgeting processes; financial management and implementation; revenue mobilization and own-source revenue generation. An effective intergovernmental fiscal transfer system therefore has to emphasize the allocative efficiency and equitable redistribution of funds to subnational governments, thereby addressing the issues of vertical and horizontal imbalances. More importantly, a good fiscal transfer design has to support: 1. 2. 3. 4. 5.
Improvement of service delivery; Equitable growth; Poverty reduction; Regional and national development; and contribute to The attainment of the Millennium Development Goals.
It needs to satisfy the criteria of adequacy and buoyancy, predictability, simplicity and transparency. Designing a good fiscal transfer system necessitates the identification of local revenue earning capacities (tax and non-tax revenue), identifying national appropriation and budgeting processes, measuring local expenditure requirements, measuring human development outcomes, designing the appropriate national and local tax systems, establishing the appropriate number of LGUs and their classifications, deciding on the extentldegree of decentralization or re-centralization, measuring the impact of
124
Brillantes and Tiu Sonco I1
decentralization to national economic growth, regionalization, poverty reduction efforts, and national economic development and growth. But there are also risks in adopting intergovernmental transfer systems. In the Philippines, for instance, the IRA formula fails the equitable redistribution function of fiscal decentralization because the LGUs that perform better still get greater shares from the fiscal transfers; further aggravating the imbalances between richer and poorer LGUs. As a result, the gap continues to widen among subnational governments, not only in fiscal capacity but also in human development outcomes. The purpose of intergovernmental transfers in strengthening decentralization should be defined, as should be the goals and objectives of fiscal transfers. The effectiveness of good fiscal transfers could be measured by their "effects on such policy outcomes as allocative efficiency, distributional equity and macroeconomic stability."" Dr. Anwar Shah advocates that for each objective of transfers, there should be a separate form of transfer design." Moreover, any endeavor in the area of decentralization should look into the wider aspects of local finance. There could be a need to harmonize and rationalize the various dimensions of local development administration such as frameworks, designs, systems and procedures, plans, accountabilities, and data generation and utilization. We recognize that the impact of intergovernmental transfers on service outcomes has not yet been well-studied, at least in Asia. Most of the studies focus on fiscal decentralization and the wider context of decentralization that link objectives and outcomes of national and subnational policies. Hence, future research should be focusing on the effectlimpact of resources, through intergovernmental transfers, on the attainment of outcomes. Furthermore, such endeavor should focus on establishing the link between intergovernmental transfers based on equalization and improved service delivery by the subnational government that might also be resulting in local and national economic growth and poverty reduction. Attempts to present the idea of harmonizing objectives and outcomes of national and subnational governments through intergovernmental transfers will require M h e r research. Comparative studies should be conducted for a critical analysis on how to effectively harmonize objectives and outcomes of intergovernmental transfers.
Harmonizing Objectives and Outcomes
Notes Bahl argues that the functions and responsibilities of local authorities should be clearly defined first, v i s - h i s national authorities. After a clear delineation of responsibilities, the funding of such services - whether by transfers from the central government or by locally generated taxes or users fees, etc - should then be addressed. One reason for the failure of decentralization strategies is the continuing practice on the part of central authorities to mandate subnational governments to perform functions without providing the necessary funds, hence "unfunded mandates." Some refer to this practice as "dumping". Kakhonem (2001) writes that allocative efficiency may be attained through better matching of public services to local preference; productive eficiency may be attained through increased accountability of local governments to citizens, fewer levels of bureaucracy and better knowledge of local costs. Cheema and Rondinelli, in their work Decentralization and Development. Policy Implementation in Developing Countries (1983), list fourteen reasons for the adoption of decentralized development planning and administration in developing countries. Good governance has been referred to as the "missing link" between anti-poverty efforts and poverty reduction. According to the UNDP (200054) "even when a country tries to implement economic policies to foster pro-poor growth and mount targeted poverty programs, inept or unresponsive governance institutions can nullify the impact." It is within this context that decentralization,properly implemented and supported, may be a way to bring about effective governance. We are cognizant, however, of the possibility that a decentralization strategy, if badly planned and implemented, can worsen inequalities (UNDP 2000:60). Decentralization may therefore be a two-edged sword. We use the term "industry standard" to suggest that many multilateral development institutions have used as reference, or taken off from, the Cheema and Rondinelli construction. This is certainly not an attempt to romanticize the notion of decentralization. If not used properly, it can exacerbate inequalities among regions and even lead to fragmentation of the state; but when used correctly, it can indeed be a potent poverty reduction strategy. This draws upon Brillantes, Innovations and Excellence in Local Governance 2003. These criteria essentially conform to the classic elements of the state as described in political science literature, i.e., people, territory and sovereignty. This would be in accordance with the time-honored dictum that there should only be taxation when there is representation of the people, in this case through the local legislative body. ADB, TA 5902. Study on Intergovernmental Fiscal Transfers for Equitable In-Country Growth. Intergovernmental Fiscal Transfers in Asia: Current Practices and Challenges for the Future, 2003. Total IRA pool is then divided among the Provinces (23%); Cities (23%); Municipalities (34%); and, Barangays (20%), then further allocated to LGUs on the basis of population, land area, and equal sharing. Ilago, Simeon. Local Government Transformation: Glimpses from the Local Finance Literature under the 1991 Local Government Code, 1997. Brillantes, Innovations and Excellence in Local Governance, 2003 Some local governments that face financial constraints are able to effect better local development through good leadership by chief executives and other officials. Some LGUs engage in other modalities of local resource generation (e.g. bond floatation, Build-Operate-Transfer (BOT), public-private partnership, etc.) activities to pump prime local development.
126
Brillantes and Tiu Sonco II
15. This section draws on Civic Participation in Local Governance in the Philippines: Focus on Subnational Budgeting and Planning by Alex Brillantes and Jose Tiu Sonco 11. 16. World Bank. Thindwa, Jeff. Entry Points for Civil Society to Influence Budget Processes. 17. Bird and Smart, 2001. Intergovernmental Fiscal Transfers: Some Lessons from International Experience. 18. Anwar Shah, World Bank Institute, during discussion at the International Workshop of Fiscal Transfers held in Swul, Korea, June 2005.
References Ahmad, Nuzhat, and Syed Ashraf Wasti. 2003. Pakistan. In Intergovernmentalfiscal transfers in Asia: Currents practice and challenges for the future (Asian Development Bank, TA 5902), Paul Smoke and Kin Yun-Hwan, eds., 176-218. Retrieved from http://www.adb.org/Documents/Books/Intergovemmental~Fiscal~Transfers/Intergovernm ental-FiscalTransfers.pdf Bahl, Roy. 2002. Fiscal decentralization in Indonesia: The first year in review and challenges ahead. Center for institutional reform and informal sector (IRIS), University of Maryland, College Park. Retrieved from http:l/128.8.56.108/iris-data/PEG/Bahasa&hl-kdk.pdf Bahl, Roy. 1999. Implementing rules for fiscal decentralization. Paper presented at the international seminar on land policy and economic development, Land reform institute, Taiwan, November 17, 1998, and International Studies Program, School of Policy Studies, Georgia State University, Atlanta Georgia, January 1999. Bahl, Roy. 2000. Intergovernmental transfers in developing and transition Countries: Principles and practice. Retrieved from http://www.worldbank.org/wbi/publicfinance/documents/TopicO8~Print~.pdf Bahl, Roy, Jaime Boex, and Jorge Martinez-Vazquez. 2001. The Design and implementation of intergovernmental fiscal transfers. Retrieved from http://isp-aysps.gsu.ed~/~rc/pofd~pofd4ahletal200 1.pdf Bahl, Roy, and Sally Wallace . 2001. Fiscal decentralization: the province - local dimension. Retrieved from http:Nisp-aysps.gsu.edu/papers/bah12001.pdf Bird, Richard M. and Michael Smart. 2001. Intergovernmental fiscal transfers: Some lessons from international experience. Paper present at Symposium on Intergovernmental Transfers in Asian Countries: Issues and Practices. International tax program, Rotman School of Management, University of Toronto. Boex, Jameson, Roy Bahl, Jorge Martinez-Vazquez, Philip van Ryneveld, and Andrey Timofeev. 2002. Developing a system of IG Grants in Tanzania. Executive Summary. Georgia State University, Andrew Young Schools of Policy Studies. Retrieved from http:Nisp-aysps.gsu.edu/projects/tanzanidexecsumm.pdf Brillantes, Alex B. 2003. Public sector reform and poverty reduction. In Growth and Institutions in Developing Asia, Ernesto Pernia and Anil B. Deolalikar, eds., 97-136. Manila, the Philippines: Asian Development Bank by MacMillian Palgrave. Brillantes, Alex Jr., and Carina Pamela Bengzon. 2002. Decentralization and people participation in the Philippines: The role of civil society and NGOS. Paper presented at the Civil Society and Democratic Governance in East Asia Academic Conference, Jeju, Korea, July 29-30. Brillantes, Alex Jr. 2004. Decentralization imperatives: Lessons from some Asian countries. Public Policy University of the Philippines, 8(1): 1-28
Harmonizing Objectives and Outcomes
127
Brillantes, Alex Jr. 2003. Innovations and excellence: understanding localgovernments in the Philippines. Center for Local and Regional Governance. Quezon City, Philippines: National College of Public Administration and Governance, University of the Philippines Brillantes, Alex. Jr. and Tiu Sonco, Jose 11. Civic participation in local governance in the Philippines: Focus on subnational budgeting and planning. World Bank Working Paper, Washington, D.C. Brillantes, Alex Jr., and Tiu Sonco, Jose 11. 2005. Policy paper on strengthening devolution through meaningful financial decentralization: Improving fiscal transfers to LGUs. Policy Research Working Papa. Unpublished. LOGODEF Ebel, Robert D. and Serdar Yilmaz. 2001. Concept of fiscal decentralization and worldwide overview. Retrieved from http://www.desequilibrefiscal.gouv.qc.ca~en/pdWebel.pdf Capuno. Joseph J. 2003. Philippines. In Intergovernmental fiscal transfers in Asia: Currents practice and challengesfor the future (Asian Development Bank, TA 5902), Paul Smoke and Kin Yun-Hwan, eds., 219-282. Retrieved from http://www.adb.org/Documents/Books/Intagovernmental~Fiscal~Transfers/Intergovemm ental-Fiscal-Transfers.pdf Cheema G. S., and Rondinelli D.A. (Eds). 1983. Decentralization and development. Policy implementation in developing countries. Beverly Hills, CA.: Sage Publishing. Indonesia Government . 2001. The Republic Of Indonesia Fiscal Decentralization: Central Local Financial Balancing As A Tool Of Fiscal Decentralization. Retrieved from http://www.gtzsfdm.or.iddocuments/dec~indgv~a~doc~isDec~A2.pdf Khemani, Stuti. 2003. Partisan politics and intergovernmental transfers in India. Development Research Group The World Bank. Retrieved from http:Necon.worldbank.org/files/25492~wps30 16.pdf Kraemer, Moritz. 1997. Intergovernmental transfers and political representation: Empirical evidence from Argentina, Brazil and Mexico. Inter-American Development Bank. Retrieved from http://www.iadb.org/res/publications/pubfiles/pubWP-345.pdf Kim, Yun-Hwan, and Paul Smoke. 2003. The role and challenges of intergovernmental fiscal transfers in Asia. n Intergovernmental fiscal transfers in Asia: Currents practice and challengesfor the future (Asian Development Bank, TA 5902), Paul Smoke and Kin YunHwan, eds., 1-19. Retrieved from http://www.adb.org/Documents/Books/Intergovernmental~Fiscal~Tmnsfers/Int~govemmenta 1-Fiscal-Transfers.pdf Lewis, Blane D. 2003. Indonesia. In Intergovernmental fiscal transfers in Asia: Currents practice and challengesfor the future (Asian Development Bank, TA 5902), Paul Smoke and Kin Yun-Hwan, eds., 138-175. Retrieved from http://www.adb.org/Documents/Books/Intergovemmental~Fiscal~Transfm/Intergovernmenta 1-Fiscal-Transfers.pdf Nam, Chang Woon and Rudiger Parshe. 2001. Looking for appropriate forms of intergovernmental transfers for municipalities in transition economies. Center for Economic Studies and Ifo Institute for Economic Research. Retrieved from http://www.cesifo-group.de/pls/guestci/download~F5588/614.PDF Rao, M. Govinda and Nirvikar Singh. 1998. Intergovernmental fiscal transfers: rationale, design and Indian experience, 1998. Retrieved from http:Necon.ucsc.edul-boxjenklcre3.pdf
128
Brillantes and Tiu Sonco I1
Romeo, Leonardo G. 2003. Cambodia. In Intergovernmentalfiscal transfers in Asia: Currents practice and challenges for thefitture (Asian Development Bank, TA 5902), Paul Smoke and Kin Yun-Hwan, eds., 60-82. Retrieved from http://www.adb.org/Documents/Books/Intergovemmental~Fiscal~Transfers/Intergovmm ental-Fiscal-Transfenpdf
Schroeder, Larry, and Paul Smoke. 2003. Concepts, International Practice, and Policy Issues. In Intergovernmental fiscal transfers in Asia: Currents practice and challenges for the future (Asian Development Bank, TA 5902), Paul Smoke and Kin Yun-Hwan, eds., 20-59. Retrieved from http://www.adb.org/Documents/Books/Intergovernmental~Fiscal~Transfers/Intergovernm ental-Fiscal-Transfers.pdf Sen, Tapas K. and Christoph Trebesch. 2004. The use of socio-economic criteria for intergovernmental transfers: The case of India. Retrieved from http://www.nipfp.org.inlworking%20paperlwp1O.pdf Srivastaval, D.K. 2003. India. In Intergovernmental fiscal transfers in Asia: Currents practice and challenges for the future (Asian Development Bank, TA 5902), Paul Smoke and Kin Yun-Hwan, eds., 83-137. Retrieved from http:/lwww.adb.org/Documents/Books/Intergovernmental~Fiscal~Transfers/Intergovmm ental-Fiscal-Transfers.pdf Tomas, Ernesto Jr. E. 1999. Budgeting for local governments. Conjucture March. World Bank Institute. 2003. Intergovernmental transfers: Concepts and policy issues. Serdar Yilmaz and Serap Bindebir, October. Retrieved from http://www.worldbank.org/wbi/publicfinance/documents/fiscalfederalism~Russi~orea% 20paper.pdf World Bank. Intergovernmental transferslgrants design. Retrieved from http://www.ciesin.org/decentralization/El World Bank . 1999. Local Revenue Mobilization and Intergovernmental Transfers. Retrieved from http://www.worldbank.org/wbi/mdf/mdfl/revenue.htm Yilmaz, Serdar, and Serap Bindebir. 2003.World Bank Institute. Intergovernmental transfers: Concepts and policy issues. World Bank Institute Working Paper, Washington, D.C.
DISCUSSANT COMMENTS THE NATURE OF EQUALIZATION: OBJECTIVES AND CONSEQUENCES
DAVID PELOQUIN~ Department of Finance, Government of Canada
PETCHEY AND LEVTCHENKOVA, HOFMAN AND GUERRA, HULL AND SEARLE, AND BRILLANTES AND SONCO
1.
INTRODUCTION
These comments seeks to synthesize the key objectives of fiscal equalization policies and to give a practitioner's view of some of the intended and unintended consequences of such policies - both from the traditional economists' vantage point of equity and efficiency issues and from the increasingly important dimension of the "good governance" implications of fiscal equalization - both for sub-national governments (SNGs) that receive equalization transfers and for intergovernmental relations more generally. The author's main experience as a fiscal federalism practitioner has been in dealing with intergovernmental fiscal arrangements (including equalization policies) in federations in developed countries -Canada and Australia. Accordingly, the paper identifies areas where the practice and objectives of fiscal equalization and related governance issues in these countries are likely to differ significantly from those that may be appropriate in non-federal countries - particularly less developed and transitional countries (LDTCs) that may be moving toward more decentralized administrative and political structures. However, the distinction between these two situations is not always watertight: even developed-country federations sometimes face equalization and other intergovernmental fiscal arrangements challenges that would be familiar to officials in LDTCs - notably moving toward greater revenue
autonomy for municipal governments (and, in Canada's case, among newly self-governing First Nations); ensuring access to borrowing in the absence of such autonomy; and applying fiscal equalization principles to current revenues and expenditure needs (as well as to SNG access to capital financing, where the latter is rationed by the central government). The paper is written around three themes -the equity, efficiency and "good governance" issues surrounding equalization - and concludes with a section on possible lessons for the design of equalization policies and the key governance arrangements that underpin them.
EQUITY ISSUES
2.1
"Fiscal" versus "Horizontal" Equity
"Fiscal" equity - a variant of the more familiar concept of horizontal equity - is the dominant rationale cited for equalization (particularly in developed countries). Typically, this objective is expressed in terms of a commitment to ensuring that equalization transfers result in the capacity of sub-national governments to provide comparable levels of public services at comparable levels of taxation. However, it is important to acknowledge the limits to what equalization can hope to accomplish. "Full" equalization - in the sense of completely eliminating fiscal disparities between sub-national governments - is (with the notable exception of Australia) almost never a real option for most countries. Fiscal disparities among sub-national governments (especially in developing countries but in most developed countries as well) are simply too large to be completely bridged. Federal or national governments (which are usually responsible for making equalization transfers) rarely enjoy large enough vertical fiscal gaps to more than partially offset these disparities through transfers. Forcing sub-national governments to share their own-source revenues with their less well-off counterparts is also not politically feasible.
In short, as a mechanism for moving toward greater horizontal equity, equalization transfers appears to be something of a "luxury good" that countries acquire more of as they grow richer - but rarely achieve completely. Moreover, as noted above, "fiscal equity" is in any event a variation on the theme of horizontal equity. Fiscal equity focuses on the capacity of
Discussant Comments
131
governments to provide public services at particular levels of taxation whereas horizontal equity is usually thought of in terms of comparable treatment of comparable individuals. Given any degree of policy autonomy among SNGs - an important goal of decentralization in its own right - it would be impossible to ensure exactly comparable treatment of comparable individuals even if one could "fully" equalize disparities between governments, simply because different SNGs will adopt different policies that focus on different clienteles. Yet it is often horizontal equity among individuals that policy makers and policy advocates have in mind when they assess the equity-enhancing properties of equalization and other policies. Still, half a loaf (some equalization) may be better than none. The key question for policy makers to answer with regard to the equity-enhancing properties of equalization is whether those policies can improve equity?
2.2
"Vertical" Equity
While the "fiscal equity7'case for equalization is well established, some of the papers in this section try to make a connection between equalization and reduction in disparities between rich and poor citizens - at least over time. In practice, vertical equity has not been the main goal of equalization but a happy side-effect, if it happens at all. Indeed, confusion surrounding the vertical equity role (if any) of equalization is the source of much potentially sterile debate, including allegations that equalization transfers end up "taking money from poor persons in rich regions and giving it to rich persons in poor regions". At best, equalization may be a necessary - though not a sufficient condition for reducing poverty or reducing concentrations of greater income and wealth. There is no guarantee that ensuring comparable capacity for sub-national governments to deliver services will translate into comparable outcomes in terms of converging socio-economic conditions. On this, the papers in this section that focus on interpersonal disparities seem to agree. A reasonable conclusion may therefore be that there is no substitute (certainly no "short cut" through equalization policies) for ensuring that SNGs have the incentives to pursue sound policies that will eventually lead to convergence of fiscal capacity and individuals' access to services. This will be achieved by striking the right combination of accountability for use of federal government grants and strong internal governance mechanisms that ensure accountability to local stakeholders for use of funds and results achieved.
2.3
Efficiency Issues
The main efficiency argument traditionally made for equalization (and noted in all the papers in this section) is that of minimizing fiscally induced factor mobility. This is the argument that factors should move in response to real differences in marginal productivity and in the costs and benefits of locating in different regions. Factors should not move due to "accidents of history or geography" that leave jurisdictions on either side of an internal border with significantly greater capacities to provide public services or significantly greater needs for such services. Yet, while the theoretical foundations for this argument may be valid, policy makers will generally ask how important the efficiency losses avoided through equalization policies may be in practice. It is possible that other factors - such as the value placed on existing social networks in one's current place of residence - may dominate mobility decisions even in the face of persistent fiscal disparities. The latter is particularly likely in those cases where disparities are relatively modest (as they tend to be in developed countries) or where it may be thought that the disparities are temporary in nature. Even there, the Brillantes/Sonco paper makes a case that in the Philippines and some other Asian countries, large horizontal disparities may be self-perpetuating (and even self-enlarging). Moreover, fiscally induced migration cannot always be laid at the feet of sub-national governments' uneven capacities or perverse policies. In many cases, it is federal or national policies (for example, agricultural subsidies and regionally differentiated social programs) that create the greatest distortions. These may be more suitable targets for policy reform. There is also the contrary argument that equalization actually impedes factor mobility that should be taking place. Essentially, this is that it may be better for a country not to equalize horizontal fiscal disparities (or to do so only in part) since they are strong signals that factors should be moving in response to, say, permanent economic shocks favoring faster economic and population growth in particular regions. In such circumstances, the faster people and businesses move to these regions, the sooner equilibrium will be restored and everybody will be better off.2 However, faster does not necessarily mean better - particularly if it is difficult to be sure that particular shocks will be permanent. In the absence of a God-like "efficient central planner" (of the type referred to in the PetcheyILevtchenkova paper) who could tell policy makers what incentives would maximize speed of adjustment at minimum cost for all, there is also a serious danger of overshooting the mark and ending up with exactly the kinds of undesirable fiscally induced migration that is the subject of much of the equalization efficiency literature.
Discussant Comments
133
In any event, this is ultimately an empirical question and there does not appear to be any solid evidence that countries with equalization have less rapid adjustment than others. Certainly, in Canada's case, fiscal disparities have been trending downward for the past three or four decades. If equalization is indeed slowing down adjustment in Canada, it does not seem to be slowing it down much.
3.
"GOOD GOVERNANCE" ISSUES
To economists, "good governance" issues are, in some sense, a subset of "efficiency" issues. However, they relate to a sufficiently important set of issues - where economics and political science overlap - that merit distinct treatment. One way of thinking about good governance in the equalization context is to think of equalization as a kind of "meta-game" - a repeated meta-game of negotiations around issues relating to access to financial resources. This lies at the core of intergovernmental relations in virtually all countries with more than one level of government. Good governance is effectively the central concern of the Brillantes/Sonco paper, which asks how fiscal arrangements (including equalization policies) can support decentralization policies aimed at fostering a degree of local government autonomy that will respond to the needs and preferences of local citizens. While equalization policies per se are not the central focus of Brillantes/Sonco (their focus is on fiscal arrangements more generally), their messages in respect of equalization appears to be that: Meaningful autonomy requires accountability in respect of local revenueraising and spending; and In the absence of equalization transfers (and ideally stable and predictable ones), the ability to respond to local needs and preferences may be much weaker in poor or high-need jurisdictions.
A closely related "good governance" objective of equalization is that of fostering healthy intergovernmental competition on an on-going basis - that is, once autonomous SNGs have been established. The logic here is that - as "laboratories of federalism" (or of decentralization): These governments will compete with one another through policy experimentation and emulation of "best practices"; and
Such competition will be most fruitful if the SNGs have reasonably comparable levels of resources with which to experiment. Yet one can still ask how much equalization is actually required. It is possible that governments "below the average" may be induced to try harder if there is more variation, and that too little variation may encourage smugness. There is also the more serious problem that, in many cases, SNGs within a country may be so different - in size, wealth, needs or various dimensions of governance "capacity" - that it is difficult to imagine a level playing field between them. If so, a "one size fits all" equalization regime may be particularly inappropriate. Finally, as noted by the HulVSearle paper, given the kinds of data needed to run a well-developed equalization regime, making all of these data publicly available can help the public compare the performance of their governments on something like a comparable basis. Notwithstanding the above, the key "good governance" issue in relation to equalization policies is that of ensuring that, as a result of such policies, SNGs do not face distorted incentives that lead them to pursue perverse policies. In particular, the most important "good governance" feature of a properly designed equalization regime is that SNGs be led away from the temptations of grant-maximizing behavior and be presented instead with incentives that do not artificially tilt decisions toward high (or low) taxation (or spending) - except in cases where there are clear spillovers (e.g., in the case of the merit goods referred to in the BrillantesISonco paper) - and incentives to promote growth. More specifically, the main design features of equalization - the size of the equalization pool, the measures of fiscal capacity andlor expenditure need, and the allocation formula used to distribute the pool among jurisdictions on the basis of their measured capacity and/or need - should be as independent as possible of actions that SNGs can take. The PetcheyILevtchenkova paper points to a possible problem with even the best equalization design - namely the ability of individual jurisdictions to materially affect the national average revenue (or expenditure) effort that are typically used to determine "standardized" measures of capacity and need (and, as a result, their equalization entitlements), simply because their weight within the national average may be substantial. While the PetcheyILevtchenkova paper makes a valuable contribution to equalization theory by rigorously identifying the conditions under which such distorted incentives may arise, the extent to which this represents an important shortcoming of equalization designs is not clear. Moreover, further 1 work along the lines undertaken in that paper would be welcome in respect of other potentially perverse incentives that are sometimes claimed in
Discussant Comments
135
respect of equalization policies (and that may be more significant in practice than the problems identified by PetcheyILevtchenkova), notably: The "rate tax-back" problem that may arise in cases where too detailed a breakdown of revenue categories results in a highly regionally concentrated revenue base - resulting in particularly strong disincentives for SNGs to raise tax revenues from such bases;3 The problem of tax base (or expenditure need) elasticity with respect to high or low tax (or expenditure) effort; and The moral hazard that may arise in cases where debt service expenses (a significant source of fiscal disparities among SNGs in some countries) are equalized - with the result that equalization policies may unduly soften the budget constraint facing sub-national governments. Other aspects of the theory of equalization incentives that could usefidly be developed further are the implications of the typically repeated nature of the equalization "game".4 The repeated nature of this "game", combined with the relentless pressures of electoral politics, may, for example, make sub-optimal Nash equilibria of the kind described in the PetcheyILevtchenkova paper less likely to occur in practice than might appear to be the case at first glance (i.e. where analysis is focused on a single-round "game") - and make them less stable when they do arise. A final, highly desirable "good governance" feature of an equalization regime is that it be as self-policing as possible. Fortunately, this is made easier by the fact that SNGs tend to be jealous of their respective shares of equalization transfers - particularly in cases where there is an exogenously fixed "pool" to be distributed among them - and are willing to do most of the policing themselves if all decisions relating to allocation are done in a fully transparent manner. In this vein, Australian experience suggests that allocation rules that reflect a "treatment by inclusion" of all sources of sub-national government revenues - including any transfers resulting from bilateral arrangements outside the equalization context per se - may be a particularly valuable technique to fbrther remove the temptation of (or at least the benefits from) investing effort in concluding "preferential side deals" with the national government. By taking into account all other transfers in determining how equalization payments are to be distributed, this technique significantly "hardens" the budget constraint faced by sub-national governments. All this, of course, presupposes an honest arbiter of the equalization "game". In particular, Australia's experience with equalization policies strongly suggests that there can be considerable advantages to having at least the measurement of disparities and the allocation rule decided by a body independent of the national government. It is less clear, however, whether
136
Pdoquin
this would hold equally true in other countries - particularly less developed and transitional countries - where administrative and analytical capacity may be limited and where the staff of an independent agency might best be employed running actual programs or the general apparatus of government. Moreover, an independent agency is only as good as its ability to remain uncormpt, pointing to the need for such a body itself to be regularly brought to account.
4.
CONCLUSIONS: SOME SOUND DESIGN PRINCIPLES FOR FISCAL EQUALIZATION?
To distil the essential messages of the papers in this section into four points, a reasonable conclusion to draw would be that - in countries where there are substantial fiscal disparities among autonomous sub-national governments: Equalization is a necessary - but not a suf$cient - condition for achieving either horizontal or fiscal equity; There is both good and bad news: the bad news is that countries pursuing decentralization strategies can probably never aspire to full equalization - but the good news is that they may not really need to do so; Equalization policies can help promote economic efficiency and good governance, providing they are well designed with these features in mind; and Even then, policy makers in decentralizing countries need to weigh the benefits of equalization against other policies that may also be efficiency- or governance-enhancing. As for the core of equalization design advice that can be derived from these papers, the main points to retain are that: Formula-driven equalization regimes are preferable to ad hoc funding arrangements. In particular:
o Multilateral arrangements are preferable to bilateral arrangements, though allowances may need to be made for the radically asymmetric situations of different classes of sub-national governments; and o Regardless of the approach taken, transparency and accountability are key.
Discussant Comments
137
As much as possible, the major features of an equalization program need to be exogenous to the actions of SNGs - hard as that may be to accomplish.
Notes The comments that follow are the views of the author and do not necessarily represent the views of the Government of Canada of the Dmartment of Finance. Here we are talking about the impact of equalization on economic convergence (e.g. in the marginal products of factors in different jurisdictions) - not the convergence in service delivery (the subject of the Hull/Searle paper) or in socio-economic outcomes (the subject of the Hofman/Brillantes paper). Similar perverse incentives may also exist on the expenditure side in cases where equalization regimes equalize both revenue capacity and expenditure needs. For example, equalization has been the focus of regular intergovernmental negotiations since 1933 in Australia and since 1958 in Canada, leaving ample time for a meta-game involving strategic cooperation to develop.
3 THE INSTITUTIONAL SETTING
Chapter 6 A FRAMEWORK FOR EVALUATING ALTERNATE INSTITUTIONAL ARRANGEMENTS FOR FISCAL EQUALIZATION TRANSFERS1 ANWAR SHAH The World Bank
1.
INTRODUCTION
Fiscal equalizations programs are fairly common features of intergovernmental fiscal relations in industrial countries. Some developing countries have also recently introduced these programs and still others are contemplating such programs. Institutional arrangements for fiscal equalization vary across countries with wide variations in the form and membership of the relevant decision making bodies. For ease of analysis, these diverse arrangements can be broadly classified into five stylized models: (1) a central government ministrylagency; (2) independent agency (grants commission) reporting either to the executive or the legislature on a permanent or periodic basis; (3) intergovernmental forums including intergovernmental cum civil society forums; (4) national legislature; and (5) sub-national government forums. These arrangements have not yet received the attention that is due in view of their importance in creating a credible and stable fiscal transfers regime and only a handful of recent papers (Searle, 2004, Boex and Martinez-Vazquez, 2004) have documented these arrangements and provided commentaries on alternate regimes. This paper carries this work hrther by providing a simple neo-institutional economics framework for assessing alternative institutional arrangements for their impacts on simplicity, transparency and objectivity of the equalization program as well as transaction costs for various parties involved. The paper is organized as follows. Section 2 briefly discusses the goals for intergovernmental fiscal relations. It further highlights variety of institutional arrangements pursued by various countries to further these
Shah
142
goals. Section 3 presents a simple framework to conduct a comparative evaluation of institutional arrangements. Section 4 presents a qualitative evaluation of alternate institutional arrangements with special emphasis on the Commonwealth Grants Commission of Australia. A concluding sections draws policy implications of this analysis.
2.
GOALS AND ALTERNATE MODELS OF INSTITUTIONAL ARRANGEMENTS FOR REVENUE SHARINGAND EQUALIZATION TRANSFERS
Institutional arrangements for fiscal transfers are typically structured to fulfill a number of objectives. These include both the process and program objectives. Program objectives seek to design a program that is consistent with general revenue sharing andlor equalization objectives. Further the design should be simple so that it is easily understood to forge a broader consensus, ownership and support. It should use uncontestable data and transfer funds in such a way to respect local autonomy while creating incentives environment that is compatible with accountability for results. These program objectives require a process of consultation with recipient governments that is open and transparent, conducive to consensus building and entails a relatively low cost of transactions for all parties concerned. The process should further aim to ensure that there is a wider public acceptance of the programs that are implemented. While the above objectives are commonly shared, specific institutional arrangements structured to fulfill those objectives vary widely across countries. Nevertheless, to simplify matters for analysis, in the following we present four stylized groupings of these arrangements. The fiRh grouping mentioned earlier namely sub-national government forums are typically formed to develop a common position on national fiscal transfer programs but do not assume decision making position on national transfers.
2.1
CentraVNational Government Agency Model
This is the most commonly practiced model in both industrial and developing countries. A central agency typically, either the president or prime minister's office or the ministry of finance or home affairs or local government or planning (including planning commission) assume sole or shared responsibilities for policy making and implementation of a system of fiscal transfers including equalization transfers. A few country examples of such arrangements are reported below:
A Frameworkfor Evaluating Alternate Institutional Arrangements
143
OfJice of the President: Kyrgyz Republic, Tanzania (Regional Administration and Local Government Unit) Ministry of Finance: China, Italy (policy only), Kazakhstan. Netherlands (shared with the Ministry of Home Affairs), Poland, Switzerland, Ukraine Ministry of Home Affairs/Interior: Italy (distribution of f h d s only), Netherlands (with Ministry of Finance), Philippines (Ministry of Interior and Local Government), South Korea (Ministry of Government Administration and Home Affairs) Ministry of Local Government: Ghana (Ministry of Local Government and Rural Development), Zambia Planning commission: India (for pladcapital grants) Ministry of Public Administration: Japan (Ministry of Finance is consulted).
2.2
National Legislature
In almost all countries with the single exception of China, national legislature must enact legislation to provide a legal basis for central-statelocal transfers. Brazil, however, represent a unique case where 1988 Constitution specifies the pool and the broad criteria for revenue sharing transfers and the upper house of the national parliament (the Senate) serves as the primary decision making body for specifics of the formula as well as monitoring compliance. The Senate regulations spell out the specific distribution criteria for the state and municipal participation f h d s (see Shah, 1991).
2.3
Intergovernmental Forums
Intergovernmental forums facilitate consultations among various orders of government and to strike a balance among competing interests and mediate conflicts. In view of this such institutional arrangements are commonplace in federal countries. In some countries such as Australia and the Republic of South Africa where an independent agency has been assigned a strong role in intergovernmental fiscal relations, intergovernmental forums are also established to review and decide on independent agency recommendations. Among the industrial countries Canada and Germany and among developing countries Indonesia, Nigeria and Pakistan rely solely on intergovernmental forums for decisions on fiscal transfers. The following paragraphs highlight salient features of these institutional arrangements.
Shah
2.3.1
Canada: Fiscal Arrangements Committee
The primary legal responsibility for the design of fiscal transfers rests with the federal government (Ministry of Finance) and final approval with the national parliament. The Federal Government of Canada, nevertheless, places a strong emphasis on intergovernmental consultation and shared decision making on intergovernmental fiscal transfers. Federal-provincial fiscal arrangement committees play a pivotal role in providing substance to such dialogues (see Figure 1). The Federal-Provincial Relations Division in the Ministry of Finance provides a secretariat for these committees. Fiscal Arrangements Committee comprises federal and provincial finance andlor treasury officials concerned with fiscal transfers. They meet periodically but exchange information and comments on a continuing basis on all technical aspects of fiscal arrangements. The recommendations of this subcommittee is fed to the Continuing Committee of Officials on Fiscal and Economic Matters comprising federal and provincial deputy ministers of finance (or treasurers). This Committee is chaired by The Federal Deputy Minister of Finance and usually meets on a quarterly basis. The final recommendations of this Committee for further action are forwarded to regular (typically semiannual) meetings of federal and provincial Ministers of Finance andlor Provincial Treasurers chaired by the Federal Minister of Finance. The fiscal equalization program is monitored and reviewed by these committees on a continuing basis with an intensive review every five years to suggest revisions for the enactment of new national legislation for the next five year period.
I
The Federal-Provincial Fiscal Arrangements
Tax I Harmonization Tax Collections Agreements Allocation Rules Reciprocal Taxation
I Fiscal Transfers CHST (Health and Social Transfers) Equalization C.A.P. Other
Figure 1(a).
I
Fiscal and Economic Policy F.M.C.'s Meetings of Finance Ministers Committees of Officials
A Framework for Evaluating Alternate Institutional Arrangements Federal and Provincial Ministers of Finance and Provincial Treasurers Chaired by the Federal Minister of Finance
Continuing Committee of Officials on Fiscal and Economic Matters Chaired by the Federal Deputy of Minister of Finance
Sub-Committee on Income Taxation
Working Group on Sales Tax
Fiscal Arrangements Sub-Committee
Committee on Equalization
Economic and Fiscal Data Sub-committee
Committee on Pensions
Figure 1(b). Federal-Provincial Fiscal Arrangements Committees 2.3.2
Germany: Financial Planning Council
The German federal system emphasizes sharing of responsibilities and joint decision making embodied in uniform federal legislation applicable to all Laender (states). The upper house of the parliament, Bundesrat, with representation from Laender governments, serves to strengthen a common approach. In fiscal relations, major decisions on the fraternal equalization transfers program are reached through a solidarity pact at a forum of federal and state leaders (presidents). Substantive inputs for reaching this pact come from the Financial Planning Council (Finanzplanungsrat) which establishes guidelines and recommendations for policy action on the financing of budgets in the short and medium term. The Council aims to reach agreement on fiscal policy coordination among federal and state governments. This Council comprises Federal Ministers of Finance and of Economics, the State Ministers responsible for Finance, four representatives of the municipalities (appointed by the Bundesrat based upon nominations by the municipal
Shah
146
associations). The Council is chaired by the Federal Minister of Finance (see Spahn, 2001) and is required to meet at least twice a year. 2.3.3
Indonesia: Regional Autonomy Advisory Board (DPOD)
The Regional Autonomy Advisory Board (DPOD) serves as important intergovernmental forum in support of the Law No.2211999 on Regional Governance and the Law No. 2511999 on the Fiscal Balance between the Central Government and the Regions. The Board advises the President on all aspects of local government organization and finance issues. The Board is chaired by Minister of Home Affairs with the Minister of Finance serving as the Deputy Chair. Other members of this Board include, Secretary of State, Minister of Administrative Reform, Minister of Defense, Chairman of the National Development Planning Board (BAPENNAS), two representatives each of the Provinces, Kota (districts) and Kabupaten (towns) and one representative each of the associations of Provinces, Kota and Kabupaten (see Searle, 2004 for details and a critique). Technical work on fiscal matters including fiscal equalization grants is conducted by the Directorate General for Center-Region Fiscal Balance of the Ministry of Finance. Work on planning grants is carried out by the National Planning Board. The DPOD reviews the recommendations of the Ministry of Finance and the National Planning Board and makes final decisions. Monitoring and implementation responsibilities lie with the Ministry of Home Affairs. 2.3.4
Nigeria: Revenue Mobilization, Allocation and Fiscal Commission
The 1999 Constitution mandated the creation of the Revenue Mobilization, Allocation and Fiscal Commission to administer the federalstate-local fiscal transfers as well as provide advice on revenue mobilization at state and local levels. The Commission is chaired by the Federal Minister of Finance and comprises of Finance Commissioners or Accountant Generals from each state. The Commission meets each month to review financial flows (see Boex and Martinez-Vazquez, 2004). 2.3.5
Pakistan: National Finance Commission
The establishment every five years of a limited duration National Finance Commission is mandated under article 160 of the 1973 Constitution. The Constitution empowers the Commission to make recommendation to the President on the pool of revenues to be distributed as well as the allocation criteria. In addition, the Commission advises on the exercise of the borrowing powers by all levels of government. The Commission is chaired
A Framework for Evaluating Alternate Institutional Arrangements
147
by the Federal Minister of Finance and comprises Provincial Ministers of Finance and other civil society members (e.g. legislators, scholars, experts, distinguished citizens etc) appointed by the President after consultation with provincial governors. The Federal Ministry of Finance serves as a secretariat for the Commission. The Commission makes its decision by consensus. If the Commission fails to reach a consensus on the formula for allocation of transfers as has been the case in recent years, then the formula that was operative in the previous five years is allowed to continue to operate until such time that a new consensus can be forged.
2.4
Independent Agency (Grants Commission) Model
An independent agency is created usually by the central government to report either to the executive or legislature on a permanent or periodic basis. Australia pioneered this model by creating the Commonwealth Grants Commission in 1933 to assess claims made by states for financial assistance (special grants) under section 96 of the Constitution. This model has since then been adopted in several other countries most notably India, the Republic of South Africa and Uganda. In view of the growing popularity of this model of transfers governance, brief remarks on a sample of such commissions are presented in the following paragraphs. 2.4.1
The Commonwealth Grants Commission (CGC) of Australia:
This commission was created in 1933 in response to dissatisfaction expressed by states especially a secession threat by Western Australia with the bilateral negotiations with the federal government on applications for special grants. The Commission in its 1936 report articulated that its assessment of the States' funding needs were to be based upon their capacity to raise revenue and any abnormal expenditure influences they faced. It stated that "special grants are justified when a state through financial stress from any cause is unable efficiently to discharge its functions as a member of the federation and should be determined by the amount of help found necessary to make it possible for that state by reasonable effort to function at a standard not appreciably below that of other states." (source Australian Treasury, 2004). The Commission's mandate was vastly expanded in 1973 (The CGC Act, 1973) when it assumed responsibility for calculating the per capita relativities for allocation of federal general revenue sharing assistance to all states and the Northern Territory and the Territory of Cocos (Keeling) Islands; financing of works and services in the capital; determination of state entitlements for local government as well as determination of state grants to local governments. In 1975, state commissions relieved the CGC of its role in determining state grants to local governments. The determination of state
148
Shah
entitlements for local governments was terminated by the Local Government Financial Assistance Act, 1986.The special grant program for selected states under section 9 of the Constitution was terminated in 1981182 and replaced by program of assistance for all states and calculation of state relativities for general revenue grants that includes tax sharing, health and special grants on a five year basis with annual updates. The Commission consists of a chairperson and a maximum of five members appointed by the federal government in consultation with the states. The Commission has a permanent secretariat of about 60 (as of October 2004) staff members. The day-to day working of the Commission is handled by a Secretary and two assistant secretaries responsible for Expenditure Analysis Branch and Revenue, Budgets and Research Branch respectively. The Commission is constituted as an advisory body and empowered to conduct it business only within the purview of the terms of references provided by the Federal Minister of Finance and Administration. It does not have the powers to initiate and pursue inquiries on its own authority. In recent years, the main references have sought Commission's advice on per capita relativities for distributing, among the States and Territories, the pool of general revenue assistance made available by the Commonwealth. For this purpose, the Commission in 2004 applied a specific principle of fiscal equalization, which says that State governments should receive funding from the pool of goods and services tax revenue and health care grants such that, if each made the same effort to raise revenue from its own sources and operated at the same level of efficiency, each would have the capacity to provide services at the same standard. (Commonwealth Grants Commission, 2004, p.x) Another important matter on which the Commission has reported in recent years is the interstate distribution of general purpose grants for local government. Although the references are provided by the Minister for Finance and Administration, their content is usually decided in negotiations between the Commonwealth and the States, conducted largely through their treasuries. A formal mechanism for this purpose is the Heads of the Australian Treasuries (HOTS) Forum which meets periodically. While the resulting Commission reports are provided formally to the Commonwealth Government, they are made available to the states immediately thereafter. The relativities recommended in those reports are considered at the Annual Treasurers' Conference. The Commission's relativities are almost always accepted by the Treasurers' Conference as preliminary relativities are publicly defended by the Commission in open adversarial proceedings in all
A Frameworkfor Evaluating Alternate Institutional Arrangements
149
states prior to their formal presentation. Only in 1981 (the Commission was asked to re-do the relativities and present a new report in 1982) and in 1982 the Commonwealth Government chose not to accept the Commission's recommendation but instead the Premiers' Council under the leadership of Prime Minister J.M. Fraser (Commonwealth Treasurer, J.M. Howard) adopted relativities different from those recommended by the Commission for the years 1982-1984 (see Commonwealth of Australia, 1995, pp.137158). 2.4.2
The Finance Commissions of India
The Finance Commissions of India comprising a chair and four members are constituted by the President every five years to meet the constitutional requirement (under Article 280 of the Constitution of India, 1949 as amended in 1992 and 1993) to redress the fiscal gaps in the revenues and expenditures of the Union (federal) and State governments arising out of a mismatch of revenue means and expenditure needs at various levels. The Finance Commission is mandated to make recommendations to the President as to: a) The distribution between the Union and the States of net proceeds of taxes which are to be, or may be, divided between them and the allocation between the states of the respective shares of such proceeds; b) The principles which should govern the grants-in-aid of the revenues of the States out of the Consolidated Fund of India; c) The measures needed to augment the Consolidated Fund of a State to supplement the resources of the Panchayats in the State on the basis of the recommendations made by the Finance Commission of the State; d) The measures needed to augment the Consolidated Fund of a State to supplement the resources of the Municipalities in the State on the basis of the recommendations made by the Finance Commission of the State; and e) Any other matter referred to the Commission by the President in the interests of sound finance. achieve this through revenue sharing and special grants to needy states. The Commission is also required to recommend allocation among states of their share of federal taxes. The first Finance Commission was established in 1951 by an act of parliament, the Finance Commission (Miscellaneous Provisions) Act, 1951 (Act XXXIII of 1951). Subsequently, these commissions have been reconstituted every five years with new terms of reference for the next quinquenniel period. According to the 1951 Act, the chairperson of the Commission must have experience in public affairs members may be
Shah
150
selected from among persons who: (a) are, or have been, or qualified to be appointed as judges of a High Court; or (b) have special knowledge of the Finances and Accounts of the Government; (c) have had wide experience in financial matters and in administration; or (d) have special knowledge of economics. The Commission members are usually a mix of politicians, retired civil servants and experts in fiscal federalism. Each Commission creates a temporary secretariat managed by a secretary appointed by the federal government usually from the Planning Commission. The Commission is disbanded upon submission of a report consistent with its terms of reference. The Commission does not have the mandate to initiate an inquiry outside its terms of references. The Commission's recommendations are not binding upon the government but under article 281 of the Constitution, these must be presented to both houses of the parliament along with government response to each recommendation. 2.4.3
Fiscal and Financial Commission (FFC) of the Republic of South Africa
The FFC was established in 1993 under section 198 of the 1993 interim Constitution. The commission was to have 18 members appointed by the President with nine members designated one each by nine provincial cabinets. The interim Constitution gave a broad mandate to the Commission in providing advice on financial and fiscal requirements of the national provincial and local governments (section 199(l)(a)). The Final Constitution of 1996 expanded the Commission membership to 22 by adding two representatives from the organized local government structure and two additional presidential appointments. Such a large membership was subsequently seen as unwieldy and an amendment to the Constitution carried out in 2001 (The Constitution of the Republic of South Africa Second Amendment Act, 2001, section 7) reduced the Commission membership to the current strength of nine to be appointed by the President in consultation with the Cabinet and Executive Councils of the nine provinces as follows: A chairperson and a Deputy Chairperson Three members recommended by provincial premiers; Two members recommended by local governments; and Two other members. The 1996 Constitution (section 214(1)) narrowed Commission's mandate to provide advice on equitable allocation of central revenue sharing to provincial and local governments; provincial taxation; municipal fiscal powers and function; sub-national borrowing and central government guarantees (sections 21 8(2), 228(2)(b), 229(5) and 230(2)). The role of the
A Frameworkfor Evaluating Alternate Institutional Arrangements
151
Commission was further clarified by Central legislation. The Borrowing Powers of Provincial Governments Act, 1966 authorized the Minister of Financed to seek Commission's advice on provincial borrowing and debt management issues. The Provincial Tax Regulation Process Act, 2001 empowers the Commission to provide comments on tax proposals by the provinces. The Intergovernmental Fiscal Relations Act, 1997 clarified the institutional arrangements and the processes for the Commissions' advice to executive and legislative organs. The Commission was given an observer status at the Budget Council, a forum of the Ministers of Finance of Center and the Provinces. Further the Act requires the Commission to provide advice on equitable shares at least 10 months prior to the commencement of the fiscal year and the Division of Revenue Bill must include comments by the national government on the Commission's recommendation. Overall the Commission enjoys a strong constitutional-legal foundation to play role of an influential advisor on intergovernmental fiscal relations. This role was carefully crafted to ensure that "it can bark, but not bite" (Wehner, 2003, p.5). 2.4.4
Uganda: Local Government Finance Commission
The Local Government Finance Commission of Uganda is mandated under the 1995 Constitution (article l94(l,4)) and the Local Government Finance Act, 2003 (section 9) to serve as an advisory body to national government (Minister of Local Government) on all matters relating to transfer of resources to local government and to advise local governments on appropriate levels of local revenues. It is expected to recommend both the total pool of transfers as well as allocations to local governments in the form of equalization and conditional grants. In addition, it monitors compliance of local governments with legal requirements associated with their taxing and spending decisions. It is also empowered to mediate financial disputes among local governments. The Commission consists of seven commissioners appointed by the President as follows: Three commissioners nominated by the district councils through the Uganda Local Authorities Association; One commissioner nominated by the Urban Councils through Urban Authorities Association of Uganda; Three commissioners nominated by the Minister of Local Government in consultation with the Minister of Finance, Planning and Economic Development.
Shah
152
The President designates two of the commissioners as Chairperson and Deputy Chairperson who work on a full time basis. The rest of the commissioners serve on a part-time basis only. A permanent secretariat headed by a secretary with 31 staff conducts the day-to-day business of the Commission.
3.
A NEW INSTITUTIONAL ECONOMICS VIE) FRAMEWORK FOR EVALUATING ALTERNATIVE INSTITUTIONAL ARRANGEMENTS FOR EQUALIZATION TRANSFERS
The previous section noted diverse institutional arrangements are used for decision making on central-state-local transfers. There is, however, no framework available in the literature to provide a comparative evaluation of these arrangements to guide future reform efforts. This section attempts to fill this void by borrowing ideas and concepts from a relatively new discipline of neo-institutional economics (see North, 1990). Under the NIE framework both the principals and their agents act rationally in their own self-interest. The access to information is costly and not uniformly available to all. In such circumstances, the agent may not secure the interests of their principals and the principals may not be able to restrain opportunistic behaviors of their agents due to the "bounded rationality" of principals and high transactions costs associated in overcoming this handicap. In the context of institutional arrangements for a fiscal equalization program, the problem manifests itself as follows. First, there needs to be national compact on equalization principles and standards so that there is a clear view on the mandate given by principals (citizens). This compact can take the form a constitutional provision, legislative enactment or an informal but universally shared consensus on the goals of such a program. This compact, however, will have to be administered by various public agents e.g., executive and legislative organs typically at the national level.. Such administration may entail commitment problems where it may be in the self interest of some agents not to follow the compact. For example, the national executive or legislative leadership may come from a region with little enthusiasm for interstate equity. Alternately current regime is committed but unable to tie the hands of future regimes and thereby durability of the compact is threatened. Enshrining of equalization principles in the constitution is often motivated by these considerations. But constitutional enshrinement limits the commitment problem but does not overcome it as
A Frameworkfor Evaluating Alternate Institutional Arrangements
153
current coalitions can be replaced by coalitions of opposing interests and policy preferences in the future2. Institutional arrangements for administering the compact also entail a number of transactions costs for principals and their agents. For principals various types of arrangements impose differential participation and monitoring costs (PMC). There are also costs associated with legislative (LC) and executive decision making (EDMC). These costs are the time and effort needed to strike a legislative compromise or an executive decision. They are higher when stakes for individual parties are high and when there are strong conflicts of interest and when there is some uncertainty as to the future revenue streams available to donor and recipient governments - a frequently recurring situation in negotiations on fiscal transfers. All institutional arrangements entail costs incurred by principals to induce compliance by their administrative agents with the compact, the so-called agency costs (AC). Agency costs arise as the administrative agents to implement the compact on behalf of the principals may not share the objectives pursued by the principals. They may undertake decisions to serve narrow self interest of bureaucratic power or personal enrichment and because of the high transactions costs, civil society or the legislature may not be able to exercise effective oversight on these decisions. In view of the difficulty of monitoring and taking corrective action ex-post, legislatures typically try to influence the appointment of executives so that they share the same goal and would not undermine the enacted legislation. In addition, they rely much more on civil society monitoring and respond to "fire alarms" raised by unhappy constituents (Horn, 1997, p.21). But such response to fire alarms may be constrained if the executive agency is given a significant degree of autonom?. Finally, there also risks and uncertainty costs (UC) associated with unstable regimes. Risks and uncertainty in fiscal relations, arises because both the potential benefits and costs of a given compact may not be fully known at the time a deal is struck. In the future, such a deal may be undone by a new coalition and constellation of interests. The analytical framework described above argues for instituting administrative arrangements and governance structures that (a) facilitate greater access to information by citizens, interested sectors of civil society (including the media and academics) and legislators that would enable them to better hold those agents involved in equalization decision-making to account; (b) minimize transaction costs associated with participation, monitoring and decision making, agency costs (costs incurred by principals and other agents to ensure that agents involved in equalization decisionmaking act faithfully to serve their mandates) and uncertainty costs; and (c) create an incentive structure for both the legislative and administrative agents to comply with their compact with the principals.
154
Shah
This is a rather complex task because of interdependencies associated with various actions. For example, as pointed out by Horn (1997, p.24), attempt to reduce agency loss between citizens and legislatures through restraining the influence of legislatures on the executive may potentially increase agency losses between legislatures and government executives. There are further difficulties in ensuring durability of legislation which can be undermined through lack of effective enforcement even if the legislation remains unchanged.
4.
COMPARATIVE EVALUATION OF ALTERNATE INSTITUTIONAL ARRANGEMENTS USING NIE FRAMEWORK
An earlier section highlighted four stylized types of institutional arrangements. A central government agency model represents one of the most prevalent arrangements. A national legislature model represents the least common of these arrangements. The first arrangement rests decision making solely in the hands of a central government agency and the second arrangement involves the legislature not just in legislation but also in executive decision making role. Other interesting options for institutional arrangements are represented by intergovernmental forums and independent agency models. In the following analytical comparisons of the two common options are made using the NIE framework. It must be noted at the outset that these two options are not necessarily exclusive choices and both arrangements can co-exist, but when they do, incremental value added offered by the independent agency must be rigorously examined. Intergovernmental Forums: An intergovernmental forum provides a framework for institutionalized but restricted political bargaining4. Bargaining is restricted as the constitution and the legal framework usually define the limits to such bargaining. There is, however, strong peer pressure to strike a bargain. Thus intergovernmental forums are usually successful in defining an explicit political compact acceptable to all parties. As such a political compact cannot be easily reached when complex criteria are put on the table, this institutional model places great premium on simplicity and "rough justice" as opposed to complex but precise justice. Conflicting interests are represented at these forums. Unless the discussions of the forum are conducted in camera, political grandstanding may prevent political compromises. Durability of such compromises is usually assured as all parties stand to loose from a deal that is unraveled. Blame shifting is also not possible as the members of the forum assume full responsibility for their decisions. The forum further enables participating governments representing
A Frameworkfor Evaluating Alternate Institutional Arrangements
155
competing interests and varying commitments on equalization, to reach a broader consensus. Independent Agency (grants commission) Model: An independent agency is usually established to seek an independent, professional, transparent and rigorous view of a complex task of developing recommendations on the determination of the pool, the allocation criteria and distribution of funds among recipient governments. The presumption here is that if such a decision is divorced from politics, the resulting criteria and the associated distribution would serve the broader interests of the nation as well as its constituent units better. These theoretical advantages are rarely achieved in practice. First, decisions on the standard of equalization e.g., the minimum level of per capita fiscal capacity to which all jurisdictions are entitled to be brought up to, cannot and should not be divorced from politics. Secondly, such an institutional arrangement creates a number of agency problems as discussed below. Mission creep: To secure its long term existence and enlarge its spheres of influence, an independent agency faces continuous imperatives reinterpreting its terms of reference to enlarge the scope of its activities. Such a mission creep goes unchecked as the politician do not want to be seen curtailing the search of such agencies for a holy grail - the ultimate formula for equitable distribution of federal funds. Incentives for complexity: An independent agency faces powerfbl incentives to seek ever more complex solutions to simple questions. This is because complexity and associated expertise fuels demand in the external market for professionals serving these agencies. The greater the complexity of formulae and associated calculations, greater is the premium placed by the market on professional possessing those skills. Interested parties submissions make it politically imperative to accommodate ever growing complexity. Outside academic experts typically clamor for further complexity to achieve more precise justice. There is no escape from this circle as part time or term employment of members of the commission limits the oversight provided by them. It takes sometime for term members to grasp the complexity of the allocation rules and by the time they can form their own judgment on their relative merits, it is usually time for them to say good bye. In any case, the staff would be resistant to any simplification and those recipient governments who benefit from the complexity and associated inequities of the system will likely block any reforms. Independent think tanks and researchers may even call for greater complexity to bring practice in conformity with the theory. In conclusion, constraining influences to keep the system simple and easily comprehensible are stunted by the very existence of an independent agency. Fire alarm oversight impractical and costly: Citizen oversight of such independent agencies becomes infeasible for several reasons. First, more
Shah
complex the distribution criteria suggested by the agency, the more difficult it is for individual citizens and civil society groups to make informed comments. Further conflicting representation by various citizen groups trying to secure local interests strengthens broad discretion granted to such agencies in the interest of a scientific a-political approach. Even "fire alarm7' oversight sought by legislatures becomes too costly and impractical as unhappy constituents make conflicting demands on their representatives.
4.1
Tentative Conclusions on a Comparative Evaluation of Intergovernmental Forum vs. Independent Agency Model
The above discussion suggests that the ultimate decision on relative merits of each institutional arrangements must be guided by an analysis of the incentive regime created by each institutional set-up and associated agency costs and their success in achieving simple, equitable and durable outcomes. Table 1 presents a comparative NIE perspective on these arrangements. The NIE framework predicts that overall transactions costs are expected to be higher and potential outcomes less desirable under an independent agency model as compared to an intergovernmental forum. These results stem from the fact that independence and autonomy offered to grant commissions weaken citizen oversight. Their drive for optimal (ideal) systems invites complexity and undermine transparency and accountability. Thus participation and monitoring costs as well agency costs rise. Intergovernmental forums on the other hand look for simple and feasible alternatives to strike a political bargain and thereby reduce transactions costs for the nation as whole. Higher transactions costs under independent grants commission nevertheless are not expected to secure better outcomes. On the other hand grants commission processes do not necessarily encourage consensus building that is achieved by forging a political compact on equalization standard. In the absence of a political compact on the equalization standard both the pool and allocation among constituent units, are determined quite independently of that equalization standard. Stability of allocation criteria is also not assured by a grants commission as the desire for perfection may lead to frequent changes in the methodology. In summary, the independent grant commission is a poor substitute for an intergovernmental forum. Its usefulness as a complementary institution to an intergovernmental forum is also quite limited in view of high agency costs and its pre-disposition towards optimal as opposed to feasible reforms.
A Frameworkfor Evaluating Alternate Institutional Arrangements
157
Table I. Comparative Conceptual Evaluation of Intergovernmental Forum (IGF) vs. Independent Agency (IA) (Grant Commission)
Transactions costs (overall): Participation and monitoring costs Legislative and executive decision making costs Agency costs Uncertainty costs Potential outcomes Political compact on equalization standard Durability of political compact Pool determined by equalization standard Allocation determined by equalization standard Stability of allocation criteria
4.2
IGF
IA
Low to medium
Low to high
High
High
Low Low
High Medium
Yes
No
Yes Yes/No
NA No
Yes
No
Yes
May be
Why Then Are Independent Agencies (Grant Commissions) so Popular?
The NIE framework suggests that independent grant commission may not represent a better institutional choice in view of the incentives regimes created by the underlying structures. These conclusions run counter to predominant view in the fiscal federalism literature that such institutions personify "best practices" in institutional arrangements. In fact, it is quite commonplace for international development agencies and leading consultants to recommend establishment of such commissions in developing and transition economies (see Searle, 2004, Boex and Martinez-Vazquez, 2004). For a NIE analyst such popularity is not surprising if one looks at the incentive structure and culture of these institutions. Independent agencies find strong support among academic scholars, think tanks and politicians by playing to the enlightened self-interest of these groups. These agencies cater to "elites" especially academic elites as they give them a forum for dissemination of their research and scholarly works. The agencies are perceived to be a-political pursuing technical excellence. Further they support the consulting industry by seeking their advice and analysis. These agencies serve as convenient tools for national and regional politicians as they are seen to be providing fair, balanced and professionally rigorous
Shah
158
analysis. For any unpopular distribution criteria, politicians have the ability to distance themselves from the analysis and instead shift blame on the agency. Further, they can avoid taking hard decisions and simply accept the agency's view as a "take it or leave it" proposition. No wonder one finds a growing chorus of professional and political views advocating independent agency approach to vital decisions on equalization transfers.
5.
FROM THEORY TO PRACTICE: DO NIE PREDICTIONS HOLD WATER?
The previous section presented in abstract a comparative analytical perspective on the working of intergovernmental forums and independent grants commissions. This section explores the same concepts based upon experiences of Canada and Germany with intergovernmental forums and of Australia and India with independent grant commissions. Note that we have abstracted from the complexity in Australia that the independent grant commission works as a complement to the intergovernmental forum (HOTS Forum). This abstraction should not bias our analysis as the recent history of the CGC in Australia demonstrates that it has enjoyed significant independence and autonomy and its recommendations have almost always been accepted by the Federal Cabinet. Table 2 reflects upon comparative experiences using NIE framework. Briefly the following lessons can be drawn from these experiences: Transactions Costs: The institutional arrangements in the four case study countries incur differential citizen participation and monitoring costs, agency costs and uncertainty costs. Intergovernmental forums typically lead to lower transactions costs for the principals (citizens) primarily due to greater transparency, simplicity and media and civil society scrutiny of programs. Agency costs are highest under the Australian program due to greater autonomy and incentives for complexity and mission creep faced by the CGC staff. The periodic grants commission in India has medium agency costs as it is constrained by its limited duration tenure. Legislative and executive decision making costs are very similar across case study countries. Overall intergovernmental forums appear to offer less costly alternatives for the principals to induce compliance from their agents. Outcomes: Program outcomes are to be judged for clarity of mandate by the principals, durability of political consensus and for simplicity and equity of the equalization transfers programs. Equalization programs in Canada and Germany are enshrined in their constitutions. The Australian program is mandated by federal law and the Indian program is primarily concerned with equitable distribution of federal revenue sharing pool and has no explicit equalization objective.
A Framework for Evaluating Alternate Institutional Arrangements
Table 2. Equalization Programs - Comparative Experiences with Intergovernmental Forum (IGF) vs. Independent Agency (Grant Commission) (IA) IGF Germany
IA Australia
LOW
Medium
High
High
Low Medium
Low Medium
Low Medium
Low Medium
Low Low
Low Low
High Medium
Medium Medium
Yes, Const. Yes Const.
Yes, Const.
Yes, Fed. Law Yes No
IGF Canada
Transactions costs (overall): Citizen participation and monitoring costs Legislative costs Executive decision making costs Agency cost Uncertainty costs Outcomes Political consensus on equalization Durability of consensus Political compact on equalization standard
Yes Solidarity pact
IA India
Type of equalization program
Paternal
Fraternal
Paternal
Pool determined by equalization standard
Yes
Yes
No
Allocation determined by equalization standard Fiscal capacity equalization Fiscal need equalization
Yes
Yes
Yes, RTS
Yes, Act. Revenues No
No but formula Yes, RTS
Stability of allocation criteria
Yes
Yes
No
Sunset clause Dispute resolution
Yes Supreme Court
Const. Court
Supreme Court
Supreme Court
Program Equity Program Complexity
Yes Low
Yes Low
May be? High
May be? High
Source: Author's perspectives
Yes
Paternal
Yes, Some
Shah
160
There is a reasonable degree of political consensus on the principles of equalization in Canada, Germany and Australia. No such consensus has yet emerged in India. What distinguishes Canadian and German programs from those of Australia and India are clarity of the equalization standard and simplicity in implementing it. Needless to say both these programs have a number of shortcomings (see Shah, 2003). The Australian and Indian programs lack clarity in equalization standard. Australia uses a comprehensive program attempting to equalize fiscal capacity as well as fiscal needs requiring highly complex calculations. Massive amounts of data are analyzed to calculate revenue disability for 18 tax bases and expenditure disabilities for 40 programs with countless relevant determinants. The procedures used are so complex that the Australian program is a black box even for a serious student. The program thrust is on absolute comparability of services across states and territories. This focus diverts states' energies to proving that "they need more to do less" as opposed to "doing more with less". While the overall approach to expenditure needs is sound and defensible, the pursuit of idealism by the CGC and constant refinements lead to super complexity and nontransparency. For highly correlated factors disabilities are artificially magnified through double counting and multiplication. Under such a program use of judgment on factors and weights is inevitable but such judgments invite controversy and compromise the credibility of the whole program. When all is said and done the results are often disappointing. As the program lacks an explicit equalization standard, its generosity is overwhelming for its Northern Territory , Tasmania and South Australia and the program is punitive for Victoria and New South Wales. The program, however is not equitable and grant allocations vary directly with most macro fiscal capacity indicators (see Shah, 2004, for a detailed critique of the Australian Program and suggestions for simplification). The Indian formula is less complex but uses arbitrary factors and weights. Curiously enough all recent commissions have insisted on using 1971 state population figures for calculation of grant shares. The rationale presented is that India adopted a population control policy in that year. This is not defensible as the state population has experienced major changes due to migration.
CONCLUDING REMARKS Comparing institutional arrangements across different countries is a daunting task. The success of these arrangements depends upon a multitude of factors. The success of governance structures for fiscal matters may depend not only on the incentives regime associated with their inner
A Frameworkfor Evaluating Alternate Institutional Arrangements
161
structures but also their interactions with other formal and informal institutions in the country. This paper presented a simple framework to understand these incentives and interactions and draw implications for their impacts on transactions costs for the society as a whole and achievement of societal objectives. An application of these concepts to the specific case of institutional arrangements for fiscal equalization transfers was carried out and the predictions based upon the theory were compared to observed experiences in major federal countries. The paper demonstrates that the simple new institutional framework presented here has a significant power for predicting potential impacts. The paper concludes, both in theory and practice, that the case for independent grants commission to enhance the transparency, equity and accountability of the intergovernmental finance system is vastly exaggerated.
Notes This paper draws heavily upon Boadway and Shah (forthcoming a,b). Earlier versions of this paper were presented at the meetings of the Government of Canada Experts' Panel on Fiscal Equalization, Montreal, Canada, September 1, 2005. The views expressed in this paper are those of the author alone and should not be attributed to the World Bank. The author is grateful to Robin Boadway, Roy Bahl, Richard Bird, Jameson Boex, Fred Gorbet, Jorge Martinez-Vazquez, David Pdoquin, Bob Searle for comments. It should be noted that the cast of "agents of the citizenry" is potentially much broader than just the "executive and legislative organs" of the central government. These agents through intergovernmental competition - especially among stateslprovinces - and of players from civil society, can help ensure transparent self-regulation of any equalization governance regime through a system of checks and balances that minimizes the risk of "capture" and the resulting domination of a narrow set of more-or-less private interests in the governance regime. In his comments on this paper, David PBoquin has argued that it is not immediately clear who the "unhappy constituents" may be in the equalization context: presumably, it is provinciallstate governments who are most likely to first raise "fire alarms" and then go about drumming up secondary alarms on the part of local civil society actors and the local citizenry (as recent Canadian experience demonstrates only too well...). Since provinceslstates are the very same prime stakeholders who would be the main clients and interveners of any autonomous grants agency, it is not clear that the latter could in any sense be indifferent to them, given the quite credible threat of "going public" and appreciably raising the political and electoral stakes when they suspect their interests are not being given fair consideration. Note these restrictions should not be such as to reduce political bargaining to a zero sum game as under such a scenario, benefits of a federal bargain will be significantly curtailed.
Shah
References Boadway, Robin and Anwar Shah. Forthcoming (a). Fiscal Federalism: Principles and Practices. New York and London: Cambridge University Press Boadway, Robin and Anwar Shah, (eds). Forthcoming (b). Intergovernmental Fiscal Transfers: Principles and Practice. Washington, DC: World Bank. Boex, James and Jorge Martinez-Vazquez. 2004. Developing the institutional framework for intergovernmental fiscal relations in decentralizing LDCs. International Studies Program Working Paper No. 04-02, Georgia State University, Atlanta, Georgia. Commonwealth of Australia. 1995. Equality in Diversity: History of the Commonwealth Grants Commission. Second Edition. Canberra: Australian Government Printing Service. Government of India, 1 2 ' ~Finance Commission. 2003. Fifty Years of Fiscal Federalism Finance Commissions of India. New Delhi: Government of India. Horn, Murray J. 1997. The Political Economy of Public Administration. New York, USA: Cambridge University Press. North, Douglas (1990). Institutions, Institutional Change and Economic Per$ormance. Cambridge, UK: Cambridge University Press. PBloquin, David. 2005. Backgrounder on Equalization and TFF Governance Issues. Department of Finance, Canada, Ottawa. Searle, Bob. 2004. Institutional Aspects of the Balancing Fund, Ministry of Finance, Indonesia, December 2004. Shah, Anwar. 2004. The Australian horizontal fiscal equalization program in the international context. Paper present at the Heads of the Australian Tresuries (HOTS) Forum, September 22, and at the Commonwealth Grants Commission, Canberra, Australia, September 23. Shah, Anwar. 2003. Lessons from The Theory and Practice of Intergovernmental Transfers. Unpublished paper, Washington, DC.: World Bank. Shah, Anwar. 1991. The New Fiscal Federalism in Brazil. World Bank Discussion Paper No. 124. Washington, DC.: World Bank Uganda, the Republic of. 2004. Local Government Finance Commission, Corporate Strategy, 2004-2008. Kampala, Uganda. Wehner, Joachim. 2003. The Institutional Politics of Revenue Sharing in South Africa. Regional and Federal Studies, 13(1): 1-30.
Chapter 7
INTERGOVERNMENTAL TRANSFERS: THE FUNDING RULE AND MECHANISMS PAUL BERND SPAHN Goethe University, Frankjiurt
1.
INTRODUCTION
No federal or decentralized system of government can function without intergovernmental transfers. Once responsibilities are assigned to the different public agencies both vertically and horizontally, a typical outcome is vertical and horizontal fiscal imbalances. It means that the assignment of revenue does not fully match the expenditure h c t i o n s for each public agency. It requires interagency transfers to allow all jurisdictions to carry out their respective public responsibilities. Intergovernmental transfers are understood in a broad sense at this point. They include direct payments, the sharing of revenues, and non-monetary transactions such as complimentary service delivery among agencies (e.g. administration) or the waiver of intergovernmental commitments. Revenue sharing, in particular the sharing of taxes, is often indistinguishable from grants and therefore deserves special attention in this context. Yet it must be clearly set apart from the sharing of tax bases. Base sharing resembles a revenue assignment that conveys rights, and hence legal entitlements, to exploit a tax base conjointly among governments. The advantage of base sharing over tax sharing lies in greater fiscal autonomy because different agencies can impose different tax rates on the shared base. Where legal entitlements from base sharing also exist for the sharing of the proceeds from taxation, this must be considered tax sharing conjointly with base sharing. In this instance tax assignment and intergovernmental transfers are necessarily blurred. Base sharing is not dealt with in this paper because the emphasis is on transfers, not tax assignment.' Tax sharing in the sense of sharing the proceeds from taxation will be discussed as a special transfer regime. It is often used to allocate h d s from the national level to subnational tiers of
164
Spahn
government, although the converse can also be true. However, tax sharing appears to require a nationally uniform tax rate to make sense, so tax sharing is conventional for taxes under the control of the national government. In classifying and analyzing intergovernmental transfers one must ask: What is the purpose of such transfers? How should the transfers be funded? and What mechanisms are commendable to achieve the objectives of a transfer? I shall analyze these questions in the following chapters in an abstract way as if one could design a transfer system from scratch. This is of course usually unrealistic as all decentralized or federal governments will have to respect their historical and political conditions, so reforms will be highly path-dependent. Yet speaking in abstract terms can raise the awareness on possible deficiencies of existing arrangements and serves as a benchmark for potential reforms. Furthermore, I try to include some international experience to illustrate the great variety and richness of institutional arrangements that have been adopted in rebalancing intergovernmental fiscal relations in different parts of the world. Finally, I sketch an incentive-neutral transfer scheme for a fictional country, Krakozhia, which aims at introducing a new scheme for its general purpose transfers using a methodology in the spirit of the mechanism developed by the Commonwealth Grants Commission of Australia.
2.
THE PURPOSE OF INTERGOVERNMENTAL TRANSFERS
It is usehl to distinguish the following types of transfers according to different purposes: (i) general transfers; (ii) specific transfers; and (iii) special transfers. The nomenclature may vary, especially as the purpose of these transfers is often not clear or mixed, but essentially these categories are found in all decentralized governments of the world. General transfers. These serve to provide general revenue to an agency's budget to fund basic operations where own revenue would not be sufficient to hlfill the agency's responsibilities. General transfers or grants are thus used to redress existing vertical and horizontal fiscal imbalances that may exist between levels of government or among agencies at any one level of government. General grants are unconditional general purpose grants to balance the budgets of autonomous or quasi-autonomous public agencies.
Intergovernmental Transfers: The Funding Rule and Mechanisms
165
Most frequently, they fund current budgets and are derived from current budgets. Providing general transfers does not mean open-end financing. There is an overall macro constraint on all public budgets that must be respected, so the resource pool is ideally well determined and fixed. Within this overall constraint those agencies with relatively better fiscal positions transfer funds (either explicitly or implicitly) to those whose fiscal position is comparably weak. Control on the spending of such transfers is carried out by the constituents of the recipient agency. General grants are often used to redress vertical and horizontal fiscal imbalances simultaneously, which introduces asymmetries to the transfer system. However, it is useful to distinguish these processes carefully to better understand the mechanics of the system. An ideal allocation formula for general transfers would always specify the vertical component first; it serves to rebalance the consolidated resources and spending needs for all agencies at any one level of government. In a second step one would specify the rules governing the interagency reallocation of resources if there is want for some horizontal equity. So for general transfers I discern two separate goals: vertical fiscal rebalancing; and horizontal equalization. As said, vertical rebalancing and horizontal equalization of budgets is often effected simultaneously through vertically asymmetric grants. An exception is Germany where vertical fiscal balance is taken up first, and horizontal equalization is addressed in a second step. A similar approach is found in Denmark for inter-municipal equalization. Instead of a single, vertically asymmetric system of transfers there are two separate processes in this case, which requires the second step, horizontal redistribution among agencies at a particular level, to be arranged explicitly in a "brotherly" fashion. Such horizontal general grants are pure equalization transfers. Politically, it is often more convenient to start from a situation of vertical fiscal imbalance and engineer interagency equalization implicitly through a vertically asymmetric allocation of funds using a well defined formula. This is common practice in most parts of the world. It avoids direct rivalry of jurisdictions at a given level of government and it conveys some extra power to the senior government. Figure 1 illustrates different forms of allocating general grants, whereby Australia and the European Union represent single stage vertically asymmetric transfer systems while Germany's equalization starts from a position of relative vertical fiscal balance. It is also obvious that senior government does not necessarily mean donor government as is clear from the case of the EU. Vertical fiscal rebalancing does not necessarily entail direct transfers between layers of government. On the contrary: In most instances it is effected through revenue or tax sharing. Indeed, tax sharing is ideal for
166
Spahn
implicitly transferring resources between levels of government, but it does not equalize among agencies at a given level. The revenue returned to, or retained by, an agency is usually proportional to the revenue collected in this agency's jurisdiction. The allocation of resources follows the principle of derivation or origin. This could be considered advantageous as it allows to clearly distinguish between vertical and horizontal redistribution via transfers. As said, Germany makes use of this: first, vertical fiscal balance between the federal and the states' budgets is attained through variations of the VAT shares going to each level of government; and horizontal equalization is carried out by the interstate equalization scheme mentioned above.2 Bosnia and Herzegovina is a federation that relies exclusively on tax assignment and tax sharing rules, and there is no formal interjurisdictional equalization.3
Approaches to regional equalizatio~ Australia
'poor'
-
EU
Germany
'!lch
Figure I . Examples of Fiscal Equalization among Regional Administrative Bodies The redressing of vertical fiscal imbalances raises the question which resources or taxes should be used in these instances. This will be dealt with more generally below when discussing the funding rules. Problems of specifying equalization transfers, respectively the elements of an implicit allocation formula, are addressed when sketching the mechanics of intergovernmental transfers.
Intergovernmental Transfers: The Funding Rule and Mechanisms
167
Specijk transfers. Specific transfers are paid so that one public agency provides particular services on legal, bureaucratic, or contractual grounds, as required by another (funds-providing) agency. As such, these transfers are generally for current services. I have argued elsewhere that contractual arrangements offer particular benefits in that they are more flexible and adaptable than legal or bureaucratic rules.4 They also enable compensation for vertical and horizontal spillover effects between and among jurisdictions. Vertical spillovers or interagency externalities occur where two or more layers of government encounter costs, or draw benefits from, some policy or action. Horizontal spillovers or externalities occur where two or more jurisdictions at the same level of government encounter costs, or draw benefits from, some policy or action. In both instances, transfers between layers of government or among public agencies can enhance social welfare from an efficiency point of view. It results from the fact that each level of government or public agency, from its own perspective, would supply insufficient amounts of the public service because it would disregard spillover effects that accrue to other levels of the public sector, or to other agencies. Contractual arrangements, including cofinancing provisions, are needed to achieve an optimal outcome for the federation as a whole. The appropriate instrument for compensating spillovers in a quidpro quolike fashion is matching grants. Ideally each agency would contribute a share of program financing that corresponds to its citizens' relative benefit from the public program to be financed. Matching transfers entail a change in relative prices and hence exhibit a substitution effect in addition to providing extra funds. This will interfere with the recipient government's policy and alter its priority setting. Where it exists, it is often effected on the basis of fixed legal or bureaucratic rules, not on negotiated contractual arrangements, and could be a source of embedded inefficiencies. There are, however, examples of contract based transfers for specific programs among jurisdictions, for instance among the cantons in Switzerland. Spillovers must not necessarily be defined in terms of service delivery however. Very often they are expressed in terms of political costs and benefits. Specific transfers then turn into political signaling instruments. If, for instance, the federal government puts racial non-discrimination on its political platform, but the provinces pursue such discrimination in the area of their competencies (say, education), a federal grant for schooling could be rendered conditional on the recipient province abandoning its discrimination policy. Specific transfers are common in many countries, but the extreme example is perhaps found in the United States where the federal government uses a host of so-called categorical grants to compensate for vertical spillovers; to signal federal policy principles; and to impose its own
168
Spahn
priorities onto the states. A specialization of grants allows the "targeting" of programs by Congress. During the 1990s, these transfer arrangements had become more and more specific, often imposing federal policy priorities onto state and local which was resented as violating the loth Amendment. However the Supreme Court affirmed these grants as being in line with the Constitution because entering into such transfer arrangements is based on a contract and hence "voluntary". There are five types of provisions that are typical of categorical or specific-purpose g r a n d observation of quality standards; permitted use of the funds; expenditure constraints; matching obligations; and transparency requirements (record-keeping and reporting). The setting of quality standards through conditional grants is a powerful instrument to achieve some degree of homogeneity in the delivery of public services in a decentralized system of government. It is usually confined to minimum standards however. Standards exceeding the delivery capacity of subnational governments would violate their budget autonomy. Clauses on spec@ uses of the transferred funds are often inefficient. The so-called specific-purpose transfers or grants, prominent in almost all countries, are sometimes binding for the recipient government in that they require spending these funds on policies to which the benefiting constituency attaches low priorities, in which case spending could represent a waste of public resources.' In other cases, the clauses are non-binding in that the recipient government would have spent a similar or even greater amount fi-om its own budget anyway. While the former case could entail inefficiencies where "forced spending" is not warranted by vertical spillovers, the latter is tantamount to a general revenue grant because it frees unconditional budget resources. It therefore entails a pure revenue effect and does not interfere with local priorities. Nonetheless, specific transfers are extremely popular and common in many countries although they risk waste of resources or, at best, constitute general revenue funds. This is because of their potential to convey policy priorities of the donor government to its electorate. Politicians are keen on the "signaling" function of such transfers to demonstrate their commitment to "specific causes". They claim to have spent money on certain policies that are out of their control, even though the money would be (or may have been) spent by the recipient government anyway. An interesting illustration of the purely political function of specific transfers provides Australia. This country uses both general and specific grants. General transfers are calculated, inter alia, on the basis of relative fiscal capacity of recipient states. Specific grants of the Commonwealth to state governments are indeed for specific purposes, but most of them are counted as increasing fiscal capacity. In that way the Commonwealth "claws back" the relative effects of specific transfers through the system of general
Intergovernmental Transfers: The Funding Rule and Mechanisms
169
grants.* Nevertheless, Australian politicians attach high value to retaining specific transfers, most likely because of their signaling potential. Special transfers. Special transfers are neither pure general budget resources nor do they compensate for interjurisdictional externalities, including the spillover of political benefits and costs. They are used to compensate for extraordinary costs, such as from local catastrophes; for the targeting of national policies; and for regional development. The case of local catastrophes is self-evident. Emergency transfers do not represent regular budget resources by nature, nor are they geared toward altering the behavior of recipient regional governments. It is also clear that national policies might want to focus on a particular region, for instance to develop its economic base, to ease poverty or to service minorities directly. Where the central government uses regional authorities as its agents, there will be special transfers that correspond de facto to the financing of national policies. So, wherever the national government pursues regional development objectives, it may use the instrument of special transfers to subnational governments or agencies. Such transfers can be unconditional or conditional, but in all cases they are exclusively determined by national priorities. Nevertheless the national government may employ preference revealing mechanisms, such as matching requirements, even for such types of transfers, in order to respond to local priorities and to control unwarranted demands for such grants. Often, such transfers are limited in time. This is helpful in avoiding grants dependency and moral hazard. A sunset date for a regional development initiative emphasizes the need for a region to become self-sufficient after a phase of support. The Marshall Plan initiative after World War I1 was indeed limited in time; the Structural Funds of the European Union are not time limited, but based on fixed criteria that result in regions eventually "maturing" from assi~tance.~ Other special grants are typically given for "bulky" expenditures such as for larger public investment projects. Special transfers linked to the investment budget of the recipient government are often labeled "capital grant^".'^ They are of course limited in time by the gestation period of the local investment project, and they are sporadic by nature. Where capital grants are purely motivated by national regional development policies, they should indeed be subsumed in the special transfers category. However, matching capital grants could also reflect spillovers, and to the extent that regional authorities possess investment responsibilities, capital transfers could also be included in an equalization formula. In these instances one would prefer to convert a "bulky" capital transfer into current annual transfers to support the operating budgets of recipient governments. This requires some rules for accrual accounting. Moreover it requires the regional authorities' ability to borrow from capital
Spahn
170
markets or to engage in corresponding leasing contracts with the private sector. The converted capital transfers could then be used to service the debt or to honor such leasing contracts. In fact, accrual accounting is still rare in the public sector and subnational governments often have only limited access to capital markets, either by statute or by reduced creditworthiness. Capital transfers for investment purposes are thus still prominent in most decentralized governments, and it is therefore appropriate to include them in the special transfers category.
3.
THE FUNDING OF TRANSFERS
As said, the funding of transfers has to respect the overall constraint on public resources. So a general principle should be the closed fimding of transfers. This is achieved either through the assignment of a fixed amount or proportion of the donor government's budget, or through tax sharing. There are certain exceptions however. For matching grants the program might be open ended, but in practice it is often limited by the recipient government's ability to commit corresponding own resources. Where this is not the case, or where there are worries about possible abuses of the transfer system, the funding of matching grants could still be limited in size. Closed funding is achieved through restrictions on the total transfer pool. It raises the question of how to define the pool and what resources to allocate to it. This question can be addressed by first looking at tax sharing arrangements, which represent a special case of transferring resources between layers of government. Lessons can be drawn from this for other types of transfers that flow directly from budgetary resources. Tax sharing. The most prominent case for funding transfers is indeed tax sharing. Here, the resource pool is formed by total revenues collected from a particular tax (or a set of taxes). The various levels of government share this revenue at given proportions and lower tiers usually receive their share based on the taxes collected in their respective jurisdiction. This way, the overall resource constraint is automatically imposed conjointly on all governments participating in the sharing arrangements. However, it would be erroneous to expect that this also preserves macro stability at the same time. There are essentially three sources of potential macroeconomic risks associated with tax sharing." The sharing of taxes is too comprehensive, including all or most of the important taxes. In this case the national government could lose its ability to use taxes as countercyclical policy instruments. If the economy is booming and tax revenues swell, the national government may want to
Intergovernmental Transfers: The Funding Rule and Mechanisms
171
sterilize some tax revenue, but an important share is being retained and spent at lower tiers of government in a procyclical manner. Brazil represents a situation where this constitutes a serious problem. Lower tiers of government participate in taxes with highly volatile returns. This leads either to variations in the provision of local public services, which could be politically stressful, or to the national government being compelled to step in to smooth the impact of such arrangements. The effect is exacerbated where subnational governments have only limited access to borrowing. Colombia and Nigeria represent cases where subnational entities benefits from the highly volatile returns on natural resources (petroleum). It entails this type of macroeconomic risks. Apart from cyclical effects, the sharing of volatile resource taxes may also involve regional inequities and dynamic inefficiencies through unsustainable local investments (construction of future "ghost towns"). Lower tiers of government are given a share of taxes that are less buoyant or are likely to dry out over time. This jeopardizes the sustainable provision of local public services and may require a revision of the transfer arrangements or cause a vertical shift in power toward the central government.12Where local public service delivery is threatened, but lower tiers can mobilize political support, there is a moral hazard risk by which the national government is compelled to provide additional resources beyond existing transfer arrangements. This could entail a softening of local budget constraints and, again, jeopardize macro economic stability.13 Tax sharing also has a dynamic dimension: it could be instantaneous in that the proceeds are allocated to the budgets of benefiting governments "as you go"; or there could be some time lag and even constraints imposed through formula apportionment and quantity ceilings. Time lags are often used for administrative reasons (transfers are effected only after the completion of the fiscal year), but there may be other motives as well, such as the smoothing of cyclical effects. Moreover, lagged transfers significantly reduce the budgetary risks of recipient governments through greater revenue predictability and transparency. It allows the better planning of subnational budgets and thus enhances the prospects for macro stability. Where formulae are used to modify the revenue impact of tax sharing, it is questionable whether we can still speak of tax sharing. For instance, Argentina operates a lagged tax sharing scheme for its provinces, but has occasionally put a lid on the annual increment; Germany operates a personal income tax sharing scheme for its municipalities, but limits the sharing entitlement to the proportional part of the progressive tax. The objectives are obvious: In the first case, the intention is to limit the buoyancy of
172
Spahn
subnational tax shares in order to avoid procyclical spending by lower tier agencies, and hence macroeconomic instability. In the second case, there are also stability concerns, but in addition municipalities are thwarted to benefit from the fiscal drag that results from bracket creep of an expanding tax base within a progressive system. These examples illustrate that there are possibilities to limit the pool even within a tax sharing arrangement. However much depends on the legal framework for such arrangements. Where tax sharing is seen to represent legal entitlements, as in Germany, there is a case for a transfer-as-you-go system without restrictions. including on the use of these funds.14 Where it is seen to represent an allowance by the senior government, as in Argentina, the "donor" may also control the conditions that reign the sharing arrangements. Budgetary resources. In many instances, intergovernmental transfers are funded directly from the general budget of the donor government. This is particularly true for general revenue transfers or equalization grants. Occasionally, the resource pool is defined by historical developments, yet it often remains at the full discretion of the budget authorities controlling this pool. In Australia, for instance, the States ceded their constitutional rights to levy the income taxes to the national government during World War 11, receiving "tax reimbursement grants" fkom the Commonwealth instead. Over the years, such grants were transformed into ordinary general revenue grants representing a share of total budgetary resources of the national government. Today the Commonwealth has again formed a pool for general revenue grants from a single tax, the goods and services tax (GST). The Australian example illustrates the great variety that exists in defining a fimding pool for intergovernmental transfers for unconditional grants. If the pool is formed on a single or narrow resource base, the same problems will arise as with tax sharing discussed above: excessive volatility or structural shifts going one way or another. This could jeopardize macro stability or sustainable public service delivery by subnational agencies. A broader definition of budgetary resources would counteract such budgetary risks. But the very fact that budgetary authorities can control the size of the pool in a discretionary way is an advantage over blind tax sharing. Moreover, the use of budgetary funds allows us to deviate fkom the principle of derivation and to allocate the transfers to recipient governments proportionately based on equity. In this way, it is possible to combine elements of vertical fiscal rebalancing with horizontal equalization. However, transfers are often not directly funded from the general budget of the donor authority, but through the budgets of its line ministries. This is particularly true for specific and special transfers. For instance a specific purpose grant for education could come from the education ministry, a specific purpose grant for hospitals from the ministry of health, etc. The idea
Intergovernmental Transfers: The Funding Rule and Mechanisms
173
is to entrust the line ministries with the setting of policy priorities in the realm of their competency, and with the monitoring of spending activities by the recipient agencies according to the conditions specified. The financing of transfers through special agencies and funds entails some inefficiency risks however. There might be a proliferation of specific purpose grants with substantial overlaps, counteracting effects and a waste of resources because of a loss of effective control. There is a risk of fragmentation of national policies and a loss of parliamentary control, especially where resources are attributed to special funds whose functions are hardly ever reviewed. The danger is all the greater if such special funds are off-budget, create turf, or are subject to political patronage. Line ministries and recipient agencies could take advantage of their information prerogatives by jointly colluding in order to maximize the transfers reaped from the general budget. Because of the greater discretionary power of the donor ministry, transfers are more likely to flow to subnational jurisdictions according to party affiliation or nepotism. Finally, there might be a greater risk of moral hazard due to strategic hold-up positions of recipient authorities in the context of specific purpose programs to which the donor government and its ministries attach high priorities.'5 Such inefficiency risks are significantly smaller for unconditional transfers funded from the general budget and allocated on the basis of objective criteria through formula apportionment. However, moral hazard might not totally be avoided even for unconditional funding from the general budget. Often, donor governments take recourse to gap-filling transfers where they are compelled, by law or political pressure, to bail out subnational budgets under distress. This may occur despite firm rules reigning the overall transfer system. Gap-filling transfers soften budget constraints of the public sector as a whole, entail inefficiencies and jeopardize macroeconomic stability. Germany has a transfer system based on objective criteria, but the federal government is bound, by constitution, to assist states in financial troubles. Russia provides transfers to its republics and oblasti on reasonably well determined criteria, but there are opaque last-round transfers to fill remaining budgetary gaps, the rules of which are hard to comprehend. Brazil operates a tax sharing system and state and municipal equalization funding is based on clear criteria, but a host of transfers based on "contracts"
Spahn
(convenios) is often used to honor partisanship. In all these instances hard budget constraints are weakened, producing regional unfairness and economic inefficiencies. Whether moral hazard in the transfer system can be avoided depends largely on a credible no-bailout commitment of the donor government. Given legal constraints and political obligations this is not always easy. But generally speaking it largely depends on the recipient's taxing and budgeting autonomy. The stronger the financial basis provided by own resources of the recipient, the more credible a no bail-out commitment of donor governments will be. Or conversely: the smaller the recipient government's own tax base, the larger its transfer dependency, the greater the need for gap-filling, and hence moral hazard.
MECHANICS OF TRANSFERS SYSTEMS There are no uniform rules for designing the machinery of a transfer system. This very much depends on traditions, political and social conditions, including value judgments. It is also constrained by the informational base and administrative procedures. One should distinguish transfer mechanisms that are wholly determined by ad hoc political decisions and transfer mechanisms that are based on well established rules. The former are often found in the realm of specific purpose transfers and special grants, and they are difficult to analyze without reference to given political objectives. I shall therefore refrain from discussing such "mechanisms" here. The latter consist of formal procedural constraints that can be analyzed independently from underlying political value judgments. Rules are often used to design general revenue grants for equalization purposes, serving to achieve transparency and reach a degree of interjurisdictional fairness among recipient governments. Such transfers should display a pure income effect. There are also rules-based transfer systems containing specific incentives so as to influence the behavior of recipient governments. Such transfers exhibit an income and a substitution effect. In this paper, I shall dwell on general grants with a view to a more general applicability. The design of transfers embedding substitution effects hinges on specific policy objectives that cannot be dealt with here. A few principles for rules-based general transfers will be sketched in the following however.
Intergovernmental Transfers: The Funding Rule and Mechanisms
175
Standard setting. Typically, rules-based transfer mechanisms will first define a reference point. This serves as a standard or benchmark against which to measure the relative positions of all or a part of the jurisdictions participating in the scheme. The standards and the relative positions are almost everywhere defined in per-capita terms, but some countries, for instance Germany, also attach "weights" to the population figures to account for regional agglomeration or low-density effects. The benchmark could be a comprehensive national average or a subset of jurisdictions with standard characteristics. It is important that the standards be measurable, i.e. the relevant information must be available and comparable in quality across participating jurisdictions. It is crucial that this information be as objective as possible and that recipients cannot unduly influence the standards through unilateral action. General transfers are designed with a view to the relative fiscal position of recipient governments. One may distinguish three different philosophies in this regard: Where there are no significant differences between the level and the costing of service delivery across subnational governments, as assumed for Canada or Germany, it is sufficient to set revenue capacity as a single standard and to equalize revenue capacity across jurisdictions. Some countries such as Switzerland or South Africa use, inter aha, regional GDP instead, which could be interpreted as a proxy for revenue capacity. For specific purpose or bloc transfers it is common to focus on expenditure needs, i.e. on indicators translated into budget equivalents through the costing of a standard level of services. This philosophy also creeps into some equalization schemes for general transfers as in South Africa or Russia, for instance. Another example is Mongolia, where heating and transportation plays an important role; it emphasizes energy costs in its transfer mechanisms. Obviously, it is also possible to include both revenue capacity and expenditure needs in a comprehensive approach to budget equalization. Australia represents the most prominent example for this approach. It has led to a highly complex system that is being criticized as overly complicated. In particular, it is argued that differences between the level and the costing of service delivery across subnational governments are said to be only small (except for the Northern Territory), which would warrant a much simpler model such as in Canada.
Spahn
Despite possible shortcomings, the Australian system has become the reference model for designing an ideal transfer system. The basic approach is sound, complete, feasible, and reasonably transparent. The proliferation of criteria that render the Australian system so cumbersome result exclusively from political haggling, not from an ill-designed scheme. On the contrary, the scheme has been able to absorb the varying political demands without sacrificing its basic philosophy as is often the case elsewhere. True, the Australian scheme for general transfers is heavy in terms of information requirements and technical expertise. This renders it difficult, if not impossible, to export it to other countries where data might be poor and administrative capacity weak. It is my contention, however, that the basic approach of the Australian system should be, and could be, followed by other countries, including developing countries, albeit in a simpler and typified form. In the following chapter I illustrate this by sketching the adoption of the scheme for a developing country in Eurasiafrica, the federation of Krakozhia. Degree of equalization. Once the benchmark is set, the degree to which the gap between the effective standardized tax capacity, needs indicator or budget deficit of any recipient government should be closed is still to be decided. This is basically a matter of value judgment. Germany for instance insists on equalizing all states at least up to 99.5 percent of the national per capita average. It leads to an open pool for the federal budget because its size will be determined by remaining fiscal gaps before federal grants. In other countries such as Australia, the transfers are proportional to the size of the pool and allocated onto the states according to relativities. Both systems are symmetrical in the sense that the rules apply to all receiving entities in the same way. However, Canada and Spain, both in their own way, operate asymmetrical transfer systems with varying degrees of equalization. In Canada the betteroff provinces such as Alberta do not participate in the equalization arrangements at all, while in Germany the 'richer' states are asked to make a contribution to an interstate equalization pool. In the latter case the fiscal capacity of 'richer' states is leveled down, while it remains untouched in the former. In Spain transfers from the national government obey different sets of rules for different categories of regions. Navarre and the Basque region, which are fiscally self-sufficient on their own resources, contribute to the national budget through upward-oriented transfers. These examples illustrate that value judgments differ greatly among countries and even within countries, which renders it impossible to establish firm rules on the degree to which equalization should be carried.
Intergovernmental Transfers: The Funding Rule and Mechanisms
5.
A BUDGET EQUALIZATION SCHEME FOR KRAKOZHIA
In this concluding chapter, an attempt is made to design a transfer scheme for a developing country, Krakozhia, with all the desired characteristics of a good transfer scheme for budget equalization through general grants: closed pool; standardized budgets; and regional fairness (symmetry). The model is borrowed from the Australian philosophy for designing general revenue grants, but the scheme had to be stylized and simplified considerably to conform with the conditions prevailing in countries with weak administrative capacity and a shortage of statistical data. Krakozhia is a country situated in the heart of Eurasiafrica that has peacefully lived with its neighbors for centuries (which renders the case even more fictitious). The federation consists of 10 rather inhomogeneous provinces: Adania, Birino, Chuchko, Dobrodzia, Eshowia, Filiasi, Gubkin, Humppila, Ilgin and Jabel. The population and population density characteristics of the federation are represented in Table 1.
Table I. Population Characteristics of Krakozhian Provinces
Adania Birino Chuchko Dobrodzia Eshowia Filiasi Gubkin Humppila Ilgin Jabel Total provinces
Population (Percent of total)
Population (Mill. persons)
Area (Sq.km.)
Population density ( 1 .OOOper sq.km.)
2.0 0.6 8.6 11.9 1.9 0.3 6.6 0.1 0.1 0.2
53.6 48.6 80.9 179.8 136.0 25.2 56.9 17.0 0.2 1.O
36 13 106 66 14 11 117 7 501 173
6.0 2.0 26.6 36.7 6.0 0.9 20.6 0.3 0.3 0.5
32.3
572.4
56
100.0
Area (in percent of total) 9.4 8.5 14.1 31.4 23.8 4.4 9.9 3.0 0.0 0.2
100.0
The Krakozhian federal authorities, afier evaluating and comparing various international grant models, have concluded that "the formula which incorporates both revenue raising capacity and expenditure needs has been found appealing in its generic form for adoption in Krakozhia." They are
178
Spahn
eager to implement the methodology developed by the Commonwealth Grants Commission (CGC) of Australia for transferring general revenue to the states, and have asked for consultancies to design the scheme. In the following, an attempt is made to develop a simplified formula inspired by that methodology. The purpose is to illustrate what a formula incorporating both revenue raising capacity and expenditure needs would look like in Krakozhia, a country where, contrary to Australia, there are huge differences in the level of economic development, of revenue potentials, and of public service needs that might require special attention beyond a policy emphasizing standard tax capacity and standard expenditure needs.
5.1
Moving Toward a Transfer Formula: Some Basic Issues
The objectives of any grants system are concisely spelt out in a paper prepared by the Krakozhian Ministry of Finance: "Objective: The prime objective of the formula is to cover the portion of each province's expenditure requirements that cannot be covered with its own revenue sources. In doing so, the formula distributes the pool of regional grant resources among the provincial governments in a way that narrows the disparities between them." Hence the transfer formula will have to address both vertical fiscal imbalance (mismatch between revenues and expenditures at levels of government) and horizontal fiscal imbalances (equalization among provinces) at the same time. Before embarking on simulations of a methodology to cope with horizontal fiscal balancing in Krakozhia, it might be useful to sketch the CGC approach. The Australian equalization mechanism functions as follow^:'^
1. The total pool of unconditional transfers is well defined and closed. It was set by the Commonwealth government prior to the 2000 tax reform (Financial Assistance Grants); and is based on the proceeds from the goods and services tax since the year 2000. 2. Distribution of the funds among the states is made according to the per capita rule proposed by the Commonwealth Grants Commission: This distribution rule, designated by the term "relativity," is established each year by complex calculations; and
Intergovernmental Transfers: The Funding Rule and Mechanisms
179
the CGC's recommendations (and the federal government's proposals on total untied transfer payments, before 2000) aretwere discussed at the annual meeting of the Ministerial Council of Treasurers . 3. The establishment of per capita relativities consists of the: Calculation of the standardized per capita expenditures of each state, i.e. the spending that the state must cover if it were to provide the average level of public services of the states overall, bearing in mind the cost of delivering such services; Calculation of the standardized per capita revenues of each state, i.e. the revenues that each state could obtain were it to apply the average taxation structure of the states overall to its own tax bases; Calculation, using these standardized amounts, of the funding that each state requires to provide a level of public services equivalent to the average for the states overall; and Calculation of the difference between this amount and the amount of specific federal transfers paid to the state (the result is the shortfall or the state's net financial requirement, covered by a transfer payment). Per capita assistance in the form of equalization thus comprises an equal per capita portion of unconditional federal transfers plus an adjustment in respect of the cost of delivery plus another adjustment that reflects fiscal capacity. It should be noted that equalization in Australia must satisfy the principle whereby each state must have the capacity - not the obligation, which would render the transfers conditional - to offer services at a level similar to that prevailing in the other states were the state in question to apply the average taxation rates of the states overall and conduct its affairs with a similar degree of efficiency. Issue I: Administrative capacity As noted above "each state must have the capacity ... to offer services at a level similar to that prevailing in the other states". This condition is largely met in Australia; it is highly doubtful for Krakozhia. The differences in administrative capacity are extremely large among Krakozhian provinces, and often even larger within a province. This might jeopardize the mandate of any one province, or parts within the province, to "provide the average level of public services ... overall". After all, the objective is not to equalize government spending as such; it is to equalize government output-the level and quality of public service delivery. It requires appropriate administrative capacity. The lack of administrative capacity can not be coped with by including a corresponding "disability factor" in the formula. It would mean to simply transfer additional funds to a region that exhibits administrative "disability"
180
Spahn
without addressing the fundamental problem itself. Transfers due to a lack of administrative capacity cannot be converted into effective output. They cannot be effectively "absorbed", but will be spent anyhow. This represents a waste of public resources and results in inefficiencies. Issue 2: Size ofjurisdiction
The size of the Australian states is more or less comparable in terms of population and area-except the Northern Territory, the Australian Capital Territory and, perhaps, Tasmania. In Krakozhia the regions exhibit much larger discrepancies in these terms (see Table 1). Dobrodzia controls roughly one third of the country's resources in terms of population and area; tiny Ilgin covers only 0.03 percent of the total area, but has the highest density of population in the country. Humppila, Filiasi and Eshowia are the regions with the lowest population density reflecting the high geographical diversity of a country with large inhospitable areas. It is obvious that this regional variation represents a major challenge for designing a grants formula based on expenditure needs and tax capacity for the average of the country as a whole. Issue 3: Per capita relativities
The large differences in area will definitely affect the calculations of transfers, but they can eventually be addressed through cost adjustment factors. The differences in the size of the population represent a thornier problem for a formula that works with per capita relativities as in the CGC methodology. This is because of the existence of fixed costs elements (overhead costs) of public administration. These costs represent a much higher burden for the smaller jurisdictions than for the larger ones. Assuming regions of more or less equal size in terms of population, and assuming all other factors equal for the time being, the CGC methodology would calculate equal per capita amounts that can be multiplied by the population of each state to obtain the total grant entitlement. So the per capita distribution by size of jurisdiction would largely form a parallel to the x-axis. The grant as a function of population would be proportional, i.e. linear without an intercept. In many countries with regions of uneven size, including the federation of Krakozhia, the intercept is highly significant however. A simple regression with actual data for the existing transfer scheme of Krakozhia reveals the following picture (see figure 2). On the basis of this regression analysis, the smaller provinces appear to be underfunded, while the middle-sized provinces are relatively better off compared to a purely population based grant with a fixed cost element. The
Intergovernmental Transfers: The Funding Rule and Mechanisms
181
larger and more populous regional governments of Dobrodzia and Chuchko are funded more or less in line with population as one would expect. Measured against this standardized benchmark, the regional pattern of grant allocations would hence be significantly different from the present one. This can best be illustrated by Figure 3.
5.2
Standardizing Recurrent Expenditure Needs
In order to obtain a standard regional budget one must first determine the amount of money needed to finance the standardized per capita expenditures of each province, i.e. the outlays the province must incur to provide the average level of public services of the provinces overall, bearing in mind the cost of delivering such services. The CGC calculates such standard needs in an extremely detailed fashion for a large number of expenditure items covering the complete recurrent budgets of the States and Territories. In Krakozhia the methodology must be considerably simpler because of data limitations. The calculations of standardized per capita expenditures can be based on a consolidated set of budget data. The Krakozhian provinces have only three types of expenditures: administration, education and health. They are represented in Table 2. For each of the expenditure items, an appropriate method for standardizing expenditure needs has to be developed. This is done on a per capita basis. The sum of the standardized per capita expenditure items times the population would then form the total standardized expenditures for each province. The CGC methodology does not only look at the spending that the state must incur to provide the average level of public services of the states overall, but also on the standard cost of delivering such services. This is done for each expenditure item separately. Because of an obvious lack of corresponding information in Krakozhia, it is proposed to use a generalized cost-adjustment factor on the aggregate of per capita expenditure needs, rather than making individual adjustments for each expenditure item individually. 5.2.1
Administration
Administrative expenditures constitute a major spending item of provincial budgets in Krakozhia: forty percent of total regional spending goes to general services. Moreover these outlays vary significantly among provinces. The larger provinces, Dobrodzia and Chuchko, spend as little as 23-27 danir per capita a year on administration; the smaller provinces spend
Spahn
Table 2. Consolidated Expenditures of Krakozhian provinces Administration Adania Birino Chuchko Dobrodzia Eshowia Filiasi Gubkin Humppila Ilgin Jabel Total provinces In percent of total In percent of total
65 75 200 320 90 40 150 30 15 15
Education Health In mill. danir
TOTAL
45 10 70 115 10 15 60 10 8 7
195 100 520 915 110 75 450 60 38 37
1000
85 15 250 480 10 20 240 20 15 15 115 0
350
2500
40.0%
46.0%
14.0%
100.0%
40.0%
In danir per capita Adania Birino Chuchko Dobrodzia Eshowia Filiasi Gubkin Humppila Ilgin Jabel Total provinces Std.dm. Std.dev./mean
33 118 23 27
44 24 29
23 16 8
40 5
10
46
142 23 270 174 87
71 36 180 174 88
31 82 266%
36 62 173%
5 53 9 90 93
41 11 34 311%
100 157 60 77 57
265 68 54 1 442 216 77 171 221 %
5 to 8 times as much. This is clearly related to the importance of overhead costs for administration and general services. It is obvious that the larger provinces will greatly affect the average per capita spending. If per capita expenditure needs for administration of all provinces were to be expressed in terms of the comparably low average
Intergovernmental Transfers: The Funding Rule and Mechanisms
185
figure for the provinces as a whole, it would not be sufficient to cover the overhead costs of smaller provinces. Thus some model calculations have to be made in order to arrive at standardized needs that take overhead cost more explicitly into account. For the following regression analysis it is assumed that outlays for administration contain a fixed cost element, and they are related to population and area. The regression results are shown in the following table: Repression statistics Multiple correlation coefficient R-square Adjusted R-square Standard error
0.990668534 0.981424145 0.9761 16758 15.08909472
Intercept Population (mill. persons) Area (sq.km.)
CoefJicients 16.67035687 18.23257937 0.408391237
Standard error 7.018771693 1.789895876 0.129889631
t-statistics 2.375 1 1029 10.18639107 3.144140398
The results are highly significant and appear to produce a reasonably good fit. The estimated coefficients can now be used to calculate the standard expenditures for administration and general services. This produces the expenditure pattern reflected in Table 3 below. The model reproduces the actual pattern of spending reasonably well, except for the smallest provinces Ilgin and Jabel. In fact these entities now spend considerably less than they would according to the standardized model. If this standardization procedure were to be adopted, Ilgin and Jabel could expect higher grant entitlements based on this single expenditure item. The relationship between actual and estimated standard expenditures can also be shown graphically as in Figure 417: The line drawn represents a purely population based estimate for provincial administrative expenditures. The inclusion of the area of each province improves the estimates remarkably and indicates higher general services outlays for less densely populated areas. Since neither population nor area are under the direct control of the respective provincial spending authorities, it makes sense to standardize administrative outlays on the basis of these two variables, i.e. the regression estimates are used to standardize general grants for administration, notwithstanding a cost adjustment factor that will apply to total spending later on.
Spahn
Table 3. Estimating Provincial Outlays for Administration in Krakozhia -
--
Estimated Actual mill. danir Adania Birino
Difference in % of actual
74
65
13.8%
48
75
-35.8%
Eshowia Filiasi
207 306 108 32
200 320 90 40
3.3% -4.3% 19.7% -19.7%
Gubkin
161
150
7.4%
Humppila
26 18 20
30 15 15
-14.5% 22.1% 35.1%
1000
1000 Std.dev.
21.6%
Chuchko Dobrodzia
Ilgin Jabel
Total provinces
0.0%
Figure 4. Administration expenditures of Krakozhian provinces - Actual and estimated
Intergovernmental Transfers: The Funding Rule and Mechanisms
5.2.2
Education
Expenditures on education represent the most important item of regional government budgets. 46 percent of total regional spending is on education. This puts the first priority on refining the methodology for standardizing expenditure need in that sector. For education, some independent information was made available for Krakozhia that is used for the present grants formula. These indicators include for all provinces: Average primary & secondary participation rate Total primary school teachers Number of classrooms in primary school Number of students in primary school Students-teacher ratio Students-classroom ratio This are valuable statistics that could also be used for standardizing education expenditures according to the CGC methodology. However one must guard against using the information as such for each province. Some indicators-for instance, the number of school teachers or the number of class rooms-represent "capacity indicators". Such indicators exhibit inefficiencies. If used directly, the transfers would induce perverse incentives such as increasing the number of class rooms or teachers, rather than service delivery, without relating it to standard needs. To avoid this, the information must always be used in a standardized fashion when calculating grants relativities, relating it to a standard curriculum adjusted for eventual cost differentials as may be applicable. But the information contained in the above set of data is in fact useful for deriving standard or average figures. There are a number of options on how to relate per student education expenditures to the variables listed above. This is not the place to go into a deeper analysis. The model used here is testing simply whether there is a relationship between education expenditures on the one hand and the number of students enrolled on the other. The regression results for this type of model are as follows: Regression statistics Multiple correlation coefficient R-square Adjusted R-square Intercept Number of students
0.990263267 0.980621338 0.978 199005
Coefficients
Standard error
t-statistics
2.508932093 0.146559482
9.287561259 0.007284171
0.270138955 20.1 2026829
Spahn
188
Table 4. Estimating Provincial Outlays for Education in Krakozhia Estimated Actual Mill. danir Adania Birino Chuchko Dobrodzia Eshowia Filiasi Gubkin Humppila Ilgin Jabel Total provinces
Difference in % of actual
91
85
7.5
7
15
-53.
290
250
16.
43 8
480
-8.8
18
10
76.8
17
20
-12.7
264
240
10.0
9
20
-57.0
7
15
-55.5
9
15
-41.1
1150
1150
0.0 42.2
The regression results again appear to yield a reasonable fit. It might therefore be appropriate to standardize the transfers for education by using the number of students enrolled. If the actual outlays for education are confronted with the estimated figures based on the number of students, the data in Table 4 is obtained. The differences between the estimated and the actual expenditure figures are dramatic in some instances. In Eshowia the estimated amount is roughly twice the actual; in Birino, Hurnppila and Ilgin it is less than half compared to the status quo. This may have several causes. First, there could be significant differences in the costs of producing education services due to differences in population density, for instance. In particular overhead costs could be higher than represented by the intercept in the equation for smaller jurisdictions. Second, there could be differences in the quality of education services. Third, there could be distortions induced by institutional factors, including corruption. This would need further inquiries that cannot be carried further in this paper. However, one particular aspect has to be taken care of. As can be seen from Figure 5, the take-up rate of education is highly uneven in Krokazhia. These differences will put significant stress on the equalization formula. They do not represent disability factors in the sense of cost differentials, but are explained by different access conditions to schooling throughout the country. If the equalization formula were based on the actual number of
Intergovernmental Transfers: The Funding Rule and Mechanisms
189
students, it could perpetuate regional unfairness; if it were based on the standardized number of students, it could over-equalize and lead to a spending spree in some provinces without generating corresponding output. In this case the transfers are better geared toward specific purpose payments conditional on some appropriate performance criteria, such as take-up rates, and with some standards for the quality of service delivery. One has to caution however that this will require additional monitoring because simply enrolling students is easy. What counts is whether the students are in fact receiving public services, or not, and at what level of quality. For the present study, estimations on the basis of actual students are used for starting the new transfer system in Krakozhia, understanding that incentives will have to be built in to increase the schooling take-up rate and to harmonize quality standards through appropriate incentives and effective monitoring. Humppila Filiasi Ilgin Adania Gubkin Jabel Dobrodzia Chuchko Eshowia Birino 0.0%
50.0%
100.0%
150.0%
200.0%
Percent of national average
Figure 5. Schooling take-up rates per population in school age, as percentage of national average
5.2.3
Health
Health expenditures represent an average of 14 percent of total spending by provincial governments in Krakozhia. The distribution of the average health costs per capita is highly uneven among the provinces. They are as little as 5 danir in Eshowia, and about 90 danir in Ilgin or Humppila, two regions with very different characteristics. The average cost per capita for all
Spahn
190
regions is 11 danir, with a standard deviation of 34 danir. The variation coeffi~ient'~ is correspondingly high: over 300 percent. This indicates extremely varying conditions for the provision of health services in Krakozhia, which - again - cannot be addressed simply by dealing out grant money. As in the case of education, it requires a thorough analysis and, perhaps, overhaul of the whole health sector and the delivery of health services. Again, some independent information has been made available for the health sector. These indicators include, for all provinces, the: Number of health centers Number of hospital beds Number of doctors Number of nurses Number of health assistants Health center coverage Unit cost of constructing a health center Under five years age mortality rate per 1,000 population The same precautionary note made for education on "capacity indicators" also applies for health expenditures. The information should never be used as such for individual provinces, but only for the total of the country and in a standardized fashion. In order to detect a common pattern for health expenditures that could be used to standardize across regions, health expenditures were regressed on a number of variables: population, the number of health centers, the number of hospital beds, the number of doctors, the number of nurses, and the infant mortality rate. None of the health-related indicators were significant. As a result, the regression was again simplified by using population figures as the only explanatory variable. The regression results for this type of model are shown as follows: Regression statistics Multiple correlation coefficient R-square Adjusted R-square Standard Error
0.964047418 0.929387424 0.920560852 10.33793001
Intercept Population
Coeflcients 8.09275238 8.320586041
Standard error 4.188232878 0.8 10870602
t-statistics 1.932259407 10.26129942
191
Intergovernmental Transfers: The Funding Rule and Mechanisms
For the illustrative model exercise presented in this paper it was decided to estimate health expenditures on the basis of population only (Table 5). It is obvious that, given the large discrepancies in per capita health expenditures among regions, the standard assessment for health expenditures based on average national costs will vary significantly from the actual distribution of health expenditures for most regions. These variation could be as large as 100 percent measured against actual budget appropriations. It is questionable whether such large differences, if converted into transfer funding, would be bearable for regional government budgets. It will hinge on the importance of the cost correction factor discussed below.
Table 5. Estimating Provincial Outlays for Health in Krakozhia Estimated Actual Mill. danir Adania Birino Chuchko Dobrodzia Eshowia Filiasi Gubkin Humppila Ilgin Jabel Total provinces Std.deviation
5.3
24 13 80 107 24 10 63 9 9 10 350
45 10 70 115 10 15 60 10 8
7 350
Difference in % of actual -46.0 33.8 14.5 -7.2 143.1 -30.4 5.7 -9.8 10.1 35.9 0.0 51.9
Costing
As mentioned before, it was deemed to be appropriate to treat cost differentials on the basis of total expenditures rather than for each individual budget item. The approach to evaluating costing differentials is based on a heroic assumption: It is assumed that the difference between the total standard expenditures and total actual recurrent expenditures is an appropriate indicator for existing cost differentials among provinces.
192
Spahn
This methodology would also close some of the gaps that will exist between the standard and the actual budgets of individual provinces. The exercise is thus based on a comparison between standard and actual expenditures as presented in Table 6. Table 6. Comparing standard and actual expenditures of provincial governments in Krakozhia Standardized Expenditures
Adania Birino Chuchko Dobrodzia Eshowia Filiasi Gubkin Humppila Ilgin Jabel Totalreeions
Admin. Service 74 48 207 306 108 32 161 26 18 20 1000
Actual
Mill. danir Educa tion
71 17 285 447 18 9 261 12 24 5 1150
Health 24 13 80 107 24 10 63 9 9 10 350
Differences % of actual
Total 169 79 571 860 150 52 485 47 51 35 2500
Total 195 100 520 915 110 75 450 60 38 37 2500
26 21 -52 55 -40 23 -35 13 -13 2 0
13.4 21.3 -10.0 6.0 -35.9 30.8 -7.8 21.5 -35.0 5.0 0.0
Std.deviation Variation coeflcient
It is interesting to note that, despite the simplifications made when standardizing each individual budget item, the total of standardized expenditures reflects a rather similar pattern to the total of actual expenditures, at least in the aggregate. The standard deviations for both relevant columns of Table 6 are almost identical. This conceals the fact that individual variations can be a high as 35 percent for some provinces (Eshowia, Ilgin). Even so, the variations are much less pronounced than those of the individual budget items, which disclose some compensatory effects through aggregation. As said before, the differences between standard and actual expenditures were taken to indicate systematic cost differentials among provinces. These had to be rationalized by using some external information explaining such differentials.
Intergovernmental Transfers: The Funding Rule and Mechanisms
The proposed procedure is as follows: 1. A cost differential indicator is calculated representing actual divided by standard total recurrent expenditures. 2. The logarithm of this indicator is regressed on the following independent variables reflecting proxies for cost differentials (no intercept): The logarithm of unit costs of constructing a school (in mill. danir); Te logarithm of unit costs of constructing a health center (in mill danir); and The logarithm of population density (1,000 per sq.krn.). 3. The estimated "cost correction factors" are then used to bring standard expenditures closer to the actual ones. This requires two steps: First the harmonic mean of these estimated cost correction factors is calculated; Second, the estimated correction factors are adjusted by dividing them by their harmonic mean to standardize them to a factor of 1. It shows which regions are above and below average. The following table exhibits the regression results for the "cost correction factors": Regression statistics Multiple correlation coefficient R-square Adjusted R-square Standard Error Intercept Log unit costs of schools Log unit costs of health center Log population density
0.697698935 0.486783804 0.197293462 0.081227937 Coeficients 0 0.075376029
Standard error #NV 0.276714959
t-statistics #NV 0.272395932
0.40751877 1 -0.059398231
0.184230933 0.045075743
- 1.317742695
2.21 1999715
Table 7. Adjusting total provincial expenditures for cost differentials
Total outlays standardized
Estimated cost Corrected cost correction adjustment Adjusted Total actual standard outlays factor factor
'
in % of actual
Adania Birino Chuchko
169
1.029
1.017
172
195
-24
-12.1
Dobrodzia Eshowia
860 150
1.034 0.999
1.022 0.987
Filiasi
52
1.038
915 110 75
485 47 51 35 2500
1.004 1.129 0.909 0.929 1.012
1.026 0.992 1.115 0.899 0.918 1.OOO
877 148 53 480 52 46 32 2500 287
450 60 38 37 2500 290
-38 37 -22 30 -8 8 -5 0
-4.1 34.0 -29.1 6.8 -12.6 21.1 -12.9 0.0
11.5%
11.6%
Gubkin Humppila Ilgin Jabel Total regions Std.deviation Std. deviation in % of mean
Note: Estimated corrections factor divided by its harmonic mean.
'
Diff.
"
196
Spahn
way, but taxpayers manage to evade the tax to some degree, this would reduce the tax potential accordingly. The tax rate is usually a proportional statutory rate, but it could also be a set of rates, for instance for different types of property, or for income brackets. "Tax effort" is finally a measure of the effectiveness of fiscal administration and tax collection, which will include an element of taxpayers' compliance. Tax effort is particularly difficult to measure as regional governments often engage in mutual "tax competition" in order to attract business activities to their region. Where the tax legislation does not allow regions to compete through the tax rate, this is often effected through leniency in administering and collecting the tax. Ideally, standardization of tax capacity has to be effected at all three levels-the tax potential, the tax rate, and the tax effort. Moreover, tax potential would have to include harmonizing eventually different legal provisions regarding the definition of the base and their social and economic relevance in any one region. Fortunately, standardization is often unnecessary because there is some degree of uniformity across regions. Where the legal provisions defining the tax base are uniform throughout the country, the exercise amounts to defining the economic relevance (potential) only. No standardization on the basis of a "harmonized" tax base is required. The same is true where regional governments have no discretion to vary tax rates as they are set by national legislation. Where the central government administers and collects regional taxes, one can usually assume that the same "tax effort" is made in every region. Where this is not the case, the regional authorities cannot be held accountable for a lack of tax collection effort. Standardization of fiscal capacity has to be effected for each single revenue item, including non-tax revenue. However it is sufficient to focus on the "big" revenue items, and include smaller revenue sources as a simple markup. For Krakozhia, we shall simplify matters by assuming a single provincial tax only. This tax is based on the same legal provisions throughout Krakozhia. Moreover it is administered and collected by the national government. This makes a good case for using actual rather than standardized tax income, which appears to be an excellent proxy for economic potentials, and hence tax capacity, in the regions.
Intergovernmental Transfers: The Vertical Sharing Dimension
197
Total provincial taxes collected in Krakozhia are 1,000 mill. danir, which is 40 percent of total expenditures. The remainder has to be covered, on average and on a standardized basis, by transfers of the national government. The ratio of own taxes relative to total spending is reasonably high to avoid transfer dependency at the aggregate level for all provinces. Although they cannot use tax instruments, which are fixed at the national level, they could be expected to adjust expenditures and employ their non-tax revenue so as to be financially autonomous under the transfer regime. 5.4.1
Non-tax revenue
Non-tax revenue includes a number of charges, fees, and fines, revenue from the sale of goods and services, income from public investments, and other capital income. By nature, this revenue category is extremely diverse and some crude method has to be used to take them into account in a simplified manner. Because these revenue items are under the full control of regional jurisdictions, it is reasonable to treat non-tax revenue as a simple markup on tax revenue. This standardizing will preserve the revenue raising autonomy of provinces in the realm of their budget authority. If there were a penalty on raising revenue through charges and fees, there would be a severe incentive to bring these charges to zero to the benefit of local taxpayers. This is avoided through standardization. The markup is 25 percent of tax revenue, bringing own resources up to 50 percent of total provincial expenditures at the aggregate level.
5.5
Integrating Budgets and Calculating Transfers
In the following, the standardized expenditure and revenue sides of provincial budgets are integrated to compare the results with the actual data. This exercise is, again, for illustration purposes only. Table 8 exemplified the procedure for integrating the budget for calculating regional grants in Krakozhia Table 8 compares actual and standard total outlays assuming that actual taxes (including non-tax revenues) represent effective taxable capacity, which produces actual and standard deficits for each province. If the national government decides to establish vertical fiscal balance by closing the budget gap on average for all provinces (1,250 mill. dank), it would be advised to use standardized fiscal gaps rather than actual gaps, since gap filling, as was discussed earlier, entails perverse incentives leading to higher outlays and lower taxes at subnational levels. Standardized fiscal gaps require budget adjustments for some provinces however since their transfer flows correspond to standardized outlays, not actual outlays. For instance Jabel will have to reduce expenditures (or
Table 8. The composition of provincial budgets (in mill. danir) Budget deficit Own revenue
Adania Birino
120
Outlays
' Actual
Standardized
Actual
Standardized = Transfer
195
169
75
49 64
Amount
% of outlays
15
100
79
85
Chuchko
262
520
571
257
309
Dobrodzia
556
915
860
359
304
Eshowia
29
110
150
81
121
Filiasi
18
75
52
57
34
Gubkin
203
450
485
247
282
Humppila
11
60
47
49
36
Ilgin
12
38
51
26
40
Jabel
24
37
35
13
11
-2
-5.0
1250
2500
2500
1250
1250
0
0.0
Total provinces
Note: 1
Uncovered budget deficit
Estimate This is assumed to also equal standardized revenue.
Intergovernmental Transfers: The Funding Rule and Mechanisms
199
increase non-tax revenues) corresponding to 5 percent of actual outlays. In Dobrodzia it is 6 percent, and in Adania 13 percent. This appears to be manageable for these provinces. Larger adjustments would be required for Birino, Humppila, and Filiasi with about 20 and 30 percent of actual outlays respectively. This would require further examinations leading, perhaps, to additional specific purpose or special transfers for these provinces.'9 Four provinces would receive more than that need to cover their actual deficit. Two of them are tiny provinces where relativities tend to be overblown, Chuchko (+lo percent) and Gubkin (+8 percent) require special attention. However, there are two larger provinces where taxes per capita fall short of the national average, so possibly the simplifications made by equating actual with potential tax revenue need further investigation and a refinement of the transfer mechanism.
5.6
Political Feasibility: The Power of the Status Quo
The foregoing sketch of an equalization scheme for Krakozhia will meet strong resistance, although the basic methodology proposed should be persuasive if one accepts transparency based on objective and non-malleable statistical data. The opposition will be mainly for political reasons because the outcome of the calculations will undoubtedly deviate from the existing pattern of intergovernmental transfers, and these deviations might be highly significant in some instances. It will mobilize political defiance by the potential losers of a reform. The status quo has thus a bearing on any reform that will touch upon existing grant relativities. As said above "all decentralized or federal governments will have to respect their historical and political conditions, so reforms will be highly path-dependent". From a political perspective there are basically three options here. Follow an objective formula-based approach although it compels one to alter existing grant relativities. Even though the approach may be refined by including additional objective information to better reproduce the status quo, it is unlikely that all discrepancies can be eliminated that way. This is because the status quo always includes ad hoc political concessions that are brought to the fore under an objective formula. Continue to follow the existing approach with its opaque ad hoc components then it is probably best to at least fix the existing grant relativities for the hture and abandon any attempt to shed light onto the transfer arrangements by searching for some 'objective' formula.
200
Spahn
Recognize both the need to move toward more objectivity and the status quo by searching to realize the formula over some transition period but 'grandfathering', to some extent, existing relative positions during that time. There is little to say on options one and two, but a few remarks on smoothening the process of transition may be in order. First of all, any grandfathering provision will increase the size of the transfer pool because those below their calculated standard will have to be leveled up while those above the standard are unlikely to be brought down. This calls for a transition period in order to contain the budgetary risks of the public sector and the senior government in particular. It is also important to fix the protected (or 'grandfathered') transfer amounts in absolute terms, or, perhaps, allowing increments that are less than proportional to the overall increase of the transfer pool. The smaller the increase of protected transfers relative to the increase to the pool, the shorter will be the transition period. This will be facilitated by strong economic growth, but also by inflation unless transfers are inflation-protected by indexing. The adoption of a transition period makes the formula complicated and 'dirty' during that period. However, one should aim at designing even the transition path in a sound and transparent way. The basic idea is to phase in the new formula-based approach, and to phase out the old ad hoc relativities. If a; are the new relativities and bi are the existing relativities, with xi a; = Ci b; =I, i indicating the province, t the time subscript and T the total number of years of the transition period, the formula-guided transition process can be written in the following way: (a) Transfel;.,=
ait + bi (T - t )
T
x [~ransfer pool], ,for t = 1to T.
This formula allows to phase in the new regime as it phases out the old. It would however require those provinces that are now above the standard to reduce, perhaps, their grants entitlements below those of the previous year unless the increase of the total pool would compensate for that. So grandfathering is incomplete in this case. Full grandfathering is achieved by the following formula: (b) Transfeq, = Max
{{art +Y T - t )
x [~ransfer pool]t
However one should recognize that this would require an extra budget to fund any excess of the transfer of the previous period over the one calculated according to the straight formula (a) above.
Intergovernmental Transfers: The Vertical Sharing Dimension
CONCLUSION It is useful to distinguish the following types of transfers according to different purposes: (i) general transfers; (ii) specific transfers; (iii) special transfers. General grants are unconditional general purpose grants to balance the budgets of autonomous or quasi-autonomous public agencies. They usually fund current budgets and are taken from current budgets. Specifzc transfers are paid so that one public agency provides particular services on legal, bureaucratic, or contractual grounds, as required by another (fundsproviding) agency. They are also used as political "signaling instruments" by the donor government where policies reach beyond the realm of its own competencies. Special transfers are neither pure general budget resources nor do they compensate for interjurisdictional externalities, including the spillover of political benefits and costs. They are used to compensate for extraordinary costs, such as from local catastrophes, for the targeting of national policies, and for regional development. The funding of transfers has to respect the overall constraint on public resources, so a general principle should be the closed funding of transfers. This is achieved either through the assignment of a fixed amount or proportion of the donor government's budget or through tax sharing. There are many possibilities to limit the pool, even within a tax sharing arrangement and there is a great variety of methods that can be used to define the funding pool for intergovernmental transfers for transfers. If the pool is formed on a single or narrow resource base, there will be excessive volatility or structural shifts that impact on subnational budgets. This could jeopardize macro stability or sustainable public service delivery by subnational agencies. A broader definition of budgetary resources would counteract such budgetary risks, but perhaps limit the national government's ability to use taxes as a countercyclical policy instrument. Often transfers are not directly funded from the general budget of the donor authority, but through the budgets of its line ministries. This is particularly true for specific and special transfers. The financing of transfers through special agencies and funds entails some inefficiency risks however. Such inefficiency risks are significantly smaller for unconditional transfers funded from the general budget and allocated on the basis of objective criteria through formula apportionment. However, moral hazard might not totally be avoided even for unconditional funding fkom the general budget. Often donor government take recourse to gap filling transfers where they are compelled, by law or political pressure, to bail out subnational budgets under distress. This may occur despite firm rules reigning over the overall transfer system. There are no uniform rules for designing the machinery of a transfer system. This very much depends on traditions and political conditions with
202
Spahn
different objectives. It is also constrained by the information base and administrative procedures. Rules are often used to design general revenue grants for equalization purposes so as to foster transparency and to reach a degree of interjurisdictional fairness among recipient governments. Such transfers should display a pure income effect. There are also rules-based transfer systems that can contain specific incentives to influence the behavior of recipient governments. Such transfers exhibit an income and a substitution effect. A few principles for rules-based general transfers are sketched out in this paper and demonstrated by applying them to a fictional case: the redesign of a general transfer system for the federation of Krakozhia. It is based on objective criteria that cannot be manipulated by recipient governments, so they avoid strategic behavior that would produce inequities and economic inefficiencies. The example is for illustrative purposes only. Although the basic methodology proposed should be persuasive if one accepts transparency based on objective and non-malleable statistical data, it is likely to meet resistance. This is mainly for political reasons because the outcome of the calculations will undoubtedly deviate from the existing pattern of intergovernmental transfers. The status quo has thus a bearing on any reform that will touch upon existing grant relativities. A realistic reform will have to take such resistance into account by 'grandfathering' the position of potential losers. It will require a certain transition period in which the new regime is phased in as the old is phase out. All other options will entail opaque mechanisms driven by ad hoc political decisions.
Notes For revenue assignment and its interactions with the transfer system see, for instance, Charles E. McLure and Jorge Martinez-Vazquez, The Assignment of Revenues and Expenditures in Intergovernmental Fiscal Relations, asserted from (http://www.worldbank.org/wbi/publicfinance/documents~cClure%2O&%2OMartinez.p df). After unification Germany has increasingly used a third-round vertically asymmetric system of federal transfers. The Republic Srpska attempts some mild inter-municipal equalization by leveraging the sharing ratios for municipalities with poorer fiscal capacity. This approach to equalization is largely ineffective because it does not compensate for fiscal deficiencies: Even a full retention rate will produce nil if the tax potential is zero. See Paul Bernd Spahn, Contract Federalism, in: Handbook on Fiscal Federalism edited by Ehtisham Ahmad und Giorgio Brosio (E. Elgar; forthcoming). For instance they imposed a 21-year-old drinking age for the states' citizens, or a 55mph speed limit on interstate highways.
Intergovernmental Transfers: The Vertical Sharing Dimension
203
See Bruce A. Wallin (2001), Forces Behind Centralization and Decentralization In the United States, Commission on Fiscal Imbalance, Commission on Fiscal Imbalance, Conference Paper, Quebec, p.4. This could however be outweighed by national benefits in the case of vertical externalities. Since the pool for general grants is closed however, the volume of specific purpose payments will have to be added to this pool to obtain the full budgetary impact. So the expression "claw back" is somewhat misleading. On the structure of the Marshall Plan see Spahn, Paul Bernd (2004) The Marshall Plan: Searching for ,Creative Peace' Then and Now, John Agnew and J. Nicholas Entikin (eds.), The Marshall Plan Today: Model and Metaphor, Routledge, London and New York, pp. 191-2 13. And for the European Union see Barry, Norman (2001), Competitive federalism: The case of the European Union, The Centre for the New Europe (CNE), Brussels. Capital grants should be distinguished from interagency lendinghorrowing within the public sector. Intergovernmental transfers resulting from interagency lendinghorrowing create commitments by the recipient government to service and redeem the debt in due course. Often, however, such interagency lending is significantly at variance from market conditions, so lendinghorrowing are blurred by implicit transfers such as subsidized interests, grace periods, etc. See also Spahn, Paul Bernd (1998), "Decentralized Government and Macroeconomic Control", in Horofumi Shibata and Toshihiro Ihori (Eds.), The Welfare State, Public Investment, and Growth, Selected Papers from the 531d Congress of the International Institute of Public Finance, Springer: Tokyo, 129-150. The latter could be desirable in some instances for instance in Bosnia and Herzegovina where the national government is extremely weak. A similar effect occurs where lower tiers of government are able to divert resources from specific transfer programs to which the national government attaches high priority. These programs are thus artificially "dried out" and the national government is compelled to step in and soften budget constraints. There are indications that this strategy has played a role in South Africa. The sharing arrangements with local governments are exceptional in this instance. See footnote 13 above. Australian Government, Commonwealth Grants Commission (2004), Annual Report 2003-04, Canberra. The x-axis (population) is shown in logarithms to spread the axis at the lower end in order to bring out the smaller regions. This transformation makes the linear trend appear in the form of an exponential function. The variation coefficient is the standard deviation divided by the mean. It could of course also indicate the existence of political-patronage under the present system.
References Australian Government, Commonwealth Grants Commission. 2004. Annual Report 2003-04. Canberra. Barry, Norman. 2001. Competitive federaIism: The case of the European Union. The Centre for the New Europe (CNE), Brussels.
204
Spahn
McLure Charles E., and Jorge Martinez-Vazquez. 2002. The assignment of revenues and expenditures in intergovernmental fiscal relations. Retrieved from World Bank Web Site accessed 22 February 2005. http://www.worldbank.org/wbi/publicfinance/documents/McClureY02O&Y02OMartinez.pdf
Spahn, Paul Bemd. 1998. Decentralized government and macroeconomic control. In The Welfare State, Public Investment, and Growth: Selected Papersfrom the 53rd Congress of the International Institute of Public Finance, Horofumi Shibata and Toshihiro Ihori, eds., 129- 150. Tokyo: Springer. Spahn, Paul Bernd. 2004. The Marshall plan: searching for ,creative Peace' then and now. In The Marshall Plan Today: Model and Metaphor, John Agnew and J. Nicholas Entrikin, eds., 19 1-213. London and New York: Routledge. Spahn, Paul B a d . Forthcoming. Contract Federalism. In Handbook on Fiscal Federalism, Ehtisham Ahmad and Giorgio Brosio, eds., Elgar. Wallin, Bruce A. 2001. Forces Behind Centralization and Decentralization In the United States, Commission on Fiscal Imbalance, Commission on Fiscal Imbalance, Conference Paper, Quebec.
Chapter 8
INTERGOVERNMENTAL TRANSFERS: THE VERTICAL SHARING DIMENSION ROY BAHL AND SALLY WALLACE
'
Department of Economics, Andrew Young School of Policy Studies, Georgia State University
1.
INTRODUCTION
There is both a horizontal and a vertical dimension to the structure of an intergovernmental transfer (Bahl and Linn, 1992). The vertical share is the total pool of funds to be allocated to subnational governments, while the horizontal shares are the amounts received by individual subnational governments. Most research (and most political attention) is devoted to the latter.2 The subject of this paper is vertical sharing. We are interested in the question of whether the claim of subnational governments (SNG) on national revenues has been increasing over time. This research paper therefore investigates three issues. The first is the trend and cross-country variation in the level of the vertical share. The second is the range of the practice in vertical sharing. Third, we offer some criteria by which the practice of vertical sharing might be evaluated.
2.
DETERMINING THE VERTICAL SHARE
How should the total amount of transfers to the subnational government sector be determined? Is there a good normative rule? This is one of the first questions to be answered in designing or reforming an intergovernmental transfer system, and it is arguably also the most politically charged question. In fact, there is no one "best" way to determine the vertical share. Where you stand on the answer will depend on where you sit. A central government official is likely to have a very different view to a subnational government official about the best approach to vertical sharing. However, whether or not the question is explicitly raised or nationally
206
Bahl and Wallace
debated, every country makes a decision about vertical sharing, and it makes that decision every year. One might begin the search for the "right" approach to vertical sharing by estimating the size of the gap that is to be filled by intergovernmental fiscal transfers. The gap for the subnational government sector is the difference between (a) the amount it must spend to provide a minimum acceptable level of government services, and (b) the amount it can raise from own revenue sources if it exerts a "normal " revenue effort. This needs-resources gap (G) might be defined more precisely as
R,
where E,
= the amount of expenditure needed to provide a minimum
=
acceptable level of (assigned) services in local government i. the revenue that would be raised fiom own sources at "normal" effort in local government i.
The vertical share (VS), is equal to
where a is the percent of the needs-resources gap that the central government proposes to cover with the transfer and CR is the total amount of revenue (or tax revenue) raised by the central (or state) government. This norm of vertical balance in an intergovernmental system is more easily conceptualized than it is measured. Particularly, the measurement of the cost of providing a minimum level of services is difficult. There have been numerous attempts to make such a measurement but few countries are successful at using this approach to defining the vertical share3. In addition, the definition of a normal level of tax effort by subnational governments is very hard to establish when tax bases cannot be measured as is often the case in developing countries. The result is that few countries use a direct, objective measurement of (G) to establish a vertical share. While a vertical share certainly will be established, it is more likely to be determined subjectively than objectively. A second issue surrounding the determination of the vertical share is affordability. Expenditure need is a subjective matter and always far outweighs the capacity (or willingness) to finance these services. So, not only must the central (or state) government take the expenditure needs of
Intergovernmental Transfers: The Vertical Sharing Dimension
207
subnational governments into account, but it must also estimate the extent to which these needs can be covered by available central and local resources. Whatever the true gap between needs and resources at the subnational government level, it is unlikely to be filled completely from central government revenues. Ultimately the vertical share will almost always be determined by a bargaining over what the subnational governments want and what the central government thinks it can afford. Political consideration will always enter into the bargaining. The result of the bargaining is shown as a in equation (2), and might be thought of as a parameter of affordability.
3.
MEASUREMENT OF THE VERTICAL SHARE
One could take a purely positive approach to answer the question of whether the vertical share has been increasing. Following equation (2), empirically, the vertical share may be defined as
(5)
where
Tr = intergovernmental transfers4 Tx = taxes raised by the government making the transfer We would like to track the size of the vertical share, over time and across countries. A first question is how one defines an intergovernmental transfer, i.e., the numerator of the vertical share ratio. Our definition of an intergovernmental transfer is a grant of funds from a government that raised the revenues of another level of government. The only database that can be used for broad international comparisons of fiscal choices (that we know of) is the International Monetary Fund (IMF) compilation, Government Finance Statistics (2003, 2003a). The IMF uses the term "grant" and defines it as "a noncompulsory transfer from one government unit or international organization to a second government unit or international organization" (IMF, 2002). This definition appears to be close to the concept we want, hence we make use of the Government Finance Statistics (GFS) database in this comparative analysis. But there are qualifications to this use of the data. It is not clear from so aggregated a definition that each fiscal instrument that we would have identified as a transfer was so identified by the IMF.' A random comparison of the IMF data against that reported in case studies where the definitions appear to be correct, give approximately the same results in some instances but not in other^.^ We will assume that in general, the classification made in GFS is correct. Tr The ratio, - , is an indicator of the priority that central governments give Tx to intergovernmental transfers. In column 1 of Table 1, we present data on
208
Bahl and Wallace
intergovernmental transfers as a share of total tax revenue of the granting government for all countries for which Government Finance Statistics reported information during the 1990s.~Using the latest year for which data are available, we estimate that for industrialized countries, the vertical share is equivalent to about 19 percent of total taxes collected by the granting government(s), although there is a great deal of variation around this average. In Denmark and Australia, for example, the vertical shares are above 20 percent, while in France and Portugal, they are reported to be less than ten percent.8The average vertical share has remained approximately the same over the past three decades. Only 9 of the 22 industrialized countries for which data are available showed an increase in the vertical share. For developing countries, the average vertical share is 13.3 percent. That the vertical share is lower in developing than in industrialized countries is no ~urprise.~ The budget pressures on central governments and the limited tax collection capacity of SNGs would cause us to predict a flypaper effect: the higher level governments have the more productive tax bases and the money sticks where it hits. However, on average and in 46 of the 72 countries reporting, there has been an increase in the size of the vertical share in less developed countries. The increasing claim of SNGs on central tax revenues suggests the importance placed on their budgetary support. The vertical share for transition countries is only slightly higher than that for developing countries (Table 1). On average, it has fallen over the past two decades. A plausible explanation is that there has been an increase in reliance on subnational governments in the transition countries to raise revenues. lo
4.
DETERMINANTS OF THE VARIATIONS IN THE VERTICAL SHARE
To better understand changes in the vertical share, and differences in the ratio across countries, we make use of the following identity:
where SE= subnational government expenditures1' Y= GDP. Tr, Tx are as defined above.
Intergovernmental Transfers: The Vertical Sharing Dimension
Table 1. Alternative Measures of the Size and Determinants of Intergovernmental Transfers IntergovernmentalTransfers as Percent of Total Tax Collections of the Higher Level Governments
Subnational GDP Government Expenditures
Subnational Taxes Government as a Expenditures Percent as a Percent of GDPC of GDP
Average for Industrialized Countries
19.0
38.1
5.5
15.4
31.3
Average change per yearb
0.0
-0.1
0.0
0.1
0.3
Number of countriesa
22
25
25
21
23
Number of countries with an increase
9
7
16
15
21
Average for Developing Countries
13.3
40.1
2.2
6.4
17.5
Average change per year
0.1
0.5
0.0
0.1
0.1
Number of countries
72
18
73
32
80
Number of countries with an increase
46
11
48
16
50
Average for Transition Countries
14.2
29.4
2.9
9.5
23.9
Average change per year
-0.2
-0.2
-0.1
-0.1
-0.3
Number of countries
22
23
22
22
20
Number of countries with an increase
10
8
10
10
7
Source: Computed from IMF (2003a). Notes: a) Number of countries for which data are available. b) Based on data from 1972-2001. The earliest and latest years of available data were used for each country. c) Taxes of higher level governments making transfers to lower level governments.
210
Bahl and Wallace
Equation (3) allows us to decompose the vertical share into a transfer dependency effect, a jiscal decentralization effect, and a revenue mobilization effect, respectively. An increase in the first two drive the vertical share up, but an increase in the third will dampen it.
4.1
The Transfer Dependency Component
The vertical share will be higher, cet. par., if transfers are the primary revenue source for SNG, i.e., if
Tr SE
- is higher. This component of the
vertical share describes the fiscal dependence of SNGs on transfers from higher level governments. The results of our analysis indicate that the transfer dependency effect increased the vertical share in developing countries, but drove it down in industrialized and transition countries. In column 2 of Table 1, we present the average values for the ratio of intergovernmental transfers to SNG expenditures for all countries which reported GFS information during the 1990s. In general we find that SNGs in industrialized countries rely on intergovernmental transfers to finance about 38 percent of expenditures. Again, however, there is a substantial variation in these shares, ranging from nearly 80 percent in Ireland to ten percent in New Zealand. Subnational governments in industrialized countries became more self sufficient during the past 25 years. Note that the average reliance on grants declined by about 0.1 percent per year, but that 7 of the 25 countries in the sample actually showed an increase. This is an unexpected result since the revenue raising powers of subnational governments tend to be greater in the industrialized countries. The opposite pattern may be observed for the less developed countries in this sample. The average dependence on transfers is slightly higher than in industrialized countries, but this average hides a very wide variation. The share of transfers in SNG spending varies from over 70 percent in Peru and Indonesia to less than 5 percent in Paraguay. Over time, the average dependence on intergovernmental transfers in developing countries has increased (Table 1). Transfers were about 30 percent of SNG expenditures in the 1970s and 1980s and rose to nearly 40 percent by the beginning of the 2000s. Fiscal dependency increased in 11 of the 18 countries for which data were available (Table 1). This pattern is partly explained by the slow growth in per capita real expenditures by SNGs. Another view of this finding is that central governments in developing countries were not willing to relinquish taxing powers to SNGs, but chose instead to finance increased decentralization from centrally raised taxes. The case of the transition countries is more difficult to translate because of the blurred lines between what is a transfer and what is a local tax, and
Intergovernmental Transfers: The Vertical Sharing Dimension
21 1
because of possible misclassifications in GFS. Using the available data, however, we show in Table 1 that the reliance on transfers for financing local government expenditures is about 30 percent in transition countries, and this reliance has been declining in favor of own source revenue mobilization over the period for which data are reported. Of the 23 countries in the sample, only 8 showed an increased reliance on transfers for financing SNG expenditures.
4.2
The Fiscal Decentralization Effect
The second component in equation (3), (SEIY), is the subnational government expenditure share of GDP. All else equal, a larger subnational government share of national output will be associated with a larger vertical share. In industrialized countries, the subnational government share of GDP averages more than 15 percent and has increased slightly over the past three decades (Table 1, column 4). In the developing countries, the subnational government expenditure share of GDP is considerably lower, averaging only 6.4 percent. The expenditure share of SNGs in GDP has increased in these developing countries, therefore having a positive effect on the vertical share. In the transition countries, the subnational government spending share of GDP is 9.5 percent on average, and has declined over the period since data have been recorded suggesting, if anything, a decreasing influence on the vertical share.
4.3
The Revenue Mobilization Component
At the margin, the vertical share will be inversely related to the level of revenue mobilization by higher level governments. To explain this, let us assume that the sequencing of government decisions is to first identify the level of fiscal decentralization (SEN), and second to decide on the extent to which subnational governments will be financed by transfers (TrISE). At the margin then, a higher level of revenue mobilization will leave more resources for central government purposes and lead to a lower vertical share. The data in Table 1 show that, on average, the rate of revenue mobilization by governments making transfers has been considerably larger in industrialized countries than in developing economies. One might interpret this as showing that, at the margin, slow improvement of revenue mobilization efforts had a dampening effect on the vertical share in the developing economies. In the transition countries, by contrast, the reduction in revenue mobilization by transferring governments has resulted in an increase in the size of the vertical share.
Bahl and Wallace
5.
EXPLAINING THE VARIATION
As noted above, there is a great deal of variation among countries within these three groupings. This moves us to a more micro analysis to address the question: why do some countries choose larger vertical shares than others? There is a substantial variation across countries in the percent of taxes of higher level governments that are devoted to intergovernmental transfers. We have used a regression analysis to explain the variation in vertical share as defined in equation (2) and measured as reported in column 1 of Table 1. The dependent variable in this analysis is measured as the average for each country over the 1990s. Data are available for 20 developed and 51 developing countries. The explanatory variables that drive the explanation of variations in the vertical share are the fiscal decentralization effect, the transfer dependency effect and the revenue mobilization effect. We treat the expenditure decentralization ratio12 as an exogenous independent variable in this estimation to capture the decentralization effect. We use several specifications of exogenous variables to explain the variation due to the transfer dependency effect. A dummy variable to indicate whether a country is less developed (LDC) or industrialized. We would expect that, ceteris paribus, an LDC would allocate a lower share of central taxes to intergovernmental transfers because the needs of the central budget are thought to be more pressing. A dummy variable for federal structure, which would be expected to predict a lower dependency by SNGs on transfers and therefore a lower vertical share. Federal countries are higher income and tend to give a greater degree of taxing power to their SNGs. A higher level of corruption should be related to a greater transfer dependence because the enhanced possibility for bribery that comes with intergovernmental transfers may cause some countries to hold to this pattern.'3 Finally, the revenue mobilization effect, with an expected negative marginal effect on the vertical share, is treated as an endogenous variable. The instruments used in the first stage equation are the agricultural share of GDP, the population growth rate and the openness ratio (the ratio of imports plus exports to GDP). The results of our analysis are shown in Table 2. All of the explanatory variables have the correct signs. The ratio of transfers to central government taxes is most strongly correlated with expenditure decentralization. The vertical share is significantly higher in countries that have committed to a
Intergovernmental Transfers: The Vertical Sharing Dimension
213
Table 2. The Determinants of the Vertical Sharea: Two Stage Least Squares Estimatesc
Equation (1)
First Stage Equation for Revenue Mobilization
Constant Expenditure Decentralization Share of GDP in Agriculture LDC Dummy Federal Dummy Population growth rate Corruption Index Revenue Mobilizationa
Notes: a) Endogenous, measured as the ratio of tax to GDP b) Import plus export value as a percent of GDP. c) t-statistics shown in parenthesis.
greater degree of fiscal decentralization. The revenue mobilization effect has the expected significant, negative effect.
Bahl and Wallace
6.
THE PRACTICE
The shares of higher level government revenues allocated to subnational governments are determined primarily by the institutional arrangements that are used for vertical sharing. These institutional arrangements may be determined as much by historical practice as by the demand for decentralization in a country. In fact, institutional arrangements for intergovernmental transfers differ widely from country to country. In some countries the intergovernmental transfer "system" is an amalgamation of several different types of transfers that may or may not fit together in a coherent way. In other countries, one form of transfer is dominant e.g., shared taxes in Indonesia and the Philippines. Either way, there is an implicit or explicitly defined vertical share for each type of intergovernmental transfer. Governments seem to have taken three basic approaches to defining the vertical share of SNGs (Bahl and Linn, 1992). The first is to share a defined percent of revenues (or taxes) of the higher-level government. This is effectively a revenue entitlement for SNGs, and it may or may not bear a relationship to expenditure needs or to local taxable capacity. The second is to use an ad hoc approach where the vertical share is defined by a discretionary decision, and may vary from year to year. The third approach is for the higher-level government to agree to cover a portion of "allowable" costs. In fact, these are three very different approaches in terms of how they are administered and in terms of their impact.
6.1
The Shared Tax
Arguably, the form of vertical revenue sharing that is most in step with the goals of fiscal decentralization is the shared tax approach. In this case, the higher level government allocates a share of its tax collections to the lower level governments. As may be seen in Table 3, there are many different versions of tax sharing. Vertical sharing by this method can significantly strengthen the fiscal position of the SNG sector. In effect, this gives state and local governments a claim on some share of national revenues and makes them partners in the central tax system. It provides some degree of certainty as to the revenue flow to local governments, and it could give local governments access to broad-based and income-elastic taxes. From the point-of-view of the central government, the shared tax approach could seriously limit fiscal flexibility. Even in the face of a serious fiscal deficit, the center would be obligated to pass down a share of its revenues to support subnational government spending.14
Intergovernmental Transfers: The Vertical Sharing Dimension
215
Two major design questions arise here. The first is the tax to be shared and the percentage of collections to be shared. The second is for the granting government to design a system that it can support by following the revenue sharing rules that it makes The choice of a shared tax will depend in part on the commitment of the central government to tax sharing. Countries that are pushing subnational governments to be important players in the public service delivery system might choose major revenue sources with income-elastic tax bases. At one extreme, countries may share all tax collections with their local governments. The cornerstone of the Indonesian decentralization program that took effect in 2001 was a 25 percent sharing of all "domestic" revenues. The Philippines allocates 40 percent of the total internal tax collections (in the third preceding year) to local governments. As may be seen in Table 3, this approach is not an uncommon choice among developing countries. A middle ground is to share specific taxes, but to choose those that are revenue productive and elastic. China, for example, designates 100 percent of the enterprise income tax, 60 percent of the individual income tax and more than 25 percent of the VAT for provincial governments. Latvia earmarks 75 percent of the revenues from the personal income tax for local governments. A third approach is to assign the revenues from less productive central taxes to subnational governments, or to assign only a small percent of collections. The property tax is a central government tax in Indonesia. All revenues, however, are turned over to the local governments. A similar situation exists in Jamaica, though less than 100 percent of property tax revenues are shared. It is not a big surprise that many advocates of fiscal decentralization are enthusiastic about shared taxes as the best approach to vertical sharing. It gives subnational governments a fixed claim on central revenues, and perhaps access to an income-elastic tax base. From the point of view of local politicians, it offers the prospects of seeing increased revenues without having to ask the voters for permission. An added desirable feature is that shared taxes are very often passed to SNGs as unconditional grants. There are, however, some significant drawbacks to using the shared tax method of determining the vertical share. First, SNG revenues will be sensitive to central government tax policy changes. For example, China shares enterprise income tax revenues with the provincial governments on a basis of origin of collection. Changes in Chinese industrial policy directly effect the revenue flow from enterprises and hence revenues received by the subnational governments. Second, a high tax-sharing rate may dampen the enthusiasm of the central government for vigorous enforcement of the tax (if it is the collection authority) thereby reducing the revenue flow to local governments. Moreover, a shared tax may limit the incentive for the central
216
Bahl and Wallace
government to reform the structure of the tax administration if the central government itself does not receive any of the revenue from the enhancement. The central government property tax in Indonesia is a case in point where the center administers the tax but the revenues accrue to the local governments. There is little incentive for the center to invest in administrative or structural reform. A similar situation arises for individual and company income taxes in China, and for the individual income tax in ~ussia.'~ Third, there is the issue of whether the subnational government share is measured against budgeted or actual central government revenue. If the base is actual revenue and if the central and local fiscal year calendars are the same, there is some degree of uncertainty in the revenue flow and fiscal planning may be compromised. If the base is the budgeted amount, both the central and subnational governments assume some risk, especially in the case of countries that are revenue dependent on products sold in world markets. The Philippines has taken an interesting approach to this issue by sharing 40 percent of central revenues collected three years earlier, i.e., 2002 sharing is based on 1999 collections. Fourth, shared taxes are a tempting target for tax avoidance. If a SNG receives only a partial share of collections within their jurisdiction, there is incentive to find ways to understate the actual amount of taxes paid. In China, local officials have found creative ways to channel collections to extra budgetary accounts and therefore avoid a protion of the tax sharing payments to the center. Some estimates place the revenues transferred to these extra budgetary accounts as being equivalent to 50 percent of budgetary collections (Bahl, 1999a). In Russia, the "dual subordination" of tax administration has resulted in a tendency of subnational governments to hold onto shared revenue instead of sending it to Moscow, thus underminding the central government (Martinez-Vazquez, 2001). Fifth, Keen (1998) makes the point that changes in the centrally controlled tax rate of a shared tax may induce subnational governments to substitute higher local taxes if central tax rates fall. In many countries, this behavior may not be relevant because subnational governments have little revenue autonomy, but it may be relevant in other cases. Keen's argument does bring into question the impact of tax reform at the central level on the transfer policy. If central tax rates are reduced, is the vertical share reduced as we11?16 Finally, tax sharing raises political problems. What happens when the higher level government, who makes the tax sharing rules, runs into a tight fiscal situation? Will it view the sharing rates as an inviolate contract, or will it cut them back to preserve some of its own programs? The record is mixed on this. In some past years, the Philippine government has provided less than its full commitment to subnational governments. Capuno, Manuel and
Intergovernmental Transfers: The Vertical Sharing Dimension
217
Salvador (2001) calculate that during the 1995-1999 period, the actual grant allocation averaged only about 12 percent vs. the 40 percent entitlement. The same kind of default on the tax sharing arrangement has occurred in Ecuador (Frank, 2003). In most industrialized countries, by contrast, the percentage sharing arrangements seem to have been protected, even in difficult economic times. The tax sharing approach addresses some issues but not others. It can be used to significantly improve vertical balance in the intergovernmental fiscal system and, if income-elastic taxes are shared, to maintain this new balance. It also has the advantage of giving the subnational governments a vested interest in the revenue performance of the central government tax system. Whether or not this feature can be used to stimulate collections at the local level is an open question. Tax sharing is not a good instrument to address the externality issue. For example, if subnational governments as a group are under-investing in primary education, a shared tax will be a less effective remedy than a conditional grant. Nor is it likely to significantly increase the accountability of local politicians to their constituents because it allows local politicians to divorce themselves from the political pain of setting the tax rate. The tax sharing approach might produce a larger pool of funds for equalization, but whether or not it leads to more equalization will depend on the method of horizontal sharing that is chosen.
6.2
Ad Hoc Transfers
A second approach to vertical sharing is for the central government to decide on the amount of transfers on a discretionary basis. Whereas the shared tax approach gives subnational governments an entitlement, in effect an ownership of some share of central revenues, the ad hoc approach sends an opposite message: the center owns all of its revenues and may or may not choose to grant some share to the local sector. This is a centralizing approach to determining fiscal balance. At one extreme, the Parliament or the President will decide annually on an allocation to the subnational government sector. This approach to determining the vertical share often involves more negotiation and political consideration than subjective analysis and both the approach taken and the amounts agreed upon may vary from year-to-year. Hence it might be labeled an "ad hoc approach." Obviously, there are great drawbacks to this approach. First, it is not transparent and it is susceptible to political manipulation. This leads to uncertainties on the part of the local government sector as it does not know what it will receive each year. Fiscal planning and effective budgeting are more difficult, if not discouraged.
Table 3. Shared Tax Determination of Vertical Structure Country
Shared Tax Arrangement
Comments
Pakistan
37.5 percent of federally collected taxes
Customs duties included in the base
Indonesia
25 percent of domestic tax revenues collected
Base does not include customs revenue
Hofman (2003).
Philippines
40 percent of internal revenue collections in the third preceding year
Base does not include customs revenues.
Diokno (2003).
Argentina
58.05 percent of net coparticipated funds
Base does not include customs revenue.
Rezk, Emesto (2000).
20 percent of federal revenues: 47 percent of income taxes and 57 percent of federal VAT taxes
Base does not include customs revenue
Afonso, Jose Roberto R. and Luiz de Mello (2000).
1I percent of current revenue and 25 percent of income taxes
Base includes foreign trade taxes
Frank, Jonas (2003).
49.5 percent of VAT and 50 percent of the revenue from income taxes
Base does not include customs revenue
Spahn, Paul Bemd and Oliver Franz (2000).
Brazil
Year
1988 Constitution
Ecuador
Germany
2000
Source
Table 3. Shared Tax Determination of Vertical Structure (Continued) Country
Year
Shared Tax Arrangement
Comments
Source
India
2000
87.5 percent of income tax and 47.5 percent of union excise tax
Base does not include customs revenue.
Rao, Govinda and Nirvikar Singh (2001).
Kazakstan
1999
Sharing rates for VAT, excise, enterprise income, and individual income are set annually. Rates differ for each Oblast
Base does not include customs revenue.
Mclure, Jr., Charles E. (1999).
Korea
13.27 percent of the national tax revenues
Mexico
20 percent of "assignable taxes:" revenues collected from most domestic taxes (main components are federal income tax, VAT, and the ordinary fees from oil).
Ma, Jun (1997). Base does not include customs revenue.
Courchene, Thomas and Alberto DiazCayeros (2000).
Table 3. Shared Tax Determination of Vertical Structure (Continued) Country
Year
Shared Tax Arrangement
Comments
Source
Nigeria
1999
24 percent of the Federation Account and 50 percent of VAT revenues.
Base includes customs revenue if they flow into the Federation Account
Martinez-Vazquez, Jorge and Jameson Boex (2001).
Russia
1998
15 percent of federal collections
Excludes import duties and 10 percent federal share of PIT.
Martinez-Vazquez, Jorge and Jameson Boex (2001).
South Africa
2000
Equitable share of nationally collected revenues
Switzerland
1992
30 percent of Federal tax on income and profits, 10 percent of withholding tax, 20 percent of tax on exemption fkom military.
Smoke, Paul (2000).
Does not include customs revenue. It is required that 85 percent 85 percent of the co-participated revenues must be used towards investment expenditure (defined as all new expenditures of municipalities except for wages, salaries, and other personnel expenditure.
Spahn, Paul Bernd.
Table 3. Shared Tax Determination of Vertical Structure (Continued) Country
Year
Shared Tax Arrangement
Comments
Source
Bolivia
1996
20 percent of taxes collected by the national government
Base includes customs duties.
Mackenzie, G.A. and Jose-Luis Ruiz (1997).
Situado Fiscal: 24.5 percent in 1996 Participacion Municipal: According to 1991 Constitution this figure should have increased to 22 percent in 2002.
Situado Fiscal: Total amount determined as a minimum share of the nation's total current revenues. Participacion Municipal: Transfer is determined as a share of total current revenues of the national government.
Ahmad, Ehtisham and Katherine Baer (1997).
Shared revenues of taxes jointly levied and collected by federal and state government is still to be defined. General non-conditional grants extended until 1997.
General non-conditional grants are determined by a formula and ad hoc adjustments. Resources come from total tax and non-tax revenues, counterpart funds, and foreign assistance to the states.
Brosio, Giorgio, and Sanjeev Gupta (1997).
Columbia
Ethiopia
1991 Constitution
Table 3. Shared Tax Determination of Vertical Structure (Continued) Country
Year
Shared Tax Arrangement
Comments
Source
Latvia
1995
100 percent of personal Base does not include income tax and property customs revenue. tax. 75 percent of natural resource tax revenues.
Martinez-Vazquez, Jorge and Jameson Boex (2001).
China
1994 Revenue Sharing Arrangement
100 percent of Personal income taxes and enterprise income tax on locally owned enterprises, collectives, private enterprises, and joint ventures; 25 percent of VAT on domestic transactions; 50 percent of Securities and exchange tax. Resource taxes.
Base does not include customs revenue.
Bahl(1999a).
4.3 percent of total government revenue was shared taxes.
The only certainty in the transfer system is that the local governments will receive at least last year's amount of shared tax.
Weist, Dana (2003).
Thailand
Table 3. Shared Tax Determination of Vertical Structure (Continued) Country Japan
Year 2003
Jordan
Papua New Guinea
1995 Provincial Reforms
Australia
2001
Shared Tax Arrangement
Comments
Source
32 percent of personal income tax. 35.8 percent of company income tax. 29.5 percent of consumption tax. 25 percent of revenue from to "Fuel tax"
Base does not include customs revenue.
Ihori, Toshihiro (2003).
Base includes customs revenues. Distribution determined by formula.
World Bank (1995).
Starting in August of 1999, no more than 30 percent of VAT revenues and 30 percent of mining levies Starting in July 2000, all collections of Goods and Services Tax (less collection costs) are distributed as untied grants to States and Territories.
Base does not include customs revenue.
Edmiston, Kelly. (1999).
The GST is a VAT applicable to all activities except some educational and health related expenditures.
Searle, Bob (2002).
Table 3. Shared Tax Determination of Vertical Structure (Continued) Country Canada
Year 1995
Shared Tax Arrangement
Comments
Source
Personal Income Tax and Corporate Income Tax are shared
Personal Income Tar: Tax on Tax rate ranges from 40 to 60 percent among nine provinces that participate. Quebec receives 16.5 percent of federal PIT. Corporate Income Tax: Provinces use federal tax base, but determine their own tax rates and credits which ranged fkom 14 to 17 percent. Does not include customs revenues.
Krelove, Russell, Janet Stotsky and Charles L. Vehom (1997).
35 percent of the PIT collected by central government
Lutz, Mark, Edgardo Ruggiero, Paul Bemd Spahn, and Emil M. Sunley (1997).
Intergovernmental Transfers: The Vertical Sharing Dimension
225
Second, the ad hoc approach makes it easier for the central government to treat the expenditure needs of the subnational government sector as having a lower priority than its own. When there is no statute or constitutional mandate for vertical sharing, the central government may view intergovernmental transfers as just another competing expenditure request (just as those from line agencies) and cuts (and expansions) are more easily made than in the case where the vertical share is defined as a percent of central government taxes. In this setting, the reduction in transfers can be another way to offload central government budget deficits. Finally, the ad hoc approach makes it easier to deny the link between expenditure responsibilities and revenue resources. While the central government may cut or increase the local revenue share each year, they are less likely to change the expenditure functions assigned to local governments. The result can be a growing vertical imbalance that can produce harmfbl effects on the level of public services provided. Another result is that subnational governments are likely to be discouraged from increasing efficiency and from becoming self-reliant if all grants are made on an ad hoc basis. Local officials will feel less in control of their budgets and less accountable to their voters for the level of services provided. It becomes very convenient to blame any service delivery shortfalls on the inadequate resources provided by the center. For this reason, many elected local government officials will look kindly toward the ad hoc approach. On the other hand, the ad hoc approach offers some advantages, particularly in developing and transition economies. First, from the point of view of the central government, it provides maximum flexibility in carrying out macroeconomic fiscal planning. The government can implement a fiscal program with a minimum regard for a fixed committed share to the local government sector. For example, if a deficit reduction program calls for a tax increase of x percent, the increase can be accomplished without having to pay a fixed share of the increment to the local government sector. If an expenditure austerity program calls for cuts in subnational government spending, the central government can accomplish this by simply reducing the transfer since it is not bound to a guaranteed tax share. In Sierra Leone, the decentralization legislation allows for this to be done "equitably" by specifying that any reduction in subnational government funding cannot be greater than the reduction in funding provided to central government budget agencies.I7 A second advantage is that this approach will enable the central government to change national spending priorities without changing the expenditure assignments of each level of government. For example, subnational governments are more likely to spend for consumption than for infrastructure purposes. An ad hoc grant will allow the center to reduce the
226
Bahl and Wallace
flow of revenues to the local sector and use the funds directly to spend on central infrastructure or other projects. Third, the ad hoc approach allows an adjustment of the subnational government claim on revenues, as the situation in the country changes. Shared tax provisions, particularly when placed in constitutions, are not easily changed. In sum, the ad hoc approach to determining the size of the distributable pool is the most centralizing approach to designing an intergovernmental transfer system. Despite some very apparent flaws it is widely used, even in some countries that feature decentralization as part of their development plan. Examples abound of the ad hoc approach to vertical sharing (See Table 4). The FFSR (Fund for Financial Support of the Regions), which is Russia's premier system of intergovernmental transfers, was in the past stated each year as a percent of central government revenues.'' However, the percent chosen was an annual central government decision, presumably based on the fiscal position of the central government. The funding rate (as a percent of total tax collections less import duties) was 15 percent in 1996 but had fallen to 13 percent in 1999. Such changes are large enough to compromise the budgetary stability of local governments.
6.3
Cost Reimbursement
A third approach to determining the size of the revenue pool for distribution to local governments is the cost reimbursement approach. Under this approach, the central government defines a service for which it will guarantee to cover some portion of the cost incurred by SNGs in delivering certain services, for example, teachers' salaries, drugs and dressings, highway construction and maintenance, infrastructure projects, etc. (Table 5). The program could be "open-ended," i.e., the reimbursement could be unlimited and cover all expenditures on the functions made by SNGs. In this case, once the eligibility and reimbursement rules are established, the vertical share may be determined by simply adding up the entitlements of the eligible units. In effect the SNGs determine their horizontal shares and this determines the aggregate vertical share. Probably the more common approach is to first determine the total amount that will be spent to reimburse costs incurred on the specified functions, based on affordability, then "cut the cloth" in terms of reimbursement and eligibility. The grant typically is "closed-ended" as far as its size to each recipient and the definition of the function being funded. The cost reimbursement approach is likely to involve a large number of
Intergovernmental Transfers: The Vertical Sharing Dimension
227
conditional grants that are controlled by the line ministries and are continued from year-to-year. The great advantage, and disadvantage, of the cost reimbursement approach to vertical sharing is its conditional nature. It specifies how much the national (state) government desires to spend on a particular public service area. Thus the cost reimbursement grants can be used to direct investment to high priority national needs. Local governments, left to their own devices, will ignore externalities and under spend on services with regional and national benefits. The disadvantage of the cost reimbursement approach to determining vertical shares is that it compromises local choice. This can retard true fiscal decentralization because it limits the budgetary discretion of local governments. The true-believer decentralist would much prefer an unconditional grant. Moreover, the cost reimbursement approach imposes an administrative cost on the central government, who must monitor the program, and a compliance cost on the recipient SNGs who must do significant reporting on their use of funds and their adherence to standards. It is more cumbersome and more costly to administer than is either the shared tax or the ad hoc approach.
EVALUATION OF VERTICAL SHARING How does a country evaluate its practice of vertical sharing? Are there criteria for a "good" practice that might be used? Are there norms, similar to those used for taxation, that might be applied? The answer on all counts is that there are not easily defined indicators of "good" or "bad" practice. It is possible to regularly reconsider the practice by benchmarking it against some important indicators of performance. Arguably the four most important, discussed below, are whether there is the right balance between transfers and local taxes, revenue adequacy, revenue elasticity and administrative costs.
7.1
Transfers versus Local Taxes
Determination of the right balance between locally raised revenues and intergovernmental transfers is a key decision in designing an intergovernmental fiscal system. A solution that is weighted more heavily toward autonomous SNG revenue-raising powers introduces more accountability at the local level, and can make it possible to impose a hard budget constraint on local governments. Because certain tax bases can be more easily assessed and collected by local governments, the overall rate of
Table 4. Ad Hoc Determination of the Vertical Share Country
Year
Ad Hoc Arrangement Non-reimbursable transfers mostly in the form of grants; and "reimbursable" discretionary transfers consisting of transfers made through FONAVI (National Housing Fund) and Treasury advances against future tax revenues.
Comments
Source
Non-reimbursable grants: mainly used to fill resource gaps at the provincial level. Reimbursable transfers: effectively nonreimbursable because of their low rate of repayment (loan recovery rate is less than 10 percent).
Schwartz, Gerd, and Claire Liuksila (1997).
Argentina
1997
Columbia
1997
Co-financing funds
Has an ad hoc nature because the 1991 constitution did not require that these funds increase at a rate in relation to the central government current revenue.
Ahmad, Ehtisham and Katherine Baer (1997).
Brazil
1997
Covenios
Ad Hoc grants used to fund education, health and social areas
Teresa TerMinassian (Ed.) (1997). "Brazil."
Table 4. Ad Hoc Determination of the Vertical Share (Continued) Country
Year
Ad Hoc Arrangement
Comments
Source
India
2000
Finance Commission decides on tax shares. In addition, the Union Excise Duty share is optional. Various ministries give grants to their counterpart states for specified projects; this amounted to 20 percent of transfers.
In 1998 Grants to States were 16.3 percent of total revenue.
Rao, Govinda and ~ i & k a rSingh (2001).
Mexico
1999
President's discretionary fund for natural disasters and salary increases (is being phased out).
3.5 percent of total transfers to subnational governments
Nigeria
1999
Discretionary Recurrent Transfers and Discretionary Capital Transfers
Discretionary Recurrent Transfers: made to meet specific recurrent needs. Discretionary capital transfers: federal grants given for specific purposes in the context of the national development plan to finance expenditures, or transfers that represent onlending of borrowing by the federal government.
Courchene, Thomas and Alberto DiazCayeros (2000). Mered, Michael (1999).
Table 4. Ad Hoc Determination of the Vertical Share (Continued) Country
Year
Ecuador
Ad Hoc Arrangement Significant discretionary transfers
Comments
Source
Sometimes transfers are actually bailouts.
Frank, Jonas (2003).
Tanzania
2003
Total amounts are determined annually as part of the budgetary process.
Numerous conditional grants
Boex, J., R. Bahl, J. MartinezVazquez, and L. Rutasitara (2003).
Pakistan
1997
Recurrent Grants
Recurrent Grants: higher level of government may subsidize a particular activity (e.g. primary education).
Ahmed, Qazi (1997).
Mutual settlements
Employed to compensate subnational governments for tax changes or the imposition of expenditures. Dominant form of nonequalizing transfers that are most often not budgeted.
Martinez-Vazquez, Jorge and Jameson Boex (2001).
Russia
Table 4. Ad Hoc Determination of the Vertical Share (Continued) Country
Year
Ad Hoc Arrangement
Thailand
FY03
All yearly transfers are in a sense ad hoc.
Indonesia
2003
Ad hoc transfer for regional wage increases
Jordan
1995
Distributes ad hoc grants (or zero interest loans)
Malawi
200 1
Resource Supplementary Grant
Comments
Source
There is no allocation rule nor formula in their yearly transfers.
Weist, Dana (2003). Hohan, Bert (2003).
Determined each year based on budget availability Main grant distributed on a formula basis. However, it's timing and frequency is erratic.
World Bank (1995). Martinez-Vazquez, Jorge and Jameson Boex (2001).
Table 5. Cost Reimbursement Determination of the Vertical Structure Country
Year
Cost Reimbursement
Comments
Source
Argentina
2000
Earmarked transfers under the form of diverse funds (energy, housing, regional disequilibria, education); as well as transfers for decentralized services, road construction, or provincial social security regimes
Brazil
1997
Revenue from COFINS (special contribution levied on enterprise turnover)
Earmarked for financing health and some other social programs.
Teresa TerMinassian (Ed.) (1997) "Brazil.".
Colombia
1997
Situado Fiscal, Participation Municipal, Co-financing Funds, and National Royalties Fund
Situado Fiscal: Transfers earmarked for expenditures on health and education. Participacion Municipal: Special transfers to local governments earmarked for education, health, water provision, sports, recreation and culture. Co-financing Funds: Transfers made mostly on a matching basis National Royalties Fund: distributes natural resource royalties to producing regions for investment purposes.
Ahmad, Ehtisham and Katherine Baer (1997).
Rezk, Emesto (2000).
Table 5. Cost Reimbursement Determination of the Vertical Structure (Continued) Country
Year
Cost Reimbursement
Comments
Source
Bolivia
1992 Census11996 Decentralization Law
1992 Census mandates that 5 percent of national government revenue to be transferred automaticallyto public universities. 1996 Decentralization law reforms established prefecturas (departmental governments) that are assigned royalties from forestry and petroleum and minerals extraction, as well as 25 percent of Special Tax on Hydrocarbons.
Revenue going to prefecturas is earmarked for road construction, rural electrification, irrigation infrastructure, environmental preservation, tourism, social assistance programs, institution building (for municipalities), other projects in conjunction with municipalities, human resource management, administration in the health, education and social assistance areas.
Mackenzie, G.A., and Jose-Luiz Ruiz (1997).
Matching transfers from Finance Commission were almost 15% of total state expenditures. Planning Commission makes grants and loans for implementing development plans. National Development Council calculates Planning Commission grants on the basis of the Gadgil formula, which is currently at 30 percent of plan outlay.
In 1998 Grants to States were 16.3 percent of total revenue.
Rao, Govinda and Nirvikar Singh (2001).
India
Table 5. Cost Reimbursement Determination of the Vertical Structure (Continued) Country
Year
Cost Reimbursement
Comments
Source
Mexico
1999
Aportaciones-Conditional Grants
46.3 percent of subnational government revenue. Earmarked transfers for education, health, social infrastructure, and other uses.
Courchene, Thomas and Alberto DiazCayeros (2000).
Pakistan
1997
Development Grants
Takes place out of the Annual Development Plan.
Ahmed, Qazi (1997).
Russia
2001
Subventions
Earmarked for capital expenditures or current expenditures allocated by the State Duma
Martinez-Vazquez, Jorge and Jameson Boex (2001).
Thailand
FY03
25 percent of grants were specific
Grants were roughly 8 percent of total govemment revenues.
Weist, Dana (2003).
Japan
2003
National Disbursements and Specific traffic safety grants.
National Disbursements: Allocated to finance part or all of the expenses related to specific expenditure programs (education, social welfare, public works, transportation, regional development, etc.)
Ihori, Toshihiro (2003).
Table 5. Cost Reimbursement Determination of the Vertical Structure (Continued) Country
Year
Cost Reimbursement
Comments
Source
Philippines
1991 Local Government Code
At least 20 percent of Internal Revenue Allotment (IRA) block grant
The only conditional nature of the IRA transfer program is that at least 20 percent must be spent on local development projects contained in the local development plans.
Diokno, Benjamin E. (2003
Jordan
1995
Central governments pay 70 percent of the cost for building and maintaining major highways.
Papua New Guinea
1995 Provincial Reforms
Administrative Grants; Staffing Grants; Infrastructure Grants; Village Services Grants; District and Provincial Support Grants.
World Bank (1995).
Administrative Grants: Administrative Costs other than salaries and allowances. Staffmg Grants: Covers salaries and Allowances of provincial and district staff. District and Provincial Support Grants: Intended to support rural action programs and urban rehabilitation programs.
Edmiston, Kelly (1999).
Table 5. Cost Reimbursement Determination of the Vertical Structure (Continued) Country
Year
Cost Reimbursement
Comments
Source
Australia
1997
Specific purpose payments that cover both recurrent and capital needs
Conditions for expenditures fall into the following categories: 1)General Program Requirements (e.g. requirement that states provide free public hospital treatment to Medicare patients); 2)Requirements that the payment be spent for a specific purpose (or passed on to other entities such as Universities, nongovernmental schools, and local governments); 3) Agreements covering service provision and program delivery; 4) Detailed conditions on the operation of joint expenditure programs. The main functional expenses financed by these grants are education, health, and housing.
Craig, Jon (1997).
Table 5. Cost Reimbursement Determination of the Vertical Structure (Continued) Country
Year
Cost Reimbursement
Comments
Source
Malawi
199711998
Special Grants and Health Grants. District Development Fund (DDF)
Special Grants and Health Grants: Issued to operate clinics. DDF: Financing facility established by the government and donor organizations to provide grants to communitydriven projects initiated at the district-level.
Martinez-Vazquez, Jorge and Jameson Boex(2001).
Canada
1996
Specific Purpose Transfers comprised of the Established Programs Financing (EPF) and Canada Assistance Plan (CAP)
EPF: Allocated Solely to health care and postsecondary education. CAP: open-ended and matching at 50 percent *Note: These programs were replaced at the end of fiscal year 1996 with the Canada Health and Social Transfers program.
Krelove, Russell, Janet Stotsky and Charles L. Vehom (1997).
Germany
1997
Federal co-financing of specific state projects
Conditional grants that operate within a complex network of interstate cooperation.
Spahn, Paul Bemd and Wolfgang Fottinger (1997).
Table 5. Cost Reimbursement Determination of the Vertical Structure (Continued) Country
Year
Cost Reimbursement
Comments
Source
HwPY
1994
Normative Grants; targeted grants; transfers from social security; grants for distressed localities
Normative Grants: The budget specified a unit costs for 27 norms (e.g. Public Housing, Social day care, cultural activities, education, etc.). After the local govemments have fulfilled the criteria for these norms, the local govemments can use the remainder at their discretion. Targeted Grants: earmarked for investment. Social Security: Expenditure related "soft" fiancing.
Lutz, Mark, Edgardo Ruggiero, Paul Bemd Spahn, and Emil M. Sunley (1997).
Conditional grants are the largest source of revenue for ordinary regions (around 44 percent of total revenues in 1993).
Two main earmarked funds provide funding for health services and local public transportation
Emiliani, Nicoletta, Sergio Lugaresi and Edgardo Ruggiero (1997).
Transfers dominated by conditional grants-in-aid
Usually close ended with matching requirements
Spahn, Paul Bemd and Wolfgang Fottinger (1997b).
Italy
Switzerland
1997
Table 5. Cost Reimbursement Determination of the Vertical Structure (Continued) Country
Year
Cost Reimbursement
United Kingdom
1997
Specific Grants
United States
1997
U.S. relies mostly on conditional grants
Comments
Source Potter, Barry(1997). "United Kingdom."
Four most important categories are health, income security, education and training, and transportation.
Ma, Jun (1997).
240
Bahl and Wallace
revenue mobilization might be enhanced. These are desirable features in a system of fiscal decentralization. The major drawback to relying more heavily on local taxation (vs. grants) is that fiscal disparities might be increased to unacceptable levels. Moreover, SNGs might be induced to go on a search for taxes, the burden of which they might export to other jurisdictions. Since labor is not hlly mobile within developing countries, the market may not automatically correct such policy moves by SNGs. The argument to weigh local government financing more heavily toward transfers can be made in terms of the greater efficiency of the central government as a tax collector. The problem with this justification for transfers is that over time it can become a self-fidfilling prophesy: if SNGs are not given taxing powers, it is not clear that they ever will improve their tax administration capabilities. Moreover, it is not clear that the central government is a more efficient collector for all taxes (McLure, 1998). Some argue that familiarity with the local economy might give an advantage to local governments in the case of property taxes and taxes on small businesses.
7.2
Revenue Adequacy
A primary justification for intergovernmental transfers is to correct vertical imbalance. If SNGs are assigned expenditure responsibilities that cannot be financed with the potential yield of the revenue sources they have been assigned, then transfers are called on to fill the gap. A first evaluation criteria for the intergovernmental transfer system is whether this gap has been filled according to government intentions. Evaluation is especially difficult here because the degree to which expenditure needs are satisfied is a very subjective matter. The government might define a floor level of expenditures to be covered, as in South Africa or some education spending in the U.S., and the coverage of these levels might be monitored. Or, the "adequate" vertical share might be defined in terms of an entitlement from centrally collected taxes, as in Indonesia and the Philippines, and this might be monitored. However, the best evaluation of the adequacy of the vertical share is whether the allocation is sufficient to permit the targeted minimum service levels to be met.
7.3
Revenue Elasticity
An often overlooked dimension of revenue adequacy is revenue elasticity, i.e., do transfers grow in step with the increase in SNG expenditure needs? This is an important consideration. If transfers do not grow with expenditure needs, vertical balance in the system may erode. This has been a problem in
24 1
Intergovernmental Transfers: The Vertical Sharing Dimension
many countries. It is particularly problematic when SNGs must rely on the higher level governments to make discretionary changes each year to maintain revenue adequacy. We have calculated the elasticity (buoyancy) of intergovernmental transfer revenues with respect to G D P ' ~for all those countries for which data are available. As may be seen in Table 6, this elasticity would appear to be greater than unity for most developing countries in the sample. On average, it is 2.7 for a period covering roughly the decade of 1990s. It is, however, significantly higher for developing than for industrialized economies. To better understand the pattern of growth, we decomposed the elasticity of intergovernmental transfers with respect to GDP
[$]
, into
where Tr = intergovernmental transfers Tx = tax revenue of the granting government The first component is the elasticity of transfers with respect to the tax revenue of the granting government(s). We might think of this as a kind of "rate effect," i.e., a central or state government that increases the share of its tax revenue paid to the distributable pool for local governments might be thought of as increasing the of distribution of intergovernmental transfers. This would result from discretionary actions such as increasing the tax sharing rate, or increasing the ad hoc allocation of transfers to the distributable pool. The second component is the elasticity of higher level government tax revenues with respect to GDP. We might think of this as a base elasticity. An elasticity greater than unity implies an automatic increase in the effective tax rate, due to some combination of a progressive tax structure, improved administration and the development of easier tax handles. So, a country with a shared tax system that makes no discretionary changes in the sharing rates would see an increase in transfers driven by the increase in GDP. As can be seen from the data presented in Table 6, the elasticity of transfers with respect to GDP is about unity for most countries in the sample. There are some significant outliers but these would appear to be mostly due to policy changes in intergovernmental transfers rather than to built-in changes in the system. The elasticity is unity or higher for 29 of the 39 countries for which data are available, though it averages 0.9. One could
Bahl and Wallace
242
argue that most countries have more or less maintained their share of transfers in GDP over this period. The implication is that central (and state) governments are satisfied with the level of the vertical share and most have not been willing to reduce it dramatically. Nor have they been receptive to programs that would pass a greater share of tax collections by higher level governments to subnational governments. This probably indicates a continuing support of subnational government finance in development strategies, but also may indicate a hesitance to give up taxing powers to lower level governments. The fact that the overall average is less than unity suggests a drag on the vertical share in some countries and less enthusiasm about increasing the share for SNGs.
Table 6. The Growth of Intergovernmental Transfers as a Percent of GDP: Rate and Base Effects for Selected Countries Country
Algeria Australia Azerbaijan Bahamas, The Belize Botswana El Salvador Ethiopia Fiji Gambia, The Georgia Ghana Guatemala Guinea Guinea-Bissau India Jordan Kenya Kyrgyz Republic Latvia Lesotho Malawi Malaysia New Zealand Nigeria Oman
Rate Elasticity
Base Elasticity
Total Elasticity
0.7 0.9 0.7 4.5 0.4 8.2 9.5 -0.3 0.1 2.5 3.4 5.4 1.5 1.2 0.4 1.6 -0.2 0.2 -0.1 0.2 1.7 4.7 0.5 0.4 -1.7 1.5
1.3 1.4 0.6 1.4 1.4 1.1 0.1 1.O 1.5 1.O 0.9 0.9 1.3 0.9 2.5 0.8 1.6 1.8 0.7 1.2 1.6 1.O 1.1 1.5 0.6 0.7
0.9 1.2 0.4 6.1 0.6 8.8 0.6 -0.3 0.2 2.5 3.0 5.2 1.9
1.1 0.9 1.3 -0.3 0.4 -0.1 0.2 2.6 4.5 0.5 0.6 -1.0 1.O
243
Intergovernmental Transfers: The Vertical Sharing Dimension
Table 6. The Growth of Intergovernmental Transfers as a Percent of GDP: Rate and Base Effects for Selected Countries (Continued) Country
Pakistan Papua New Guinea Philippines Portugal Senegal Solomon Islands Sri Lanka Sudan Swaziland Turkey Uganda United Kingdom Zambia Zimbabwe Average
Rate Elasticity 2.4
Base Elasticity 1.2
Total Elasticity 2.8
3.5 1.8 2.8 5.8 0.7 2.2 0.4 5.0 0.1 3.2 0.2 0.2 1.9 2.0
1.8 2.0 1.4 1.3 1.4 1.O 0.0 1.2 1.1 0.6 1.3 1.2 1.7 1.2
6.1 3.7 3.9 7.7 0.9 2.3 0.0 6.2 0.1 2.1 0.3 0.3 3.3 2.1
The (potential) natural growth in transfers, the base elasticity, is shown in column 2 of Table 6. In most countries in the sample used, the level of taxation increased either in proportion to GDP or at a greater rate. The implication of this result is that even if subnational governments had only maintained a constant percent of the tax revenues of higher level governments (i.e., no rate effect), the vertical share would have increased. SNGs in countries with a "fixed percent" tax sharing system might be expected to maintain revenues in such a case, whereas those in countries that rely on ad hoc systems or closed-end reimbursement system would need to depend on discretionary actions to maintain the vertical share. The rate elasticity of the vertical share averages less than unity. On average, countries took discretionary actions that kept the growth in transfers to a level below the growth in taxes collected by higher level governments. For most countries in this sample, however, elasticity was about unity. Discretionary actions taken by higher level governments more or less indexed the overall elasticity to GDP. But was there a pattern to this indexing? As revenue mobilization was driven up (or down), how did countries react with respect to the effort they made in allocation to intergovernmental transfers? We describe the pattern in Table 7 where we have placed countries into four groups. Countries in the bottom right quadrant had both a base and rate
Bahl and Wallace
244
elasticity greater than unity. This means that there was automatic growth in tax revenues that was more than proportionate to GDP, and discretionary increases in intergovernmental transfers stimulated decentralization financing even further. Countries in the bottom left quadrant compensated for a strong base growth by making discretionary cuts (or attrition cuts) in intergovernmental transfers. Countries in the top left quadrant deemphasized decentralization altogether by passing on part of a slow growth in taxes to local governments with a discretionary cut in the sharing rate. Finally, those countries in the top right quadrant compensated for a slow growth in the base by increasing the sharing rate. There seems no rhyme to this classification. Neither industrialized nor developing countries cluster in any one category, nor do countries with high vs. low levels of fiscal decentralization.
Table 7. Rate and Base Elasticity Cross-Classification Rate E sticity Below Unity
Azerbaijan Kyrgyz Republic Sudan
Algeria Australia Belize Ethiopia Fiji Guinea-Bissau Jordan Kenya Latvia Malaysia New Zealand Solomon Islands Turkey United Kingdom Zambia
Above Unity
El Salvador Georgia Ghana Guinea India Malawi Nigeria Oman Uganda Bahamas, The Botswana Gambia, The Guatemala Lesotho Pakistan Papua New Guinea Philippines Portugal Senegal Sri Lanka Swaziland Zimbabwe
Intergovernmental Transfers: The Vertical Sharing Dimension
7.4
Administrative Costs
Vertical sharing schemes need not impose a large administrative cost. Tax sharing schemes impose relatively little administrative cost because the matter is straightforward: the distributable pool receives an earmarked share of collections. So long as tax collection data are accurate and up to date, the SNG share is easily determined. Neither is an ad hoc determination of the distributable pool a complicated issue, in that the amount is fixed by political decision. There may, however, be considerable costs associated with the bargaining that accompanies this determination, and there may be delays in arriving at the awards because of the difficulty in reaching a decision. Conditional, cost reimbursement grants are another matter. Whether the vertical share is based on an open-ended or closed-ended conditional transfer, it implies a significant administrative cost. Determination of the total amount for the distributable pool in the case of an open-ended scheme would require an estimate of the eligible expenditures made by each participating local government. Properly done, the monitoring of "eligible" expenditures can be a costly exercise under both an open-ended and a closed ended conditional transfer.
CONCLUSIONS The vertical share component of an intergovernmental transfer is the claim of the subnational governments on the revenues raised by the higher level government that makes the transfer. Available data suggests that the vertical share is, on average, about 19 percent in industrialized countries and between 13 and 14 percent in developing and transition countries. It has been stable in the industrialized countries over the past two decades, increasing in the developing countries and declining in the transition countries. The results of our analysis show that the commitment to expenditure decentralization is a prime reason why the vertical share is larger in some countries than others. All other things being equal (i.e., the level of fiscal decentralization and the dependence of subnational governments on transfers), higher levels of resource mobilization tend to dampen the vertical share. There is a wide variation in the practice of vertical sharing. The most decentralized version gives subnational governments a guaranteed claim on central government taxes. Many countries use this form of revenue sharing and there are many different variations in terms of the taxes shared and the sharing percentages. The more centralized approaches to vertical sharing are ad hoc (political) distributions and cost reimbursement (conditional) grants.
246
Bahl and Wallace
One way to evaluate vertical sharing in an intergovernmental transfer system is to measure its revenue-GDP elasticity (buoyancy), just as one would for a tax. Our analysis shows that the buoyancy is greater than unity for most countries but there seems to be no pattern to identifying the determinants of the variation in this elasticity, i.e., whether it is due to a base effect (growth in revenues of the central government) or a discretionary effect.
Notes We are grateful to Robynn Cox and Bayar Tumennasan for valuable research assistance. See, for example, Ma (1997), Martinez-Vazquez and Boex (2001) and Boadway (2004). The system of expenditure "norms" of the transition countries was theoretically an attempt to estimate expenditure needs at all levels of government. However, the expenditure "needs" calculations were pushed aside by budget pressures and were often ignored (Bahl, et a1 1999, Martinez-Vazquez and Boex, 2004). In some countries such as Japan, and the United Kingdom, more explicit needs calculations are made each year (see Alm, 1999). South Africa uses an index of expenditure needs in its formula for distributing its "equitable shares7' grant among local governments (Reschovsky, 2003.) Where there are three levels of government, as in a federalism, we define the level of transfers as being the sum of those made by central governments to state governments, and those made by state governments to local governments. We define the level of taxes analogously in such a case, i.e. as the sum of central plus state taxes. Particularly bothersome is the possibility that an intergovernmental transfer may be misclassified as a SNG tax. We define an own source revenue as the case where the local government is free to determine either the tax rate, the tax base, or both. So, for example, we would treat a central government tax, that is allocated in part or fully to the local government where collection takes place, as an intergovernmental transfer. If a SNG is given some freedom to levy a sur-rate on a central government tax base or central tax collections, we would treat this as a local own source revenue. In some cases, e.g., China, GFS (IMF, 2003) classifies shared taxes where there is no local autonomy in rate or base determination, as "local taxes". To the extent this is widespread in the GFS volume, the results here understate the importance of intergovernmental transfers in fiscal systems. The definitions used in this paper for data extracted from GFS are as follows: Intergovernmental transfers (Tr) are the amounts shown as "grants from other levels of government" in the revenue table for subnational governments. If grants from other levels of government are not shown in the subnational government revenue, we use the variable "grants to other general governments", taken for the expenditure table of the granting government, as the measure. This may introduce some error, because "other general level of government" might not be subnational governments. In most cases where there were data reported for both, however, the reported amounts for transfers made did match up with the reported amounts for transfers received. The detailed Appendix tables that show actual data used are available from the authors. Again, however, note that there is a substantial variation in these shares, ranging from 50 percent in India to less than 3 percent in many countries (Appendix Table 2). Another explanation is that the revenue sources are misclassified. subnational government expenditures (SNE) are "state plus local government expenditures" (as reported in the expenditure table) net of any transfer from the state to
Intergovernmental Transfers: The Vertical Sharing Dimension
12. 13. 14. 15. 16. 17. 18. 19.
247
the local government, where the amount of the transfer is taken from the local government revenue table. The "expenditure decentralization ratio" is the percent of SNG expenditures in total government expenditures. For a good discussion of the relationship between corruption and decentralization, see Tumennasan (2005). In practice, central governments sometimes have changed or ignored current legislation when difficult times arrived. See Dahlby (1996) and Keen (1998) for a discussion of other potential strategies of subnational governments. Andersson et. al. (2004) provide an interesting case study of the vertical externality hypothesis for Sweden. Sierra Leone, Local Government Act 2004, Section 47(3). For a good discussion of the Russian FFRC, see Martinez and Boex (1999). These are calculated as arc elasticities over the beginning and end period for which data are available.
References Afonso, Jose Roberto R., and Luiz de Mello. 2000. Brazil: An evolving federation. IMFIFAD Seminar, International Monetary Fund, Washington, D.C. Ahmad, Ehtisham, and Katherine Baer. 1997. Columbia. In Fiscal Federalism in Theory and Practice, Teresa Ter-Minassian, ed., 457-503. Washington, D.C.: International Monetary Fund. Ahmed, Qazi Masood. 1997. Design of intergovernmental fiscal relations and international finance institutions' allocations for rural development. Paper present at the forum on technical consultation on decentralization in Rome. Alm, James. 1999. International experience with expenditure norms. Andrew Young School of Policy Studies Working Paper, Georgia State University, Atlanta, GA. Andersson, L., Aronsson, T., Wikstrom, M. 2004. Testing for vertical fiscal externalities. International Tax and Public Finance 11: 243-263. Bahl, R.., and Linn J. F. 1992. Urban Public Finance in Developing Countries. Washington, D.C.: World Bank, Oxford University Press. Bahl, Roy. 1999. Fiscal Policy in China. Ann Arbor, Michigan: Institute and University of Michigan Press. Bahl, R., Wallace, S., Mikesell, J., Kourliandskaia, G. 1999. Leningrad Oblast. International Studies Program Working Paper, Georgia State University, Andrew Young School of Policy Studies, Atlanta, GA. Boadway, Robin. 2004. The theory and practice of equalization. CESifo Economic Studies, 50 (1): 21 1-254. Boex, Jameson, Bahl, Roy, Martinez-Vazquez, Jorge, Bahl, Roy and Rutasitara, L. 2003. Final report: Developing a system of intergovernmental grants in Tanzania, Tanzania Local Government Reform Program. Georgia State University, Andrew Young School of Policy Studies. Processed. Brosio, Giorgio, and Sanjeev Gupta. 1997. Ethiopia. In Fiscal Federalism in Theory and Practice, Teresa Ter-Minassian, ed, 504-525. Washington, D.C.: International Monetary Fund.
Bahl and Wallace Capuno, J.J., T.C. Manuel and B.T. Salvador. 2001. Estimating the Ira, centrally provided local public goods services and other central fiscal transfers. Report submitted to NEDA. Pasig City: NEDA-RDCS. Courchene, Thomas and Alberto Diaz-Cayeros (2000). "Transfers and the nature of the Mexican federation." In Achievement and challenges of fiscal decentralization lessons from Mexico, Marcelo M. Giugale and Steven B. Webb, eds., 200-236. Washington, D.C.: International Bank for Reconstruction and Development/World Bank. Craig, Jon. 1997. Australia. In Fiscal Federalism in Theory and Practice, Teresa TerMinassian, ed., 175-200. Washington, D.C.: International Monetary Fund. Dahlby, B. 1996. Fiscal externalities and the design of intergovernmental grants. International Tax and Public Finance, 3: 397-412. Diokno, Benjamin E. 2003. Decentralization in the Philippines after ten years: What have we learned? What have I learned? Asian Development Conference 2003: Seminar A. Edmiston, Kelly. 1999. A critical evaluation of Papua New Guinea's 1995 provincial reforms. Unpublished report on Papua New Guinea's fiscal decentralization. Emiliani, Nicoletta, Sergio Lugaresi and Edgardo Ruggiero. 1997. Italy. In Fiscal Federalism in Theory and Practice, Teresa Ter-Minassian, ed., 249-284. Washington, D.C.: International Monetary Fund. Frank, Jonas. 2003. Decentralization. In Ecuador: An Economic and Social Agenda in the New Millennium, Vicente Fretes-Gibils, Marcelo M. Giugale, and Jose Roberto LopezCalix, eds., 479-491. Washington, D.C.: International Bank for Reconstruction and Development/World Bank. Hofman, Bert. 2003. Managing Indonesia's rapid decentralization: Achievements and challenges, Asian Development Conference 2003: Seminar A. Ihori, Toshihiro. 2003. Political Decentralization and Fiscal Reconstruction in Japan, Asian Development Conference 2003: Seminar A. International Monetary Fund . 2003. Government Finance Statistics Yearbook. Washington, D.C.: IMF. International Monetary Fund (2003a). Government Finance Statistics CD 2003. Washington, D.C.: IMF. International Monetary Fund .2002. October, Government Finance Statistics Manual 2001 Companion Material, Washington, D.C.: IMF. Keen, Michael . 1998. Vertical tax externalities in the theory of fiscal federalism. IMF Staff Papers Volume,45 (3): International Monetary Fund. Krelove, Russell, Janet Stotsky and Charles L. Vehorn. 1997. Canada. In Fiscal Federalism in Theory and Practice, Teresa Ter-Minassian, ed., 201-225. Washington, D.C.: International Monetary Fund. Lutz, Mark, Edgardo Ruggiero, Paul Bernd Spahn, and Emil M. Sunley. 1997. Hungary. In Fiscal Federalism in Theory and Practice, Teresa Ter-Minassian, ed., 660-679. Washington, D.C.: International Monetary Fund. Ma, Jun. 1997. Intergovernmental fiscal transfers in nine countries lessons for developing countries. WPS 1822, World Bank, Washington, D.C. Mackenzie, G.A., and Jose-Luiz Ruiz. 1997. Bolivia. In Fiscal Federalism in Theory and Practice, Teresa Ter-Minassian, ed., 423-437. Washington, D.C.: International Monetary Fund. Martinez-Vazquez, Jorge and Boex, F. Jameson. 1999. Fiscal decentralization in the Russia federation during the transition, International Studies Program Working Paper, Andrew Young School of Policy Studies, Georgia State University, Atlanta, Georgia.
Intergovernmental Transfers: The Vertical Sharing Dimension
249
Martinez-Vazquez, Jorge and Jameson Boex. 2004. The Design of equalization grants: theory and applications. International Studies Program Working Paper. World Bank InstituteIGeorgia State University Andrew Young School of Policy. Martinez-Vazquez, Jorge and Jameson Boex (2001). Russia's transition to a new federalism. Part of Fiscal Decentralization Series on Russia: World Bank Publications. McLure, C. E., Jr.. 1998. The revenue assignment problem: Ends, means and constraints. Public Budgeting, Accounting and Financial Management, Winter: 652-683. McLure, Jr., Charles E. 1999. The design of intergovernmental grants, In Fiscal Transitions in Kazahtan. Asian Development Bank. Mered, Michael. 1999. Nigeria. In Fiscal Federalism in Theory and Practice, Teresa TerMinassian, ed., 598-61 1. Washington, D.C.: International Monetary Fund. Potter, Barry. 1997. United Kingdom. In Fiscal Federalism in Theory and Practice, Teresa Ter-Minassian, ed., 342-358. Washington, D.C.: International Monetary Fund. Rao, Govinda M. and Singh, Niwikar. 2001. The political economy of center-state fiscal transfers in India, Stanford Center for International Development Working Paper No. 107, Stanford University. Reschovsky, Andrew. 2003 Intergovernmental transfers: the equitable share. In Restructuring Local Government Finance in Developing Countries: Lessons from South Africa, Roy Bahl and Paul Smoke, eds., Wallace E. Oates, series ed., Studies in Fiscal Federalism and State-Local Finance. Northhampton, MA: Edward Elgar. Rezk, Ernesto. 2000. Federalism and decentralization under convertibility: Lessons from the Argentine experience. Unpublished paper. Schwartz, Gerd, and Claire Liuksila. 1997. Argentina. In Fiscal Federalism in Theory and Practice, Teresa Ter-Minassian, ed., 387-422. Washington, D.C.: International Monetary Fund. Searle, Bob, Federal Fiscal Relations in Australia 2001, International centre for economic research, Working Paper No. 0112002, Torino, Italy. Smoke, Paul. 2000. Fiscal decentralization in east and southern Africa: Selective review of experience and thoughts on moving forward. IMF: Paper prepared for conference on fiscal decentralization. Spahn, Paul Bernd and Oliver Franz. 2000. Consensus democracy and Interjurisdictional Fiscal Solidarity in Germany. Journal of Economic Literature. Spahn, Paul Bernd and Wolfgang Fottinger. 1997. Germany. In Fiscal Federalism in Theory and Practice, Teresa Ter-Minassian, ed., 226-248. Washington, D.C.: International Monetary Fund. . 1997. Switzerland. In Fiscal Federalism in Theory and Practice, Teresa TerMinassian, ed., 324-341. Washington, D.C.: International Monetary Fund. Spahn, Paul Bernd. 1997. Intergovernmental transfers in Switzerland and Germany. In Financing Decentralized Expenditures: An International Comparison of Grants, Ehtisham Ahmad, ed., Cheltenham, U.K.: Edward Elgar. Ter-Minassian, Teresa. (Ed.) 1997. Brazil. Fiscal Federalism in Theory and Practice. Washington, D.C.: International Monetary Fund. Tumennassan, Bayar. 2005. Fiscal decentralization and comption in the public sector. Unpublished Doctoral Dissertation. Andrew Young School of Policy Studies, Georgia State University, Department of Economics. Weist, Dana. 2003. Thailand's decentralization progress and prospects. Asian Development Conference 2003: Seminar A. World Bank. 1995. The Hashemite Kingdom of Jordan: Intergovernmental fiscal relations and municipal finance, Management Sector Study.
DISCUSSANT COMMENTS
GIORGIO BROSIO Department of Economics, University of Torino
SHAH, SPAHN, AND BAHL AND WALLACE
The three papers by Bahl and Wallace, Shah and Spahn have a clear complementarity. Bahl and Wallace deal with vertical sharing arrangements. They show and analyze how countries decide the division of their total national fiscal resources between different tiers of government, present some recent trends, and provide convincing explanations for them. Shah's paper is one of the first contributions to the literature concerning the political institutions responsible for the decision-making process and the implementation of a main component of vertical sharing, namely equalization grants. Spahn's paper is more composite. Its first part provides an illustration of the various transfer instruments available, and of the objectives they serve. The second part illustrates, with a simple numerical example, a widely used model of equalization grants that is based both on expenditure needs and fiscal capacity. The papers, particularly those of Shah and Spahn, although very different in focus and in analytical properties, refer mostly to federal type systems and to the allocation of resources between the federal level and the intermediate level - federated states or regions. Bahl and Wallace's paper refers to all subnational governments and concentrates the attention on the determination of the central government share against the share for all subnational governments. There is no specific consideration of the distinct problems that arise when local (municipal) governments are separately considered from states andlor regions for sharing purposes. I start my comments by providing a few cases of why distinct consideration of the position of municipal government is needed. There is no doubt that specific consideration of local governments, in addition to the intermediate tier, adds to the complexity of the analysis. Determining a
Brosio
sharing rate of total national fiscal resources between the federal government, on the one hand, and all subnational governments on the other, is a theoretically complex issue. Bahl and Wallace think that it verges on impossibility. In any case, it becomes more much complicated when the sharing has to take place distinctly between the federal, the regional, the departmental and the local level, as happens in countries such as France, Italy and spain'. Most federal systems adopt a two-stage process for sharing. There is a system for sharing national resources between the federal government and the states. Then, each state has to provide its solution to the sharing of its resources between itself and its local governments. Non federal or, more specifically, non classical federal systems with more than two levels of governments - Brazil and South Africa provide examples of an approximation to these systems - have to solve simultaneously the sharing issue with more than two levels. A similar problem applies to the choice of the appropriate institutional framework for revenue sharing. Intergovernmental bodies, such as the premiers' conferences praised by Shah, work in a distinct, more functional and easier way in federal systems, particularly when the number of federated states is small, as in Australia and Canada. Their working is much less efficient and less determinant when they also have to work for local governments. One reason for their different working is that the sheer number of jurisdictions does not allow all of them to be represented simultaneously. Conferences thus become bodies of representative democracy - usually small local government units elect their representatives - instead of being direct assemblies, as is the in Australia and Canada. Thirdly, numbers inevitably constrain the choice of the specific equalization model. These models become quite sensitive to small changes of indicators and parameters when they reach a minimum level of sophistication. Furthermore, as Spahn's paper shows, allocations to small states are extremely sensitive to small changes in the context considered for the allocation. A sceptical reader can very easily check this statement by making simulations using the model and the numbers provided by Spahn. When equalization is applied to the local government level, differences between jurisdictions simply skyrocket. How can we accommodate, that is how to equalize with the same model, the city of Paris with a few million inhabitants and the municipality of Montgenevre in the Department of Hautes Alpes, with about 50 inhabitants? Or in Italy how do we accommodate within a single model the circumstances of Rome and those of ClaviCre, a municipality of less than 70 inhabitants located in the Piedmontese Alps? As mentioned, Bahl and Wallace prefer to leave aside the issue of selecting the analytical criteria to be used in the determination of the sharing
Discussant Comments
253
rate and concentrate their attention on the analysis of empirical evidence about sharing in both developed and developing countries. This is quite interesting and well presented information. Moreover, the paper provides sound policy orientations applying to the mechanisms used for revenue sharing. Certainly, as Bahl and Wallace maintain, there is no "one best way to determine the vertical share". Having no criteria spelled out, however, their paper has problems providing answers to the crucial issues that it explicitly considers, such as that of the adequacy of revenue for local governments and, consequently, the adequacy of the elasticity of this revenue with reference to the trends in subnational expenditure. These problems show that economists should try to provide an analytical framework for the determination of the sharing rates. In fact, an infant literature is making its entrance on this topic. A possible way of looking at the issue of vertical shares would, in my view, be to look at the weights assigned by individuals in their utility fknction to the expenditure of the central and the subnational governments. For example, if defence is assigned to the national government while health care is assigned to the subnational governments, and if all subnational government units have the same per capita average GDP - thus excluding the need for equalization - and if VAT is the only source of revenue, then its collections should be shared between the national and the subnational levels according to the relative weights of defence and health in the utility k c t i o n of individuals, independently of the level of government that administers and collects the tax. In fact, there is no reason why vertical sharing means that resources have to go from the central level to the local level. They can also go in the other direction, as shown by the EU. Obviously, individual utility hnctions have to be aggregated and this is a formidable challenge to the analyst. Thus, more realistic assumptions and deeper analysis are needed than the simple hint provided here. In my view, the point is simply that, once we are able to provide an analytical basis for the determination of the sharing rates, many of the problems concerning their implementation are easier to solve. For example, Bahl and Wallace have a strong preference for classical revenue sharing, where the sharing rates - for a number of individual taxes andlor for total tax collections - are determined for the medium term by some legal text. Bahl and Wallace reject ad hoe systems of grants where the sharing rates are determined ex post in a rather casual way and where they vary from year to year according to the choices of the central government. Their preference is clearly justified on most of the grounds one can consider. The same preference would be even more justified on the basis of the previously suggested approach for the general determination of the sharing rates. In other words, if we were able to single out a criterion such as the weights of the various national and local expenditures in the individual
Brosio
utility functions, then ex ante sharing rates would have to be introduced to guarantee the attainment of these weights. The same considerations would apply, with even greater strength, when considering the choice between revenue sharing and simple cost reimbursement of local functions. As Bahl and Wallace very correctly argue, the latter is, in fact, a system that recognizes an arbitrary power to the central government to determine the size of local government expenditure. The relevance of Shah's paper becomes evident when we realize that the intergovernmental relations literature has almost neglected the assessment of the positive and normative properties of the political/ institutional arrangements responsible for the decision-making and the implementation process of equalization grants. Shah's contribution is an important step in filling this gap. It provides an illustration of the arrangements in a number of federal and regional countries and a discussion of their merits with reference to the analytical framework provided by the neo-institutional economics. Shah questions the merits of independent bodies and commissions which are generally widely praised in the literature. Possibly, an important source of this praise derives, according to Shah, simply from the fact that these bodies provide positions searched for by academics and experts who have a private interest in the evaluation of the relative merits of alternative arrangements. At the same time, Shah shows some preference for intergovernmental bodies with the participation of appointed officials, such as prime ministers conferences, or analogous bodies. The problem, as I see it, is that the two institutions - independent commissions and intergovernmental political bodies - are essentially not substitutes, but rather complement each other. This is because of the tasks that they are asked to perform. Intergovernmental political bodies are mostly a locus for decisionmaking. With reference to revenue sharing issues, they have to produce an agreement on the sharing rates. Intergovernmental bodies are prevalent and play a relevant role in parliamentary systems where the second chamber is not selected according to the pure version of the federal principle. More precisely, they are prevalent in bicameral systems where federated states do not select their delegation to the second chamber. This is the case of Australia and Canada. On the other hand, in Germany and South Africa, where the members of the second chamber are appointed by the governments of the states, intergovernmental bodies have a much smaller role to play in decision-making. In these cases, in fact, the real intergovernmental decision-making body concerning intergovernmental issues is the Senate. Decisions about sharing rates provided by intergovernmental bodies may be a specification of a broad andlor vague constitutional mandate. Or, they can provide the basis for a formal decision by the legislative bodies or, as is the case in some countries, for a ministerial decree. However, following the
Discussant Comments
255
increasing importance of equalization grants, national parliaments want to play an increasingly substantial role in the definition of the sharing rates and in the formula for their allocation. Parliaments increasingly resent a rubberstamping role with reference to decisions taken elsewhere. In any case, intergovernmental political bodies can reach agreements on vertical and also horizontal sharing rates, and on the principles and the main ingredients of the formulas for their allocation. But they cannot produce specific formulas unless these formulas are very simple, and especially they cannot implement them since they are not bureaucracies. Germany provides an illustration of a system where the simplicity of the revenue sharing system almost eliminates the necessity of choosing between alternative institutions for its implementation. The German system has relatively simple aims. Per capita revenues of each Land have to be no lower than 95 percent of the national average, while population is the basic indicator of need. Since the system is enshrined in the basic legislation, equalization does not need a specific bureaucratic body or independent commission to implement it. In fact, implementation is done mainly through direct horizontal transfers among the Laender (another reason for this "fraternal" system of equalization is that Laender are responsible for administration and collection of all national taxes with the exception of tariff duties). The implementation of constitutional mandates or ordinary laws, (or in non fully-fledged democratic countries - Presidential decrees) concerning revenue sharing and transfers is devolved either to ministries or independent commissions. Clearly, borderlines between political and technical roles are quite difficult to draw and implementation can be a determinant of the final results, depending on the choice and use of information. Australia provides an example. Legislation there mandates that the equalization system is aimed at providing each state with the resources that will allow it to provide a level of public services comparable to that of the other states, after considering relevant expenditure needs and taking into account reasonably equal levels of efficiency in the provision of services and reasonably levels of tax efforts. Although the aims are clear, their implementation is not discretion-proof and it is left to the judgement and the technical skills of the Commonwealth Grants Commission - the independent body responsible for the allocation of equalization grants. As illustrated in Shah's paper, a number of countries have apparently similar bodies, but at a closer look the diversity of functions is huge. The South African and Ugandan bodies, for example, can make only recommendations and proposals, while their participation in determining the final allocation of grants is minimal. In other words, they are extremely weak institutions both from the decision-making and the execution point of view. They also have completely different memberships: while academics
256
Brosio
prevail in the Ugandan commission, representatives of distinct levels of government sit on the South African commission. When discussing the merits of autonomous commissions, it is thus difficult to hold sufficient variables constant to make legitimate comparisons. In general, the strongest idea behind recommending recourse to independent commissions with some decision-making power in the design and execution of revenue sharing is, in my view, the introduction, as in the old Athenian democracy of Aristotle, of some form of government by the sages, or the wise men. Provided that these persons are available, the system is not totally wrong, especially in countries where typical political bodies at the legislative or executive level - are not able to arrive at sufficient agreement and accommodation between highly conflicting positions, as can be the case with ethnically or otherwise divided countries.
Notes 1.
There are even more difficulties in these countries, since we have to consider an additional tier: namely the European Union. As is known, sharing fiscal resources between Member States and the EU is a thorny issue that is made more complex by the fact that a substantial share of EU resources are allocated to regional governments of Member States.
4 CHALLENGES IN IMPLEMENTING EQUALIZATON
Chapter 9 EXPENDITURE-BASED EQUALIZATION TRANSFERS RICHARD BIRD* AND FRANCOIS VAILLANCOURT*
*'
*Rotman School of Management, University of Toronto,**Economics Department, Universitk de Montrkal
1.
INTRODUCTION
Intergovernmental transfers are a major source of revenue for sub-national (regional and local) governments (hereafter SNG), representing 60 percent of their total revenue for developing countries and 33 percent for OECD countries (Shah, 2004). The continued and growing decentralization observed in many countries calls for a better understanding of the design, role and impact of fiscal transfers. Prominent among the objectives commonly attributed to intergovernmental fiscal transfers is 'equalization' although exactly what this term means is often rather obscure and may differ from country to country or even over time within any one country. Our focus in this paper is on those transfers specifically labeled as equalization transfers, and in particular on the extent to which and the method by which differences in expenditure 'needs' can and should be formally incorporated into such transfers. The paper is divided into three main sections. In the first, we set out briefly the standard theoretical case for both a general equalization transfer and for the incorporation of expenditure needs as a key factor in the design of such transfers and discuss how this case may be implemented. In the second section, we review how and to what extent expenditure factors are actually incorporated in a few selected equalization schemes around the world. Finally, in the last section we consider critically the relevance of assessing differentials in expenditure needs in determining equalization transfers. In effect, in the first section we discuss why it may be argued that such 'needs' should be taken into account in principle. In the second section, we show that some countries actually do so in practice in varying ways and to varying extents. In the final section, we focus on the
260
Bird and Vaillancourt
circumstances in which it seems advisable to take needs into account in the design of the general equalization formula, and ask whether the game is worth the candle. Our aim is to provide some general guidelines that may be useful to the many countries currently facing the need to establish sounder intergovernmental fiscal transfer systems in order to ensure that the outcome of decentralization is broadly economically and socially beneficiaL2
2.
EQUALIZATION AND EXPENDITURE NEEDS
2.1
The Theoretical Case for Equalization
Following Buchanan's seminal paper (1950), an important body of work developed on the use of equalization transfers to correct the equity and efficiency distortions that (from a certain perspective) may be considered inherent to a decentralized governmental str~cture.~ We do not purport to review this literature in depth.4 Rather, we shall simply set out in this section the main concepts relevant to the aim of this paper. Initially, as did Buchanan (1950), we focus on fiscal capacity alone, introducing expenditure needs considerations subsequently. With respect to equity, the central aim of an equalization transfer is to enable SNG with different abilities to raise revenues to provide comparable levels of services with comparable levels of own-taxes. The taxable capacity of SNG depends both on the tax bases open to them and on the territorial distribution of those bases (e.g. natural resources). Since no country is completely uniform, a fundamental characteristic of a decentralized state is that SNGs have different fiscal capacities and are hence unable to provide the same level of public services at the same tax rates. Since a 'poorer' SNG in the sense of an area with a smaller (per capita) tax base will raise less revenue at a given tax rate than a 'richer' SNG, two identical individuals living in these two different SNGs will not receive the same level of public services, even though they pay the same amount of tax. Alternatively, to obtain the same level of services, the SNGs must impose different tax burdens. Decentralization thus fails to produce horizontal equity in the sense of treating 'equals' equally (Boadway, 2001). Horizontal equity in this sense is of course a normative objective that rests on value judgments that may not be accepted by all. The underlying assumption is that two people who are otherwise equal but live in different regions of the same country are to be considered 'equal' -- even though they live in different regions. We shall return to this point. A more comprehensive approach to horizontal equity utilizes the concept of net fiscal benefit (NFB) or the difference between taxes paid and public
Expenditure-Based Equalization Transfers
26 1
services received (Boadway, 2001). The NFB approach is superior to focusing on tax rates alone because it allows for different levels of taxation and public services in different SNGs according to the preferences of voters. Fiscal or horizontal equity in the sense discussed above requires that NFBs be the same for similar individuals across all SNG jurisdictions. SNG A may impose higher tax rates than SNG B, but if A also provides a correspondingly higher level of public services, NFBs may still be equal for similar individuals in both regions. Boadway (2004, p. 215) recognizes that for this case for horizontal equity to be applicable to a country with heterogeneous regions, and particularly to a federal country, "requires a consensus that social citizenship or solidarity among all citizens apply with equal force nationwide as opposed to being region-specific." The theoretical equalization literature simply assumes that this condition holds: that is, that not only are governments at all levels benevolent but also the correct standard for judging decentralization is how closely it replicates the outcomes of a (benevolent) unitary state. Indeed, the principal theoretical rationale for equalization is that it is necessary precisely for this purpose, although this aim is often obscured by the emphasis many authors place on the unconditional nature of equalization transfers and hence the freedom of SNGs to deviate in practice from national uniformity or unitary outcomes (which are not necessarily the same). As we discuss below, the usual representative tax system (RTS) equalization payments are unconditional, thus permitting such regional differences. Nonetheless, it should be clearly understood that the basic rationale for equalization schemes is to enable SNGs to replicate the outcomes of a unitary state - if they choose to do so. Those who have a different view of what a federal state is supposed to be may perhaps be less enthusiastic about the need for equalization. While many countries have implicitly or explicitly set minimum national standards of revenue and public service, and have established equalization schemes as one means of obtaining this objective, the United States conspicuously has no such scheme. Presumably, this is not an oversight but a deliberate policy choice, perhaps reflecting deep philosophical and political differences between the 'exceptional country' and much of the rest of the world. On the other hand, contrary to the way it is sometimes viewed by American analysts (e.g. Oakland, 1994), equalization is not concerned at all with vertical equity between persons. It is also not, as noted by Buchanan (1950, p.596), in any sense a "charitable contribution from the rich to the poor [...] or subsidy from the citizens of the more favored regions."' On the contrary, the purpose of equalization is simply to make sure that the fiscal structure is as neutral as possible regarding the place of residence, not only to achieve horizontal equity as discussed above but also to improve the efficiency of resource allocation.
Bird and Vaillancourt
262
Since labor and capital respond to both market incentives and fiscal incentives when making location decisions, differences in the abilities of SNGs to provide equal levels of public services at equal tax rates provide an incentive for migration to the area providing higher NFBs. Economic efficiency, however, requires that people and economic resources be distributed according to their private productivity and not to their expected NFBs. Equalization must therefore be designed in such a way as to eliminate fiscally induced migration that would result in a misallocation of economic resources. To ensure that economic resources migrate in response to market incentives and not fiscal balance, SNGs should, it is argued, be able to provide similar services at comparable levels of taxation. Equalization transfers thus provide one of those rare instances in public economics where equity and efficiency considerations coincide (Shah, 1994). Although the fiscal efficiency of equalization in this sense is debatable,6 Wilson (2003) offers some interesting recent empirical evidence of the efficiency cost of fiscally induced migrati~n.~
2.2
Implementing the Theory
Taxonomically, one may distinguish between implicit and explicit equalization, gross and net equalization, and vertical and horizontal equalization. Although there is a large literature both on the extent to which transfers in general, even those not explicitly intended to do so, equalize, we shall not consider such 'implicit' equalization further here. Of course, implicitly, even an equal per capita transfer is equalizing since revenues are not raised equally across all regions.8 Similarly, many targeted grants such as the US Medicare program that distributes grants to states on the basis of two equally weighted factors (per capita income and per capita spending) may be equalizing to some extent. For example, Laurent and Vaillancourt (2003) found a small negative correlation between per capita state product (or income) and such transfers. Here, however, we focus on explicit equalization schemes, consisting of formula-based unconditional redistributive transfers among governments. A gross scheme is one that, so to speak, 'levels up' in the sense that SNGs below some target (a national norm, the national average, the top SNG, or some percentage of one of these) are brought up to the level of the target. In contrast, under a net scheme, better-off SNGs are also 'leveled down' to the target, usually through horizontal equalization. A horizontal scheme thus implies transfers between SNG, while a vertical scheme implies transfers from the central government.9 In principle, the same equalization result can be produced by either approach, but a horizontal scheme requires much more political cohesiveness between regions and may therefore be more difficult to implement in a country with heterogeneous regions. Such a system may
Expenditure-Based Equalization Transfers
263
also create undesirable disincentives for SNG tax effort. On the other hand, a vertical scheme requires the central government to occupy a larger share of tax room than its expenditures require, which can itself raise important political issues. At the same time, however, it may facilitate the harmonization of taxes, internalize fiscal externalities and ensure that central attempts to influence regional government behavior through conditional grants take place, as it were, on a 'level playing field' (Boadway, 2004). Of course, especially in developing countries in which SNGs have only limited capacity to collect taxes, a vertical scheme might be the only feasible approach (Vaillancourt, 200 1). Many other issues need to be decided in implementing any equalization transfer. For example, the size of the total 'pool' to be distributed under any equalization scheme may be determined in various ways. It may be set annually as part of the normal budgetary process, as in many developing countries. It may be determined: as some given proportion of central revenues (e.g. Argentina); on the basis of central collections of one or more particular taxes (e.g. Australia, Germany); or it may, as in Canada, be paid out of general central revenues but with the amount paid set in the budget of the central government or determined by a formula driven by other factors such as the level of SNG revenues, the growth of GDP or the relative differences between the taxable capacities of SNG. Equalization can be implemented either by direct central transfers or by some form of tax sharing (Dafflon and Vaillancourt, 2003). The amount distributed to different SNGs may be determined on the basis of expenditure differentials andlor fiscal capacity and perhaps also fiscal effort, or it may simply fill budget gaps (as in some countries of the former Soviet Union and Eastern Europe). Although our focus in this paper is on the role of expenditure equalization, we shall first outline briefly how fiscal capacity is commonly handled in equalization schemes.
2.2.1
Fiscal capacity
The fiscal capacity of a SNG is defined by its ability to raise revenues from its own tax bases. Assuming that 'own' tax bases are clearly defined, as Shah (1994) and Martinez-Vazquez and Boex (2001) demonstrate, a number of methods may be employed to determine fiscal capacity in this sense.'' Perhaps the simplest to implement are measures based on current or past years' revenue collections. Unfortunately, such measures raise serious problems. While potential ability to raise revenue is not directly affected by tax rates," fiscal effort, taxpayer compliance and actual revenues are affected. Using current revenue collection as a measure of fiscal capacity provides SNGs with an obvious incentive to impose lower tax rates, to assess tax bases at a lower level or to make less effort to collect taxes in
264
Bird and Vaillancourt
order to receive higher equalization grants. Although using past collections would seem to alleviate this problem there remains a problem of timeinconsistency since SNGs may (reasonably) expect that current increases in revenues obtained by increasing rates or collection effort will reduce future transfers. Alternatively, such macroeconomic indicators as income or output may be used to measure fiscal capacity. Although difficult to obtain at the local level in most countries, such data, however imperfect, are often available at the regional level. But this does not necessarily imply that macro measures of fiscal capacity are good indicators. The measured personal income in a given SNG, for example, may be a reasonable measure of the average ability of its residents to pay taxes, but it may not be an accurate indication of the ability of the SNGs to impose taxes - for example, because SNGs may be prohibited from levying personal income taxes. Gross Regional Product (GRP) is a more comprehensive measure, representing the value of goods and services produced within a region and hence the incomes received by the owners of the economic resources (land, labor, natural resources and capital) used in the region, regardless of where they may live. But the composition of GRP is also clearly relevant since some economic sectors (e.g. mining) are easier to tax than others (e.g. agriculture). Similarly, informal activity is more difficult to tax than manufacturing activity (particularly in large enterprises). SNGs with identical GRP may thus have substantial differences in taxable capacity even in the unlikely circumstance that they are hlly empowered to levy all possible taxes. A third possible approach is what is called the representative tax system (RTS), which measures the amount of revenue that could be raised by a SNG if it used 'standard' tax bases and 'standard' (usually average) tax rates. Obviously, to use this approach, information on tax bases and tax revenues for every region is needed. The equalization entitlement for a region is determined by the sum for all tax sources of the following formula: Ex,i = POPx - tna, i . [PCTBna, i - PCTBx, where Ex,i is the equalization entitlement of region x for a revenue source i, POPx is the population of region x, tna,i is the national average tax rate of revenue source i, PCTBna,i and PCTBx,i are the per capita tax base of revenue source i for the national equalization target and region x respectively. This formula may be adjusted to equalize according to a predetermined standard (median, arithmetic mean or other norm). In theory, the RTS approach appears to provide a complete and accurate method to measure the fiscal capacity of SNGs. It may, as in Canada, be broadened to include non-tax revenues (user fees, royalties). Provided the tax base information is appropriate and all SNG are equally able to exploit
Expenditure-Based Equalization Transfers
265
all their 'assigned' tax bases freely - matters that are not always easy to determine in practice - this approach, although very data intensive, seems clearly superior to either the macroeconomic or actual revenue approaches to measuring fiscal capacity for purposes of equalization.
2.2.2
Expenditure differentials
The fiscal capacity of SNGs, however, deals only with the last term on the right side of the net fiscal benefit equation:
NFB
= public
services received - taxes paid
Equalizing NFBs requires not only adjustments to ensure equivalent revenue-raising capacity but also adjustments to ensure that the provision of equivalent public services requires the same expenditure in different SNG, if two regions have the same fiscal (potential) capacity but differ in terms of expenditure 'spending requirements', the NFBs of residents will be lower in the region in which public services are more expensive. Expenditure differences in providing public services reflect two factors: cost differences and need differences.'* Cost differences are differences in the cost per unit of a 'standardized' public service. They may arise from climatic or geographic features, density or distance factors, or differences in labor cost across regions. Costs should be calculated using real (not nominal) private sector wages for equivalent inputs and not on the basis of public sector wages which may reflect such political factors as the government's political philosophy or the relative strength of workers' unions (Courchene, 1998). Need differences -- differences in the number of units of standardized service required per capita -- usually arise owing to demographic reasons such as the age structure of the population and different participation rates in social programs by persons of different ages. Three questions arise with respect to implementing equalization schemes incorporating expenditure differentials: how are 'standardized' expenditures determined; how are needs differentials measured; and how are cost differentials measured? We shall take these up in reverse order, leaving for later discussion the key question of whether such differentials should be taken into account at all. Many different factors linked to cost differences have been proposed for inclusion in equalization formulas. Some make intuitive sense: climate (snowfall, heavy rain), population densitylurbanization, frequency of natural
266
Bird and Vaillancourt
disasters (floods, earthquakes), location (remoteness), topography (mountainous or desert regions). Others are perhaps less obvious such as the proportion of land occupied by public infrastructure (e.g. national parks), he1 costs, and indicators of development level. Some are general (e.g. electricity consumption, number of telephone lines) and some relate to particular public services (road length and condition, environment and wildlife preservation services, irrigation and drainage, coastal services, fire protection, etc.). Similarly, among the plausible factors related to needs differences are such things as: the share in the total population of such dependent populations as infants, elders (health care) and school age children; and the share of population with special needs irrespective of their age, such as new immigrants (language skills, acquisition, integration into society) or less developed groups (e.g. aboriginal populations). Many other factors presumably related to need are found in equalization schemes around the world: poverty indicators, single parent families, illiteracy, infant mortality, life expectancy, elementary school enrollment rate, public transit, female labor force participation rate, population, unemployment rate, social assistance payments, corrective services, number of drug-addicts, etc.13 Of course, to a considerable extent the relevance of many of these indicators depends on the role SNGs play in delivering public services. For instance, if it is the central government or the private sector that provides health care, the share of infants or elders may not be relevant in determining transfers. Broadly, expenditure needs may be measured in three ways (MartinezVazquez and Boex, 2001). One method is to estimate the cost of providing a standardized set of public services. To do so requires, first, that someone usually the central government but possibly an intergovernmental body determines what services are to be included and what 'standards' are to be met. Secondly, considerable effort is required to obtain the necessary detailed data, involving as a rule both field visits and considerable knowledge of the price of inputs and the factors affecting the scope of the services to be provided. Even when such complex methods are used to assess expenditure differentials, as in the case of Australia, some analysts have characterized the results as "somewhat crude, imprecise and subjective" (Shah, 1996, p. 103). A simpler alternative is to rely on historical expenditure patterns and use observed average costs for various expenditures. Two problems arise with this approach. First, as in the case of many transitional countries emerging from a history of central planning 'norms', past observed expenditures on particular activities may not reflect current policy objectives. Secondly, expenditures that seem the same in the data may in fact be quite different. For example, expenditure on education can be broken down by level (primary, secondary, etc.) and then within levels by inputs such as salaries,
Expenditure-Based Equalization Transfers
267
books, etc. How can one interpret the fact that books in SNG A are 20 percent more expensive than books in SNG B even though both jurisdictions buy the same number of books for the same number of pupils? Is it because A overspends on fancier books (perhaps because it tries harder to keep up with new pedagogical approaches) or is it because A may be teaching to a different language group and thus face higher unit costs for otherwise identical books? Of course, as with all these approaches, the emphasis is entirely on inputs and not on outputs, and the relation between inputs (better books, higher-paid and perhaps more qualified teachers) and outputs (better educated students) is seldom well understood. A third possible approach is to set out a representative expenditure system (RES) analogous to the RTS on the revenue side. Shah (1996) proposes a five-step way to determine such a system: 1. Disaggregate SNG expenditures into major h c t i o n a l categories such as health services, education, transportation and communication, etc. 2. Determine the influence on spending levels of cost and need indicators such as those listed above through regression analysis. This step is critical and difficult, requiring thorough understanding of not only differences in service areas, populations and local needs but also the objectives of public policy and the production functions (input-output relationships) of public services. As discussed further below, it is also critical to understanding possible strategic behaviors of recipient governments (Shah, 1994, Petchey and Levtchenkova, 2006). 3. Establish the per capita standardized expenditure of SNG for each category, employing national average values for the fiscal capacity indicators. What this procedure does is establish how much an SNG would spend, given its needs and costs profile, for each specific expenditure category if it had 'average' revenue. Since the weight of each factor was obtained empirically through regression analysis using data on all regions, in principle this method thus has the advantage of requiring objective standards. 4. Estimate the standardized per capita national expenditure (PCSEna,i ) for each category by evaluating the regression results at national mean values for all variables. 5. Using the equation below in the same way as RTS, the equalization grant to which each SNG is entitled can then be calculated:
EEx, i
= POPx
.[PCSEx,i - PCSEna, i]
EExj represents the equalization entitlement of region x for the spending category i, POPx is the population of region x and PCSEx,i and
Bird and Vaillancourt
268
PCSEna,i are the per capita standardized expenditure for region x and at national average for spending category i. In principle, then, to ensure that NFBs are equal across regions, not only the resources potentially available to finance public services but also the cost of providing those services must be taken into account. A comprehensive equalization scheme should, it seems, clearly include both fiscal capacity and expenditure differential measures. Before considering how meaninghl this conclusion actually is for countries concerned with developing a system of equalization transfers, however, we shall first review briefly the practices followed in a few selected countries to illustrate the extent to which expenditure differentials are actually taken into account in real-world equalization schemes.
2.3
Expenditure-based Equalization in Practice
We do not attempt here to present a complete picture of equalization systems around the world. We focus on the national level, neglecting subnational entities such as American or Indian states, or Canadian or Pakistani provinces, that are as large as many countries and that frequently engage in more or less generalized needs-based equalization in areas such as school finance (American states) or general expenditures (some Pakistani provinces). Moreover, we cover only six selected countries -- of various sizes, developed and less developed, with decentralized units that vary considerably in strength, and with varied institutional arrangements governing intergovernmental relations and of various sizes. Even this limited sample, however, probably represents fairly well most of the key ways in which expenditure factors are taken into account in practice.'4 Even with respect to these countries, the account presented here is very condensed and focused on only a few points of particular interest in the present context.15 Tables 1 and 2 set out the main points that emerge from this brief survey. Table 1 summarizes some key features of the equalization systems compared, and Table 2 depicts the expenditure factors taken into account in these systems. The balance of this section discusses briefly each of the six selected countries in turn. We turn to the lessons we draw from this brief survey in the final section of the paper.
2.3.1
Australia
Australia is characterized by a large vertical imbalance between revenues and expenditures at the national and SNG levels. The central government has control over all major revenue sources and collects 70 percent of total public sector revenues but is responsible for only 50 percent of total
Expenditure-BasedEqualization Transfers
269
government's expenditures. In contrast, the state governments, while spending almost 45 percent of total expenditures, only have control of 25 percent of revenues (Craig, 1997). Unsurprisingly, transfers from the center to the six states and two territories account for nearly 40 percent of SNG revenue in this highly centralized federation. Transfers from the center to the states take two broad forms, conditional grants called Special Purpose Payments (SPP) and a formal equalization program administered by the Commonwealth Grants Commission (CGC) that currently consists of an unconditional grant and a grant intended specifically for state health care services (HCG). About half of all transfers to the states are SPPs intended to achieve national policy objectives, mainly for education and health care, in areas of state expenditure. The principle underlying the general purpose grants within the equalization program is that "State Governments should receive funding from the Commonwealth such that, if each made the same effort to raise revenue from its own sources and operated at the same level of efficiency, each would have the capacity to provide services at the same standard" (CGC, 2004, p. 4). The CGC, established in 1933 as a permanent and independent authority, every five years determines state equalization entitlements employing a very comprehensive approach using 18 revenue and 41 expenditure categories (Rye and Searle, 1997). Since, as seen in Table 2, Australia has by far the most developed expenditure-based system, we include in Table 3 a full list of these categories (equivalent to those set out in the first step suggested by Shah (1996) and discussed earlier). The expenditure requirements of different states are estimated both on the basis of spending patterns and a variety of factors. Among what may be labeled the 'disability' factors used to determine states 'relativities7 are the proportion of population in the 5-15 age group, government school enrolment, non-government school enrolment, per capita demand for inpatient hospital services, and the unemployment rate. Among the cost factors are the proportion of population living in remote areas, the proportion of indigenous people, wages and salaries levels, and a building price index. A major change was made in the Australian system in 2000, when the total distributive pool was set as the amount of the Goods and Services Tax (a VAT) collected in the previous year, plus an amount determined by the central government specifically for health care services (HCG).
2.3.2
Brazil
Brazil is exceptional in Latin America in the extent to which it is a decentralized federation in terms of the size, tax autonomy, and expenditure share of its SNGs. It is also exceptional among federal states in the extent to which it operates a 'federal7 fiscal system with
Table 1. Equalization Transfers in Six Countries Country
Number of regions
Number of equalization programs
Australia
8
Germany
16
Switzerland
China
26
Distributive pool
Fiscal capacity categories
Expenditure differential categories
Fiscal effort
1
Federal VAT
18 (RTS)
41
No
3
Federal VAT
3 (RTS)
Horizontal sharing
3 (RTS)
Supplementaryfederal grants from general revenue
Variable
Federal conditional grants from general revenue
3 (RTS and macro)
FDT, withholding tax, custom duties (petrol and motor fuel) and National Bank's benefit
3
Cantonal contributions to social security
3
3
No 2
2
Gap-filling (general revenue) Determined ad hoc by the central government
13 (RTS)
Central VAT
I3 (RTS)
General revenue
1 (number of civil servants)
Yes
Table I . Equalization Transfers in Six Countries (Continued) Number regions
Number of equalization programs
35
3
Country India
Brazil
27
2
Note: Compilation by the authors
Distributive pool
Fiscal capacity categories
Expenditure differential categories
Fiscal effort Yes
Total central taxes
2 (RTS and macro) plus gap-filling
2
General revenue Specific purpose grants (general revenue)
2
2
Federal personal and corporate income tax and VAT
No
For states
2
For municipalities
1
Table 2. Expenditure Factors in Equalization Schemes Disability factors
Australia
China
Germany
India
Switzerland
a
a
Needs-based factors Population (equal per capita) Inverse of population share Population density Population dispersion Ratio of specific age groups to total population Sex Population control measures Low-income households and individuals Social security beneficiaries Low official language fluency Non official language background Number of students Literacy rate Humanitarian immigrants Number of general practitioners Indigenous or minority population Urban population Remote population Agricultural population
a a a a
a
a a a
a a a a a
a
Brazil
Table 2. Expenditure Factors in Equalization Schemes (Continued) Disability factors
Australia
China
Germany
India
Cost-basedfactors
Geographic area (mountainous, desert, arable, etc.) Temperature, rainfall, soil, etc. Natural hazards Border, revolutionary or civil war areas Road length Proportion of unpaved or sub-quality roads Sinuosity factor of regional roads Deck area of bridges, culverts and tunnels Harbor services Number of government units Number of districts, counties and municipalities Average salaries (public) by occupation Average salaries (non-public) by occupation Average effective rent rates Number of government vehicles Fuel consumption per vehicle per year Unit fuel price Vehicle damage coefficient Regional standard per capita heating expenditure Proportion of electricity and cost by plant type (hydro, steam, internal combustion, etc.)
a
a
a
a a 0 0
a
a a
0
a
Switzerland
Brazil
Table 2. Expenditure Factors in Equalization Schemes (Continued) Disability factors Average price of a three minute phone call Price of Internet connection Average price of transporting 50 kg of goods by freight Average price of a one-way economy fare plane ticket (intra-regional travel) Number of business locations Number of residential dwellings Number of first home owners Number of national park visitors
Note: Compilation by the authors
Australia
China
Germany
India
Switzerland
Brazil
Expenditure-Based Equalization Transfers
Table 3. Categories employed in Australian Equalization Grants, 2004 Revenues: 1 . Payroll Tax 2. Land Revenue 3. Stamp Duty on Conveyances 4. Financial Transaction Taxes 5. Stamp Duty on Shares and Marketable Securities 6. Gambling Taxation 7. Insurance Taxation 8. Heavy Vehicle Registration Fees and Taxes 9. Other Vehicle Registration Fees and Taxes
10. Stamp Duty on Motor Vehicle
Registration and Transfers 11. Driver's License Fees 12. Revenue Replacement Payments - Petroleum 13. Revenue Replacement Payments - Tobacco 14. Revenue Replacement Payments - Alcohol 15. Other Taxes 16. Interest Earnings 17. Mining Revenue 18. Contributionsby Trading Enterprises
Expenditures:
Pre-school Education Government Primary Education Non-government Primary Education Government Secondary Education Non-government Secondary Education Vocational Education and Training Higher Education Transport of Rural School Children Hospitals Nursing Homes Mental Health 12. Community Health 13. Public Health 14. Police 15. Administration of Justice 16. Corrective Services 17. Public Safety and Emergency Services 18. Family and Child Welfare 19. Aged and Disabled Welfare 20. Other Welfare 2 1. Housing Source: CGC 2004.
22. 23. 24. 25. 26. 27. 28. 29.
Electricity and Gas Water Supply and Sewerage Freight Non-urban Passenger Transport Other Trading Enterprises Other Concessions Culture and Recreation National Parks and Wildlife Services 30. Aboriginal Community Services 3 1. Superannuation 32. Other General Public Services 33. Primary Industry 34. Mining, Fuel and Energy 35. Tourism 36. Manufacturing and Other Industry 37. Roads 38. Urban Transit 39. Debt Charges 40. Depreciation 4 1. Regulatory and Other Services
276
Bird and Vaillancourt
respect not only to federal-state relations but also to federal-local relations. Its 5000 or so municipalities and 26 state governments play a very important role both in collecting revenue and in delivering public services. Two constitutionally-based transfers from the federal level to the state and municipal levels respectively are the heart of its intergovernmental fiscal system. Under the Fundo de Participacao dos Estados (FPE) the federal government distributes 2 1.5 percent of net revenues of the three main federal taxes, the personal and corporate income taxes and its (limited) VAT, to the states, with 85 percent of the funds designated specifically for the three poorer regions of the country (the North, Northeast and Center West). Within each group of regions (the three mentioned plus the wealthier South and Southeast regions), 95 percent of the FPE hnds are allocated on an equal per capita basis and a measure of per capita income (macro indicator) and 5 percent on the basis of geographic area (cost) (World Bank, 2002). A second very important revenue-sharing arrangement is the Fundo de Participacao dos Municipios (FPM). This transfer consists of 22.5 percent of net revenues of the same three federal taxes. The main allocation criterion is population, with 10 percent of the funds distributed to state capitals based on their share of the combined population of all state capitals and the remaining 90 percent to the rest of the municipalities. Of this 90 percent, 86.4 percentage points are allocated to states on the basis of their share of national population and then divided among municipalities according to a scale in which a minimum share is provided to small localities (population less than 10,188) and a maximum to cities with a population over 156,216. The final 3.6 percent of the FPM is distributed to the larger municipalities just mentioned, taking into account both population and per capita income.I6 In addition, two other grants are intended to achieve some degree of equalization. The FUNDEF, a find created to finance SNG spending on education, provides earmarked grants to states and municipalities intended to reduce disparities across states in wages and salaries, capital outlays, and operations and maintenance. The second program, the SUS, is a transfer aimed at financing health care provided at the municipal level. Based on cost differentials measured by past budget allocations on a per capita basis, it is earmarked for specific programs such as basic and preventive health care, pre-natal care, oral hygiene and immunization. The SUS has been used in recent years to increase budget allocations for poorer regions (Afonso, 2000, and Afonso and de Mello, 2000). Although Brazil has no formal equalization for either fiscal capacity or expenditure differentials, the net effect of the system described above is to achieve a substantial degree of equalization among states (World Bank, 2002).
Expenditure-Based Equalization Transfers
2.3.3
China
The disparities found among China's 27 provinces and autonomous regions and 4 large municipalities are among the largest in the world (Ma, 1997). It is thus not surprising that among the objectives of a major Tax Sharing System (TSS) reform in 1994 were simpler and economically more appropriate revenue and expenditure assignments, the reduction of the perverse incentives for SNGs arising from the previous system of intergovernmental finance, and more equalization among the highly diverse regions (Zhang and Martinez-Vazquez, 2003).17 Although not a federation, China has developed a complex system of intergovernmental transfers in its attempt to achieve these goals. At present, equalization transfers represent only 20 percent of central government transfers to the provinces, with most transfers being derivationbased VAT payments. In total, China has nine equalization transfers, with the most important (80 percent of total) consisting of pre-tax sharing system grants, transitory period grants, equalization grants for minority regions, and grants for increasing wages of civil servants. The pre-TSS transfers were determined by the gap between the revenues and expenditure requirements of a province measured in a base year. Such transfers may flow from or to the center depending on the deficit or the surplus of the provincial governments. Although there are obvious problems with this 'gap-filling' approach, Zhang and Martinez-Vazquez (2003, p.22) consider the pre-TSS grants to be the most equalizing transfers in the present system. To a considerable extent, of course, this result reflects the relatively equalizing nature of the pre-TSS system of intergovernmental finance. The transitory period grants, unlike the pre-TSS grants, grew steadily from 1995 to 2001, although their total amount is determined annually on an ad hoc basis by the central government. Once determined, however, the grantl8 is allocated by a formula relying on a measure of 'standard7 provincial revenue and expenditure multiplied by a provincial coefficient. Fiscal capacity is determined using an RTS approach with 13 different revenue sources, taking into account also such factors as tax rebates, other general-purpose grants and remittances to the central government.19 Similarly, on the spending side, a detailed RES approach is used with 12 expenditure categories. Spending in the first seven categories -- government administration, agriculture, forestry and irrigation, culture and sports, education, health care, and "other" -- is determined by using measures of personnel and office expenditures. Personnel expenditures are based on a 'standard7 number of civil servants and a standard per capita salary for a given sector in a given region. The standard number of civil servants for a specific category is determined (by the central government) on the basis of population,
Bird and Vaillancourt
278
agricultural population, number of government units, number of students, area of residential districts, number of districts and counties, and area of arable land. Per capita salary is determined by the average local per capita salaries paid at different government levels (province, prefecture and county). Similarly, 'office' expenditures depend on measures of standard expenditure for fuel, maintenance of vehicles, heating and other office expenses. Fuel and maintenance expenditures are determined by the number of vehicles, fuel consumption per vehicle per year, unit fuel price, annual actual maintenance expenditure (determined by ratio of sub-quality roads on total length of roads and a vehicle damage coefficient), plateau area and road condition adjustment coefficients. Standard expenditure for heating depends upon the number of civil servants and standard per capita heating expenditure of the region (one of five groups of regions divided according to the average daily temperature of the capital cities in each region). For the other five expenditure categories included in the transitory period grants -pensions, urban maintenance and construction, social security and welfare, grain risk fund, and agricultural support -- the main variables used to determine standard expenditure requirements are urban population, area of completed urban zones, number of urban vehicles, real local expenditure for urban maintenance and construction, urban road distribution. A special grant for minority regions (designated areas within particular provinces) was created in 2000, allocated in part by the same measures of standardized revenues, expenditures and provincial coefficients and in part by the growth rate of the central share of the VAT.^' Finally, although the grants for increasing standard wages of civil servants are obviously conditional in nature, they too are equalizing since they are directed to the western provinces (among the poorest in the country). This transfer, motivated by the (nationally-mandated) 2001 wage increases granted to civil servants, is based on the number or civil servants in each region times the mandated salary increases. The same regional coefficients, varying from 20 to 80 percent, are used, with minority regions benefiting fkom an additional 5 percentage points. 2.3.4
Germany
In Germany, all major tax sources (personal income tax, corporate income tax and VAT) are shared between the central and state governments. The extensive influence of the central government and the strong cohesion among the states on policy, legislation and tax issues characterize what has been called the "unitary German federation" (Spahn and Fottinger, 1997, p. 226). Germany's constitution commits all levels of government to "the establishment of equal living conditions in the federal territory" (Article 72 I1 GG), which is much stronger than the requirement for some 'minimum
Expenditure-Based Equalization Transfers
279
level' or 'same standard' of public services found in federations such as Canada (and implicit in the theory of equalization set out earlier). Germany also provides the clearest federal example of a horizontal equalization . ~ ' introduced in 1951 to compensate some scheme, the ~ i n a n z a u s ~ l e i c hFirst states for "special burdens" with respect to refugees, harbor maintenance and so on, since 1955 this scheme has been enshrined in Article 107 of the Constitution to equalize the fiscal capacity of the states (Ma, 1997). Actually, three different transfer programs serve equalization purposes in Germany. The first program is based on the VAT, of which 49.5 percent goes to the states, with 75 percent distributed on a per capita basis and the remaining 25 percent distributed to states with below-average per capita personal and corporate income tax and local business tax revenues. These transfers are intended to raise the tax capacity of the receiving states to 92 percent of the national average. The second program is the Finanzausgleich under which the sum of payments received by below-average states is always equal to the sum of disbursements of the above-average states. An "equalization yardstick" is defined by taking the average tax revenue per capita weighted by a population density factor, adding 50 percent of a standardized measure of municipal revenue and allowing for "special burdens" for some northern states. The population density factor is graduated between 1.0 and 1.3, except that the city-states of Bremen, Hamburg and Berlin are given a 1.35 factor.22 The special burden compensation is a lump-sum correction with the only important category being the provision of harbor services. In short, the equalization yardstick is a fiscal capacity indicator accounting for two expenditure differential factors. The yardstick is compared to the actual financial situation of each state and the difference is equalized down in progressive steps for above-average states and compensated up to 95 percent of national average for the poorer states. The third equalization program consists of supplementary federal grants given to financially weaker states, both eastern and western, to ensure they reach at least 99.5 percent of the national average fiscal capacity. Other special grants (for many purposes) complete Germany's transfers system, making it the most equalizing system found in federal countries - although one that has often been criticized for penalizing fiscal effort in rich states and encouraging fiscal irresponsibility in poorer states (Spahn and Franz, 2000). 2.3.5
India
In India, intergovernmental fiscal transfers have been used since 1919 to correct vertical and horizontal imbalances. India's transfer system now consists of three programs: those established by the Finance Commission;
280
Bird and Vaillancourt
grants and loans for state development plans authorized by the Planning Commission; and a set of 220 specific purpose transfers from various central ministries to the states (Rao, 2004). For over 50 years, Finance Commissions have been responsible for determining state budgetary requirements and suggesting methods of revenue distribution every five years. The Eleventh Finance Commission reported in April 2000, and the tax devolution arrangements and unconditional gap-filling grants it recommended accounted for 65 percent of total intergovernmental transfers in 2001-02. For the 2000 to 2005 period, the distributive pool has been set at 29.5 percent of net proceeds of total central taxes (Rao, 2004). Equalization transfers are formula-based using fiscal capacity, expenditure differential and fiscal effort indices: 10 percent is based on population23(equal per capita); 62.5 percent on income per capita (macro); 7.5 percent on area, thus favoring lower density regions (cost); 7.5 percent on an index of infrastructure; and 12.5 percent on 'tax effort' and 'fiscal discipline.' The second set of transfers, representing 30 percent of the total, consist of two programs determined by the Planning Commission. Thirty percent of these funds are intended for the 'special category' states (eleven mountainous states of the north and northeast) to provide assistance for specific projects and plans submitted to and approved by the federal government. The other 70 percent go to the major states, 30 percent as grants and 70 percent as loans. These funds are distributed 60 percent based on population, 25 percent on income per capita, 7.5 percent on fiscal performance (tax effort, fiscal management and national objectives), and 7.5 percent on 'special problems7 (e.g. population control, literacy, land reform, etc.) (needs). The final 5 percent of transfers consists of the 220 specific purpose grants mentioned earlier. On the whole, India's complex system, while criticized by some as providing undesirable incentives with respect to the fiscal management of the states, appears both to have been significantly equalizing and to have contributed to achieving a degree of cohesiveness in a large and diverse country 2.3.6
Switzerland
Swiss cantons enjoy a high degree of autonomy for both historical and political reasons. With important variations among the 26 cantons in terms of population, language, religion, size, geography and economic activity, Switzerland, although a small country in terms of population, is among the most decentralized federations. Equalization programs have been incorporated in the constitution since 1958 though without any explicit
Expenditure-Based Equalization Transfers
28 1
objective to ensure a minimum or standard level of service provision among the different cantons. Rather, the very pragmatic objective has been "to render cantonal disparities politically acceptable so that remaining differences do not endanger the cohesion of the Confederation" (Dafflon, 2004, p. 15). Equalization accounts for 10 percent of cantonal revenues and is accomplished through three broad programs. The first consists of a variety of federal conditional grants to cantons. About a third of the federal aid programs that benefit cantonal expenditure categories have an equalizing component. These equalizing transfers, representing 20 percent of federal expenditure and 15 percent of cantonal revenue, are based on a financial capacity index (E) that is computed using both fiscal capacity and expenditure differential indicators. Three revenue-raising capacity indicators are used: national income per canton (macro) - 30 percent; tax revenue per capita for both canton and municipalities (RTS) - 30 percent; and a tax effort or burden index (RTS) - 20 percent. Two expenditure differential indicators are employed: proportion of agricultural land in mountainous regions (cost) - 10 percent; and population density (need) - 10 percent. As in Australia and India, but in contrast to Germany, cantons with low population density benefit &om equalization transfers since it is thought that cantons with higher population density reap economies of scale in the provision of public services. The second equalization scheme is a revenue sharing arrangement between the Confederation and the cantons under which 30 percent of the federal direct tax (income, corporate profits and capital) is distributed among cantons, 17 percent on a derivation basis and 13 percent using the financial capacity index. In addition, cantons receive 10 percent of the federal withholding tax (5 percent on a per capita basis and 5 percent on the financial capacity index), 12 percent of custom duties on motor fuel (of which 42 percent is equalized) and 213 of the Swiss National Bank's profit (518 on a per capita basis and 318 equalized)24. Finally, the third equalization scheme relates to cantonal contributions to social security expenditures. Contributions to the three federal programs (old-age and survivors insurance (AVS), disabled pension scheme (AI) and family allowance in agriculture (AFA)) are scaled using a modified version of the financial capacity index. Following a decade of negotiation between the cantons and the Federation, equalization in Switzerland is about to undergo a major reform, assuming the new equalization law succeeds in the referendum required to make the necessary constitutional changes. Among other features, the new system will introduce new formulas with respect to both the expenditure and revenue sides of equalization, employing an RTS approach to fiscal capacity
Bird and Vaillancourt
282
and using socio-demographic indicators as well as geographic factors on the expenditure side (Dafflon, 2004).
EXPENDITURE EQUALIZATION: EVALUATION While it is difficult, and perhaps dangerous, to generalize on the basis of this selective review of equalization systems, three conclusions seem to emerge: 1. The theoretical logic of equalization set out in the first section does not appear to have had much impact on actual equalization systems. Why? 2. Only Australia, of the countries considered here, appears to have adopted the apparently most logical (NFB = RTS+RES) approach to equalizing expenditure. Why? 3. Many countries seem to take equalization into account in a variety of 'case-specific' ways such as special allowances for particular regions. Should this approach be considered a supplement to, or a replacement for, more formal expenditure-based equalization schemes? Drawing not only on the explicit features of the countries discussed above but also on our own personal experience with intergovernmental finance systems in a wide variety of countries in recent decades, we argue in this section that it is not surprising that neither the logic of expenditure equalization nor the practice of RES has had so little impact in practice, and that the simpler and more pragmatic approach to equalization observed in most countries is not only understandable but desirable.
3.1
Why Is No One Listening?
Recall the question we raised earlier: why should the outcome of a unitary state be the desired one? Of course, in many countries in the past - and indeed in many even today - the general shape of the state has in effect been imposed by a small elite. Still, in democratic countries such as Spain or Belgium, it is clear that there have been decisive moves towards increased decentralization in the last 25 years or so, either with (Belgium) or without (Spain) a formal move towards federali~m.~'Presumably, democracies can also choose to move towards centralization, implicitly or explicitly. It thus does not seem to us unreasonable to assume that at least in a democracy an observed high degree of decentralization is in a real sense the result of choice.
Expenditure-Based Equalization Transfers
283
When resources are reasonably free to move within a country, observed differences in economic well-being between regions and thus in levels of public services can, equally reasonably, be assumed to be desired at least to some extent by their residents. Some such differences, for example, may well be compensated in less visible ways - for example, by differences in social capital (family ties, community support), access to 'free' natural resources (fishing, hunting, better climate) and the value of fixed assets (paid-up ancestral houses). Of course, if additional funds to make life even better are freely obtainable from a benevolent central government, the residents of 'disadvantaged' regions will certainly accept them. But it is not at all as clear as the standard theory of equalization asserts that this is the only decent way to run a federal state. Even in countries that are neither federal nor democratic, path dependence often rules. Conditionality, for example, is not consistent with the theoretical case for equalization, but it is not surprising to find, as Slukhai (2003, p.9) notes with respect to transition countries in Eastern Europe that "...the state treats equalization transfers as an element of its influence over local governments." In China, for example, much of what is called 'equalization' is really aimed at ensuring that central government objectives (salary increases) are met. Similar policies may be found in many developing countries, perhaps particularly those coming from a centralplanning tradition andlor those fearing regional separatism in one form or another. They are also common in developed countries when it comes to state-local (as opposed to federal-state) relations. He who pays the piper frequently wants to call the tune and can, in most cases, do so regardless of the name of the transfer payment. Economists may be frustrated when those in power seem not to listen to their advice. Politicians succeed, however, by listening to those who matter with respect to keeping them in power, a group unlikely to include many academic economists. Even in established federal democracies, the level and structure of transfers can often be best understood as the outcome of prolonged political negotiation between contending (and in some sense equally 'sovereign') governments - negotiation that in part no doubt reflects what Breton and Scott (1978) called the differing 'span of concern' to which governments with different territorial jurisdictions properly respond. It may still be useful in some respects to compare the resulting system against a 'baseline' (unitary state) comparison as the theory of equalization in effect does, but there is no presumption that this baseline case sets a relevant normative standard.
Bird and Vaillancourt
3.2
Why Does No One Do it Right?
The question posed in the title to this sub-section may seem both unnecessary and wrong. The question seems unnecessary since, if as just argued almost no one is doing 'it' - equalization - for the reasons assumed in the theoretical discussion, then it seems unnecessary to consider how it might be done 'right' if anyone wanted to do it. On the other hand, Australia does appear to come close to doing 'it', so the question seems wrong. We suggest, however, first, that even if countries wish to equalize in the sense assumed in the theory, that the Australian 'model' is unlikely to provide a useful guideline for both conceptual and practical reasons and, second, that even if this model were to be followed, its results would be unlikely to accord with the theoretical objective. Since Petchey and Levtchenkova (2004) essentially deal with the second of these points, we shall comment here only on the first. Almost all who have studied the RTS-RES approach agree on two points: first, it is formally the most satisfactory way to meet the normative objectives of the theoretical equalization model and, second, that it is difficult and costly to obtain the necessary data (in a form that all parties agree is satisfactory), especially for expenditures. While we continue to think that the RTS approach is, on the whole, nonetheless probably still the best way to deal with the revenue side of the on balance we think that it is unlikely to be worth the cost of introducing a parallel RES system in most countries. In federal countries, as Bird (1986) shows with respect to Switzerland, for example, experience tells us that, so to speak, formula follows fimction. That is, what is important is not that the formula used for equalization purposes is 'correct' but that the results of applying it are politically viable. When the results produced by any particular formula become sufficiently unsatisfactory from the perspective of significant political actors, the formula gets changed. Introducing still more elements in the formula on the expenditure side may be a useful way to, as it were, reduce the annual political turmoil about who gets what by confusing everyone as to what exactly is going on, but those who think they are losing out are unlikely to stay confused for very long. Equalization transfers may be an essential element of federal 'glue' in some countries but in the long run, in this as in other aspects, transparency should trump complexity when it comes to such central political elements of the intergovernmental fiscal system. In more centralized countries, however, in which transfers are essentially one of the tools used by the central government to get its objectives achieved in a situation in which, for whatever reason, important activities (e.g. health and education) are in the hands of SNGs, then an RES approach may make more sense. Instead of a variety of conditional grants on the U.S. model, with different degrees of equalization from grant to grant, a country might
Expenditure-Based Equalization Transfers
285
choose to have a more uniform equalization system focused on the provision of uniform national standards in some key areas both to achieve national objectives and also to ensure in effect a more 'level playing field' for other transfers intended to induce specific actions by recipient governments.27
3.3
Pragmatism and Asymmetry
On the whole, therefore, expenditure-based equalization seems more likely to find a welcome in unitary than in (truly) federal countries. Even in such countries, however, many of the purported objectives of such policies can be more simply achieved by special 'regional' transfers rather than by attempting to incorporate a variety of complex factors in a 'general' equalization scheme. Some instances of such 'special' transfers were mentioned in the countries surveyed above. It is not hard to find others. In Canada, for example, not only is there a separate formula for the three sparsely-populated northern territories but, unlike the case with the provincial equalization system, this formula incorporates substantial 'expenditure' elements in recognition of the extreme cost and need differentials between the territories and the provinces. Many other countries to some degree or other incorporate similar factors in their financing arrangements for 'special states' such as border areas or those inhabited by minority populations. In the name of transparency, such 'asymmetric' arrangements often seem superior to attempts to incorporate every region, no matter how different, in a uniform formula. A possible argument against the approach we suggest - essentially, deciding what a transfer system is supposed to do and then doing it as clearly and simply as possible - is what may perhaps be called the 'universality' argument. Those concerned primarily to direct more resources to the poor and deserving may think this can best be achieved by establishing an elaborate general transfer system that churns huge amounts of money around the system, much of which ends up where it started, in the name of channeling a relatively small additional amount to those who (the proponents of this approach think) need it. We are not sympathetic to this approach. Politicians who do not understand who gets what are unlikely to be long in office, and if they and their constituents do not support regional redistribution trickery is unlikely to make it more sustainable.
286
Bird and Vaillancourt
Notes We thank Philippe Wingender for his excellent contributions to this paper. The importance of good transfer design is evident from any review of transfer systems around the world: see, for example, Slukhai (2003) on transition countries and International Symposium (2001) on Asian countries. See for example Scott (1952), Courchene (1978) and Boadway and Flatters (1982). Boadway (2004) provides an excellent recent survey of this literature. Nor is equalization intended to rectify regional 'imbalances' or eliminate regional disparities, which is just as well since it can be argued (Courchene, 1978) that equalization is more likely to perpetuate than to reduce such disparities. See, for example, Petchey and Levtchenkova (2006). Watson (1986) is an earlier examination of the same issue. Martinez-Vazquez (2002) and Boadway (2003, p.258) consider that the equal per capita feature of Canada's main social transfer (the CHST) ... "is implicitly a perfect system of equalization" and Laurent and Vaillancourt (2003) find a negative relationship between per capita income and CHST transfers for Canada. We do not discuss the problem of how such transfers are financed (see Musgrave 1961). It is by no means always clear what is an 'own' tax base: is it a tax administered by an SNG? Or one whose rate and/or base is determined by the SNG? Or simply one the proceeds of which accrue to the SNG? All possible combinations of these characteristics exist somewhere: where is the line to be drawn? For helpful though not conclusive discussions of these points, see OECD (1999) and Ebel and Yilmaz (2003). Indirectly, tax potential may be affected since labor supply and financial decisions can be affected by tax rates. Boadway and Hobson (1993, p.92) note that this distinction is not always clear. For example, if there are ten snow storms (IOcm) per year in region A and five (20cm) in region B, leaving in total the same amount (100cm) of snow on the ground, then if there is a fixed and a variable cost to each snow removal operation A will incur higher expenditure even if the marginal cost of an additional cm is the same in both regions. Does A have higher costs or greater need (more storms)? For examples, see Shah (1996, p.103), Ma (1997, p.1 I) and Bird (1986, p.144). For much more information, on many more countries, see, among others, Shah (1994a), Ahmad (1997), Ter-Minassian (1997), and Bird and Tarasov (2004). There is of course a huge literature on each of these countries - only a few items of which are cited here -as well as a good deal of ongoing work which we make no attempt to summarize. To illustrate : we are aware of several recent studies on various aspects of equalization in China (Bahl and Martinez-Vazquez, 2003; Wong, 2003). For an earlier analysis of the Chinese case by one of the present authors, see Bird and Chen (1998). In addition to the FPM, states are required to share 25 percent of their own broad-based VAT with municipalities, although largely on a derivation basis. The balance of this section is based largely on Zhang and Martinez-Vazquez (2003). There are actually two grants, a general and a special grant, with the latter based on the higher costs of providing public services to minority regions, border areas, and counties considered as revolutionary and civil war areas. It should be noted, however, that in almost all cases the rates and bases of SNG taxes in China are set centrally and uniformly. For a detailed treatment of China's complex fiscal treatment of its 'minority' regions, see Wong (2003). 21. Denmark provides an example of such a system in a unitary state (Lotz, 1997).
Expenditure-Based Equalization Transfers
287
22. Interestingly, while Australia considers remote and scarcely populated regions to be more demanding in terms of expenditure requirements, Germany's Finanzausgleich favors highly populated areas. 23. 1971 population is used in this calculation, reportedly to encourage birth control!(Rao,2004). 24. In the first case, the general index of financial capacity, E is multiplied by total cantonal exvenditures on roads (share of canton I in total for Switzerland) to calculate the amount received by each canton; in the second, E is multiplied by the population of each canton (Dafflon,2004). 25. For recent detailed examinations of these two cases, see Bayenet and de Bruycker (2003) and Garcia-Mila and McGuire (2003), as well as the briefer discussion in Bird and Tarasov (2004). 26. For discussion of problems with this approach, see Bird and Slack (1990) and Smart (1998). 27. This is more or less the schema laid out for Colombia by Bird and Fiszbein (1998).
References Afonso, Jose Roberto R. 2000. Intergovernment transfers: Brazilian experience. Paper presented at International Seminar on Fiscal Federalism, SCHP-CEPAL- ILPES, Cancun, Mexico. Afonso, JosC Roberto R. and Luiz de Mello (2000) "Brazil: An Evolving Federation." IMFIFAD Seminar on Decentralization, Washington, November. Ahmad, Ehtisham, ed. 1997. Financing Decentralized Expenditures: An International Comparison of Grants. Cheltenham, UK: Edward Elgar. Bahl, Roy, and Jorge Martinez-Vazquez. 2003. Fiscal federalism and economic reform. International Studies Program Working Paper No. 03-14, Andrew Young School of Policy Studies, Georgia State University, Atlanta. Bayenet, Benoit, and Phillipe de Bruycker. 2003. Belgium: A Unique Evolving Federalism. World Bank Institute, Washington. Bird, Richard M. 1986. Federal Finance in Comparative Perspective. Toronto: Canadian Tax Foundation. , and Duanjie Chen. 1998. Intergovernmental fiscal relations in China in international perspective. In Taxation in Modern China, Donald J.S. Brean, ed., London: Routledge. ,and Ariel Fiszbein. 1998. Colombia: The central role of the central government in fiscal decentralization. In Fiscal Decentralization in Developing Countries, Richard M. Bird and Francois Vaillancourt, eds., Cambridge: Cambridge University Press. , and Enid Slack. 1990. Equalization: The representative tax system revisited. Canadian Tax Journal, 38: 913-27. , and Andrey V. Tarasov. 2004. Closing the gap: fiscal imbalances and intergovernmental transfers in developed federations. Environment and Planning C: Government and Policy, 22: 77-102. Boadway, Robin. 2001. Equalization revisited: Its role and design. Paper presented at the 3oth Annual Conference of the Atlantic Provinces Economics Association, Fredericton, New Brunswick. . 2003. Options for Fiscal Federalism. Royal Commission on Renewing and Strengthening Our Place in Canada, Newfoundland and Labrador, St. John's.
288
Bird and Vaillancourt
. 2004. The theory and practice of equalization. CESifo Economic Studies, 50 (1): 2 11-254. ,and Frank Flatters. 1982. Equalization in a Federal State: An Economic Analysis. Ottawa: Economic Council of Canada. , and Paul A.R. Hobson. 1993. Intergovernmental Fiscal Relations in Canada. Toronto: Canadian Tax Foundation. Breton, Albert, and Anthony Scott. 1978. The Economic Constitution of Federal States. Toronto: University of Toronto Press. Buchanan, James M. 1950. Federalism and fiscal equity. American Economic Review, 40: 583-599. Commonwealth Grants Commission (CGC). 2004. Report on State Revenue Sharing Relativities, 2004 Update. Canberra: Commonwealth of Australia. Courchene, Thomas J. 1978. Avenues of adjustment: The transfer system and regional disparities. In Canadian Confederation at the Crossroads: The Search for FederalProvincial Balance, M. Walker, ed., Vancouver: The Fraser Institute. . 1998. Renegotiating Equalization: National Polity, Federal State, International Economy, C.D. Howe Institute Commentary 1 13, C.D. Howe Institute, Toronto. Craig, Jon. 1997. Australia. In Fiscal Federalism in Theory and Practice, Teresa TerMinassian, ed., Washington, D.C.: International Monetary Fund. Dafflon, Bernard. 2004. Federal-Cantonal Equalisation in Switzerland: An Overview of the Present System and Reform in Progress, Working Paper No 356, BENEFRI Center for Studies in Public Economics, University of Fribourg. , and Franqois Vaillancourt. 2003. Problems of equalisation in federal system. In Federalism in a Changing World - Learning from Each Other, Blindenbacher R. and A. Koller, eds., Montreal: McGill-Queen's University Press. Ebel, Robert D., and Serdar Yilmaz. 2003. On the measurement and impact of fiscal decentralization. In Public Finance in Developing and Transitional Countries, Jorge Martinez-Vazquez and James Alm, eds., Cheltenham, UK: Edward Elgar. Garcia-Mila, Teresa, and Therese J. McGuire. 2003. Fiscal Decentralization in Spain: An Asymmetric Transition to Democracy, World Bank Institute, Washington, D.C. International Symposium on Intergovernmental Transfers in Asian Countries: Issues and Practices (2001) Tokyo: Asian Public Policy Program, Graduate School of International Corporate Strategy, Hitotsubashi University. Laurent, Stephen, and Franqois Vaillancourt. 2003. Federal transfers in Canada and the United States, 1989-1990 to 1998-1999: How equalizing are they? In National Tax Association, Proceedings 2002 Annual Conference, 203-2 12. Lotz, Jorgen. 1997. Denmark and Other Scandinavian Countries: Equalization and Grants. In Financing Decentralized Expenditures: An International Comparison of Grants, Ehtisham Ahmad, ed., Cheltenham, UK: Edward Elgar. Ma, Jun. 1997. Intergovernmental fiscal transfers: A comparison of nine countries. Paper prepared for the Macroeconomic Management and Policy Division, Economic Development Institute, Washington: World Bank. Martinez-Vazquez, Jorge. 2002. Equalization goals, allocation factors and international practices. Paper presented at ASEM International Workshop on Intergovernmental Fiscal Relations in East Asia, Bali, Indonesia. Martinez-Vazquez, Jorge, and Jamie Boex. 2001. The design of equalization grants: theory and applications, Part One: theory and concepts. World Bank Institute, Washington.
Expenditure-Based Equalization Transfers
289
Musgrave, Richard A. 1961. Approaches to a fiscal theory of political federalism. In Public Finances: Needs, Sources, and Utilization, National Bureau of Economic Research, Princeton: Princeton University Press. Oakland, William. 1994. Fiscal equalization: An empty box? National Tax Journal, 47: 199209. OECD. 1999. Taxing Powers of State and Local Government Paris. Petchey, Jeff, and Sophia Levtchenkova. 2006. Fiscal capacity equalization and economic efficiency. In this volume. Rao, M. Govinda. 2004. Changing contours in federal fiscal arrangements in India. Paper presented at International Symposium on Fiscal Decentralization in Asia Revisited, Hitotsubashi University, Tokyo. Rye, C. Richard, and Bob Searle. 1997. Expenditure needs: institutions and data. In Financing Decentralized Expenditures: An International Comparison of Grants, Ehtisham Ahmad, ed., Cheltenham, UK: Edward Elgar. Scott, Anthony D. 1952. Federal grants and resource allocation. Journal of Political Economy, 60: 534-538. Shah, Anwar. 1994. A fiscal needs approach to equalization transfer in a decentralized federation. World Bank Policy Research Working Paper No 1289, World Bank, Washington. , 1994a. The Reform of Intergovernmental Fiscal Relations in Developing and Emerging Market Economies. Policy and Research Series, Washington: World Bank. . 1996. A fiscal need approach to equalization. Canadian Public Policy, 22(2): 99115. . 2004. Lessons from international practices of intergovernmental fiscal transfers. Paper presented at XVI Regional Seminar on Fiscal Policy CEPALIECLAC, Santiago de Chile. Slukhai, Sergii, ed. 2003. Dilemmas and Compromises: Fiscal Equalization in Transition Countries. Local Government and Public Service Reform Initiative. Budapest: Open Society Institute. Smart, Michael. 1998. Taxation and deadweight loss in a system of intergovernmental transfers. Canadian Journal of Economics, 3 1 : 189-206. Spahn, Paul B., and Wolfgang Fottinger. 1997. Germany. In Fiscal Federalism in Theory and Practice, Teresa Ter-Minassian, ed., Washington, D.C.: International Monetary Fund. Spahn, Paul B., and Oliver Franz. 2000. Consensus democracy and interjurisdictional fiscal solidarity in Germany. Paper Presented at IMF Fiscal Affairs Department Conference on Fiscal Decentralization, International Monetary Fund, Washington. Ter-Minassian, Teresa. Ed. 1997. Fiscal Federalism in Theory and Practice. Washington: International Monetary Fund. Vaillancourt, Franqois. 2001. Intergovernmental Finance in Laos. World Bank, Washington. Watson, William G. 1986. An estimate of the welfare gains from fiscal equalization. Canadian Journal of Economics, 19: 298-308. Wilson, Leonard S. 2003. Equalization, efficiency and migration: Watson revisited. Canadian Public Policy, 29(4): 385-396. Wong, Christine. 2003. Ethnic minority regions and fiscal decentralization in China: The promises and reality of asymmetric treatment. Working paper, University of Washington. World Bank. 2002. Brazil: Issues in Fiscal Federalism, Report No 22523-BR, Washington. Zhang, Zhihua, and Jorge Martinez-Vazquez. 2003. The system of equalization transfers in China. International Studies Program Working Paper No. 03-12, Andrew Young School of Policy Studies, Georgia State University, Atlanta, GA.
Chapter 10
DESIGNING INTERGOVERNMENTAL EQUALIZATION TRANSFERS WITH IMPERFECT DATA: CONCEPTS, PRACTICES, AND LESSONS JAMESON BOEX AND JORGE MARTINEZ-VAZQUEZ' Department of Economics, Andrew Young School of Policy Studies, Georgia State University
1.
INTRODUCTION
The design of intergovernmental equalization transfer mechanisms, whether as the result of the introduction of a new transfer scheme or as part of the revision of an existing one, is a key element of local government finance reform around the world. While the basic elements and principles of designing intergovernmental fiscal transfer schemes apply universally, less developed and transition countries (LDTCs) often face the additional challenge of designing their transfer mechanisms in the absence of substantial relevant data on local fiscal, demographic, and socio-economic variables. The absence of the necessary data to adequately quantify local expenditure needs and fiscal capacity in order to allocate formula-based equalization grants in an efficient, equitable and transparent manner forms an additional hurdle in the implementation of a sound system of intergovernmental fiscal relations in many developing and transition economies. Quite often, the only data many LDTCs appear to have in order to compute suitable measures of subnational fiscal capacity and expenditure needs are actual subnational revenue collections and actual subnational expenditures. Of course, local revenues may provide a poor reflection of fiscal capacity (due to variations in local fiscal effort) and are generally not an appropriate allocation factor for a transfer formula due to the negative incentive effects on local revenue collections. Similarly, actual local expenditures are not an appropriate measure of need since their use would
292
Boex and Martinez- Vazquez
provide incentives for overspending or may simply freeze inequitable historical patterns of local e ~ ~ e n d i t u rIn e . ~reality, these difficulties have hardly deterred LDTCs in the midst of decentralization processes to seek ways to design a system of equalization grants. In addition most policy experts would agree that even in the absence of the desired data, a formulabased equalization transfer scheme is still preferable to a discretionary, negotiated grant approach. The main issue we want to address in this paper is the challenge faced by fiscal policy practitioners in many LDTCs: how to formulate a system of intergovernmental equalization transfers with imperfect data, or at least without the luxury of first-best data. This paper seeks to achieve two objectives. First, we seek to provide a comprehensive overview of the alternative approaches that can be taken to measure local expenditure needs and fiscal capacity. Facing the difficulty of designing equalization grant systems with imperfect data, many countries have moved ahead with ingenious and sometimes nearly appropriate methodologies to quantify expenditure needs and fiscal capacity. We cover many of those experiences in the context of the countries where these have been tried. Second, we explore the implications of designing an equalization transferformula in the presence of less-than perfect data. At the present time there is a dearth of comparative evidence and generally very limited guidance in the existing literature regarding what "second-best" approaches can be used in the design of intergovernmental equalization. Our goal is to help fill this vacuum in the literature by assessing different grant mechanisms that are used in LDTCs where we lack first-best measures of expenditure needs and fiscal capacity. In order to assess the performance of the different approaches we rely on local government finance data for the State of Georgia in the U.S., going back to the 1960s. At that time, Georgia was among the poorest U.S. states and in many ways living conditions were similar to those now existing in developing countries. The advantage of analyzing historical Georgia is that there is practically full availability of first-best data, enabling us to compare alternative measurement approaches as part of the design of equalization grants. In doing so, we consider the performance of alternative measures of fiscal capacity and fiscal need in terms of equalization vis-A-vis their "first best" measures. By simulating and comparing the performance of various transfer schemes in Georgia, we uncover the relative strengths and weaknesses of the different approaches to grant design and seek to translate these fiscal patterns into policy implications for the design of intergovernmental grant schemes in LDTCs.
Designing Intergovernmental Equalization Transfers
2.
UNCONDITIONAL VERTICAL AND EQUALIZATION TRANSFERS
The general intent of unconditional transfer mechanisms is either to improve the vertical fiscal balance by providing general-purpose funding at the subnational level or to improve the horizontal fiscal balance by compensating for fiscal disparities across regions.3 Fiscal disparities arise from two main sources. First, local or regional governments may differ in their fiscal capacity, that is in their economic base and therefore in their ability to raise a particular level of revenue with standard rates and administration effort. Second, local governments may also differ in their expenditure needs. Even when localities or regions have the same fiscal capacity or ability to raise revenues, they may differ in the costs they face to provide a standardized basket of public services due to differences in needs arising from different demographic profiles (e.g., the percent of the population of school age or retired), geographical and climatological conditions, incidence of poverty and unemployment, and so on. The differences in expenditure needs among regions may also arise due to differences in costs or price levels related to the provision of a standard basket of public s e r ~ i c e s . ~ In international practice there are countries that use formulas to equalize both fiscal capacity and expenditure needs (including many developed and transitional countries), countries that use mechanisms that equalize only fiscal capacity, and countries which equalize only expenditure needs differences across subnational governments (Table 1). Typically, to achieve an adequate level of equalization, it will be necessary to equalize both fiscal capacity and expenditure needs (Box 1). Box 1. Equalizing Expenditure Needs, Fiscal Capacity, or the Fiscal Gap
Unconditional transfer formulas in many countries emphasize local expenditure needs in determining the horizontal allocation of resources. For instance, in addition to a fixed amount or "equal share" for each state, Nigeria's formula considers variations in population, land area, and socio-economic characteristics as key determinants in its allocation formula (Alm and Boex, 2002). Likewise, South Africa's provincial "equitable shares" formula is based on six expenditure needs components (Reschovsky, 2003). A focus on the equalization of subnational expenditure needs makes sense if subnational governments have relatively little expenditure autonomy (as is the case in South Africa), or if there are only limited variations in subnational fiscal capacity. Needs-based transfer formulas tend to be more common in LDTCs.
Boex and Martinez- Vazquez
294
Transfer schemes in a smaller number of countries focus exclusively on fiscal capacity equalization. For instance, Canada's equalization transfer scheme aims to equalize the revenue potential of its provinces, by providing a transfer to those provinces whose ability to collect own-source revenues falls below a certain threshold. An exclusive focus on the equalization of subnational fiscal capacity makes sense if there are substantive variations in subnational fiscal capacity, and if variations in expenditure needs are either limited or, alternatively, addressed by targeted transfer schemes. See Bird and Vaillancourt in this volume for a discussion of whether and how Canada addresses variations in expenditure needs. Unconditional transfer schemes in many countries pursue both equalization of expenditure needs and subnational fiscal capacity at the same time, by seeking to (partially or wholly) "fill the gap" between subnational expenditure needs and the resources available at the subnational level (including own fiscal capacity plus other transfers). This approach is pursued in Indonesia, Germany, Russia and others countries (Table I), and is common in many federal countries and larger unitary countries.
Table I. Equalization Goals, Allocation Factors and International Practice Goals Enable similar levels of service affordability
Factors Expenditure needs indicators (separately or in a combined indicator), or national expenditure standards
Enable similar levels of fiscal resource availability Enable similar levels of service at similar levels of taxation
Fiscal capacity indicators or "representative revenue system" Fiscal gap = Expenditure needs - Fiscal capacity, OR some other combination of need and capacity
Distribution on an equal per capita basis
Population
Country examples India, Italy, Nigeria's Federation Account, South Africa's Equitable Shares, Spain, Uganda's Unconditional Grant Canada's Equalization Grant Australia, China Germany, Indonesia, Japan, Korea, Latvia, Netherlands' Municipal Fund, Russia, Uganda's Equalization Grant, United Kingdom Some transfers in Canada, Ecuador, Estonia, Germany, Hungary, and England
Designing Intergovernmental Equalization Transfers
3.
295
ALTERNATIVE APPROACHES TO MEASURING FISCAL NEEDS, FISCAL CAPACITY AND FISCAL EFFORT UNDER PERFECT AND IMPERFECT DATA
Different allocation methodologies may be appropriate depending on the exact objective and nature of the transfer scheme. The final allocation of hnds among deserving jurisdictions may rely on a proportional distribution formula (by which subnational governments receive transfers in proportion to certain allocation factors), equalization below a certain threshold (by which only jurisdictions that fall below a certain threshold receive equalization transfers), or a variety of other computational approaches. But regardless of the formula mechanism used, the outcomes and ultimate success of the horizontal allocation approach depend critically on how expenditure needs and fiscal capacities are measured. Desirable properties such as incentive compatibility and lack of manipulability impose constraints on the measures of expenditure needs and fiscal capacity that can actually be used. As we remarked above, the task of building an equalization grant formula in less developed and transitional countries is complicated by the lack of data to implement the more proven and accepted methodologies used, for example, in many OECD countries. In this section we review different methodologies that are used in the international practice to measure expenditure need and fiscal capacity with very different levels of data availability, while still managing to arrive at more or less workable equalization formulas. Although the different approaches can be discussed in abstract and general terms, whenever possible we discuss them in the context of actual country experiences and practices.
3.1
Alternative Approaches to Measuring Expenditure Needs
In the context of equalization grant design, expenditure needs generally refer to the local government outlays that would be necessary to provide a particular standard of service. In practice, there are many options for measuring differences in expenditure needs across subnational governments, often dictated by the availability of data. The term "expenditure needs" can be misleading unless related to specific standards of service or to some overall envelope of e ~ ~ e n d i t u r eWithout s.~ those bounds, perceptions of what may be a need can easily escalate to completely unaffordable expenditure levels. Naturally the potential level of
Boex and Martinez-Vazquez
296
outlays should be related to the areas of responsibilities of local governments and is expected to vary across jurisdictions according to the population consuming the services (particular characteristics of that population such as age or income) and the costs of delivering the standard service. These costs are generally associated with the relative cost of living in the jurisdiction and they may or may not be independently affected by geography, climate and other specific conditions in the jurisdiction. There are several problems with this conventional definition of expenditure needs. It is often the case that standards of service have not explicitly been set, or if they have, they may involve such physical detail or may imply such high, unaffordable levels of expenditure that those standards are rendered useless. In the case when referring to some unknown or unusable standards, the conventional definition of expenditure needs becomes circular and empty. In many LDTCs, this vacuum tends to be filled in practice by ad hoc decisions, incrementalism, or by discretionary adjustments and updating of historical expenditure levels. There are two hdamental dimensions to defining and quantifying expenditure needs. The first dimension is the definition of the aggregate level of affordable local expenditures or the local fiscal envelope, which imposes a budget constraint on the otherwise ambiguous concept of needs.6 The second dimension is the determination of the approach used to apportion (either from the top down, or by adding from the bottom up) those need measurements across subnational governments for all areas of expenditure responsibilities assigned to them. Not surprisingly, there are a variety of approaches that have been used in different countries to quantify local expenditure needs, with differences arising from each country's specific budgetary traditions and from the different degrees of data and information available. Although the various approaches used to quantify local expenditure needs involve, to different extents, empirical or practical elements and a priori conceptual thinking, these approaches range from extremely crude to quite sophisticated, including: 1. Lagged expenditure values; Equality or equal per capita expenditure norm; Weighted indexes of expenditure needs; Per-client (top-down) financial expenditure norms; Traditional (bottom-up) physical expenditure norms; and Regression-based Representative Expenditure Systems (RES).
2. 3. 4. 5. 6.
3.1.1
Lagged expenditure values
In the absence of data and a conceptual framework for quantifying expenditure needs, the use of past values of expenditures is the simplest and
Designing Intergovernmental Equalization Transfers
297
most parsimonious approach to quantifying local expenditure needs. The extent to which a local government's lagged expenditure levels accurately reflect the jurisdiction's expenditure needs depends to a large degree on the expenditure and revenue autonomy that local governments enjoy. If local governments have a large degree of discretion over their budgetary decisions, one could argue that local expenditure patterns reveal variations in local fiscal needs or variations in the demand for local government services. These are the basic arguments used for the United States in several now classic papers (Borcherding and Deacon, 1972; Bergstrom and Goodman, 1973). However, if local budget decisions prior to the introduction of equalization grants were made in an arbitrary or discretionary fashion under a centralized system of local finance, there is the risk that variations in local expenditures may be more a function of central political whim than true variations in local needs. Thus, at the very least, caution needs to be employed before one relies on historical expenditure levels as an indication of local expenditure needs. Needs measures using lagged expenditure values comes in different forms and shapes. In some cases, the computations are just based on the prior year's budget data for local governments. In other cases, the overall envelope of expenditure is quantified in reference to a base year, for example the year prior to the most recent budget year or the year prior to significant decentralization reform in which subnational governments were assigned new competences or expenditure responsibilities. The measure of the overall (aggregate) local expenditure need may be defined, for example, as the total expenditures undertaken by the central government for those same expenditure responsibilities in the year before they became decentralized. Historical patterns may again be used to apportion the relative expenditure needs among subnational governments. Several countries have used and continue to use this approach. For example, in Spain, progressive decentralization of expenditure responsibilities to its regions over the past decades has always taken as a measure of expenditure needs the actual levels of expenditure that were realized by the central government in a particular area prior to the decentralization of the function. The continued use of these values for the purpose of computation of equalization grants and other transfers has created problems of fairness because of changes in the demographics in the intervening years. In Russia, from 1994 through 1996, expenditure needs were based on baseline data from 1993 and, in 1997, the Ministry of Finance started using expenditure data for 1991 (the year prior to the break from the Soviet Union). The switch to 1991 was based on the belief that that year's expenditures were more representative of actual expenditure needs. The base year data were adjusted for changes in legislation for the years in
298
Boex and Martinez- Vazquez
between in order to approximate the levels of expenditure needs for all regions for the current fiscal year. The use of lagged values of actual expenditures to quantify expenditure needs has the advantages of simplicity, being affordable and potentially avoiding the perverse incentive effects that would be associated with using the current actual level of expenditures. But the approach also has important disadvantages. Besides becoming obsolete over time due to demographic and other changes, as in the case of Spain, any initial misalignments and inequities in the provision of services by the central authorities would be prolonged rather than remedied by the equalization grants.7 Some potential remedies for the obsolescence problem associated with lagged expenditure measures include the indexation of the overall level of expenditures to inflation or a variety of other indexes. The problem with aggregate indexation is that the regional composition of expenditure needs is kept as it was in the past. Another remedy is to select a number of years in which measured expenditure needs remain fixed. After that period a new updated base year for lagged expenditures is chosen. The adjustment can be made ad hoc irregularly and can be done according to a fixed announced rule on a periodic basis, which is the practice in Vietnam where the preannounced periods of three years are called "stability periods."8 We must note, of course, that adjustment on a periodic basis introduces (through a process of learning) perverse effects on subnational government behavior: the more local governments spend, the higher would be the measure of expenditure needs to be used in future equalization formula applications and therefore the higher the equalization transfers they may expect (all other things equal). Overall, the lagged approach is not to be recommended and, in the longer run, may be an unsustainable strategy for measuring expenditure needs. However, it can play a useful role in the shorter term, especially when the equalization formula needs to be devised and introduced right away. A period of two or three years using lagged or base-year expenditures can give enough time for the government to develop and adopt a superior approach.
3.1.2
Equality or per capita equality
In the absence of any data on local government characteristics whatsoever, one possibility is to assume that each local government jurisdiction has an identical level of need. In this case the formula applied is that of equality (also known as equal shares), by which each local government authority receives exactly the same amount of resources, regardless of its population or any other characteristics. A "formula" based exclusively on equal shares is applied in the distribution of transfers to the village-level in Nepal, where every Village Development Committee (VDC)
Designing Intergovernmental Equalization Transfers
299
receives exactly 5 million rupees.9 The logic behind the "equality" approach is that - if all else is unknown - the best approximation of local expenditure need is that each local government jurisdiction has an identical level of need; each government unit thus receives a lump sum amount identical to the transfer pool divided by the total number of jurisdictions. Although this approach is remarkably simple and transparent, it has very serious shortcomings. The most significant are that it can lead to the possibility of causing significant variations in per capita resource availability (especially if there are large variations in population size among jurisdictions), it fails to encourage formation of jurisdictions of minimum efficient scale, and it can provide an incentive for further fragmentation. Thus, an equal shares approach is generally not recommendable even in the short run." Whereas there might be certain fixed costs associated with the operation of local governments, the cost of delivering a standard basket of services to a larger population is generally more costly. Even in the absence of consistent or accurate population data for all jurisdictions, the correlation between population and expenditure needs may be recognized by classifying local jurisdictions by sizes into several categories (e.g., Class A, B, C, etc.), and by providing larger shares (greater lump sum grants) to jurisdictions that are classified as more populous. For instance, Bangladesh uses this basic approach to provide capital transfers to municipalities and recurrent grants for operational expenses to Union Councils (World Bank, 1997). Rather than using absolute equality, the horizontal allocation formula or approach can also pursue equality in per capita terms. This requires data on population by jurisdiction, which in most countries are available on a regular basis from a decennial population census. If data on the geographical distribution of population are available, the criterion of an equal allocation per capita is superior to that of absolute equality, since jurisdictions are not the object of finance per se; instead, individual residentslvoters are the ultimate clients of local government services. As a result, many countries use population as an important factor in amving at expenditure needs and in some cases it is the sole factor in the allocation formula (i.e., local governments receive a transfer exclusively in proportion to their population).11 Nevertheless, the presumption that local expenditure requirements are the same for all individuals regardless of where they reside or regardless of their personal characteristics may not be justified. For instance, the cost of delivering identical services may vary across areas while the demand for local public services may vary with demographic profiles or other characteristics of the jurisdictions.
Boex and Martinez- Vazquez
3.1.3
Weighted indexes of relative need
A third, and probably the most commonly used approach to quantifying local expenditure needs - particularly in LDTCs - is to estimate some type of index of relative expenditure need. While some equalization schemes explicitly define such an index of expenditure needs, other transfer schemes implicitly achieve the same when a weighted-factor mechanism is used for the purpose of allocating equalization grants (Table 2).12
Table 2. Computing a Basic Index of Expenditure needs Step 1.
Determination of the aggregate level of subnational expenditure needs (SEN) The computation of a basic index of expenditure needs may be based on historical data, or based on data spec$ed in the budgetforecast.
Step 2.
Selection of expenditure needs factors Expenditure needsfactors are selected spanning the main drivers behind dzferences in expenditure needs and may includepopulation, land area, and/or other local characteristics such as poverty, cost of living and so on.
Step 3.
Computation of each locality's relative need share for each factor The share ofpopulation ( X I )for each local government in the entire population is ( X l i /ZXli), the share of land area ( X 2 )for each local government is (X2i /CX2i), and so on.
Step 4.
Determination of the relative importance or weights of each needs factor The weight for the population factor is a1 and the weightfor land area-factoris a2 and so on.
Step 5.
Calculation of the Index of Expenditure Needs for locality i NeedIndexi= a1 .(XlilZXl)+a2.(X2i/ZX2)+ ... + a n . (Xni 1Z Xn ) where the weights add to one (a1 + a2 + ... + an = 1)
Step 6.
Based on the Index, some mechanisms quantify the relative expenditure need for locality i as: Relative Expenditure Need i = Need Index i SEN (where SEN is the aggregate subnational expenditure needs envelope) Other mechanisms directly determine the size of the transfer based on weighted needs as: Transfer i = Need Index i . Transfer Pool
Designing Intergovernmental Equalization Transfers
30 1
These indexes attempt to capture, from the simplest to more complex ways, the factors that determine cost differences in delivering a standard package of local government services. In many countries, the factors included in such needs indexes rely on demographic variables (for example, reflecting the special needs of the young and the elderly), other factors such as the level of poverty and unemployment, and differences in the local price level or cost of living.I3 Other formulas - particularly many capital development grant schemes in LDTCs - predominantly or exclusively rely on a number of poverty-related needs measures such as infant mortality rates, illiteracy or other proxies for poverty. The list of variables or factors entering the index can be short or long, depending on data availability and the specific design of the equalization formula. An example of the simplest multi-factor formula is provided by the formula used (until recently) for the equalization Grant in Uganda which relied on two factors: population, with a weight of 0.85, and land, with a relative weight of 0.15. A more complex quantification of subnational expenditure needs is achieved by the Federation Account formula in Nigeria, which allocates resources to states in proportion to ten specific needs factors grouped in five needs categories (in addition to a measure of fiscal effort). On the other extreme, the allocation formula used for the general-purpose Municipal Fund in the Netherlands includes 51 individual local needs measures ranging from demographic characteristics to the age of the building stock and residential density. One problem with the index-based approach to quantifying expenditure needs is that the inclusion (and exclusion) as well as the weighting of particular factors is often subject to political pressure. Local government officials and parliamentarians are prone to fight for the inclusion or heavyweighting of factors that may be particularly beneficial to a group of specific regions or just one region. Alternatively, policy makers or advisors may simply be misguided in their desire to include certain allocation factors in order to measure expenditure needs. Specifically, capital development funds in some of the least developed countries (for instance the District Development Fund in Malawi and District Development Grant in Nepal) are allocated based on a index of poverty-related indicators (illiteracy, infant mortality and so on) that may in fact have little or no relation to local governments' true expenditure needs. Another tendency is for policy makers to be inclined to "over-design" the measure of expenditure needs by including a large number of factors. However, a much longer list of factors is not necessarily more effective in capturing variations in local expenditure needs than a more compact list. Yet, including more components does make the needs measure more difficult to maintain and less transparent. Conceptually, determining particular weights for each one of the factors in the needs index is a simple task, as weights ought to represent the relative
302
Boex and Martinez- Vazquez
contribution o f each factor to the overall measure of need in monetary terms. However, in reality, these decisions tends to be made in an opaque and quite arbitrary manner. If data were available and local expenditure patterns could be relied o n to reveal variations in local fiscal needs or variations in the demand for local government services, simple statistical techniques could b e used to identify proper weights (not unlike what is done with "representative expenditure system" methodologies). In most LDTCs those conditions do not hold. Even though other approaches are available to approximate weights for each o f the factors in the index (see Box 2), in practice these weights are often determined in an ad hoc fashion and again often responding to political pressure.
Box 2. Determining Weights for an Index of Relative Need Several approaches can be used to arrive at a particular set of weight factors in a scientific or objective manner. One approach is to utilize micro cost data to establish how the costs of delivering standard services changes across jurisdictions. Alternatively, one can rely on historical budget allocations to arrive at relative weights for a local needs index. For instance, in Latvia the relative index of need was constructed consistent with the process described in Table 2 as the weighted average of four criteria: (1) Population of local government; (2) Number of children ages 0-6; (3) Number of children ages 7-18; and (4) Population above retirement age (Martinez-Vazquez, 2000). The total resource envelope for local governments (to which the relative index of need is applied to get at the expenditure need of each local government) was negotiated every year between the central governments and the Union of Local Governments. Originally, the relative weights were derived using 1994 aggregate budget data on local expenditures (the year before the introduction of the decentralized grant scheme). For instance, the weight on the factor for children under 6 years was computed as the ratio of aggregate local expenditure on kindergarten programs in 1994 to total local expenditures across all jurisdictions in that year. Similarly, the relative weight on the factor for youths fiom 7 to 18 was defined as the ratio of school expenditures to total expenditures; while for the elderly the relative weight was determined as the ratio of expenditures on social welfare to total expenditures. For population, the relative weight was defined as the residual as the ratio of general services (all but education and welfare) to total expenditures. Two additional needs factors were added in 1998 (the number of children in orphanages and the number of elderly in retirement homes) and all relative weights were updated. Although Latvia computes an index of relative needs, we should note that its approach in many ways is equivalent to defining per-client financial norms or a traditional representative expenditure system (RES) approach.
Designing Intergovernmental Equalization Transfers
303
Another approach is to use statistical analysis to try to find out the implicit weights that the historical data show. For this purpose, one can regress actual (historical) expenditure data as a dependent &able on a set of explanatory variables containing all the factors included in the index of relative need. (Depending on the explanatory variables used, this approach could be considered equivalent to a regression-based RES approach, as discussed further below). However, this regression approach can be problematic for several reasons. First, all the necessary data may not be available. But even if they are, it is quite unlikely that the regression coefficients will produce anything like factor weights adding to one as needed in the construction of the index. Even when the regression coefficients are restricted to add to one, we may be far from being able to interpret the coefficients as weight factors in the index of relative need. An example of these difficulties is provided by a proposal for the Republic of Georgia in which expenditure needs are approximated by an index with two factors: population (Pi) and mountainous population (Mi); these are the only two possible factors for which there is consistent available information for all local governments (Martinez-Vazquez, 2004b). By taking actual expenditures for each local government (Exp i), in the most recent year (2003) as a measure of the expenditure needs of that local government, then we can try to find the implicit weight for P and M by running a regression on the two factors, (Pi / Z P) * SEN for population, and (Mi / C M)*SEN for mountainous population, where SEN is the measure of local government aggregate expenditure need for 2003. This latter is quantified as total local government expenditures in that year. Forcing the intercept of the regression to be zero, which we would need to estimate an "expenditure need" equation, we find:
where we note that the coefficient on the mountainous population factor takes a negative value. When we further force the regression coefficients to sum to one, we obtain similar results:
There are many reasons why the regression coefficients take those values, but the simplest one is that during 2003 higher public expenditures at the local level took place in the capital Tbilisi and other cities, which have no mountainous population. Thus, in 2003, other factors such as existing infrastructure or political considerations may have been more determinant factors of actual expenditures at the local level, demonstrating the limited usefulness of the regression-based approach if historical allocations are not reflective of objective expenditure needs. One difference between the index approach used in Latvia and the one discussed for the Republic of Georgia is that the latter attempt to capture
304
Boex and Martinez- Vazquez
variations in the costs of producing local services rather than number of local clients. As a result, the assignment of weights for Georgia's need index requires a somewhat different type of reasoning. Consider again the case of Georgia with two expenditure need factors: the number of total population (P) and mountainous population (M). In this example, assigning weight a to the first factor and 1- a to the second factor, can be interpreted as share a of aggregate expenditures being related to standard costs of providing public services in an average locality while the remaining share 1-a comes from additional costs incurred in mountainous areas. The meaning of a becomes clearer if we consider the expenditure need formula: Needi= (Pi/P)-a.SEN+(Mi/M).(l-a)-SEN Because mountainous population is included in total population, it would be counted twice when apportioning the aggregate level of subnational expenditure needs (SEN) among localities. First it is accounted for with weight a as part of the total population, and then it is accounted for with weight 1- a specifically as mountainous population. Thus, mountainous population is accounted for with a total weight of a+(l- a)=l, while non-mountainous population is given a weigh of only a. For all these reasons, the relative index approach to expenditure needs requires careful assessment and thorough discussions with all stakeholders to ensure that the main causes for substantial differences in the demand and costs of local public service delivery across jurisdictions are captured in the index. Ideally, the index formula should be based on a relatively limited number of factors as the inclusion of too many variables reduces transparency. It is also more costly and difficult to update a larger number of variables on a regular basis and larger designs with many factors can introduce more opportunities for political manipulation. In short, a balance needs to be struck between simplicity and transparency, and the need to find factors that equitably reflect the true fiscal need of local governments. Variables used as factors should accurately reflect needs, come from an independent source, and be free of manipulation by either central government or subnational governments. The use of "productive inputs" such as the number of local government employees, school buildings and hospital beds as allocation factors cause inefficiencies since they provide incentives that could distort expenditure decisions. A final note with respect to the computation of expenditure need indexes regards the frequency of updating the measure. With better data, methodological fine-tuning and political pressure, the index formula can be changed frequently, in some cases yearly. Yet, the resulting lack of
Designing Intergovernmental Equalization Transfers
305
institutional stability reduces the revenue predictability, making budget planning so much harder for local governments. Recent examples of LDTCs that have adopted a relative index approach include Malawi, Indonesia and the Russian Federation (Box 3). These countries provide a sampling of the different emphasis given to the factors to be included in the formula as well as the different, more or less ad hoc, approaches used to determine the weights. Among developed countries, Australia has one of the most sophisticated applications of what could be considered an index approach. Essentially the approach in Australia is to start with an equal per capita transfer which is adjusted for differences in delivering costs (for a standard set of services), as well as for differences in revenue capacity across subnational governments. The adjustments or "relativities" include 41 expenditure categories.
Box 3. Recent Examples of Index of Relative Need Approach to the Estimation of Expenditure Needs: Malawi, Indonesia and the Russian Federation Malawi District governments in Malawi receive grants from the District Development Fund (DDF), which is a decentralized development financing facility established by the Government of Malawi with support from a number of donor o&pnizations, including the UNDP and the UNCDF (Boex, Kampanje and Mwadiwa, 2001). The DDF is apportioned among rural districts on the basis of a horizontal allocation formula that is perceived to reflect local expenditure needs, thus implicitly arriving at a weighted-needs index: First, 30 percent of the DDF resources are allocated to district councils based on the "equal shares" principle. This reflects the assumption that all districts regardless of population size of other characteristics - are assumed to have a certain fixed need for capital infrastructure. Second, 70 percent of the DDF is allocated based on four factors, all equally weighted, notably:
Population, Land area, Illiteracy, and Infant mortality. Indonesia As part of Indonesia's equalization mechanism (DAU), a straightforward index of expenditure needs is constructed - again consistent with the approach outlines in Table 2 (Brodjonegoro and Martinez-Vazquez, 2004). Based on four expenditure needs factors, initially (in 2001) the factor weights adopted were 0.25 for all factors. By 2002, the measure had evolved to:
306
Boex and Martinez- Vazquez Expenditure Needs =ALE (0.4 PI + 0.1 AI + 0.1 RPI + 0.4 CI),
where: ALE: Average Local Expenditures = (Total Local Expenditures + Deconcentrated Funds) / Number of Local Governments PI: Population Index = Local PopulatiodAverage Local Population AI: Area Index = Local AreaJAverage Local Area RPI: Poverty Index = Number of Local Poor PeopleIAverage Local Poor People CI: Construction Price Index = Local Construction Price Index400 Indonesia's measure of expenditure needs is slightly different from similar needs indexes; rather than being based on the relative share of each factor (e.g., the relative share of the national population that resides in a local government), Indonesia's index is based on the ratio between each factor and its average value (e.g., local populatiodaverage local population). We should note, however, that computationally these two approaches are filly equivalent.
C h a n ~ i nthe ~ Composition of the Index of Budget Expenditures (IBE) in the Russian Federation In Russia's equalization fund (the Fund for the Financial Support for the Regions, or FFSR), the index of budget expenditures (or expenditure needs) is used to "normalize" or adjust per capita revenues. The formula assigns equalizing transfers to regions for which the normalized per capita revenue falls below some threshold. Up to 2004, the computation of the Index of Budget Expenditures used the formula: IBEi = (Ri / Ni) / (R / N), where Ri stands for the sum of expenditures needs assessed separately for 15 expenditure categories; R stands for the sum of Ris across all regions; and Ni and N stand for the regional and national population respectively. For all expenditure categories except housing and utilities, the expenditure need for a particular service was derived from a per client norm by applying two adjustment coefficients to account for differences in service need and production costs. The cost coefficients were the same for all the 14 categories (excluding housing and utilities) and were derived by taking into account wage costs, price level, utility costs, and accessibility of the region's localities by transport. The Index of Budget Expenditures will be calculated differently starting in 2005, as the weighted sum of three subindexes for relative differences in wages kiw, housing and utilities costs kih, and the price level kip respectively: IBEi=uw. kiw+ uh . kih+ (1-uw-uh) . kip Each of the three subindexes will be calculated through an additive andor multiplicative aggregation of 3-4 different indicators and normalized by the population-weighted average so that the population-weighted sum of each subindex across all regions equals to one. The subindex for relative differences
Designing Intergovernmental Equalization Transfers
307
in wages kiw takes into account three different forms of wage allowances mandated by the federal government and the proportion of population residing in small settlements. The subindex for relative differences in housing and utilities costs kih takes into account the federal norm for housing and utilities costs set for the region (both in nominal value and relative to personal income) and accessibility of the region's localities by transport. The subindex for relative differences in price level kip takes into account the consumption bundle price, proportion of population under and over working age, the proportion of population residing in small settlements, and accessibility of the region's localities by transport.
3.1.4
Per Capita (Per Client) Expenditure Norms
Some countries avoid the challenges of finding and statistically manipulating data to arrive at some measure of expenditure needs by adopting a system where the local government sectoral allocation is devised in a "top-down" manner. Under this approach, the national government through a priority setting process that cuts across sectors (generally through Cabinet-level discussions) - determines sectoral resource envelopes including resource envelopes for local government programs and activities. The local government programmatic allocations are then divided by the aggregate number of clients served, generating an expenditure norm that is used as a means of computing the per-client expenditure needs. This can be used directly to allocate intergovernmental grants, or can be entered in an equalizing grant formula. A straightforward example of a per-client expenditure norm is the Quality Basic Education (QBE) transfer in the State of Georgia, USA, where a per-student financial norm is determined from year to year based on resource availability in the annual budget process. For instance, in 2005, each school district in the state was assured instructional resources equivalent to $1600 per enrolled student.14 In this example, the expenditure norm for primary education is derived by deciding what share of total hnding available to the state should go to local governments for this activity, after considering all funds available and all other competing demands on the funds, and dividing the available resources by the total number of primary students in the state. Furthermore, when costs vary substantially between local jurisdictions (as is the case, for instance, in a country like Indonesia), the actual amount of the norm per capita could potentially be adjusted upwards or downwards by multiplying the norm by a series of indexes to reflect regional differences in the cost of provision (Table 3).
Boex and Martinez-Vazquez
308
Table 3. A Simple Top-Down Norm-Based Determination of Expenditure Needs -
-
- - - -
- - -
--
--
-
-
Step 1.
First, the central government sets the envelope or overall amount of expenditures for local governments that will be used for the calculation of needs in the following year.
Step 2.
Next, the total amount of local expenditures is divided into separate aggregate expenditure functions (education, health, and so on) to determine the notional amounts of that would be spent on the aggregate expenditure functions.
Step 3.
The next step consists of arriving at a basic expenditure norm for each main functional expenditure category by dividing the notional amounts by the objective criterion or number of clients in the entire country (population, or number of students and so on).
Step 4.
Given the basic norm for each expenditure category, the expenditure need for each local government is equal to the product of three items: a) the basic expenditure norm (per capita, or per student, etc); b) the population or the number of students, etc. in the locality; and c) a vector of coefficients that work to adjust the expenditure need up or down for special needs andfor differential costs of service provision.
Client-based expenditure norms or even the sectoral or programmatic envelopes for local government sectors can be prescriptive, forcing local government to spend according to these envelopes and norms; they can be indicative, merely becoming guidelines that are helpful to the local governments in setting their own budget allocations; or they may just simply be notional and used only for computational purposes There are several advantages to a client-based expenditure norms approach. One is the relative simplicity of the process, both in terms of decisions and data requirements. This approach also avoids the perverse incentives in expenditure decisions that existed in the past in transitional countries when they were using physical norms (discussed below). Another advantage is that despite the use of norms, there is a guarantee that the notional or prescriptive measure of expenditure needs will always be affordable. Defining expenditure needs in this "top-down" manner is a realistic and effective mechanism for using public resources in accordance with national priorities while recognizing the government's overall resource constraints. There are some potential shortcomings as well with this approach. The sectoral or programmatic trade-offs made at cabinet level, involving national priorities and other considerations, may not necessarily have a clear
Designing Intergovernmental Equalization Transfers
309
connection to real service needs (Martinez-Vazquez, Boex and Ferrazi, 2004).15 There is also the risk that the budget formulation process could potentially result in a non-transparent process where incremental resources are provided to sectors and programs that are subjectively felt to be "needed" or desired at the center without clear and direct links to sectoral outcomes. In developing country contexts, political affiliations of cabinet members and legislative allies can play a particularly dominant role in such priority setting. A recent example of a country adopting a top-down per capita expenditure norm approach is Ukraine (Martinez-Vazquez and Zeikate, 2002). In fact the steps described in Table 3 fit well the Ukrainian approach. The United Kingdom has used a similar approach but with some peculiarities (Box 4). To some extent the "per client" approach is also used in the Czech Republic for the transfers that finance the functions delegated by the central government to local governments (Oliveira and Martinez-Vazquez, 2001). It should be noted that in the case of the Czech Republic the decisions on norms are made at the top but this is not a top-down comprehensive approach as is the case in Ukraine. The approach in Vietnam is similar: expenditure needs of the local government are derived of the basis of the prevailing system of expenditure norms, which are supposed to cover all current expenditures (salaries, operation and maintenance, and so on) and (some minimum amount of) capital expenditures. The norms are adjusted for different regions depending on geography and remoteness, and are primarily specified as monetary amounts per capita. 3.1.5
Traditional (bottom up or physical) expenditure norms
If one were to ask a local government accountant (not familiar with the finer points of fiscal decentralization) to quantify the expenditure needs of his or her local governments, it is almost certain that he or she would think of a bottom up approach where the local governments financial "needs" would be equated to costs of providing the current level of local services, or the summing up of expenses required to offer a standardized basket of subnational government services. The essence of the traditional approach to expenditure norms or standards is to compute a detailed costing of the inputs needed to deliver a standard (national or minimum) level of public services. The correct application of this approach requires that the "standard levels" have been defined and that cost or price information exists for all those items. Although in some way quite intuitive, this approach is quite demanding because of the type and detail of information required. The approach also requires very explicit procedures for how to cost all aspects of the expenditure responsibilities of subnational governments.
310
Boex and Martinez- Vazquez
Box 4. Expenditure Needs in the United Kingdom The British government starts by assessing how much expenditure local government as a whole will have to finance through grants and Council Tax. This amount is known as Total Standard Spending. The Government helps meet about three-fourths of the Total Standard Spending by distributing grants. The funding from these grants is known as Aggregate External Finance. The difference between Total Standard Spending and Aggregate External Finance is the approximate amount local authorities would need to raise through Council Tax if they spent at the level of Total Standard Spending. To calculate the Standard Spending Assessments (SSA) for each jurisdiction, the Government takes into account population, social structure and other demographic characteristics. The SSA is the government's assessment of the appropriate amount of expenditure that will allow the local authorities to provide a standard level of services, on the bases of demographic, social and geographic characteristics. There are seven basic fields taken into account to calculate the SSA: education, social services, highway maintenance, age of the population, number of children and special characteristics of them, fire and police and other services. To arrive at the assessment in each field, more detailed information is taken into account. For example, for education, factors taken into account include the number of school pupils, school pupils with special needs, cost differentials across districts and so on.
A classical example of defining expenditures needs through a myriad of ineffectual physical norms is provided by the budgeting practices of the former Soviet Union, where the Planning Bureau, the Ministry of Finance, line ministries and other federal agencies employed literally thousands of budgetary norms to be applied at all levels of government and by all implementation agencies, such as what type and quantity of food will be provided to different types of patients in public clinics and hospitals (Martinez-Vazquez, 1994). Because these norms were too complex and there was no mechanism to assure their affordability, they were not actually applied. Nevertheless, and not surprisingly, many local governments and implementing agencies insisted on using them as the basis for their budget requests. A large number of the countries created fiom the dissolution of the Soviet Union, including Russia, and some former planned economies in Central and Eastern Europe, experienced significant difficulties in shedding the old practices over their years of transition during the 1990s. The real inability to make those traditional bottom-up expenditure norms operational and affordable led to the slow but systematic abandonment of the physical standards approach of old Soviet times in these ~ountries.'~
Designing Intergovernmental Equalization Transfers
311
Despite their shortcomings, traditional expenditure norms are still relied on in defining expenditure needs in a variety of LDTCs. For instance, it is common (especially in unitary countries) that line ministries and other central agencies decree a series of national standards and minimum norms for physical facilities and service delivery, which supposedly need to be satisfied by subnational governments (Alm and Martinez-Vazquez, 2002). It is in the presence and awareness of these standards and norms that the concept of "expenditure needs" almost necessarily evokes the summation and quantification in monetary terms of those norms. Whether this is conceptually the right approach has been much debated, most recently for example in Tanzania and Indonesia (Boex and Martinez-Vazquez, 2004; Martinez-Vazquez, Boex and Ferrazi, 2004). What is certain is that such an approach may be too data intensive and too costly for most LDTC.'~ Because the minimum standards of service delivery are generally specified by sectoral ministries or agencies with their narrow agendas and incentives to push standards up, another significant disadvantage of this approach is that there is no guarantee that the expenditure needs so derived are affordable within the overall budget resource envelope. The insufficiency of h d s invariably requires a downward adjustment of the estimated local budget needs, thus not complying with the specified standards. Lack of achievement of the unaffordable standards frequently becomes a source of frustration for subnational government officials and lead to voters' protest. This pretty much was the situation in the former Soviet Union and is still a common occurrence in some transitional economies, as well as in many developing economies. Another risk with the use of many specific expenditure norms for services is that they may be specified in physical terms, such as the number of hospital beds in the case of health services. As we have commented before, this type of physical norm can easily lead to perverse incentives in expenditure behavior. Few economies have the data or the administrative capacity to deal adequately with highly detailed expenditure norms in this bottom up approach. However, this approach is prevalent in Denmark, Japan, the Netherlands and Sweden, for example. In all of these countries the system identifies a large number of services provided by local government to account for the expenditure need, with units of measurement, unit costs, and modification coefficients for each of them.'* In the same vein as traditional expenditure norms, a common application of the bottom-up approach in determining local expenditure needs involves the direct summation of actual or implied expenditures. For instance, as is common practice in many African countries, both Uganda and Tanzania (prior to 2004) determine the wage needs of local authorities based on the actual wage bill for local teachers and other local government staff (Boex et al., 2003; Uganda LGFC, 2003). China, an important transitional country,
312
Boex and Martinez- Vazquez
has been using this type of simple bottom-up approach in the quantification of expenditure needs. For example, the determination of standard expenditures in China's general equalization grant (also known as the transitory grant - because its formula was supposed to be transitory), the measure of expenditure needs fundamentally involves classifying local expenditures into 12 categories (education, health, government administration, etc.) and then using simple methodologies, such as adding up personnel and office outlays, to arrive at expenditures needs.I9 As already noted, the main problem with adding up outlays is that it sends perverse incentives for expenditure decisions at the local level. An important problem in China's subnational sector is overstaffing. Russia's recent experience provides an example of how to progressively refine the measurement of expenditure needs based on individual per capita norms to incorporate the elements that can account for cost differences across subnational jurisdictions. A problem with adding up expenditure standards is budget affordability or feasibility. However, several approaches can be used to tackle this problem. The current approach used in South Africa for the S grant (for municipal basic services) provides a good example of how this can be done. The S grant is set to equal the cost of providing basic municipal services to the members of all the poor households in each municipality. Thus for municipality i, the grant is Si = (a )(b)L Hi; where a is a phase-in parameter (with 0
3.1.6
The Representative expenditure system approach
The Representative Expenditure System (RES) approach is a sophisticated and rather complex expenditure needs measure that was developed in the United Sates by the now defunct U.S. Advisory Commission on Intergovernmental Fiscal Relations (Rafuse, 1990a,b and Tannenwald 1998 and 2002). The computation of the index of fiscal need for each subnational government involves several steps. First, the methodology identifies what the main drivers are for expenditures in each of the major expenditure functions assigned to subnational governments (for example, for general administration it may be population, for elementary education the number of students, or for roads, miles of highway and traffic intensity or vehicle miles). The second step identifies determinants of the costs of providing a
Designing Intergovernmental Equalization Transfers
313
given level of service other than the prices of the inputs used by the subnational governments (these are known as the "workload" factors). If there are several determinants, a weighted average of the determinants is constructed as a composite workload factor. In the third step, the total subnational expenditures in each expenditure function is multiplied by each subnational government workload factor to arrive at how much each subnational government would have spent in that expenditure category if it had provided the standard level of service. In the fourth step the estimated per capita "standard" spending for each fbnction is adjusted for the relative costs of the inputs for that expenditure finction. In the fifth step, the adjusted per capita standard spending levels for each function are added to arrive at the subnational government per capita spending on a "standard expenditure package". Finally, these total figures are indexed to the actual total subnational per capita spending to arrive at the index of need for each subnational government. A modification to the traditional RES is a regression-based Representative Expenditure System (Shah, 1994). The regression-based approach has the advantage of being more flexible than the traditional RES, as there is no need to link specific spending categories to specific client-bases. Instead, the regression estimates will verify the existence and strength of linkages between local expenditures and multiple needs bases at the same time. Yet, while the regression-based approach arguably yields the most precise measure of local expenditure needs, the measure is often less wellunderstood and considered less transparent due to the technical and the indirect nature of the link between specific expenditures and specific needs.
3.2
Alternative Approaches to Measuring Fiscal Capacity
Fiscal capacity has been defined in a variety of ways but perhaps the most accepted definition is as the level of revenues that a subnational government could potentially raise using its assigned taxing powers and tax bases if it were to use the average level of tax effort across all subnational governments in the country. The obvious reason to base fiscal capacity on potential revenues so defined (as opposed to actual revenues) is that basing equalization transfers on actual local collections would induce perverse behavior by subnational governments: those governments that make a higher effort and collect more would see reduced equalization transfers. Another important reason for not using actual revenues is that actual collections might provide a poor reflection of fiscal capacity if there are variations in local fiscal effort. Of course, these concerns presume that subnational governments have autonomy or discretion to exercise lower or higher tax
314
Boex and Martinez- Vazquez
effort by changing enforcement effort, tax rates, tax bases, or even the ability to introduce taxes or not. This being said, it is important to realize that in many countries subnational governments have quite limited tax autonomy and that the bulk of their revenues come from tax sharing in national taxes or from different forms of transfers. This is especially true of most LDTCs. In the case of revenue sharing and intergovernmental transfers, fiscal capacity can in most cases be safely measured by the actual or forecasted level of shared revenues and transfers. Two important qualifications need to be made to this statement. First, there are countries, such as is the case in Russia, Ukraine and many other former Soviet republics, where tax administration officials have a de facto "dual subordination," meaning that they respond to their official bosses in the hierarchical structure, ultimately the central authorities, but also respond for a variety of practical reasons to the local a~thorities.~' Where this situation of dual subordination exists it will not be safe to assume that the effort to collect shared revenues cannot be affected by the local authorities. In some cases it will also be necessary to estimate potential shared revenues to avoid perverse behaviors by subnational government officials. Second, it is also assumed that local government authorities do not have the power to actually affect the size of the tax bases for either their own or shared taxes.21 Notwithstanding the fact that in many LDTCs own taxes represent a small part of subnational government revenues, it is important in the measurement of fiscal capacity not to give disincentives to local tax effort and to look for ways to measure their fiscal capacity that preserve the right incentives. This is particularly the case since own-source revenues are not only an important source of financing for local government, but also serve as an important accountability mechanism for local officials. Several approaches are used in the international practice to quantify subnational fiscal capacity, which are reviewed and assessed immediately below, including:
1. Lagged own revenue collections 2. Basic proxies for the local ability to tadability to pay (such as personal income of Gross Regional Product) 3. More sophisticated measures of fiscal capacity, such as the Representative Revenue System 3.2.1
Lagged revenues
A first available proxy or measure for the fiscal capacity of a subnational jurisdiction is the (per capita) level of own revenue collections or lagged values of revenue collections, for example the revenues collected last year. Similar to the problems encountered with actual expenditures as a measure
Designing Intergovernmental Equalization Transfers
315
of fiscal need, actual revenues provide an unsatisfactory measure of capacity for two main reasons. First, there are several elements that can create a gap between the amount of revenue raised by a regional or local government and the potential ability of that subnational government to raise revenue: (i) two local governments with the same fiscal capacity may collect different amounts of revenue as a result of applying different tax rates or defining taxable income in different ways, by for example, granting different levels of exemptions; (ii) two local governments with the same fiscal capacity may collect different amounts of revenue due to variances in the enforcement effort with which revenues are collected; (iii) two local governments with the same fiscal capacity may collect different amounts of revenue as a result of different levels of taxpayer compliance (for the same enforcement effort). Thus, while tax rates, enforcement effort and taxpayer compliance all affect the actual level of revenue collections, they do not affect thepotential ability of regional or local governments to collect revenues. Second, it is clear that using actual revenue collection as a measure of fiscal capacity in an equalization formula would provide a negative incentive for subnational revenue generation, as it would reduce equalization transfers for subnational governments that collect more revenues. In addition, using past collections or lagged revenues does not satisfactorily address the problem of negative incentives. Although the impact of the incentive is reduced, it is likely that sooner or later subnational governments will "learn" that higher collections translate into lower transfers. Despite these problems, past revenue collection can be used in some creative ways that more or less reduce the negative incentive effects to tax effort. For example, Vietnam freezes the estimates of fiscal capacity based on historical collections for a number of yeas in their "stability periods." Another example of one of these creative approaches is provided by Ukraine's current methodology for estimating fiscal capacity for its equalization grant system (see Table 4). The essence of Ukraine's approach is that fiscal capacity is defined first in relative terms by computing the ratio of actual (own) revenues per capita of the jurisdiction to the average (own) revenues per capita for all jurisdictions for a base year (Martinez-Vazquez and Zeikate, 2002). Fiscal capacity is then derived from multiplying the ratio coefficient for each jurisdiction by the aggregate revenue forecast (for all local governments). This approach dilutes the impact that actual collections behavior by a jurisdiction may have in its estimate of fiscal capacity. The incentive effect is completely eliminated in Germany's approach to fiscal capacity, which is measured as average tax revenue per capita multiplied by the population of each state.22However, the constant per capita revenue norm means that Germany's fiscal capacity measure does not take on board any subnational variations in tax bases.
Boex and Martinez- Vazquez
316
Table 4. Computing a measure of fiscal capacity: Historical average of the locality's per capita revenue relative to the national average per capita revenue
Step 1.
Select revenue sources and time window
Select revenues sources, the collections of which can serve as a proxy for the locality's overall revenue-raising capacity. Select time period for which data on the yield of the chosen revenue sources are available (and are relevant for the present) Step 2.
Dejine the Index of Relative Fiscal Capacity
The Index of Relative Fiscal Capacity Ki can be defined as the C -Per Capita Local Yield historical average of 1Per Capita National Yield where the summation is performed over the chosen revenue sources. This index reflects the historical average (two or three years, for example) of the locality's per capita revenue relative to the national average per capita revenue from the selected sources. Step 3.
Compute Fiscal Capacity
Fiscal capacity for locality i equals: Fiscal capacity i = Ki * Aggregate Local Revenue Forecast This amount reflects the amount of collections that each locality is expected to have, given its historic revenue base and the macroeconomic forecast.
3.2.2
Basic proxies for ability to tadability to pay
Moving away from fiscal capacity measures based on actual revenue collections, a widely used approach to measuring fiscal capacity is to use a proxy or a combination of proxies capturing the general ability of subnational governments to raise taxes, whatever those taxes actually are. One such basic measure is the per capita level of personal income, which is widely available and has the advantage of simplicity. Of course, to the extent that local taxes are applied to commercial property or business income and transactions, personal income may fail to accurately measure a local government's revenue potential. Another widely used proxy measure for fiscal capacity is gross regional product (GRP), which is the subnational equivalent of Gross Domestic Product at the national level (GDP). GRP may be considered a more comprehensive measure of the fiscal capacity than per capita income
Designing Intergovernmental Equalization Transfers
317
because GRP includes income generated within a region irrespective of the location of residence of the worker or producer. A third proxy for the local ability to tax is the Total Taxable Resources (TTR), which is a modified version of GRP and excludes certain items such as central taxes and transfers from the measure of fiscal capacity since they do not provide a potential tax base. All these proxies and others take specific shape in different country contexts. For example, Uganda uses household expenditure data as a proxy for revenue capacity while South Africa uses average monthly income per capita in each municipality as a measure of their fiscal capacity. In India, the tax capacity of each state is approximated by the difference between the per capita income of the state and the highest per capita income state. In the case of Switzerland, fiscal capacity is estimated as an index of several weighted factors including per capita National Income in the individual Canton and several measures of tax burdens and per capita tax revenues of the Cantons (Dafflon, 2001). 3.2.3
Representative revenue system
Some countries (for example, Australia, Canada, the United States, and more recently Russia) use or have used a multi-dimensional measure of fiscal capacity known as the Representative Revenue System ( R R S ) . ~ ~ Similar in nature to the Representative Expenditure System discussed earlier, the basic idea underlying the RRS is to calculate the amount of revenue that a region would collect given its tax bases if it were to exert average fiscal effort. This is done by collecting data on revenue collections and tax bases for each of the taxes under consideration for every subnational region. Based upon information on all tax bases for every region, as well as the national average, and fiscal effort for each of the taxes, one can compute the amount of revenues that each jurisdiction would collect under average fiscal effort. This amount is then considered to quantify the fiscal capacity of each jurisdiction. The main benefit of RRS is that computations are made at a disaggregated level and based on detailed knowledge of (proxies for) the statutory tax bases. Naturally, the accuracy of the RRS is determined by data availability. For instance, in Canada, the fiscal capacity of a province is a measure of its ability to raise revenues from more than 30 revenue categories - including personal income tax, corporate income tax, sales taxes, property tax, and other sources. The revenue potential for each province is computed based on the size of respective tax bases and the assumption that each province applies an average level of tax effort (i.e., average effective tax rates). Provinces with revenue raising ability, or fiscal capacity, below a threshold or 'standard' amount receive equalization payments from the national
318
Boex and Martinez- Vazquez
government.24 In many LDTCs, the required data are not available to compute similarly detailed RRS assessments. Instead, it is possible to apply a simplified RRS (based on a single proxy tax base) or a modified RRS to arrive at reasonable measures of fiscal capacity (see Table 5 and Box 5, respectively). Overall, the RRS is a more thorough and complete method of measuring the fiscal capacity of a region. It is based on disaggregated data and takes into account variations in effective tax rates among various tax components and non-tax revenue sources. As a result, fiscal capacity as measured by the RRS can be considered a more accurate representation of a region's true fiscal capacity. However, by the disaggregated nature of the computations, the measure is data intensive and, therefore, it is not always possible to use it or use it fully. The main benefit of the Representative Revenue System (RRS) as a measure of fiscal capacity is its accuracy. However, the method's intensive data requirements may prevent its implementation in many LDTCs. An alternative solution that would maintain much of the accuracy of the RRS while reducing its data requirements is by introducing regression analysis to the RRS method (Martinez-Vazquez and Boex ,1997). The use of regression analysis in the Representative Revenue System dramatically reduces the data requirements for the measurement process. Rather than collecting data on the actual collections and tax bases for every single tax component, the regression-based RRS method only requires information on the total amount of revenues collected for each jurisdiction plus data on a series of proxies for the tax bases for each subnational government. While the integrity of the data still needs to be guarded, the requirements on the proxies for the tax bases are somewhat less strict. Most importantly, there is no need to group revenue items into separate components and to specifically match each tax component with a standard tax base.
3.3
Do Measures of Fiscal Effort Belong in an Equalization Formula? If So, How Should It Be Measured?
A further policy decision in the formulation of equalization grants is whether to incorporate tax effort as part of the intergovernmental transfer mechanism. This is a controversial issue. If local governments are presumed to deliver pure private goods (excludable and rival) without significant interjurisdictional externalities, the generally accepted principle is that the transfer systems should neither encourage nor discourage tax effort by subnational governments.26 As the argument goes, there is no a priori justification for the public sector to reward those subnational jurisdictions
Designing Intergovernmental Equalization Transfers
Table 5. A Simplified Representative Revenue System (RRS) Step 1.
Select proxy measures for the tax base (T) Select a measure of a subnationaljurisdictions' own-source revenues (i.e., regional revenue collections) and a proxy for an applicable measure of the subnational tax base, such as aggregate personal income or gross regional product.
Step 2.
Define the Average Effective Tax Rate (AETR) The average effective tax rate can be defined as: AETR
=
(xi Own source revenues i) 1 (Xi Tax Based
This coefficient reflects the average share of own revenue collections (in relation to the subnational tax base) that is collected across all jurisdictions. Step 3.
Compute Fiscal Capacity Fiscal capacity for jurisdiction i equals: Fiscal capacity i
=
AETR
* Tax Base i
This amount reflects the amount of collections that each jurisdiction would have if it exerted an average level of fiscal effort in collecting own-source revenues.
Box 5 The Modified RRS approach in Indonesia Fiscal Capacity = LOR + (PT + LTF + PIT + 0.75*NRS) LOR : Predicted Local Own Revenue = a + b*GRDP Services PT : Property Tax Revenue Sharing LTF : Land Transfer Fee Revenue Sharing PIT : Personal Income Tax Revenue Sharing NRS: Natural Resources Revenue Sharing GRDP Services: The service component of gross regional domestic product This example of Indonesia shows a partial use of a regression-based RRS approach to estimate capacity for local tax revenues, while the capacity measure for all shared revenues is taken to be the next year's. This is justified because local governments control local taxes but have no control at all on shared taxes with the central government.25
320
Boex and Martinez- Vazquez
that decide to tax themselves more and thus consume a large share of their incomes through the local government level. In fact, the introduction of tax effort in the formula opens the formula to potential manipulation by local governments, with the potential outcome of redistributing funds away from more needy local governments to high spenders and wealthier local governments. These theoretical arguments notwithstanding, the fact is that numerous countries incorporate some measure of tax effort in their equalization formulas. One main argument used is that the transfer system can be used to try to stimulate subnational governments to use revenue raising capacity when little of this capacity is used, with subnational governments relying instead on intergovernmental grants and revenue sharing.27Thus, if there is a serious problem with the lack of revenue mobilization, the question is whether or not the equalization formula should promote or give incentives to stimulate tax effort or revenue performance. One answer may be that encouraging subnational tax effort may be justified at least on a temporary basis, but clearly this issue will not be resolved here. Among developing countries, the following introduce a tax effort element in their transfers system: Ghana, ~ndia:~Nigeria, Mexico, Colombia, Nicaragua, Ecuador, Venezuela, and Sri ~ a n k a . ~ ' Fiscal effort can be defined as the degree to which a government or subnational region utilizes the revenue bases available to it. As such, the level of fiscal effort is affected by the level of the tax rates applied (if subnational governments have discretion over rate), by the level of exemptions granted (again if subnational governments have discretion over the structure of the tax), and by the tax enforcement effort exerted by the tax administration authorities. Thus, if we want to encourage higher tax effort by subnational governments on a temporary basis, it is not a simple matter to measure it. The level of fiscal effort is appropriately measured as the ratio of the actual amount of revenues collected to some measure of fiscal capacity. But as we have seen in the previous discussion, estimating fiscal capacity is not an easy task. Therefore many of those countries that incorporate fiscal effort in their equalization formulas resort to a measure of increases in own source revenue collections relative to the previous year(s), such as in the case of Nigeria and ~olombia.~'The disadvantage of this approach is that subnational governments that initially exerted a high level of fiscal effort have little room to increase their measured fiscal effort further, and thus this approach may not reward those jurisdictions with higher levels of fiscal effort. Another disadvantage of this mechanism is that it certainly rewards subnational governments that had previously low levels of fiscal effort. Furthermore, this approach fails to recognize that changes in local revenue collections over time are often more the result of changes in the local tax
Designing Intergovernmental Equalization Transfers
321
base well beyond the control of local officials rather than actual changes in local fiscal effort. Other developing countries do a better job at estimating tax effort. For example, Nicaragua's tax effort is only measured through property tax revenues, the most important source of revenues over which local governments have discretion. To arrive at tax effort, the revenue from the property tax in the previous year is divided by a measure of potential property tax revenue, which is calculated using a system that estimates the amount of taxable property in the municipality times its corresponding tax rate.32
SIMULATING DIFFERENT TRANSFER APPROACHES: LOCAL GOVERNMENT FINANCE IN GEORGIA CIRCA 1960 A key challenge for policy makers and policy analysts assessing intergovernmental transfer schemes in developing countries - which were developed in the absence of "perfect" data - is that practitioners or analysts are unable to compare the bctionality of the system at hand to the performance of the first-best transfer system that would have been possible under perfect data availability. One possible way to learn something about which "second-best" measures of fiscal capacity and expenditure needs might perform well in the context of a developing economy would be to compare the performance of a several second-best approaches against the counter-factual of a transfer system for a developing economy where "perfect" data is available. One approach followed in some earlier studies (for instance, Vaillancourt 2001) has been to explore historical data for local government finances in a relatively lesser-developed part of a developing country. Such an approach assures the adequate availability of data to construct a "first-best" counter-factual, which is generally impossible in a developing country. For the purpose of this study, we argue that around the year 1960, the State of Georgia in the U.S. and its local government (finance) structure presented many of the features of a "typical" middle-income developing economy today. The State's economy was largely rural (74 percent), with over one-third of the population living below the poverty line, and one-third of the housing stock lacking basic amenities such as piped water (Table 6).33 As far as the State's local government structure in 1960, Georgia was divided into 159 counties, with its structure of local government having changed little since the 1850s. Anecdotal evidence suggests that the size of counties in Georgia was historically based on the desire for every resident to
322
Boex and Martinez- Vazquez
Table 6. Local Government (County) Characteristics, Georgia, 1960 Average
Std. Dev.
Minimum
Maximum
Population Land Area (sq. mi.) Population density (pop / sq. mi.) Percent population under 18 Percent population over 65 Percent population rural Percent housing w/o piped water Infant Mortality Rate (per 1000) Median Family Income (US$) Per Capita Personal Income (US$) Poverty Rate Source: Computed by the authors based on U.S. Bureau of the Census (1960);
University of Georgia (1963); and Georgia Department of Public Health (1962). be able to commute to his or her county seat by mule-drawn cart in a day. Although (unlike in many developing economies today) certain local government services in Georgia were provided by municipalities and school districts, the sub-county nature of municipalities and the county-wide natureof all but a few school districts allows for the analysis to proceed at the county-level by aggregating all necessary local data into county-based local government areas. Local governments in Georgia - then as now - provided a variety of typical local government services such as primary and secondary education, some public health services, parks and recreation, local infrastructure and community development. Data from the 1962 Census of Governments show that local services were funded by a combination - roughly evenly split of own-source revenues and intergovernmental grants received from the state government (Table 7). In turn, local own-source revenues were roughly evenly divided between local tax revenues and non-tax sources, with property taxes accounting for an average of about one quarter of total local revenues. Transfers were provided through a variety of different targeted and sectoral grant mechanisms. Although local governments in developing countries typically enjoy less revenue autonomy than their historical counterparts in Georgia, the overall environment for local government
323
Designing Intergovernmental Equalization Transfers
Table 7. Local Government Finances in Georgia, 1962 (Per capita dollars) Std.
Local Revenues of which Intergovernmental Grants of which Local Own Revenues of which Property Taxes o f which Other Taxes
Local Expenditures of which Education of which Capital Expenditures
Average
Dev.
Minimum
Maximum
128.76 69.41
23.89 16.70
15.06 9.84
204.00 122.65
59.34
24.57
5.23
162.38
30.24 3.65 129.46 70.10 18.02
12.84 2.37 27.05 11.27 16.08
2.38 0.15 13.22 7.99 0.59
102.16 15.05 240.65 98.23 116.50
Source: Computed by the authors based on U.S. Bureau of the Census (1960); Census of Governments (1962).
finance in Georgia in 1960 was altogether not so different from many decentralized developing economies today. Based on the fiscal and other data available for Georgia counties around 1960, in the remainder of this section we set out to compute various different measures of local expenditure need and fiscal capacity which would typically be used in the development of a formula-based unconditional grant system. In the next section, we then compare the performance of these approaches with different levels of technical sophistication and dataintensity, and identify which "second-best" measures correlate most closely to the "first-best" approach.
4.1
Alternative Measures of Local Expenditure Needs
Based on the data available for Georgia for 1957-1960, we were able to compute a total of eight alternative - and progressively more sophisticated measures of local expenditure need, namely:
1. 2. 3. 4. 5. 6.
Actual local expenditures and lagged expenditures. Equal per capita expenditure norm. The proportion of poor households. The proportion of households without piped water. An index of need based on infant mortality. An index of expenditure needs (similar to a Human Development Index) based on poverty, water access and infant mortality.
324
Boex and Martinez- Vazquez
7. A "traditional" Representative Expenditure System (RES). 8. A regression-based Representative Expenditure System (RES). Perhaps the most plain and apparent proxy for a local government's level of expenditure needs is the local government's actual expenditure level. County governments in Georgia engaged in local spending at an average rate of $129 per person, ranging from a minimum of $13 per capita to around $240 per person (Table 7). However, there are two major problems with the reliance on actual expenditures as a measure of local expenditure needs. First, as already noted above, the inclusion of actual expenditures in a transfer formula would provide local governments with a perverse incentive to increase expenditures beyond their efficient level. Second, a perhaps more importantly, local expenditure levels may simply be a poor reflection of local expenditure needs, as discrepancies should be expected to exist in actual expenditures and the cost of an average basket of local government services. Differences in local preferences, variations in the demand for local services, and spending constraints dictated by local resource availability may all contribute to variations in actual local spending, while leaving unaffected a local government's expenditure needs. Since many countries seek to mitigate the incentive issue in part by using lagged local expenditures collections, we also compute lagged local expenditures (extracted from the 1957 Census of Government) as an alternative proxy for local expenditure needs. An alternative approach to quantifying local expenditure needs that would avoid the efficiency and equity issues of actual (or lagged) local expenditures is reliance on an equal per capita expenditure norm. Such a per capita expenditure norm would assume that each local resident has an identical level of expenditure need, and thus would not vary at all across local jurisdictions. Based on aggregate actual local expenditures of $609.4 million and a state population of 3.9 million, one could posit that as the basis for computing fiscal needs, each local government has expenditure need equal to a norm of $154 per person. A third methodology to compute a "second-best" measure of expenditure needs relies on several socio-economic and demographic characteristics that are typically associated with poverty reduction and local expenditure needs. For instance, if the population in a local government jurisdiction is relatively poor, the need for local government services (a function of both demand and cost) is expected to be greater. Particularly in countries that are pursuing a poverty reduction strategy, reliance on local poverty as an indicator of local government need may be an attractive option. As such, relying on available census data, we computed an index of local expenditure needs in proportion to the number of residents below the poverty line in each county.34Thus, for instance, if a local government contains 5 percent of the state's poor
Designing Intergovernmental Equalization Transfers
325
households, this measure assumes this county to have 5 percent of the aggregated expenditure needs at the local government level. Since poverty counts (or even good poverty estimates) at the local level are not available in many developing or transition countries, policy practitioners may seek to rely on secondary indicators of poverty for which data may be available, such as housing characteristics, infant mortality figures or illiteracy. Likewise, based on available data for housing in Georgia (in particular, the number of housing unit that had no piped water) and the infant mortality count in each jurisdiction, we computed alternate local expenditure needs measures in proportion to these factors.35 Naturally, each of these preceding local expenditure needs measures only take into account one possible measure of local expenditure needs. Mirroring practices in many developing countries, for the purpose of our analysis of local government finance, we construed a local needs index (similar to a Human Development Index) which takes into account the weighted average of three available needs measures (poverty, water access and infant mortality). Mirroring another common practice in developing countries, the relative weights are arbitrarily assigned to equal one-third for each of the three factors.36' A potentially more realistic (but correspondingly much more dataintensive) approach to quantifying local expenditure needs is presented by the "traditional" Representative Expenditure System (RES). While this approach might be used in practice in the context of some developing countries, the quality of its results will depend crucially on the availability of expenditure data by function, and the availability of data on plausible clientbases for each expenditure category. For instance, in the case of 1960 Georgia, we were only able to identify two local spending categories. Local education was the only specific spending category for which we were able to obtain both spending data and a plausible corresponding client-group (namely the population of school-aged population). The financial (per-client) norm for education under this simple traditional RES suggested a local expenditure need of $260 per school-aged child (computed by dividing the aggregate local education spending of $276 million by the 1.06 million school-aged children). The only other spending category included in this traditional RES comprises all non-education spending, and uses population as its client-base ($84 of non-education spending per resident). The final - and arguably the most complex - approach to quantify local expenditure needs is a regression-based Representative Expenditure System. Based on the available Georgia data for 1960, we experimented with a number of demographic, social and economic variables as potential explanatory variables for per capita expenditures in Georgia, including total population; percent of the population under 18 and over 65, respectively; percent of the population that lived in a rural place; the poverty rate;
326
Boex and Martinez- Vazquez
population density; infant mortality rate; and the enrollment rate. Guided by the relative significance of independent variables, the most successful RES model we were able to identify took on the form:
where population (POP) is measured in thousands, METRO is a dummy variable for counties within metropolitan areas, and all other independent variables (RURAL, NOWATER, and P O P 4 8) are specified in percentages. TAXBASES represents a vector of local tax bases (including the value of taxable property, personal income and local payroll) to control for the impact of fiscal capacity on local expenditures. In computing the measure of local expenditure need using the regression results, the impact of fiscal capacity on expenditures is removed by assuming that all counties have an average level of fiscal capacity, so that: LOCAL NEED, = 99.2 + 0.046 PO? - 0.19 RURAL, - 18.6 METRO,
- 0.30 NOWATER, + 0.65 POP < 18,
+ TAXBASES
The explanatory power ( R ~ of ) the estimated model equals 0.45, and all independent variables were found to be statistically significant at 1 percent or greater. Based on this regression-based RES approach, local expenditure needs were estimated for each local government as the predicted value of per capita expenditures multiplied by each local government's population base. The estimated local expenditure needs based on the regression-based RES is compared with the other seven proxies of local expenditure need in Section 5 of this paper.
4.2
Alternative Measures of Fiscal Capacity
Similar to the computation of the different measures of local expenditure need based on the data available for Georgian counties in 1960, we also computed several alternative measures of local fiscal capacity, ranging from less to more sophisticated. These proxies include: 1. 2. 3. 4. 5. 6.
Actual and lagged local revenue collections; Poverty as a proxy for local fiscal capacity; Regional income level as a proxy of local fiscal capacity; Average per capita personal income; Traditional Representative Revenue System; and Regression-based RRS
Designing Intergovernmental Equalization Transfers
327
Again similar to the preceding discussion of local expenditure needs, the most obvious measure for a local government's level of revenue potential or fiscal capacity is the local government's actual own-source revenue collections. As highlighted in Table 7, own-source revenue collections in Georgia in 1962 ranged from $15 per person to $204 per capita, with an average level of local revenue collections of $129 per person. Yet, as was the case for the use of actual expenditures, the reliance on actual revenues as a measure of fiscal capacity would raise significant equity and efficiency issues. First, actual revenues may not reflect actual fiscal capacity - which should measure the potential of local governments to collect revenues. Use of actual local revenue collections in an unconditional transfer formula would reward well-endowed local governments who exert little of no fiscal effort, while penalizing poor local governments who exert a high level of fiscal effort despite their weak revenue base. Second, as discussed earlier, reliance on actual revenue collections as a measure of fiscal capacity in a transfer formula would provide local governments with an incentive to reduce own source revenue collections in anticipation of a larger grant. In order to circumvent this incentive issue, we also computed lagged revenue collections (along the lines of Ukraine's fiscal capacity measure) as an additional second-best measure of fiscal capacity. As an alternative measure of fiscal capacity, data could be used for average personal income or median household income. Since household wealth arguably forms an important - if not the most important - local tax base, it is reasonable to expect that higher household incomes relate to a higher level of local revenue collection. This measure is further particularly expedient in developing countries where such data is periodically available from household expenditure (or income) surveys. For Georgia around 1960, county-level data on total personal income was extracted from Georgia's Statistical Abstract (1963), with per capita levels of personal income ranged from $505 to $3,334, averaging $1,200. While the necessary personal income data may be available for local government areas in some developing countries, other countries may not produce such income data below the regional level. In order to determine whether there is value in using regional income as a proxy for local income as a measure of fiscal capacity, we compute average personal income per person for Georgia's nine economic regions (State Economic Areas, as defined by the Bureau of the Census in 1960) as an alternative measure of fiscal capacity for each of the counties located in the respective state economic areas. This approach results in a significant reduction of the range of fiscal capacity estimates from a minimum $1,006 to a maximum $2,066. Another possible, albeit admittedly very indirect, proxy for a local government's revenue potential could be the share of the local population that is non-poor. In case there is no income of economic data available
328
Boex and Martinez- Vazquez
whatsoever at the local level, this proxy might be warranted on the assumption that there is a correlation between the relative number of presumptive taxpayers and local revenue collections. A conceptually much more accurate (and conversely, more data demanding) approach measuring fiscal capacity is to relate local revenue collections to specific tax bases through a Representative Revenue System (RRS). A traditional RRS computes fiscal capacity by first, defining as many revenue categories as local revenue for which data on collections and the size of the tax base are available; second, measuring the overall average effective tax rate (AETR) for each different revenue category (as a ratio of aggregate local collections and the aggregate tax base), and third, computing the revenue potential for each subnational government by applying the various AETRs to the different local tax bases in locality. In the traditional RRS based on the historical Georgia data, we are able to identify only two distinct revenue categories: property tax collections and all other revenues. The average effective tax rate for local property taxation is $4.70 per $100 of taxable property, based on an aggregate assessed value of taxable property of $4.2 billion and aggregate local property tax collections of $198.3 million. The presumptive tax base for all other local government revenues is personal income, for which the average effective tax rate was computed at $2.59 per $100 of local personal income. Although additional data were available on tax bases by county in Georgia from various census sources (including payroll and retail sales), we were unable to include these in the traditional RRS approach because we have no specific data on corresponding revenue collections. However, the regression-based RRS approach (like the regression-based RES) does not require direct linkages between specific revenue instruments (or categories) and specific tax bases. Instead, the regression itself will quantify the relative (direct and indirect) impacts of the various tax bases on the total amount of local revenue collections. Based on the available data for Georgia, the regression model with the highest explanatory power for per capita local revenues (PC REV) resulted in the equation:
+ 0.0027 PC PERSINC, + 0.12 PC PAYROLLi + NEEDS with a predictive power of 0.73, and all revenue parameters statistically significant at the 10 percent level or better. NEEDS reflect a vector of local needs measures (included in the RES regression above) which are controlled for and held constant for the purpose of computing local fiscal capacity. Inclusion of additional potential tax bases as independent variables (e.g.,
Designing Intergovernmental Equalization Transfers
329
local retail volume) did not result in significance or an increase in the model's predictive power.
4.3
Which Are the "True" Measures of Fiscal Need and
Capacity We started out this paper with the observation that unconditional grant systems generally aim to provide fbnding to equalize fiscal needs, fiscal capacity or the "fiscal gap" between needs and available resources. In order to achieve any of these three objectives, we need to consider which measures of expenditure needs and fiscal capacity should be considered "first-best". The perception by some is that although actual local revenues and expenditures are unsuited as measures of needs and capacity in a transfer system due to their perverse incentive effects, they otherwise come quite close to reflecting actual needs and capacity. This would generally imply that measures such as lagged or base-year expenditures and revenues would be appropriate "true" measures of fiscal need and capacity. While we believe that correlation with actual expenditures and revenues is an important indicator of a well-performing proxy of needs and capacity respectively, there is no conceptual reason to consider that these measures are the best available proxies for need or capacity. Since there is no independently verifiable "true" or best measure of either expenditure need or fiscal capacity, we have to make a subjectivejudgment. Instead of considering actual (or lagged) expenditures or revenues as the most appropriate indicator of fiscal need or capacity, we believe that the technically most sophisticated techniques (notably local expenditure needs computed using a regression-based Representative Expenditure System and local fiscal capacity measured by a regression-based Representative Revenue System, respectively) quite possibly provide the best possible measures. While all other measures of need and capacity are based on assumed relations between client- and tax-bases on one hand and expenditure needs and revenue potential on the other, regression-based RES and RRS quantify the strength and provide statistical feedback on the certainty of the relative relationship between tax base and collections. As such, we believe that regression-based RES and RRS - which are among the most data-intensive approaches - are most likely to provide us with the closest or first-best approximation of local expenditure needs and fiscal capacity. 37
Boex and Martinez- Vazquez
RESULTS The relatively plentiful data for county-level governments in Georgia in 1960 allow us to quantify nine different measures of local expenditure needs
and seven different measures of fiscal capacity. By combining the alternative measures, we could quantify the fiscal gap using 63 different combinations of fiscal needs and capacity measures. The question of interest, of course, is how well do each of the different measures compare against each other, and how well do they perform against the presumed first-best measures that rely on regression-based methodologies with data that are often unavailable in developing countries? Our analysis here consists of three parts. First, we compare the various fiscal needs measures, and see how they compare to each other. Naturally, specific attention is paid to which second-best measure performs best in approximating the first-best measure of local expenditure needs. Second, we similarly look at the comparison and performance of the various fiscal capacity measures. Third, we quantify the first-best fiscal gap (local expenditure needs minus fiscal capacity) and compare its incidence pattern to the performance of the 62 second-best combinations that we were able to simulate based on the available data. In our analysis, we compare the results of the different methodologies in two different ways. First, we compute the simple correlation coefficients for the various per capita levels of local needs and capacity. The correlation coefficient allows a statistical comparison of the different methodologies, albeit without taking into account the impact of variations in jurisdiction size on the ultimate a~locations.~~ Also, the correlation coefficient is not helpfbl (as we note below) in assessing the performance of the per capita normbased measure of local needs (since it contains no per capita variations). Therefore, in addition to the correlation coefficient, we have computed the total absolute deviation from the first-best option (referred to here as the "Index of Fit" or IOF) for a standard size transfer pool.39For each needs or capacity measure, the IOF indicates the amount of funds that would have to be reallocated in order to achieve the first best incidence in the distribution of a transfer pool of $1,000,000. For the purpose of analysis, we compute the IOF for the different second-best measures in Tables 8 and 9 not only against the first-best measure, but also against actual local expenditures and revenues, respectively.
5.1
Assessing the Different Measures of Fiscal Need
The different proxies for local fiscal need are compared in Table 8, based both on the correlation of per capita needs (in the top panel of the table), as well as based on the Index of Fit (in the bottom panel). The first observation
Designing Intergovernmental Equalization Transfers
33 1
regarding the various measures of local expenditure needs is that - based on both the correlations as well as the IOF results - regression-based RES is not only arguably the best local needs measure on conceptual grounds, but that the first-best measure also achieves the closest correspondence with actual local expenditure patterns (r=0.54; IOF=$l7l,OOO). Also notable is that there are substantive variations among the performance of the different measures. Although some combinations of measures reveal substantive positive correlation, many of the stronger positive correlations in the table are in fact spurious (particularly between the composite index and its components). More remarkable is the weak or seeming lack of correlation between many of the measures and the negative correlation between many of the measures with the regression-based RES approach and actual expenditures, respectively. Similarly, analysis of the IOFs supports the contention that the performance of the expenditure needs measures varies widely. Based on the index of fit from the first-best regression-based RES approach, the ranking of remaining measures of local expenditure needs range from per capita financial norm approach; the traditional RES; actual per capita expenditures; the expenditure needs index based on infant mortality; poverty; the composite needs index; the water access measure; and finally, lagged expenditures. Highlighting the poor performance of the worst-performing expenditure needs measure (lagged expenditures) is the fact that the index of fit between it and the first-best measure (based on a grant pool of $1,000,000) is an incredible $91 1,000. Perhaps a surprising result of the simulations is that the best-performing alternative measure of local expenditure needs (even if we consider - despite efficiency concerns - actual and lagged expenditures) is the per capita expenditure norm, which allocates local expenditure needs exclusively in proportion to population. While this measure is in fact the least demanding in terms of data requirements, it clearly outperforms more technically complex measures such as the traditional RES and the composite index of local needs, and even performs better than actual expenditures. The policy implication of this finding - if confirmed by similar studies in other countries - would be that reliance on a simple per capita financial norms would be an appropriate and relevant measure of local expenditure need in the design of an unconditional (general purpose) transfer formula. Based on the analysis contained in Table 8, poorly performing measures of local expenditure needs include the poverty measure, the composite needs index, water access and lagged expenditures. In fact, these four measures have an IOF from the first-best measure from slightly below $500,000 to over $900,000 (based on a simulated grant pool of $1,000,000). The weak performance of the poverty measure reveals a general weakness of efforts to allocate resources in a "pro-poory7fashion: the objective of (local) public spending is to fund government services and infrastructure to all residents
Table 8. Comparison of Different Measures of Local Expenditure Needs for Georgia, 1960 Regr. RES
Trad. RES
Pov. Index
No Water
Infant Mort.
P.C. Norm
Index
Lag Exp.
Act. Exp.
Correlation Coefficient. Per C a ~ i t aMeasures Regression RES Traditional RES Poverty Index No Water Index Infant Mortality -0.37
0.68
0.78
0.90
0.48
1.OO
Per Cap. Norm
0.00
0.00
0.00
0.00
0.00
0.00
1.OO
Lagged Expend.
-0.27
0.38
0.51
0.50
0.00
0.50
0.00
1.OO
Actual Expend.
0.54
-0.1 1
-0.26
-0.36
-0.06
-0.34
0.00
-0.05
1.OO
Composite Index
Index of Fit (Rank) from: Regression RES
--
0.085
0.471
0.880
0.260
0.484
0.059
0.911
0.171
Rank
--
2
5
7
4
6
1
8
3
0.171
0.226
0.581
0.978
0.351
0.596
0.202
1.017
--
1
3
5
7
4
6
2
8
--
Actual Expend. Rank
Designing Intergovernmental Equalization Transfers
333
both rich and poor, not just the poor. Likewise, the poor performance of the composite needs index suggests that combining several different povertyrelated measures into a single index - a common practice in many developing countries - does not necessarily improve the performance of the needs measure. Vaillancourt (2001) concludes the same in regards to various composite indices simulated for Newfoundland and Prince Edward Island.
5.2
Assessing Different Measures of Fiscal Capacity
The comparison of fiscal capacity measures provides an overall picture quite different from the analysis of fiscal needs measures. Arguably all simulated measures of fiscal capacity function as at least an adequate proxy of (the first best measure of) fiscal capacity; all (cross-)correlations are positive, most are substantive, and the indexes of fit indicate that all measures provide a reasonable measure of fiscal capacity. To be fair, we ought to note that the data requirements needed for any of these measures are still relatively hard to obtain in many developing and transition countries. We again note that the regression-based Representative Revenue System (RRS) not only arguably provides the best measure conceptually, but is the fiscal capacity measure that correlates the closest with actual local revenue collections (r=0.83; IOF=$l54,OOO). If we were to exclude actual revenues as a viable alternative based on efficiency grounds, the order of second-best options based on IOF ranking from best to worst is: traditional RRS; personal income; average regional personal income; the percent of the population that is non-poor; and lagged revenues. Yet, even the worst-case scenario performs relatively well; only $223,000 would have to be redistributed from and to counties in order to get to the first-best allocation of resources.
5.3
Assessing Different Measures of the Fiscal Gap
If our ultimate purpose in designing a formula-based grant is to equalize the fiscal gap between fiscal needs and fiscal capacity, what lessons does the simulation based on the Georgia data bear? In order to explore this question beyond the separate analysis of fiscal needs and capacity measures, Table 10 quantifies the Indexes of Fit between the "first-best" Fiscal Gap (needs minus capacity; restricted to prevent negative fiscal gaps) and all other second-best combinations of fiscal needs and capacity that we simulated based on the available data. As such, Table 10 allows us to compare the performance of all different combinations of need and capacity measures in achieving the first-best allocation pattern. The grey areas in the table are not
334
Boex and Martinez- Vazquez
attainable in a second best world due to data limitations or due to incentive concerns with using actual expenditures and revenues. If we first consider which local expenditure needs measures (contained in the columns in Table 10) generally perform best, the performance of the per capita financial expenditure norm and the traditional Representative Expenditure System stand out as performing significantly better than any of the other expenditure measures. This is consistent with the analysis of the fiscal needs measures above. Furthermore, in the context of fiscal gap measures, both of these factors perform substantially better as measures of fiscal needs than either actual local expenditures or lagged expenditures (but without the associated incentive problems). Having identified and selected the two most likely expenditure needs measures, the best-performing fiscal capacity measure in quantifying a second-best fiscal gap (given the two selected expenditure needs measures) is in fact regional personal income. While this ranking is not universal across all other measure of local expenditure needs, this finding is consistent for the two main needs measures under consideration. At the same time, we observe again that the performance of the different fiscal capacity measures is much more similar, and that given the selection of the fiscal need measure, virtually all measures of fiscal capacity seem to perform at least adequately. Another observation of interest is that, in equalizing the fiscal gap, personal income, the non-poor index and regional personal income levels which all fared relatively poorly as measures of fiscal capacity in isolation (Table 9) - achieve a good approximation of the fiscal gap. The performance of these proxies is in fact better than more intuitive second-best measures of fiscal capacity (such as actual collections or lagged collections) or conceptually more appealing alternatives (such as the traditional RRS). In other words, our results suggests that just because a potential allocation factor is not the best measure of fiscal capacity (or fiscal need for that matter), this does not mean that it may not be the best suited for quantifying the fiscal gap to be equalized. Thus, depending on the selection processes and criteria, policy makers may arrive at one of several acceptable choices for a second-best fiscal gap measure in the historical Georgia exercise. Interestingly, the gap between actual expenditures and actual revenues (which local governments would consider their actual fiscal gap) proves to be a fairly poor indicator of the first-best fiscal gap measure, ranking only 19th behind numerous other plausible second-best options. Likewise, lagged expenditures and revenues performed quite poorly as proxies for expenditure needs and fiscal capacity, respectively. In contrast, measures of fiscal needs and capacity which have relatively limited data requirements seem to perform very well in quantifying the fiscal gap to be filled by the simulated equalization grant in Georgia.
Table 9. Comparison of Different Measures of Local Fiscal Capacity for Georgia, 1960
Regress. RRS
Trad. RRS
Personal Income
Regional Pers. Inc
Non-Poor Index
Lagged Own Rev
Actual Own Rev
Correlation Coefficient, Per Capita Measures
Regression RRS Traditional RRS Personal Income Regional Pers. Inc. Non-Poor Index Lagged Own Rev. Actual Own Rev. Index of Fit (Rank) from:
Regression RRS
--
0.079
0.085
0.176
0.185
0.223
0.154
Rank
--
1
2
4
5
6
3
0.154
0.167
0.162
0.270
0.268
0.166
--
1
4
2
6
5
3
--
Actual Own Rev. Rank
Regr.
Trad.
Poverty
No
Infant
P.C.
L
Table 10. Index of Fit (total absolute deviation) from first-best fiscal gap measure (Rank in p
Regression RRS Traditional RRS Personal Income Regional Personal Inc. Non-Poor Index Lagged Own Revenue Actual Own Revenue Average
Designing Intergovernmental Equalization Transfers
CONCLUSIONS Before drawing any lessons from the simulations for Georgia in 1960, we should caution against over-generalizing the results. Although we believe the results obtained for Georgia hold some general conclusions for the design of intergovernmental grant systems in the presence of imperfect data (particularly for developing countries), we recognize that the specific results of the current exercise are unique to the data availability and the social, economic and fiscal conditions prevalent in Georgia in 1960. While some of our results and conclusions were consistent with earlier studies (e.g., Vaillancourt 2001), such exercises should be further replicated to confirm their general applicability, preferably on the basis of actual developing country data. However, in general, we believe this exercise has contributed to the body of knowledge on transfer design. First and foremost, it has demonstrated the importance of measurement issues and the selection of allocation factors in the process of designing and implementing an intergovernmental grant scheme. Table 10 highlights the huge variation of outcomes that can be achieved with the same formula but with different measurement methodologies, particularly when it comes to the measurement of local expenditure needs. In addition to more careful analysis of proposed measures of fiscal need and capacity, much greater attention should be paid to the collection of sound data that could serve as an appropriate basis for these measures. A second general conclusion fi-om the current study is that the best measures of fiscal needs and capacity are not necessarily data-intensive. In fact, one of the best-performing formulas in the current exercise relied exclusively on two variables: population, in order to construct the per capita financial expenditure norm, and regional household income, in order to compute the fiscal capacity measure. Indeed, more computationally complex measures of need and capacity did not necessarily fare any better, and in fact often performed worse. Given the poor performance in our simulations of many commonly used measures of expenditure need and fiscal capacity (particularly measures based on lagged expenditures and revenues as well as pro-poor composite indexes), the selection of these measures may actually reduce the effectiveness of the transfer schemes in achieving their policy objectives.
Boex and Martinez- Vazquez
338
Notes We would like to thank Ina Simonovska, Maryclaire Ngari, Renata Puwanty and Nandya Yuwono for their research assistance with the historical county-level data for Georgia and Andrey Timofeev for useful discussions and inputs. As discussed in greater detail below, the relationship between expenditures and need is tentative. Actual local expenditures may reflect expenditure needs, but only if local governments have substantive budget autonomy. Furthermore, local expenditure patterns may also affected by variations in fiscal capacity or variations resulting from transfer allocations. Even if the purpose of the transfer mechanism is to improve the vertical fiscal balance, as is the case with general or uniform revenue sharing in central government taxes, their design cannot ignore their impact on the horizontal fiscal balance (Bird and Tarasov, 2002). Thus, the choice of tax instruments for revenue sharing matters. For example, revenues from the personal income tend to be more evenly distributed throughout the national territory than revenues for natural resource taxes. The rule or formula behind the revenue sharing also matters for horizontal fiscal imbalances (for example, using the derivation principle versus a capitation rule.) The potential unequalizing impact of revenue sharing is what led to the Soviet Union-type practice of "regulating" (or customizing) different revenue sharing rates for different subnational governments. These were highly discretionary practices. In modern times, Vietnam still "regulates" revenue sharing rates but it does so through an explicit formula that imitates those used in equalization grant systems themselves. See Martinez-Vazquez (2004) Of course, the sources of these differences in expenditure needs may not be independent. Harsh geographical and climatological conditions are typically accompanied by higher prices of goods and services as well as salaries of public employees. The term "expenditure need" will also be referred to here as "fiscal need." In some actual occurrences, such as in the case of old Soviet norms discussed above, the overall envelope for affordability may have not been present. But in the absence of a mechanism to translate such unconstrained budget norms or needs into "affordable" budget norms, it is impossible to arrive at a practical measurement of subnational need. Similarly, basing local need on lagged expenditure patterns would prolong the favorable treatment of capital city region or certain ethnic regions, such as, for example, has been alleged was the case for the island of Java in Indonesia during the Suharto era. The existence of stability periods in Vietnam is more significant for the computation of fiscal capacity. See the discussion below. Many other more complex transfer formulas include an equal shares component. For instance, disbursements from the Federation Account to the state level in ~ i ~ e r are ia based 40 percent on equality (Alm and Boex, 2002). The new equalization formula in Indonesia also has a sizable equal shares component (Brodjonegoro and MartinezVazquez, 2004). There are exceptions to the general rule. For instance, Boex and Martinez-Vazquez (2004b) argue that Nepal's VDC block grant in fact conforms to the incidence pursued by policy makers, since most smaller villages tend to be mountainous and have higher expenditure needs. In reality equality or per capita-equality measurements of expenditure needs are in fact simply the most basic form of index measure of local needs, an approach to measuring expenditure needs that is discussed next. Some countries use an index approach mixing expenditure need and fiscal capacity factors. For example, India's current statutory unconditional grants (from the 11th finance commission valid until 2005) are distributed according to an index with the following composition: Population (10%); distance of per capita income from the highest
Designing Intergovernmental Equalization Transfers
13.
14.
15.
16.
17. 18.
19. 20.
339
income state (62.5%); infrastructure (7.5%); area (7.5%); fiscal discipline (7.5%) and tax effort (5%). Cost-of-living differences may be proxied by type of location (mountainous or island population) or climate (artic territories, desert areas, and so on).There are two relatively common ways to incorporate price levels in a weighted index of expenditure needs. One could simply incorporate a cost of living (COL) index as one factor in the index; alternatively, one could multiply the relative need index for each jurisdiction by the jurisdiction's COL. See Rubenstein (2000) and Governor's Education Finance Task Force (2004) for more details on the finance of QBE in Georgia. In fact, QBE provides slight variations in expenditure norms based on whether students are "regular" students; special-needs students; or talented-and-gifted students, as well as a limited number of other factors. An important question is how the priorities are identified and quantified. In countries like Ukraine and Latvia these priorities are based on political decisions for marginal resource changes (increases and decreases) to last year's allocations. Conceptually, these priorities may be determined through some estimate of service standards which the government is committed to achieving. However, to the extent that the norms are just notional and not prescriptive, the choices ultimately made by local governments in the use of the available resources may be efficient. Nevertheless, in many of these countries there continue to be official and unofficial calls for a return to the old system of norms. In some countries, such as Russia, governmental commissions labored unsuccessfully for years to revive a new system of norms. See Martinez-Vazquez and Boex (2001). In some other countries, such as Ukraine, a new system of per capita norms were developed for budgetary and equalization purposes that largely avoid the problems of the past. The Ukraine example is further discussed below. Denmark and Japan are two examples of data intensive approaches. For example in Sweden, the standard costs per capita are calculated on the basis of 15 indexes (using on average between 4 to 5 variables for each index. Some indexes use regression analysis for their determination and others use formulas. The indexes that account for expenditure needs or costs of providing services for municipalities are: 1) Child care (age structure, parental employment, population density); 2) Care of the elderly (age structure, balance of the genders; occupational background, rural areas); 3)Individual and family care (single women with children, migration, non-Nordic national, density of the settlement); 4)Computsory schooling (rural areas, age structure, home language); 5) upper secondary school (age structure, study program preferences); 6) Water and sewage (sparseness of settlement, geological conditions); 7) Street and roads (traffic, weather conditions); 8) unemployment rates; 9) Constructions costs; 10) heating costs; 11) Frigid zone allowance; 12) additional expenditure due population loss; 13) Additional expenditure due weak population base (population within a radius of 30 and 90 Km); 14) Administration, travel and rescue services (population, population density, degree of urbanization). See Zhang and Martinez-Vazquez (2002). Besides the general (or transitory) grant other grants with equalization objectives include the pre-tax sharing system grants, grants for minority regions, grants for increasing wages of civil servants, and others. In many instances, local authorities can provide housing, salary bonuses and other perks to central tax administrators. In the former Soviet Union most government employees were subject to formal dual subordination. Vietnam nowadays also suffers from similar problems (see Martinez-Vazquez, 2004). China created in 1994 separate tax administrations for central and local taxes to avoid the problems associated with dual subordination (mostly that tax administrators would defend the interests of the subnational governments at the cost of central government interests) but dual subordination problems have persisted. (World Bank, 2002.)
340
Boex and Martinez- Vazquez If subnational authorities can actually affect the size of the tax bases, it may not be wise to use actual shared revenues in the measure of fiscal cavacitv. In addition. if subnational authorities can actually affect the size of the tax bases, using the conventional measure of potential revenues (for a given tax base) may not be safe either since this type of calculation would still induce perverse subnational government behavior; this time not so much in exercising less tax effort but in actually working to reduce economic activity and tax bases where they count for the equalization formula. City-states have a weighted factor of 1.35 on the population. The RRS was developed at the now-defunct U.S. Advisory Commission on Intergovernmental Relations. See Akin (1973) and U.S. Advisory Commission on Intergovernmental Relations (1986, 1988, and 1993). The 'standard' measures the average fiscal capacity of five 'middle income' provinces Quebec, Ontario, Manitoba, Saskatchewan and British Columbia. Notice that only 75 percent of the natural resource revenue sharing is taken into account in the formula. This is a concession to the natural resource rich provinces, which in the past felt exploited by the central authorities. See Brodjonegoro Martinez-Vazquez (2004). Stimulating revenue performance can be accomplished in a number of ways besides introducing incentives in the equalization formulas, such as introducing matching grants. We will not further discuss these issues here and rather continue to concentrate on equalization transfers. For instance, Prud'homme (2003) argues that local residents are fooled by some type of fiscal illusion, and demand a below-optimal level of public services in the presence of a large transfer system. In China this takes place through the VAT tax rebate component of the transfer system. India's current equalization formula gives tax effort a 5 percent weight. Some developed countries also use tax effort factors; for example Australia and Canada, but the latter only in its Territorial Formula Financing for the Far North and Northwest territories, which are generally less developed. Colombia's "fiscal efficiency" factor in the formula is measured as the growth in local revenues per capita on the last 3 years. Of course, this presumes that local property valuations are performed in a consistent and unbiased manner. Vaillancourt (2001) selected the Canadian provinces of Newfoundland and Prince Edward Island for a similar exercise. His selection was based on the fact that while good historical data are available for the provinces, the income levels in these provinces during the 1950s (which were in fact somewhat lower than average income levels in Georgia in 1960) approximate income levels of current middle-income developing countries such as Morocco, Ukraine and China. Since the definition of the poverty threshold in the U.S. in 1960 varied according to family composition, we were unable to determine the exact number of poor residents by county based on the available data. For instance, the family poverty threshold for a family of two persons under age 65 for 1960 was set at $1982. (U.S. Census Bureau, 2005). Correspondingly, we approximated the number of persons below the poverty line in each county based on the percent of families that had an income of under $2000 (which is reported by the census) multiplied by the number of county residents. The infant mortality rate (Georgia Department of Public Health, 1962) was multiplied by each county's population in order to arrive at the infant mortality index. As noted in Section 3 of this paper, the selection of relative weights need not be done on an arbitrary basis. Putting regression-based representative revenue and expenditure systems to good use requires that certain conditions are met besides data availability, for example adequate levels of expenditure and revenue autonomy at the local level. These conditions will not
.
>
,
d
Designing Intergovernmental Equalization Transfers
34 1
always be met. Thus even when data are fully available it may not be appropriate to try to implement an equalization formula based on representative systems. This paper does not suggest the use of representative systems as an ultimate goal in equalization formula design, but rather it uses this approach in a controlled environment to assess how well simpler methodologies may perform in arriving at expenditure needs and fiscal capacity measures. 38. As a result, if a measure is a particularly poor predictor for fiscal capacity or needs of highly populated local government jurisdictions, then the correlation coefficient will over-state the performance of the measure at hand. 39. Thus, the total absolute deviation (or index of fit) is computed as the sum over all counties of the absolute difference for each county between (1) the simulated allocation of $1,000,000 proportion to the first-best measure of needs (or capacity) and (2) the same grant pool allocated based on an alternative measure. The range of the IOF is from zero (perfect fit; no redistribution needed) to 2 (completely lack of fit; the entire fund needs to be redistributed in order to achieve the incidence of the first-best approach).
References Akin, G. 1973. Fiscal capacity and the estimation method of the advisory commission on intergovernmentalrelations. National Tax Journal, 26 (2): 275-90. Alm, J., and J. Boex. 2002. An overview of intergovernmental fiscal relations and Subnational Public Finance in Nigeria. International Studies Program Working Paper No. 02-01. Andrew Young School of Policy Studies, Georgia State University, Atlanta, GA. Alm, J., and J. Martinez-Vazquez. 2002. On the use of budgetary norms as a tool for fiscal management. Public Finance and Management, 2(3): 387-435. Bahl, R., and P. Smoke, eds. 2003. Restructuring Local Government Finance in Developing Countries: Lessonsfrom South Africa Cheltenham, U.K.: Edward Elgar. Bergstrom, Theodore C., and Robert P. Goodman. 1973. Private demands for public goods. American Economic Review, 63 (3): 280-96 Bird, R., and A. Tarasov. 2002. Closing the gap: Fiscal imbalances and intergovernmental transfers in developed federations. International Studies Program Working Paper No. 02-02. Andrew Young School of Policy Studies, Georgia State University, Atlanta, GA. Boex, J., R. Bahl, J. Martinez-Vazquez, and L. Rutasitara. 2003. Final Report: Developing a System of Intergovernmental Grants in Tanzania. Tanzania Local Government Reform Program. Boex, J., R. Kampanje, and R. Mwadiwa. 2001. Malawi Intergovernmental Fiscal Transfers Study. Government of Malawi / UNCDF. Boex, Jameson, and Jorge Martinez-Vazquez. 1998. Reforming the Fund for the Financial Support of the Regions: An Analysis of the New Proposal for an Equalization Mechanism in Russia. Georgia State University Consortium Russian Fiscal Reform Project, Moscow, Russia. Boex, J., and J. Martinez-Vazquez. 2004a. Introducing formula-based block grants as an alternative to national minimum service standards. Technical Note: Tanzania Local Government Reform Programme. Boex, J., and J. Martinez-Vazquez. 2004b. The determinants of the incidence of intergovernmental grants: A survey of the international experience. Public Finance and Management, 4(4).
342
Boex and Martinez- Vazquez
Borcherding, Thomas E., and Robert T. Deacon. 1972. The demand for the services of nonfederal governments. American Economic Review, 62 (5): 891-901. Brodjonegoro, B., and J Martinez-Vazquez. 2004. An Analysis of Indonesia's Transfer System: Recent Performance and Future Prospects. In Can Decentralization Help Rebuild Indonesia?, J . Alm, J . Martinez-Vazquez and S. Mulianyi, eds., Cheltenham, U.K.: Edgar-Elgar. Dafflon, B. (2001). Fiscal Federalism in Switzerland: A survey of constitutional issues, budget responsibility and equalization. Department of Political Economy University of Fribourg, Switzerland. Georgia Department of Public Health. 1962. Georgia Vital and Morbidity Statistics. Atlanta: The Department. Governor's Education Finance Task Force. 2004. Understanding Education Funding for Georgia's Public Schools The Quality Basic Education Act. Presentation: November 2004. Martinez-Vazquez, J. 1994. The challenge of expenditure assignment reform in Russia. Environment and Planning C: Government and Policy, 12: 277-292. Martinez-Vazquez, J. 1999. Fiscal Decentralization in Kazakhstan: Recent developments and Agenda for Reform. Mimeo, Andrew Young School of Policy Studies, Georgia State University. Martinez-Vazquez, J. 2000. Latvia's Equalization Fund. Mimeo. International Studies Program, Andrew Young School of Policy Studies, Georgia State University. Martinez-Vazquez, J. 2004a. Making Fiscal Decentralization Work in Vietnam. International Studies Program Working Paper, Andrew Young School of Policy Studies, Georgia State University, Atlanta, GA. Martinez-Vazquez, J. 2004b. A proposal for an equalization grant system for the republic of Georgia. International Studies Program Working Paper, Andrew Young School of Policy Studies, Georgia State University, Atlanta, GA. Martinez-Vazquez, J., and J. Boex. 1997. Fiscal capacity: An overview of concepts and measurement issues and their applicability in the Russian federation. International Studies Program Working Paper No. 97-03. Andrew Young School of Policy Studies, Georgia State University, Atlanta, GA. Martinez-Vazquez, J., and J. Boex. 2001. Russia's transition to a new federalism, World Bank Institute Learning Resources Series, Washington D.C.: The World Bank. Martinez-Vazquez, J., and S. Zeikate. 2002. An Assessment of the Implementation of the New Formula Based Intergovernmental Transfer System in Ukraine. Mimeo, International Studies Program, Andrew Young School of Policy Studies, Georgia State University. Martinez-Vazquez, J., J. Boex, and G. Ferrazi. 2004. Linking expenditure assignments and intergovernmental grants in Indonesia. International Studies Program Working Paper No. 04-05, Andrew Young School of Policy Studies, Georgia State University, Atlanta, GA. Oliveira, J.C., and J. Martinez-Vazquez. 2001. Intergovernmental fiscal relations in the transition: Czech Republic. World Bank Technical Paper no. 517. Washington D.C.: The World Bank. Prud'homme, R. 2003. Fiscal decentralisation in Africa: a framework for considering reform. Public Administration and Development, 23 (1): 17-27. Rafuse, R. 1990a. Representative expenditures: Addressing the neglected dimension of fiscal capacity. Washington D.C.: U.S. Advisory Commission on Intergovernmental Relations. Rafuse, R. 1990b. A walk on the expenditure side: 'Needs' and fiscal capacity. Intergovernmental Perspectives US. Advisory Commission on Intergovernmental Relations magazine: 25-30.
Designing Intergovernmental Equalization Transfers
343
Reschovsky, A. 2003. Intergovernmental transfers: the equitable share. In Restructuring Local Government Finance in Developing Countries: Lessons from South Africa, Roy Bahl and P. Smoke, eds., Cheltenham, U.K.: Edward Elgar. Rubenstein, R. 2000. Provision of an equitable public school finance structure in Georgia. Fiscal Research Program Report No. 43. Andrew Young School of Policy Studies, Georgia State University, Atlanta, GA. Shah, A. 1994. A fiscal needs approach to equalization transfer in a decentralized federation. World Bank Policy Research Working Paper No 1289, World Bank, Washington, D.C. Tannenwald, R. 1998. Come the devolution, Will states be able to respond? New England Economic Review, MayIJune: 53-73. Tannenwald, R. 2002. Interstate fiscal disparity in 1997. New England Economic Review, 1733. Uganda Local Government Finance Commission LGFC. 2003. Allocation Principles, Formulae, Modalities and Flow of Central Government Transfers. Report, Kampala: LGFC. United States Advisory Commission on Intergovernmental Relations. 1986. Measuring State Fiscal Capacity: Alternative Methods and their Uses, Information Report M-150, Washington DC: ACIR United States Advisory Commission on Intergovernmental Relations. 1988. State Fiscal Capacity and Effort, Information Report M-170, Washington DC: ACIR. U.S. Advisoty Commission on Intergovernmental relations. 1993. State Fiscal Capacity and tax Effort--1 991. Washington D.C.: U.S. Government Printing Office. U.S. Census Bureau. 1957. Census of Governments 1957. Washington D.C.: U.S. Government Printing Office. U.S. Census Bureau. 1960. Census of Population and Housing 1960. Washington D.C.: U.S. Government Printing Office. U.S. Census Bureau. 1962. Census of Governments 1962. Washington D.C.: U.S. Government Printing Office. U.S. Census Bureau, Housing and Household Economic Statistics Division. 2005. Table 1. Weighted Average Poverty Thresholds for Families of Specified Size, 1959 to 2003. Annual Social and Economic Supplements. <www.census.gov> University of Georgia (1963), Georgia Statistical Abstract 1963, Graduate School of Business Administration. Division of Research. Athens, GA: University of Georgia. Vaillancourt, F. 2001. Simulating inter-governmental equalization transfers with imperfect data. Paper presented at the November 2001 meetings of the National Tax Association, Baltimore. World Bank. 1997. Bangladesh Municipal Finance Management Sector Study, Washington: The World Bank. Zhang, Z., and J. Martinez-Vazquez. 2002. The system of equalization transfers in China. Ministry of Finance, China and International Studies Program Working Paper, Andrew Young School of Policy Studies, Georgia State University, Atlanta, GA.
Chapter 11
A MODEL FOR PUBLIC INFRASTRUCTURE EQUALIZATION IN TRANSITIONAL ECONOMIES' SOPHIA LEVTCHENKOVA AND JEFFREY D. PETCHEY School of Economics and Finance, Curtin University, Australia
1.
INTRODUCTION
Many transitional economies have a shortage of infrastructure for the provision of public services (e.g. health, education) that are important in developing human capital and increasing the rate of economic growth. Moreover, the regional distribution of the available public capital for the provision of these services is often uneven, implying that citizens also have inequitable access to the public services that are available. A vivid example is South Afiica, a country making the transition from a minority rule political system with substantial government intervention in the economy to one that is more reliant on markets, democracy and democratic institutions. South Africa has an overall shortage of public infrastructure for services such as education and health and, in particular, an extreme variation in the distribution of the available infrastructure across regions2. Service standards are high in some areas while in others there is virtually no infrastructure at all. This shortage of infrastructure and its regionally disparate distribution is known in South Africa as the 'capital backlog' problem and implementing policies to alleviate the problem is a priority of the national government for both economic and social stability reasons. As in South Africa, national governments in other transitional economies may also have access to some supplementary funds for additional spending on infrastructure. These funds might be provided by donor countries, international aid institutions or from their own budgetary resources. However, even if some additional resources are available, policy-makers face two other major issues. The first is how they should allocate the pool of
346
Levtchenkova and Petchey
supplementary funds to each service (e.g. education or health) and each region in a way that addresses the inequity between regions and at the same time raises the overall level of public infrastructure. The second is whether the pool of supplementary funds available is sufficient to meet the policy goal of substantially reducing capital backlogs over some realistic period of time. This is an especially important question in South Africa. In this paper, we construct a model that can be used as a simulation tool by policy makers to help them answer these questions. Specifically, we suppose that there is some pool of h d s provided by the national government for additional expenditure on public infrastructure. The h d s are available over some specified period of time and, in the simulations conducted for South Africa later in the paper, this period is taken to be ten years. The model then constructs infrastructure estimates over this period for each service of interest and each region on the assumption that current levels of capital expenditure continue. The next step is to construct estimates of the desired level of infrastructure over the same period using international benchmarks. For most transitional economies, South Africa included, the desired level of infrastructure lies above the actual (the difference being a capital backlog). We then show that over time it is possible to increase the level of regional and service specific infrastructure available until it is equal to the regional and service specific desired level. During this process, some degree of equalization in public capital stocks is also achieved. Indeed, this is assumed to be the ultimate aim of the policy-makers and, therefore, the model developed for their use. We call the expenditure track that would be followed in doing this an ideal or initial transition path. It represents the path that could be followed by the stock of infrastructure given sufficiently large additional capital expenditures. Of course, if the pool is big enough the transition from what is available now to the desired capital stock could be achieved more quickly. The initial transition path, constructed mathematically within the simulation model, allows us to calculate service and region specific 'capital needs' in each year of the transition process. The need is the difference between the stock of capital associated with the ideal transition path and the actual stock (the one that would apply without any extra expenditures). Generally, these needs differ across regions (for the same service) because each region starts with a different capital backlog inherited fiom the past (in South Africa the differences are large). The capital needs calculated in this way are then used to estimate region and service specific 'shares' of the available pool of supplementary knds in each year. These shares are then applied to the available pool to determine the region, service and year-specific allocation of the pool. Since, in general, the pool available is not sufficient to fill the calculated needs in each
A Model for Public Infrastructure Equalization
347
year, the actual process of transition falls below the ideal transition path. For this reason, the model also produces an 'actual transition path'. This is the track that is actually followed by the level of infrastructure under the scheme of supplementary expenditures. As noted, the actual transition path lies below the ideal one but above the level of infrastructure that would prevail in the absence of intervention. In this case (the one that is likely to apply in transitional economies), by the end of the period of supplementary expenditures, capital backlogs still persist but they are not as large as they would be in the absence of the scheme. Moreover, the regional inequities in levels of infrastructure will have been partly 'equalized'3. Of course, as discussed, if the pool of funds is sufficient to allow the actual transition path to follow the initially specified ideal path, the backlogs will have been eliminated and full inter-regional equalization achieved by the end of the program period. Before progressing with development of the model, some other points require additional comment. The first is that the contribution of the paper is a practical one, namely, to provide policy-makers in transitional economies with a practical and easy-to-use tool to assist them in making decisions over the regional and service-specific allocation of scarce national funds to meet basic public infrastructure needs for services such as education and health (important in creating human capital). Second, the pool of funds available for supplementary public capital expenditures is assumed throughout our analysis to be determined outside of the model. We concentrate on the issue of how to allocate a given pool among different regions and services and the consequences (for backlogs) of allocating pools of different sizes. Nevertheless, determining how large the pool should be is an important matter and in many countries this needs to be answered by macroeconomic forecasting models. The forecast of the available pool would then be an (exogenous) input into our model that determines the regional and servicespecific allocation. Even so, the model is of considerable use in providing information on what the pool of funds ought to be since, within it, we calculate the total need that must be filled in each period if capital backlogs are to be eliminated (or substantially reduced) over some period of time. Thus, the model answers the question of whether the resources available to a particular economy are sufficient to address its public capital backlog problems or whether the backlogs are too large to be handled within a time that is considered acceptable. As will be seen, this is important information that comes from the simulations for South Africa. Finally, we think of the model being applied in centralized (unitary) countries with distinct regions where the national government provides the major public infrastructure directly. Alternatively, the model might be applied in a decentralized economy where the regions have sub-national
348
Levtchenkova and Petchey
governments that are simply spending agents of the central government. This is the case in South Africa where, although the Regions have constitutionally mandated spending responsibility for education, health, welfare, transport and housing, they have virtually no taxation powers, no access to capital markets and receive all of their funding as a grant from the national government. In this case, one can think of the spending as being nationally determined even though it is channeled through 'notional' provincial governments. It simply means that the spending is not undertaken directly by the center but is undertaken indirectly via conditional grants to the regions. There is one last possibility, that of a federation with independent subnational governments. In this case, there is an added complexity - regions may have governments that undertake their own (largely independent) spending on public infrastructure. In this case, the center can still allocate conditional grants to the sub-national governments, but there is the problem of modeling the latter's response to these grants before the net impact of the central spending on public infrastructure formation can be assessed. This is something that our approach does not do and we do see the model being applied in its current form in federations. The outline of the paper is as follows. Section 2 develops the model formally by deriving the main formulae of the model. Section 3 applies the model to South Africa and presents the results of various simulations. Also provided is a brief discussion of the input data constructed for the simulation. Conclusions are presented in Section 4.
2.
THEORY
Consider an economy with a national government and i = 1,.....,I heterogeneous regions that might represent economic areas, zones containing a particular ethnic group or a political jurisdiction such as a Province or State. The national government is assumed to be directly responsible for providing s = 1,......,S public services, such as health, education, welfare and public transport4. Suppose also that the national government wishes to increase the stock of service and region-specific public infrastructure used in providing these services (e.g. schools, roads, hospitals) over some period of time beginning in period z = t (this might be the current period) and ending in some future period T = t + n . For example, if n is equal to 10, decision-makers may look to design a program to raise the level of infi-astructure over a ten year period. To achieve this, we assume that the national government has access to some pool of funds, denoted by P, ,that is to be used to achieve its goals. The problem faced by the center is
A Modelfor Public Infrastructure Equalization
349
how to allocate the pool of funds among regions and services in a way that takes account of the different levels of public capital stock inherited from the past and different levels of on-going public capital expenditure. We do not analyze how the pool is funded by the national government (this could be from borrowings, tax revenue or foreign aid). The first step in constructing a model that will give policy makers the required information is to develop some notion of a 'desired' region and service-specific capital stock that is ideally to be achieved by the end of the implementation period, z = t + n . In practical applications, such estimates could be obtained from international comparisons. For example, in the application of the model to South Africa in Section 3 we use estimates of actual stocks of public capital for Australia to construct estimates of desired region and service specific public capital stocks for South Africa. This idea can be developed more formally if we denote K:+, as the region and service-specific desired stock of public capital that is to be achieved by the end of the period over which the policy is to operate, z = t + n . It proves convenient to construct this estimate using the Perpetual Inventory Method (PIM), a commonly used method for constructing capital stock estimates (see Holtz-Eakin (1991)). The difference between our work and previous applications of this method is that we apply it to create future estimates of capital stocks rather than generate stock estimates from some previous period to the present. Our application of the approach takes an estimate of the desired stock of capital in the first period of the program,
K*,"
i,t and then uses estimates of desired net capital expenditure and depreciation to construct an estimate of the desired capital stock for each ensuing period. Thus, we are able to construct an estimate of the desired service and region specific stock of public capital from period T = t to period z = t + n (the final year). Formally, we define the desired stock of public capital for service s in region i at the end of the program as 3
The first term on the right hand side, K:;:
is the (real) desired capital
stock at time z = t (the first period of implementation), 6 is the rate of depreciation (which we suppose for convenience to be time, service and region invariant) and k:;: is a (real) desired net capital expenditure forecast for each period, service and region5. The set up assumes that an estimate of the value of ICY;: is available at the beginning of period t and thus its
350
Levtchenkova and Petchey
depreciation starts during that period and continues through all the subsequent periods. The estimate for k:;: relates to the end of each period z and therefore starts depreciating from period t onwards. The next step in developing the model is to estimate the actual level of service and region-specific capital in each period of the program. Generally, this will deviate from the desired level with some regions being above and others below. Using the PIM, the actual capital stock at time .c = t + n is given by
where K:,,,,
is the (real) actual capital stock at time t
+ n for service s,
K;,, is the (real) actual capital stock at time t for service s (this is an estimate of the actual stock of capital at the start of the scheme) and k:,, is the forecast flow of service and region-specific nationally-mandated capital expenditures between t and t + n (in the absence of the supplementary expenditures). For any particular region and service, the desired public capital stock is below or above the actual stock in any period. For transitional economies it is more likely that the desired stock exceeds the actual stock implying a 'capital stock backlog' (or deficit) in each period. This situation is illustrated in Fig. 1 where the line ac represents the desired capital stock from the first period of the program to the final period and bd is the actual capital stock over the same period. The capital backlog in the first period z = t is the distance ab while the backlog at the completion of the period of interest is cd. Note that, as drawn, the backlog increases over the period, though this need not be the case. Whether the gap increases or decreases depends on actual net capital expenditure in each period relative to desired net capital expenditure (with common depreciation)6. We are now in a position to develop the remaining mathematics of the model. Recall from the earlier discussion that the aim of the national government is to allocate a pool of h d s to each service and region so as to raise the actual stock of capital to the desired region and service-specific level by the period z = t + n . In other words, we want to supplement existing capital spending by the center so that the actual capital stock follows the line analogous to bc in Fig. 1 rather than bd. The slope of this line does not have to be constant. Any line that starts at b, ends at c and lies between bd and ac is acceptable. For simplicity, we assume it to be linear.
A Modelfor Public Infrastructure Equalization
2=t
r=t+j
r=t+n
Time
Figure I. Desired and Actual Capital Stocks for Service s in Region i The following discussion concentrates on developing this aspect of the model. It requires that a number of steps be taken. The first one is to develop an expression that shows the extent to which existing capital expenditures would need to be supplemented over the program period in order to ensure that the actual and desired capital stocks are the same by z =t +n. This expression is then used to show the supplementary expenditures required in a representative period 2 = t + j where 15 j 5 n - 1 The supplementary expenditure requirements for this (see Fig. 1)'. representative period j are then used as the basis for the firther development of the model. In this regard, suppose that there is some annual flow of supplementary net capital expenditures by the center in region i for service s between t and t + n. As noted earlier, these supplementary expenditures are to be provided from the pool of central funds. The expenditures are added to the existing actual net capital expenditures for service s in the region in question and are denoted as n;, . With the supplementary expenditures included, equation (2), which describes the actual capital stock in a given period, becomes
352
Levtchenkova and Petchey
The (depreciated) sum of supplementary expenditures that is required from time t to t + n to ensure that the capital stock in region i for service s is equal to the desired stock by year t + n is found by equating expressions (3) and (1). With rearrangement, this yields
The left hand side of (4) is the total flow of supplementary expenditures required to implement the transition along a path similar to bc (Fig. 1). The first term on the right hand side, B: = (K:;: - K t t ) , is the capital stock backlog inherited in period one and depreciated throughout the program. t+n-1
The term E: =
(k:
- k:,)(l-
6)ttn-1-7 is the (depreciated) difference
T=t
between desired and actual net capital expenditures. Thus, the right hand side of (4) has two components: one related to the capital backlog and the other to the deficiency in on-going capital expenditures. The total flow of supplementary expenditures required to achieve equality between actual and desired stocks is the sum of the two. To proceed, we must expand the lefi hand side of (4) and express it in terms of the supplementary expenditure required in a representative period j8
This expression is a key one and requires careful explanation and interpretation. As noted, the term on the left hand side, nf,t+j-l,is the supplementary expenditure required in year j. For convenience, from now on we call this a 'capital need' for period j. As explained earlier, the first term on the right hand side is the inherited capital backlog, and Eq is the depreciated difference between the actual and desired expenditures. t+n-1
The term
qt+j-l =
n:, (1- 6)"'-'-'
is the total depreciated sum of
r=t+j
future needs yet to be hnded fiom period j to the end of the program. The presence of this term reveals an interesting feature of the model, namely, that the capital need in a representative period j depends on the sequence of future needs. The logic of this can be seen by noting that since the goal is set at K;:+, , the greater the need to be met later in the program, the lower
A Model for Public Infrastructure Equalization
353
the need to be filled now, and vice versa. The model requires that the sequence of future needs be defined at the start. Indeed, the sequence actually defines the line bc in Fig. 1 and we think of it as the 'initial transition path'. The term 'initial' is used because it represents the path that we initially aim to follow in order to achieve the desired outcome. As shown later in the simulation for South Africa, depending on the size of the pool of funds available, pr , the actual process of transition may not follow this originally defined path. Instead, an 'actual transition path' will emerge as a result of implementing the program (this path is distinct from the initial transition path). Of course, if the pool of funds is sufficient in each period to meet all of the assessed needs, then the actual and initially specified transition paths coincide. But in practice the pool of funds is insufficient and the actual process of transition lies below (this is the case for South Africa). Nevertheless, to solve the model for needs in each period one must specify the initial transition path. The last term in (5) is the sum of previously distributed supplementary expenditures where,
g;t+r-l =
ntt+r-l Pt+r-1. This appears in the equation C n L l i,s
because the capital need for period j is dependent on the needs already filled by the policy program in previous periods. Before commencement of the program, this term is zero since no supplementary expenditures have been made. The following discussion shows how the g terms are calculated. In order to simplify we demonstrate the calculation for year 1 of implementation. The estimate of g for each successive period is done in a similar fashion. In this regard, we define the total capital expenditure need for region i at time z = t as the total need for each service, calculated in (5) above, summed across all services. The total need for all regions at time z = t is the need for each region summed over all regions. The total actual supplementary expenditureprovided to region i in year z = t is simply
Equation (6) defines the share of the total pool of funds available for the program and allocated to region i (where the pool 6 is assumed to be exogenous). The size of 6 is determined by fiscal and political constraints. If region i has relatively high capital needs, perhaps because it has relatively large backlogs or relatively low on-going actual capital expenditures, it will receive a relatively larger allocation from the pool. The scheme therefore
Levtchenkova and Petchey
354
directs scarce national funds available for supplementary spending on capital to the most needy regions and services. The actual allocation for service s in region i at time t (the term used on the right hand side of (5)) can be written as
The approach is one that calculates the capital need for each service, province and time period from the 'bottom up' using (5). These needs are then used to construct shares that allocate each dollar from the given pool P, to the most needy services and provinces. The magnitude of the allocation that flows to each service and province is scaled up or down depending on national fiscal constraints and the size of the pool. Thus, the scheme can be designed independently of the policies that determine the size of the grant pool since its primary purpose is to calculate the shares of the pool that are to be allocated to each region and service9. In any period j, if the previous g's have been equal to the assessed needs then the actual transition path is the same as the initially specified one, that is, the actual transition follows the path bc in Fig. 1. However, if the previous g's are insufficient, the actual transition path is below bc. This implies that equality between actual and desired capital stocks is not achieved by the end of the program period, t+n. In this case (the one that applies in South Africa for instance), the program would have to be extended to ensure completion. The actual capital stock accumulated by the (initially envisaged) end of the program period, t+n, will then be defined by
As can be seen from the expression above, since the supplementary funds distributed are insufficient when compared to calculated needs, the final capital stock will fall short of the desired amount. More generally, the capital stock that emerges in each period of the program is t+j-1
K;,+j = K:, (1 - 6)' +
(kf1,T + gy1.T )(I - 6)t+J-1-T
(9)
This expression traces out the actual transition path that is followed under the program. Note, once again, that if the distributed grants equal the needs, (9) becomes (3): the actual and initially specified transition paths coincide. However, for most transition economies it is likely that the pool of funds available to raise capital formation at the regional level will be insufficient.
A Model for Public Infrastructure Equalization
355
As will be shown below, this is the case for South Africa under reasonable assumptions about the size of the pool.
3.
SIMULATIONS FOR SOUTH AFRICA
Having developed the key expressions for the model it is now possible to construct a computer-based simulation program that can be applied to any transitional economy for which there is appropriate data, to produce estimates of how one would allocate supplementary capital expenditures to heterogeneous regions. The Financial and Fiscal Commission (FFC), the body with constitutional authority for inter-governmental matters in South Africa, agreed to assist with the construction of an input data set. The simulation model considers all of the nine South African provinces: Western Cape, Eastern Cape, Northern Cape, Free State, Kwazulu-Natal, North West, Gauteng, Mpurnalanga, and Northern Province. The services included are education, health and welfare. The simulation is based on the scheme being implemented in 2002 with a transition period of 10 years. Construction of the data set proceeds in three basic stages that can best be appreciated by referring back to Fig. 1. From there it can be seen that to construct the line bd in Fig.1 requires estimates of time, service and province-specific actual capital expenditures from 2002 to 201 1 (those that would be undertaken by the center without the supplementary funding). These were obtained from official Reserve Bank of South Africa capital expenditure forecasts, extrapolated to the year 201 1. Also required is an estimate of the initial capital stock. This we obtained by taking Reserve Bank estimates of aggregate capital stocks and applying capital expenditure weights to derive the service and province-specific capital stock estimates for the first year, 2002. An estimate of the public capital depreciation rate was obtained from Levtchenkova and Petchey (2000) for Australia. The stocks then are obtained using the PIM. Constructing data for the line ac in Fig.1, the one representing the desired capital stock, proceeded as follows. First, we used per capita estimates of service specific capital stocks for Australia to construct estimates of the respective service and province-specific desired capital stock for South Africa for the year 2002. We then constructed estimates of desired capital expenditures for each service and province for the period 2002 to 201 1 based on Australian expenditure data. Applying the depreciation rate used for the construction of the actual series, we were then able to construct service and province-specific estimates of the desired capital stock for each year from 2002 to 2012. The initial estimate for the pool of funds that might conceivably be made available by the South African Government to fund supplementary capital
356
Levtchenkova and Petchey
spending on public infrastructure for health, education and welfare is R5 billion for 2002. This is the equivalent of about $US0.580 billion, accounting for just 0.5 percent of South Africa's GDP in that year. The FFC estimate of the size of the pool for 2002 was then extrapolated for the remaining periods of the scheme. The theoretical model described in Section 2 was then programmed within Excel. Combined with the constructed input data set, this enabled us to calculate for South Africa (for each service and province), the initial backlogs, the initially specified and actual transition paths, and the degree to which the goal of equality between actual and desired capital stocks is met by the end of the policy program. These outputs of the model can best be summarized diagrammatically. Before doing this, however, it is useful to examine the size of the capital backlogs at the beginning and end of the scheme. The comparison is interesting since it gives us a snapshot of how successful the policy is in achieving its aim of eliminating the capital backlogs, for a given pool of b d s (Table 1). The data indicate that for all provinces and services the backlog actually increases over the program period. This is because the pool of R5 billion allocated as supplementary expenditure in the first year (and escalated by 10 percent in the following years) is inadequate to achieve the aims of the policy. The pool is insufficient to redress the backlogs inherited at the start of the program and does not even make up for the on-going gap between actual and desired capital expenditures (taking into account depreciation). Nevertheless, the backlogs emerging after 10 years of the policy would be even larger without the program. This raises an important issue for South Africa. If the FFC believes that a pool of R5 billion is the one that is feasible within existing resources, then our simulations suggest that the country will not be able to achieve any satisfactory reduction in its backlogs, in terms of the overall magnitude or regional disparities, even over a ten year period. It is also interesting to examine the shares of the pool that would be allocated to each region (summed across the three services) from 2002 to 201 1. This gives us valuable information on how the scheme ranks the regions of South Africa in terms of public capital need (Table 2). We can see from the shares that the scheme identifies Kwazulu-Natal as the most 'needy' province since it is allocated the highest share in every year. The richest province is found to be Northern Cape. Also, notice that the relative ranking of provinces according to the calculated shares does not change much over time. This is because of the problem noted above that the size of the pool of funds available is so small that it does not achieve any notable results in moving the regions towards the desired capital stock or changing their relative ranking (i.e. equalizing citizens' access to public capital on a regional basis).
A Model for Public Infrastructure Equalization Table 1. Initial and Final Backlogs (Rand million) Province WC EC NC FS KN NW GT MP NP
Education Initial After 10 Backlog years 11117 18384 15191 25423 2612 4135 7756 12790 23136 38871 8008 14538 18194 31632 6698 12082 15619 24515 108331
182370
Health Initial After 10 Backlog years 8140 19954 14428 32156 1731 4129 5787 13555 3 143 28368 6940 17377 7354 25613 3831 12733 8454 2297 1 59808
176856
Welfare Initial After 10 Backlog years 514 2207 1217 3768 106 432 44 1 1508 1177 4603 712 2079 1599 4540 475 1561 942 2924 7183
23622
Notes: The abbreviations for the names of provinces are: WC - Western Cape, EC
- Eastern Cape, NC - Northern Cape, FS - Free State, KN - Kwazulu-Natal,NW North West, GT - Gauteng, MP - Mpumalanga, NP - Northern Province. We now provide a more comprehensive presentation of the output data of the model. This is done in diagrammatic form for one particular region in South Africa, the Province of Gauteng, and one service, educationlo. This region is selected because it includes the city of Johannesburg and cities such as Soweto, the largest population concentrations in the country. In Fig. 2 below we plot the desired capital stock, the actual capital stock in the absence of the program, the initially specified transition path and the actual transition path. Again, the national pool of funds is R5 billion in 2002 and grows by 10 percent per year over the ten year program. As indicated above, the actual transition path lies below the initially specified path'1. Because at each stage of the program the pool is inadequate to fill the previously identified capital needs, the gap between the desired and actual capital stock with supplementary expenditures continues to widen over the policy period (though less so than it would without the policy). As before, the simulation shows that a pool of R5 billion is insufficient to yield any reasonable degree of transition by the end of the program. With this level of resources, it would require additional implementations of the scheme with an increased pool size to eliminate the backlogs. Completing the transition to a high income economy with a level of public service provision and equality of access found in high income economies would take a generation (e.g., thirty to forty years).
Levtchenkova and Petchey
358
Alternatively, policy-makers might increase the size of the pool to raise the actual transition path. Using the simulation model, it is a simple matter to find the pool of funds required to achieve a greater degree of transition. The ability to do this is of considerable importance in transitional economies, South Africa included, where policy-makers often have no information on the amount of resources they would need to reduce public capital backlogs. In Fig. 3, we present a plot, once again for education in Gauteng, where by the end of the program period just over half of the backlog is eliminated. Achieving this result requires a pool of R20 billion in 2002, four times the amount that the FFC expected would be made available at the start of the scheme. This is the equivalent of about $US2.3 billion for 2002 or 2 percent of South Africa's present GDP.
CONCLUSION In this paper we have developed a model that would allow policy-makers in transitional economies to allocate supplementary capital expenditures on public infrastructure in such a way that raises the overall level of infrastructure and provides greater equality in its regional distribution, that is, equalizes to some extent. The proposed model has important features, most notably its ability to take account of the inter-temporal nature of capital arising from its durability and an ability to provide a flow of services over many years. This means that the model must take account of past capital expenditures and hence the capital stock built in previous years. The model must also take account of what is to be provided in the future (the transition path) while allowing for the fact that capital depreciates over time. Without accounting for this interdependence within the formal structure of the model, it would be impossible to follow any meaningful transition path for the regions of the economy in question. The paper also showed how the model can be made operational if one has an input data set for a particular economy. This was demonstrated by applying the model to the Republic of South Africa and providing selected results from simulations for that country. The simulations show that, given the magnitude of the capital backlogs in South Africa, and deficiencies in on-going capital expenditure, as well as the size of the pool currently proposed (0.5 percent of GDP), little progress would be made in raising the level of public infrastructure or in improving its regional distribution (that is, only limited equalization would be achieved). We then showed that if South African policy-makers could increase the pool to 2 percent of GDP, about half of the present backlogs would be eliminated by the year 201 1 and a much greater degree of equalization achieved in the availability of public infrastructure, and hence the services provided using that infrastructure.
A Model for Public Infrastructure Equalization
Notes The background research for this paper was undertaken under the auspices of an Australian Research Council (ARC) Large Grant. The authors would like to thank the ARC for its support. In South Africa's case, this is a result of decisions made by previous Apartheid regimes. In this sense, we can think of the model as a public capital equalization scheme since it aims to reduce disparities in citizen access to basic public services important in the creation of human capital. In the application to South Africa this is reasonable. Although, technically, the provinces are responsible for these services, in practice they have virtually no tax powers and no access to capital markets (borrowings) and are essentially spending agents of the national government. As noted, equation (1) is an application of the PIM to future periods. Note also that: (i) the depreciation rate can be made variant with respect to time period, province and service without changing the qualitative results; and (ii) time is treated as a discrete variable and the capital stock undergoes a change once within a period of time (each year). Thus, time takes on only integer values. For convenience the desired and actual capital stocks are assumed to be linear functions of time but in general this is not the case. j includes the first and next to last period of the policy program. In deriving ( 9 , we assume the program of supplementary capital expenditures starts in time period t. Therefore, t is year 1 of the program. Time period t+j coincides with year j of the program. Nevertheless, as mentioned briefly in the Introduction, the model also provides important policy information on the magnitude of the capital needs of a particular country in any given year that can be used to inform any debate over the size of the pool. This information is obtained by summing all needs, as calculated from (5), across regions and services. The resulting number is a measure of the total capital need in period T for the economy in question (it is likely that the number far exceeds the resources available for spending on capital). In this sense, therefore, the scheme gives information on what funds a country would need to allocate to public capital, in each year, in order to correct for the shortage of public capital. The simulation model can produce an analogous plot for each service and Province in the country. As discussed in Section 2, we assume that the initial transition path is linear.
References Financial and Fiscal Commission. 2000. Preliminary recommendations for 2001: A costed norms approach for the division of revenue consultation document. Ter-Minassian, Teresa. (Ed.) 1997. Fiscal Federalism in Theory and Practice. Washington, D.C.: International Monetary Fund. Holtz-Eakin, D. 1993. Province-specific estimates of province and local government capital. Provincial Science and Urban Economics, 23: 185-209. Levtchenkova Sophia. and Jeffrey Petchey, 2000. Regional capital stock data for Australia. The Australian Economic Review, 33(2): 193-7.
DISCUSSANT COMMENTS
SERDAR YILMAZ World Bank Institute
BIRDAND VAILLANCOURT, MARTINEZ -VAZQUEZ AND BOEX, AND LEVTCHENKOVA AND PETCHEY
1.
INTRODUCTION
Intergovernmental transfers, as essential components of intergovernmental fiscal arrangements, are extensively studied in the publicfinance literature. However, no matter how much guidance the theory provides on the objectives and design of an intergovernmental transfer system1, the implementation of any proposed system is a difficult task, especially when the system strives to achieve equalization. The chapters in this section give us a broad view of the extent and nature of the difficulties in implementing an equalization transfer system. They offer practical tools and different approaches to mitigate potential stumbling blocks in designing an equalization system. Richard Bird and Francois Vaillancourt examine the theoretical case for equalization and explore the extent to which differences in expenditure needs across different localities can be incorporated into the equalization transfer systems. The authors describe theoretical underpinning of equalization as "to enable SNG [subnational governments] with different abilities to raise revenues to provide comparable levels of services with comparable levels of own-taxes." However, the authors question the applicability of such an horizontal equity objective in countries with heterogeneous regions, particularly in federal countries. They argue that the theoretical literature on horizontal equity, as illustrated in the concept of net fiscal benefits (NFB), makes an unattainable premise: "...not only are governments at all levels benevolent but also the correct standard for judging decentralization is how closely it replicates the outcomes of a (benevolent)
unitary state." Equalizing NFBs across subnational governments requires that the system design takes into account expenditure differentials. The authors' review of various equalizations systems from selected countries show that expenditure need and cost differences are rarely incorporated into the intergovernmental transfer system. This outcome, according to the authors, is not surprising-- because of the difficulties associated with data collection, the costs of introducing expenditure needs into the equalization transfer system are not worth additional equalization benefits. This is the central argument of the chapter, and it has important implications to developing country policy makers who are advised by international consultants to design an equalization system by estimating the difference between expenditure needs and revenue potential. The next chapter by Jorge Martinez-Vazquez and Jameson Boex takes on the challenge of designing an equalization transfer system with scarce data. Data scarcity is particularly important for developing countries, where fiscal decentralization can be an important tool in achieving growth and development. Most developing counties lack data on fiscal, demographic, and socio-economic variables, therefore it is difficult to quantify local fiscal gap, the difference between expenditure needs and revenue capacity. In this chapter, the authors provide an overview of different approaches to measure expenditure needs and revenue capacity and conclude that the best measures of fiscal gap are not necessarily data-intensive. In the third chapter, Sofia Levtchenkova and Jeffrey Petchey provide an analytical tool to estimate hnds needed for capital investments to be used to address inequalities in the provision of public services across regions. The authors estimate the total amount for the distributable capital pool to redress vertical imbalance and allocate scarce resources on public infrastructure in such a way that takes regional inequalities into consideration. Their work provides a valuable tool for one of the most controversial issues in intergovernmental transfer system process: estimation of the size of the total pool to be distributed under any equalization scheme. In what follows, I offer a brief review of the three issues that are central to these three chapters: estimating fiscal gap, data needs and availability, and estimation tools.
2.
ESTIMATING THE FISCAL GAP
Bird and Vaillancourt look at the "expenditure needs approach" to intergovernmental transfer design from both theoretical and practical perspectives and offer examples from a selected set of countries. The chapter concludes that while "[a] comprehensive equalization scheme should, it seems, clearly include both fiscal capacity and expenditure
Discussant Comments
365
differential measures" nevertheless "...it is unlikely to be worth the cost of introducing a parallel RES [representative expenditure system] system in most countries." The authors argue that incorporating expenditure needs into a general equalization system makes the system complicated and incomprehensible. Instead, they favor designing a special transfer system to address needs and costs differentials across regions. An important objective of an intergovernmental transfer system design is minimizing horizontal imbalances. A horizontal imbalance occurs when subnational governments face different revenue capacities and cost structures for carrying out a similar service financed through a similar revenue source. Horizontal imbalances may arise because subnational governments face (i) an unequal distribution of revenue bases, natural resources, and wealth; (ii) variations in the socio-economic characteristics of population, and (iii) differences in the geography and climate across jurisdictions that lead to disparities in economic opportunities (ACIR, 1990b). A well-functioning intergovernmental system should provide all necessary public services to all citizens at comparable tax rates-that is to say, people in different parts of a country should have equal access to public services and the tax burden they are asked to bear should be uniform across jurisdictions. Moreover, a well hctioning transfer system should allow the country to reap the advantages of a decentralized delivery of public services, while, ensuring that the design of public services conform to general notions of efficiency and equity. In addressing the horizontal imbalance issue, there is a need to estimate the fiscal gap across subnational governments. The fiscal gap of a region is the difference between the cost of providing a standardized basket of public goods and services and the revenues the subnational government can raise from its own sources to finance this cost. Intergovernmental transfers should focus on eliminating or reducing the fiscal gap of subnational governments. Therefore measuring fiscal gap is essential in addressing interregional disparities and implementation of an equalization program. Measuring fiscal gap, however, is not an easy task. There are two major constrains: (i) The lack of reliable data related to both the expenditures and revenues of subnational governments is the most important technical constraint in estimating fiscal gap (Yilmaz, Hegedus and Bell, 2003). (ii) Measuring fiscal gap requires developing a methodology that captures not only revenue capacity but also expenditures needs of local governments. Therefore, in analyzing expenditure needs, it is very important to establish expenditure norms (Alm and Martinez-Vazquez 2002). One good example of a fiscal capacity methodology that attempts to incorporate expenditure side of the public finances into the fiscal gap concept is the United States Advisory Commission on Intergovernmental Fiscal Relations' work on representative expenditures (ACIR, 1962, 1986, 199Oa, 199Ob).
366
Yilmaz
The second paper in this section by Martinez-Vazquez and Boex provides a brief summary of the main issues in the estimation of fiscal gap and discusses different ways of measuring expenditure needs and revenue capacity.
DATA NEEDS AND AVAILABILITY A common theme across the three chapters is the lack of data needed to design an equalization system or to analyze the impact of the existing system across subnational jurisdiction. Data availability is an important stumbling block in the path to decentralization in developing countries, where fiscal decentralization may be central to improving public service delivery and quality. As the decision makers design a fiscally decentralized system, they must have adequate information so that they can allocate resources most effectively. Without proper information, intergovernmental fiscal relations will necessarily involve conflicts between the different levels of governments. Data availability is a critical issue because it brings transparency and credibility to public sector reforms in most countries. For instance, in most countries the transfers are negotiated and the unreliability of the system is influenced by the soft budget constraints. Some countries are moving away from such a system and toward a system of formula-based normative grants. However, formula-based grant systems can easily be manipulated if the system cannot be fed with reliable, timely data.
4.
TOOLS FOR ESTIMATION
An important question in designing an equalization system is the estimation of resources needed to address the vertical imbalance. Elimination of vertical imbalance requires information about the size of the total distributable pool, which in return requires a methodology for estimation. In most countries, the distributable pool is determined without taking into account the magnitude of the subnational governments' financing needs. Levtchenkova and Petchey develop an analytical tool that can be used in many countries for the estimation of financing needs on capital infrastructure. Unless policy makers use these kinds of analytical tools in designing a transfer system it is quite impossible to have a stable transfer system that promotes good planning and efficient service delivery effort. Consequently, recipient governments will neither have good budgeting practice nor will they face an appropriate hard budget constraint.
Discussant Comments
Notes 1.
Theoretically, intergovernmental transfers, which compose of a significant portion of subnational governments' revenues in all countries, have three principal objectives: (i) to equalize vertically (improve revenue adequacy); (ii) to equalize horizontally (interjurisdictional redistribution) (iii) to minimize inter-jurisdictional spillovers (externalities).
References Alm, James, and Jorge Martinez-Vazquez. 2002. On the use of budgetary norms as a tool for fiscal management. International Studies Program Working P a p a No. 02-15. Andrew Young School of Policy Studies, Georgia State University, Atlanta, GA. United States Advisory Commission on Intergovernmental Fiscal Relations.1962. Measures of State and Local Fiscal Capacity and Tax Effort. Washington, DC: ACIR. United States Advisory Commission on Intergovernmental Fiscal Relations. 1986. Measuring state fiscal capacity: Alternative methods and their uses, In Information Report M-150. Washington, DC: ACIR. United States Advisory Commission on Intergovernmental Fiscal Relations. 1990a. State fiscal capacity and effort. In Information Report M-I 70. Washington, DC: ACIR. United States Advisory Commission on Intergovernmental Fiscal Relations. 1990b. Representative expenditures: Addressing the neglected dimension of fiscal capacity, In Information Report M-174. Washington, DC: ACIR. Yilmaz, Serdar, Jozsef Hegedus and Michael E. Bell. 2003, Subnational Data Requirements for Fiscal Decentralization. Washington, DC: WBI.
THE RELATIONSHIP OF EQUALIZATION TO OTHER POLICIES
Chapter 12
REVENUE SHARING, NATURAL RESOURCES AND FISCAL EQUALIZATION BOB SEARLE Consultant on Fiscal Decentralization, Formerly the Commonwealth Grants Commission, Australian Government
1.
INTRODUCTION
As applied in Australia, horizontal fiscal equalization (HFE) is achieved when untied funds from the central government are distributed to the States so that: if each made the same effort to raise revenue from its own sources and operated at the same level of efficiency, each would have the capacity to provide services at the same standard'. Not all nations try to achieve this degree of equalization and often define it in a much more limited way. Some countries, like Canada, have decided that 'full' equalization is impossible to h d because of the great diversity of natural resource revenue capacities available to their provinces. Others, like Indonesia, have so much revenue sharing between the central and local government that the pool of funds available from which to achieve equalization is not large enough to overcome the impacts of these other policies. However it is defined, HFE is ultimately about equalizing (or at least moving towards the equalization of) the capacity of sub national governments to provide services. It usually takes account of differences in the capacity of the recipient governments to raise their own revenues, and is increasingly taking account of differences in levels of need for and unit costs of services provided by those governments. The direct objective is the equalizing of fiscal capacity, but the end objective is much more about moving to a more even distribution of (equalizing) service provision
372
Searle
standards throughout a nation. Governments, after all, are not established to raise revenues but to give order to society and to provide public goods. Equalization can be achieved at any level of service provision or quality. All jurisdictions at a level could be equalized through grant funding to a level that would enable them to provide a service that is below United Nations or other internationally acceptable standards and we would still have equality of service^.^ The situation in many developing countries is precisely this. Many such nations, however, see HFE as part of a mechanism to achieve a pre-defined level of services as well as an equalization in the provision of those services. Indonesia, for example, is currently investigating how to achieve what it terms 'minimum service standards' within its equalization based funding system. It is possible to combine these two policy objectives but it requires additional resources being made available to the recipient governments (or increased efficiencies being made by them) to fund the difference in costs between the current level of services and the required level of services. There are a number of ways in which one can view the conflicts within the allocation of funds across the public sector of a nation. Vertical fiscal imbalances (VFI) are often considered to be the primary concern, but it is often what is done to reduce VFI that produces the unexpected limitations on governments to achieve HFE and other objectives such as a national convergence in standards of living and socio-economic outcomes. The allocation of own-revenue sources (with sufficient autonomy) to the different levels of government is the primary way of handling vertical imbalances in fiscal capacity. It is relatively easy to measure the percentage of total public sector revenue collected by each level or sphere of government and thus know the level of vertical imbalan~e.~ This distribution of revenue sources is sometimes seen as so important, however, that it is considered in isolation from (or at least prior to) the allocation of functional responsibilities to those levels of government. It is when revenue sources are being allocated that the decision is made about what, if any, revenues will be shared and what level or levels of government will have access to revenues raised from natural resources. After all the taxes and other revenue sources have been allocated to a specific level of government, and the responsibilities for service delivery have also been decided, there is often a realization that the logical result of the individual decisions has left too large a vertical fiscal imbalance in favor of the central government. It has more funds than it needs to meet its responsibilities while the lower levels of government have insufficient to meet their needs. This situation results from a conflict between what might be best for overall accountability of sub national government (to give local government greater expenditure responsibilities) and what might be most efficient in raising national public sector revenue (to give the central
Revenue Sharing, Natural Resources, and Fiscal Equalization
373
government greater revenue raising capacities). Whatever the cause, the result is that to reduce VFI nations very frequently give sub national government the rights to revenue from natural resources and/or enter into revenue sharing between the central and sub national governments. These are both legitimate decisions of central governments in their attempts to reduce VFI, but are too often made in isolation of other policy objectives. Revenue sharing from natural resources may also be based on the demands of the producing sub national jurisdictions, which see the resources as their own (as opposed to a national resource) and/or on compensation for negative externalities such as pollution associated with the extraction of the resources. These issues are discussed in more detail later. This paper addresses the relationships the VFI-based decisions have to the equalization objective. Because of differences in the 'own-source' fiscal capacity of sub national governments, any injection of additional revenues into that sphere of government to reduce VFI is most likely to increase the levels of horizontal fiscal imbalance, thus increasing the pool of funds necessary to achieve HFE. Decisions on overcoming VFI are too often isolated from the decisions on the extent to which HFE is to be achieved. In Decentralizing Indonesia, the World Bank concluded that 'the decisions on how to treat natural resources should therefore go hand in hand with a debate on the acceptable level of inequality'4. A similar statement can be made for revenue sharing decisions.
2.
REVENUE SHARING
Revenue sharing is an effective mechanism for overcoming VFI as it transfers revenue-raising capacity readily from one level of government to another. It can take several forms but can always be seen as the equivalent to allocating a revenue source between more than one level of government it is simply another of the allocation decisions. Revenue sharing can be largely understood as:
1. Governments at more than one level sharing the revenue collected from a tax base; 2. Governments at more than one level working together to use the same tax base to raise separate elements of the same tax (under this 'piggybacking' arrangement, each lower level jurisdiction might even be given freedom to adjust the definition of the tax base and set their own tax rate); or 3. Governments at more than one level using the same tax base to raise revenues from different taxes.
374
Searle
However, in a stricter and more commonly used interpretation, revenue sharing refers to point 1 exclusively. Points 2 and 3 may alternatively be interpreted as part of the assignment of 'own' revenue sources to those governments. In line with the common understanding, this paper is based largely on an assumption of a revenue sharing outcome that has both (or all) levels of government sharing a revenue base through a single tax impost and in set proportions of the h d s collected. The discussion is, however, often applicable to other scenarios. Leaving aside for the moment the relationship between revenue-sharing and HFE, there are a number of issues that arise in applying any revenue sharing scheme. These issues need to be considered carefully as they relate to the distribution of authorities in the revenue raising effort and the establishment of incentives that can lead to greater or lesser efficiency in the use of the revenue bases involved. Apart from deciding exactly what is meant by revenue sharing, the more important of these questions seem to be:
1. What revenue sources should be shared; 2. Should the revenues raised be tied to particular purposes, for either or both levels of government; 3. What level or levels of government should decide on the tax rate (and how should they come to any joint decision on this); 4. What level or levels of government should be responsible for the revenue collection; 5. What level or levels of government should be responsible for defining the tax base and granting exemptions from the shared tax; 6 . Whether it is actual collections or 'face imposts' that are to be shared; 7. How is the revenue to be shared between the levels of government, and the individual jurisdictions at any one level; 8. The frequency and means by which the distribution of revenue should take place; and 9. The means by which any disagreement between governments relating to the revenue sharing are to be handled. As will be seen, the questions are often closely inter-related and should not be decided in isolation from one another. Any revenue-sharing system must be developed as an overall package. Nevertheless, it is beneficial to discuss each of the questions separately to clarify the issues and the interrelationships.
Revenue Sharing, Natural Resources, and Fiscal Equalization
2.1
What Revenues?
Identifying the revenues that are to be shared is important because of its relationship with other policies of the central government, the sub national governments and the nation as a whole. Depending on how collection authority is shared, for example, efficiency in tax collections can be either increased or decreased by the choice of what revenues are shared. Sharing the wrong base, for example Corporate Income tax or even a VAT, can result in unfair and inefficient apportionment of revenues5. This is particularly important when dealing with mobile or otherwise elastic revenue bases where decisions on tax rates can cause locational and other changes to economic activity as firms (or individuals, but this is probably less important) relocate or change their actions to minimize tax imposts. The overall objective when considering what revenues to share is to ensure that the total revenue of each sector of government is likely to increase at the rate at which it is intended their service provision levels, and therefore costs, should i n ~ r e a s e .Achieving ~ this is a delicate issue as it rests on the level of fiscal stress each sector is under - how does their current level of service provision relate to community expectations? Each level of government will be competing to achieve a sharing agreement that, as well as adjusting to the required level of VFI, maximizes its likely revenue growth. The incentive for central governments is usually to ensure that the revenue going to sub national governments does not grow any faster than total public sector revenue. The incentive for sub national government is to ensure that the shared revenues are those that should grow at about the rate of the national economy (such as income tax and VAT), or those that are expected to grow very fast because of some known future event (such as the scheduled development of mineral or oil fields). Depending on the intended basis of distribution of the revenue between sub national governments, there might also be competition between those governments. Sharing on the basis of location of origin (also known as the 'derivation basis') can cause friction between sub national governments and severely limit the ability to implement national objectives of equity. Although tax sharing on a derivation basis is probably the most common approach in the international arena, some countries use a 'formula' approach: for example, Spain uses estimates of aggregate consumption to share VAT and Germany uses population with the same objective.
2.2
Revenue Hypothecation
The most common international experience with revenue sharing is to distribute the funds unconditionally. This fits the view of revenue sharing as a way to address VFI caused by a shortfall in local governments' own-source
376
Searle
revenue. But whether the revenue flow resulting from any shared base or collection should be hypothecated (or earmarked) to a particular fimction by either level of government often depends on the source of that revenue. A shared forestry royalty, for example, is much more logically hypothecated to reforestation that any shared revenue from income tax. It is important, however, to decide in advance if all or part of the shared revenue is to flow to each or any level of government for a particular reason. If hypothecation is imposed on sub national governments as a condition of the revenue sharing, the flow of funds becomes equivalent to a tied grant. In some circumstances, hypothecation can result in very different results to those intended. While there might be benefits in hypothecating forestry royalties to reforestation, for example, there is little point in doing so if the areas that need reforestation are now denuded and producing no royalty revenues, and the sharing is based on the region of royalty origin. If hypothecation is being considered, the question is whether revenue sharing is the best way to proceed, or if it is better for the central government to collect all the revenue and distribute whatever share is agreed as a tied grant based on need. The second option provides a much better capacity to meet needs where they lie. But hypothecation does have its merits. One is that by establishing a separate fimd and hypothecating revenue to it before determining how that amount is to be shared, costs associated with the economic activity resulting in the revenue can be met appropriately once the need is identified. For example, reforestation might be met from a Trust Fund established from the royalties as the trees are felled but distributed only after the forestry activities have been concluded and the reforestation costs are known.
2.3
Deciding the Tax Rate
Deciding on the rate of tax is obviously a delicate task. The different levels of government that are to share in a revenue source often have different attitudes to what the rate should be. These attitudes depend on the importance of the tax to their budget, the way they perceive the tax in terms of its economic impact on their area and whether they are the collection authority or not. As indicated above, this can be a set of decisions taken by the individual governments involved but, for the purpose of this paper, it has been assumed that there is some co-ordination between interested parties. It is easy for a non-collecting government to place a greater onus on the collecting government by arguing for (or imposing) an increase in the tax rate, knowing it will not have to face the full wrath of the taxpayer for the increase. The question of setting the tax rate is therefore closely related to the question of what level of government is accountable to taxpayers in
Revenue Sharing, Natural Resources, and Fiscal Equalization
377
taking the tax rate decision and, to some extent, to what level of government is collecting the revenue.' For individual jurisdictions, the sector of the economy from which the revenue is collected is also a consideration. Sub national governments with little secondary industry will argue that taxes on this type of industry should be higher, while those with the secondary industry will clearly have opposite views. Denuded areas requiring reforestation may argue for higher forestry royalties while the current forestry areas will argue for lower royalties. Such circumstances can be overcome through the establishment of a separate fund andlor changes to the sharing arrangements within the recipient jurisdictional level. In summary, the pressure for higher tax rates on shared revenues will depend on how each sub national government and the central government sees the net benefit to its area, taking account of the increased taxes on their citizens and the government's increased capacity to provide services for those citizens. Not all governments will have the same expected net benefit. In arriving at a negotiated rate of tax, it is often beneficial for sub national governments to have some mechanism for co-ordination of their thoughts, such as an association, to prevent the central government dominating the decision. Once a tax rate has been set, there can be legal issues associated with how the decision is implemented. Sometimes, while only one level of government has the legal power to impose the tax and is therefore the collection agency, the task of setting the rate is given to another level. This can create difficulties if the imposing government disagrees with the rate that has been decided (or any change to vary it) and the government that made the decision have no power to enforce it. In Australia, where all the VAT (known as the Goods and Services Tax) collected by the central government is distributed to the States - a 100 per cent sharing - the Agreement reached is that: a proposal to vary the 10 per cent rate of GST will require: i. The unanimous support of the six States (and two Territories); ii. The endorsement of the central government; and iii. The passage of relevant legislation by both houses of the central parliament8. Clearly, an individual State that wants to vary the tax rate has very little power, and the combined will of all States can be thwarted by the central government9. Because of the large VFI in favor of the central government, the States find it difficult to say no to the more fiscally powerful center. They are left with little real power to alter the rate of tax on what amounts, on average, to over 25 per cent of their revenue.
Searle
In any revenue sharing circumstances, what is required is a workable means by which the tax rates on shared revenue sources are set (or changed) and implemented. Both levels of government need to have real power to have their arguments heard, yet self-serving incentives need to be avoided in arriving at the decision. This said, the typical revenue sharing arrangement in the international practice is one in which the central government sets the rates and collects all revenue, and then shares that revenue with sub national governments according to some rules set in either the constitution or central government legislation.
2.4
Collecting the Revenue and Granting Exemptions
The level of government that collects the tax must be Mly compensated for the costs of collection, but must not be in a position where it can benefit unduly because it is the collection agency. Most frequently, the government imposing the tax is also the collecting agency because of legal requirements, but the authority to collect can be delegated to a more appropriate level of government.'0 What is important is that the incentive is to use the tax base efficiently and not to give undue exemptions etc. In poorly designed systems, the collecting level of government can manipulate the process so that all the 'down-side' of collection shortfall lies with the non-collecting governments. It is too easy for a collecting authority to grant tax exemptions if it knows it will not suffer the full (or any) impact of those exemptions. This is much more of a problem, of course, where the collecting government receives a smaller share of the collections than others. In effect, it only costs the collecting government, say, 30 per cent of the collectable tax to give an exemption, and this might be less than the additional collection cost. The circumstances that result in exemptions must be laid down very specifically in the agreement and there must be a mechanism established to adjudicate on disputes with taxpayers. Clearly, the incentives in collection and exemption should be based on total net public benefit, not simply the net benefit derived by the collecting government.
2.5
What is Shared?
It is important to specify what is to be shared. Assuming it is the 'receipts' in some form that is shared, does this mean actual revenue collected (either gross or net of collection costs), the total of the tax impost notices sent to tax payers, or the estimated collection at some time prior to the collection year - usually set by the budget preparation timetable?
Revenue Sharing, Natural Resources, and Fiscal Equalization
379
To use actual collections is logical but administratively difficult as it requires an adjustment some time after the end of the fiscal year to which the collections and transfers relate. Transparency between actual collections and the value of the transfers is lost because of the inter-temporal adjustments. If the distribution is based on gross collections, then it might be expected that the collecting government would have a larger share because it has to cover the costs of collections. If the sharing is to be based on net collections, there must be some agreed way of controlling and measuring the expenses. Should the agreement go so far as to give the non-collection governments some right of 'audit' over the claimed collection costs? To use estimates is administratively simpler in managing the distribution but there are often sound budgetary reasons to be conservative in estimating revenue collections and it is necessary to have some way of overcoming the resultant possibility of underestimation. If estimates are used and there is no post-period adjustment, the collecting agency can become the 111 beneficiary of (or contributor to) any difference between actual and estimated collections. In these circumstances, there is a very high incentive for them, if they are the estimating authority, to make even more conservative estimates. There is also the issue of which government does the estimating. Clearly, there can only be one if this is the basis of the revenue sharing but governments will have different ideas on how to arrive at the estimate. To use the total of tax impost notices creates an incentive for the collecting government to make sure it receives its agreed share of the 'collectable tax' and then pass all the detriment of non-collectables and exemptions on to the non-collecting government. Under these circumstances, the 'agreed' levels of sharing are not achieved. If joint utilization of a tax base is seen as revenue sharing, another set of questions present themselves. Taking 'piggy-back' taxes for example, is there to be total harmonization of the tax effort (tax base and taxpayer definition, assessment procedures, collection methods, appeals and auditing of payments) by the sharing parties, is there to be one tax notice or two, one collection agency or two, and so forth? These questions may be easier to answer where receipts are shared, but they are non-the-less important. In cases where the nation has limited human capacity to manage both the central and sub national revenue collections, as occurs in many developing countries, there is a strong case to have a simple piggy-back arrangement with as much uniformity as possible and a single revenue collection agency. In other cases, the tax base might be shared but be used in different ways by different levels of government. In Australia, for example, the value of property is used as a tax base by both the State and municipal levels of government. Typically, municipal governments levy a property rate on all properties while the States levy a land tax on non-fann commercial
Searle
380
properties and industrial properties over a certain value. While it is not the process in Australia, these circumstances could be beneficially handled by a joint collection and distribution authority.
2.6
Sharing the Revenue
Between levels of government. As mentioned above, overcoming or ameliorating VFI is usually the reason for revenue sharing and, irrespective of what might be best policy, is most frequently seen as a political issue. As a result, the percentage of the revenue received by each level of government will most likely be decided in a political rather than an economic environment. It is more complex where there is more than one level of beneficiary local government because, in those circumstances, there is not only the competition between the center and sub national government, but also between the different levels of sub national government. It would be best if there could be a measure of the level of fiscal stress that each level of government finds itself in and the shares of total public sector revenue be based on these. That, however, is both very difficult and very unlikely as the levels of fiscal stress experienced by any government are an exogenous issue affected by the behavior of that government.'1 Between sub national governments. How the revenue is to be shared between jurisdictions at any level is probably the most difficult issue of all. It relates closely to many of the other questions and, as a result, some aspects of it have already been discussed. A basic consideration is the philosophy behind the revenue sharing and the political objectives in the fiscal decentralization within the nation. As has been seen, it is easy to produce a distribution that is contrary to the type of decentralization the country is trying to achieve. At one extreme is the distribution to recipient jurisdictions based on where the revenue is paid or collected - the origin (or derivation) basis. This is often what is argued for by (richer) sub national governments, but it often also presents problems. Depending on the economic activity and the revenue source, there can be differences between where the activity leading to the tax is undertaken, where the taxpayer lives and where the tax is paid. A forestry company, for example, might be extracting timber in a remote province, but be owned by a taxpayer who lives just outside the jurisdiction of the capital city and pays his tax at an office in the capital. If the revenue is required for reforestation in the location of economic activity, can that be achieved by applying an 'origin' basis? What is the 'origin' of the tax? Are there data available on which to base the distribution so that it is paid to the locations of economic activity and thus achieves the aim of h d i n g reforestation? But what if the revenue was to be used to fund basic education and health services? The place of origin might now be related to the
Revenue Sharing, Natural Resources, and Fiscal Equalization
381
taxpayers residential address (or the origin of his service provision). Can the data provide for each of these alternative definitions of origin, which differ with the aim of the redistribution? At the other extreme to the derivation basis, there is a pooling of the revenue to be shared and some general principle applied to its distribution between local governments. If the basis of distribution were to be equal fiscal capacity, for example, this would be the best way to overcome both VFI and HFE problems. However, an agreement to apply HFE principles to shared revenue is often very difficult to achieve because of the political considerations and the arguments of 'origin'. The greater the extents to which individual locations see themselves as 'locations of origin' for the revenue, the more difficult it is to pool the funds and distribute them according to a general principle. In Indonesia, for example, fiscal equalization was a major objective in the reforming legislation passed in 199912,but there was also a desire to return some natural resource revenue to the areas where the mining or extraction occurred, with this differing by type of mining and by area. Because of these joint aims, the distribution mechanism for mining and oil-based revenue that was written into the decentralization legislation in 1999 included a large number of elements, differentiating between provinces where revenue originated and provinces where it did not originate; and (within provinces) municipalities where revenue originated and municipalities where it did not. In each of these groups, the distribution to the originating areas was based on the value of production or contribution to GST from the area, and the distribution to non-originating areas was based on population. The system is extremely complicated and requires data on the contribution to GSP of individual industries by municipality. Obviously, this is very difficult to measure with any accuracy and many disputes have arisen as a result. It is important that the system is not too complex or too difficult to manage. The revenue sharing arrangements between sub national governments have to be understood by all parties concerned and be as transparent as possible so that taxpayers know what is happening.
2.7
The Frequency of Transfers
The frequency and means by which the distribution of revenue should take place is often a matter of contention in poorly designed revenue sharing systems. The collecting government can do its job well, but then delay the payment to other governments for as long as possible to improve its cash holdings. The incentive here is obvious and the agreement should overcome it by specifying clear time requirements and penalties on the collecting authority for late payments.
382
Searle
The best option is for the transfers to be made in accordance with the rate of collection. Whether the frequency of transfer should be daily, weekly or monthly will depend on the type of revenue and the administrative structures and technology available to manage the collections and transfers. If the shared revenue is one that is payable only at the end of each month, for example, then monthly payments might suffice. If the shared revenue flows evenly to government throughout the year, daily payments might be best. By managing bank accounts and using electronic banking technology, it is possible to arrange for such revenues to be transferred on a daily basis if that is required, but this would be unusual and may create auditing concerns that over-ride the perceived benefits. The existence of a national treasury system with regional offices and the efficiency of the government financial management system might also influence the ability to make payments more frequently. Whatever frequency of payments is arranged, it should be such that it gives all governments, collecting and non-collecting, close to the same flow of cash receipts. This raises the question of what rights the non-collecting governments should have to monitor the rate of revenue collections. It would seem reasonable for all recipient governments to have some rights to audit the collection and transfer system to ensure themselves that it was working, and being operated, in accordance with the agreement they had reached. However, such co-operation between governments in these matters is not usual.13
2.8
Disagreements between Governments
The last issue to be discussed under this topic is the dispute settlement clause. Any government that is sharing a revenue source must have some way of overcoming disputes with the other(s). In most cases, the government with the authority to collect the revenue has a stronger hand than the others, but this should not be the cause of it having a non-planned advantage in actual revenue received. The source of power or delegation for revenue sharing will greatly influence the type of legal recourse governments (especially local governments) have. Legal recourse of local governments may be enshrined in the constitution (Switzerland), prevented by the federal structure of the nation (Australia) or given at the discretion of the central government (Russia during the 1990s). Whatever the legal basis of any agreement between governments, the courts are often best placed to rule on any disagreements that occur. Assuming the nation has sufficient separation of powers between the legislature and the judiciary, an impartial decision should be achievable through the court system and access of all parties to such a mechanism is beneficial.
Revenue Sharing, Natural Resources, and Fiscal Equalization
2.9
Conclusion on Revenue Sharing Policy Determination
A revenue sharing agreement should be a formal written document with legal status if possible, and should cover all the issues raised in this paper. There are often good reasons to share revenue bases or collections between levels of government but to do so raises many questions that need to be agreed between the parties. Nevertheless, the agreements should remain as administratively simple as possible. In the longer-term, a well designed agreement will reduce administrative costs and tax levels, while maintaining tax collection rates. The negotiations leading to revenue sharing arrangements will benefit from being more wide-ranging than is generally understood and require a broad knowledge of public sector objectives at both the national and sub national levels. The negotiations will sometimes be difficult because of conflicting incentives, but all elements of them should be completed before the arrangement is entered into. Many of the problems referred to in this discussion arise from competition between governments that are parties to the revenue collection arrangements. If those parties can agree to the establishment of an independent agency to manage the agreement and collect and distribute the revenues, many possible problems can be avoided.
2.10
Revenue Sharing and Fiscal Equalization
Leaving aside some specific matters associated with the sharing of natural resource based revenues, there are a number of issues that are general to revenue sharing and a fiscal equalization system. They revolve around how the revenue capacity of the sub national governments will be assessed for the shared revenues; and what impact those assessments are likely to have on the distribution of grants between sub national governments, the volume of funds necessary to achieve the level of equalization required and the volatility in the level of funding for individual local governments. The revenue capacity assessments. Because shared revenues are tantamount to the allocation of part of the tax base to sub national governments, the capacity of those governments to raise that revenue should be included in the equalization assessment, together with their capacities from all other own-source revenues. The shared revenue is available to them to meet the expenditure needs of their communities. At present, this is not always the case even in systems that employ a representative tax system (RTS) approach to equalization because not all nations include all revenue sources available to the sub national government. Increasingly, however, this inadequacy is being overcome.
384
Searle
In most instances, the sharing of revenue between the central and sub national government requires the maintenance of a data base that can be used in the assessments needed for an equalization system. Assuming the sub national governments are all subject to the same sharing arrangement, the measurement of tax bases will measure the agreed way in which the revenue is to be distributed. Where the distribution is on an 'origin' basis, it will be necessary for the central government to hold data on that basis. This becomes the automatic measurement of the relative capacity of each council and the actual revenue received by each local government should be an accurate measure of both their absolute and relative capacities to raise the shared revenue. In fact, in all cases where there is a uniform basis of distribution as between sub national governments, and the distribution accords with the agreed method, the actual revenue received by each will be an accurate measure of their capacities, even if the agreed method cannot be measured by objective data. It is only where the actual distribution has not been done according to the agreement that there will be a need to arrive at some measure of capacity that is not based on actual receipts. These conclusions might be negated if there are different sharing arrangements for different jurisdictions at the same level of sub national government. If those different arrangements are a true reflection of the inherent capacities of the sub national jurisdictions to raise the shared revenue, actual receipts can still be used as the measure of capacity. If, however, the differences are due to political or other reasons not related to inherent capacities, there is a question of whether the equalization assessments should 'override' the agreement in its endeavor to achieve overall equity for all governments. This is the same question as arises when considering how tied grant revenue should be handled within the revenue equalization process. In theory, the answer is that if equalization is the only objective asked of the grant-determining agency, there should be adjustments made to 'override' these differences. In these circumstances, it is necessary to find a common objective base on which to measure the capacity of all sub national governments. If the grant determination process is not to negate other objectives of government (no matter how inappropriate in terms of equalization), actual revenues should continue to be used as the measure of capacity. The impact on grant distribution. In practically all cases, revenue sharing will increase the differences in capacities of individual sub national governments to raise revenue. This will increase the differences in grants between those governments, probably in both per capita and absolute amounts and, as a result, probably also increase political pressure to reduce the extent of equalization. Disgruntled local politicians tend to argue not
Revenue Sharing, Natural Resources, and Fiscal Equalization
385
about equalization of total fiscal capacity, or even total per capita revenue, but about equalizing the per capita size of grants. Increasing differences in the size of grants (even if only in per capita terms) tends to lead to those sub national governments that receive less making claims that there is too much equalization or that the grant distribution is being determined incorrectly. While HFE is the objective within the grant distribution, anything that changes the level of VFI is likely to increase the differences in per capita grant funding. Because any change in the responsibilities or revenue capacities of sub national government will not be distributed evenly across those jurisdictions in per capita terms, any change in VFI will result in a change in the calculations necessary to achieve equalization. In all but extreme circumstances, this will increase the differences in per capita grant funding. The same can be said, of course, to result from granting more tax autonomy to local governments in the general revenue assignment exercise. The funds required for equalization. If the equalization system is openended as in canada14,revenue sharing can have very detrimental impacts on the central government budget because any increase in the funds necessary to achieve equalization is provided by the central government. In closed systems like that in Australia, equalization is achieved from the same sized pool of funds irrespective of changes in the need for redistribution. This is achieved by redistribution between the States rather than changing the central government budget result or the distribution of resources between the central and sub national levels of government. A closed system is based on a premise that the equalization system should not result in changes to overall public sector outcomes, whereas an open system allows it to do so. Revenue sharing and grant volatility. The usual starting point in an allocation of revenue sources between levels of government is that the level with the greatest fiscal capacity has the best tools for overcoming fluctuations in revenue collections. This is usually the central government and, as a result, it is therefore given the more volatile revenue sources. It is also these more volatile revenues that are shared if a further step in reducing VFI is undertaken. Revenue sharing therefore often results in greater annual movements in sub national jurisdictions' relative revenue capacities, and thus greater volatility in the size of their grants - both in the absolute size of the transfer and in its distribution. One aspect of a grants distribution system that brings it into disfavor very quickly is volatility of annual outcomes. Sub national governments cannot plan and budget to an acceptable degree of accuracy if they cannot estimate their grant sufficiently well. In many circumstances, a local jurisdiction would sacrifice an increase in revenue over the long-term for more predictability and stability on an annual basis. Where the central government has all the fiscal liability, such as in Canada, this is not as great a problem as in closed systems where the zero
386
Searle
sum game can result in substantial reductions in funding for some governments. One way of handling these annual variations is to base the grant distribution on the calculations over a number of past years on a rolling basis to spread the impact of annual fluctuations in capacity. This method can be effective but complicates the equalization calculations and reduces the transparency of the fiscal transfer system.
3.
NATURAL RESOURCE REVENUE
Like revenue sharing, the way revenue from natural resources is handled can have a big impact on Vertical Fiscal Imbalance. Natural resources are also a revenue source that is often hotly argued by sub national governments as belonging to them. The difficulty with accepting this argument is that the receipts from natural resource activities are usually unevenly distributed between jurisdictions and over time, and can thus cause great complications when applying policies aimed at equalization of fiscal capacity and service provision. Whether natural resource revenues should be allocated to sub national government (and if so to what extent if revenue sharing is an option) is therefore a major issue in most fiscal transfer systems. The questions that arise in coming to conclusions in relation to natural resource revenue include: 1. Why do we raise revenue from natural resources; 2. What level or levels of government should have access to the revenue; 3. How should we raise the revenue if we are to satisfy these desires; 4. What level of government is best placed to raise the revenue; 5. How should the revenue be distributed between the levels of government; and 6. How should the revenue be distributed between jurisdictions if any of it is provided to sub national governments?
There are many policy considerations to these questions and some general discussion of the issues will be beneficial to the decision making process. But first, we should try to decide what we mean by natural resources.
3.1
What are Natural Resources?
Economists have developed the concept of natural capital to answer this question. They present it as being analogous to made capital and define it as the value of a country's existing stock of natural resources, including fisheries, forests, mineral deposits, water and the environment. Natural
Revenue Sharing, Natural Resources, and Fiscal Equalization
387
capital produces goods and services, just like other capital. The answer to this question therefore seems easy - anything that nature has provided and that has a value in the market place is a natural resource. When considering the revenue capacities of governments, however, this definition may be both too broad and too narrow. Land, per se, is not considered a natural resource in terms of revenue raising by government, but it is used extensively as a revenue base - why differentiate between land that can grow wheat and land that has minerals under it, or even on the surface? There seems to be an element missing from the definition. Is it whether or not the extraction is replenishable? On the other hand, plantation timber and the produce of aqua-farming are sometimes seen as natural resources for revenue raising purposes and these cannot be said to result solely from nature - how do they differ from general agricultural production? For the purposes of this discussion we do not need to be specific in our definition, but will include oil and gas, mineral extraction, forestry, fisheries and hydro-power under a definition of natural resource^'^ and think about classifying them in several ways that will assist in our development of an optimal policy on revenue collection authority. Exhaustible or replenishable. For the most commonly thought of natural resources - minerals, oil and gas - it is assumed that there is an exhaustible supply. Certainly this is the case in any one location. The economic activity taking advantage of the resource uses it until there is nothing of value left (given its current price) and then moves to another site. Forestry, fisheries and hydro-power might be seen as replenishable resources. These resources are useable on a permanent basis if they are well managed and the environment in which nature provides them does not change dramatically. With this difference in mind, replenishable resources might be thought of as no different from industrial and commercial activities and taxed accordingly, especially if the cost associated with establishment and management of the resource is a private expense. Where the capital expense associated with the establishment of the resource (for example, the building of a hydro water storage) andfor the cost of management is a public expense, this is tantamount to the public ownership of a commercial activity and the enterprise might be seen for revenue generation purposes as being equivalent to any other government business activity. Impact on infrastructure requirements. Infrastructure costs money and all activities that utilize natural resources require infi-astructure. Infrastructure such as roads, railways and ports we might term economic infrastructure as it relates directly to the economic activity. Infrastructure such as schools and health facilities for the people involved in undertaking the activity we might call social infrastructure as it relates more directly to the needs of society and only indirectly to the economic activity. Where the cost of this
infrastructure lies varies from case to case. The first company to commence extraction in an area might put in the economic (and even the social) infrastructure at its own expense. It may then retain ownership and maintenance responsibility of the economic infrastructure but pass ownership of the social infrastructure to the government (either the central or sub national government level, depending on the responsibility allocation existing in the nation). In developing countries, it is not unknown for the company to continue to own and run even the social infrastructure (thus obviating the social expense). Neither is it unknown for the economic infrastructure that has value as a public asset to be handed to a government free of charge, either on its completed or at the end of the firm's economic activity in the location. This might be seen as the equivalent of an up-front or final lump-sum payment of revenue to the government. The infrastructure can be either permanent or temporary in nature. It will be as permanent as the period of the mineral extraction or as temporary as the construction period of the facility that enables a long-term economic activity - a new water storage. A road and railway system may be maintained and operated in perpetuity but a town at a construction site may be dismantled when the project is completed. A town at a mining site may be built and then abandoned after 15 or 20 years because the deposit of mineral value runs out, but the port built to export the material might still be in use because of other extraction sites being operated by other companies. Some infrastructure can be income generating for the government. If the government builds or owns the ports and the railways, these might be net revenue generating for the life of the economic activity or thereafter. Social infrastructure such as schools and hospitals is most unlikely to be net revenue generating and will require long-term maintenance and then replacement if they are required as permanent assets. Most important from a natural resource revenue allocation perspective, the infrastructure may or may not be located where the economic activity is located. The mine may be in one jurisdiction, the roads and railways pass through several others and the port is in another. There are even examples where the entry and infrastructure to an underground mine is in one jurisdiction but the resource is being taken from under another. How does the concept of 'origin' fit into these scenarios? If a fishery is operated in national as opposed to local waters16, the port and other facilities are obviously in a local government area, but where is the place of origin if that is to be used to distribute the revenue base from fisheries? Ownership of the exploited resource. Ownership of natural resources is subject to both tradition and law, depending on the nation and the culture. Even in a single nation, it can differ between resources and change over time. It can differ depending on whether the resources are on the land or below the soil and, if below the soil, it may depend on how deep they are
Revenue Sharing, Natural Resources, and Fiscal Equalization
389
located. Ownership can lie either with an individual or with a community. If the land is owned by a community, it may be the traditional indigenous landowners of the specific parcel of land, or a group of people less directly attached to that land. These may be the indigenous people of a wider area, or all the residents of a local municipal council, a province or the nation as a whole. The issue of ownership frequently creates problems in negotiations concerning natural resource revenue. Where it is based on law, ownership is seen as being 'dictated' and people argue that cultural and traditional aspects of ownership have been over-ridden. Where it is based on culture and tradition, there are often different views as to how these define the owner or owners, and there are expectations that the law will provide clarity. Are the owners those that live in the area today, or that group plus past (but living) residents and future residents? Does moving to the area automatically give new residents some ownership rights? This is not the paper to discuss these issues and, for simplicity, we have assumed ownership is an accepted position. So to what extent should the exploitation of the natural resource benefit those other than the owners of that resource? If a private individual or firm owns a mineral deposit (as against the rights to exploit it), is a government entitled to raise revenue from it through export charges or extra raiVroad use charges etc? If an indigenous community or a community limited by a geographic boundary (a municipality) owns a resource, should the wider community also benefit from its exploitation? How might 'future owners' benefit from the exploitation of resources by the current group of owners, and does this have any impact of how revenues might be collected or used? Impact on the environment. Different types of natural resource activity also have different impacts on the environment that result in different levels and types of expenses. Natural fisheries might have only environmental management costs but strip mining of bauxite will have wide-ranging costs of land reclamation and uranium mining may have very long-term costs associated with protection of the community from radioactive materials. Environmental costs can be incurred before, during and after the economically beneficial activity. They may be the responsibility of the company, the sub national government or the central government. International or national market. Firms active in natural resource industries can operate in either a domestic or an international market. Those operating in a domestic market are much easier to manage as taxpayers, but is that market a local or a national market? Those operating in an international market are subject to much greater price volatility and can move profit from one nation to another to minimize tax payments. At what level of government are the taxpayers best managed?
390
Searle
Clearly, there are many ways we can classify natural resources and each has some relationship to the wider question of public sector revenue raising. No single way of looking at natural resources should take precedence over another, and their consideration should be combined with other variables that lead to the holistic environment in which revenue collection and allocations might be decided.
3.2
Why Raise Revenue from Natural Resources?
We need now to consider why revenue is raised from activities associated with natural resources. There seems to be three reasons. To coverpublic cost. We have seen that there are likely to be some public costs associated with any economic activity exploiting natural resources. These can be either up-front one-off costs or continuing costs, but are likely to be a mixture of both, varying according to the activity of the firm. They can also be costs of the sub national government or the central government. They may be reduced by a 'construct and handover' agreement with the firm or fully funded by the firm through user charges (such as port handling fees). They may be within the jurisdiction in which the extractive activity is taking place, or beyond its boundaries. Whatever the arrangement, a jurisdiction should not be out-of-pocket due to infrastructure costs it incurs because of a natural resource activity in its area, unless for social or political reasons it has decided to subsidize the firm's activities. In some ways, the public sector revenue required to cover the expense of publicly funded infrastructure might be seen as being hypothecated revenue. Ownership would seem to have little relevance to this aspect of the revenue allocation. To fund speczjk objectives. In some circumstances, natural resource revenues are used to fund specific services or types of infrastructure17. This hypothecation of the revenue seems to have some association with concepts of ownership, yet also some control by non-owners. It would be unlikely that a jurisdiction that owned the asset would want to hypothecate its own revenue resulting from the sale. A government's normal budgetary procedures will, if they are operating effectively, allocate funds to community priorities anyway. It is more likely that hypothecation, where practiced, is imposed on the beneficiary jurisdiction by a more powerful (central) government, perhaps to ensure that the revenues are 'well used'. In these circumstances, it is also possible that the people the central government represents have less ownership connectivity with the resource. Where it is towards infrastructure development, hypothecation could also be seen to have something to do with the benefit to future owners. It could be a means by which non-owners are preventing the current group of owners from using all the funds, to the detriment of future owners.
Revenue Sharing, Natural Resources, and Fiscal Equalization
39 1
In a sense, hypothecation may be nothing more than the use of the b d s as general revenue because the fields of expenditure are very largely unrelated to the economic activity, and might be totally unrelated to it. The hypothecation may simply be some external authority's idea of where there is a need for public expenditure. It could be recurrent or capital expenditure and its distribution might be based on need or political considerations. For the recipient jurisdictions, hypothecation is just another way of another government attaching strings to money it provides. To add to general revenue. Natural resource revenues are more generally used simply as a supplement to unhypothecated general revenue used to fund the range of services a recipient government provides. This is where the ownership question becomes much more important because most such revenue is collected through some form of royalty or taxing away pure economic rents. Who has the right to collect such revenues? Are royalties the equivalent to the sale of a capital asset by an owner? If we see them as the sale of an asset, what is the justification for governments, where they are not the owners, collecting taxes from natural resource activity through such means as railway and port user charges to achieve super-profits as an alternative to royalties? Whatever means is used and with whatever justification, revenues from natural resources are very often (most often) seen as a source of general revenue.
3.3
Revenue Raising from Natural Resources
Like the time sequence of public expenses associated with the utilization of a natural resource, there are also different possibilities in the time sequence of public sector revenue flows from natural resources. Depending on the arrangements between the producer and the government(s) concerned, it is possible to have an up-front payment andlor a payment over the period of the economic activity. Up-frontpayments. An up-front fee is the equivalent to either the sale of an asset or the cashing out of a future revenue flow. Where it is a right to explore, for example, it is the sale of an asset that we can separate fiom a mineral deposit because the exhaustible asset has not yet been proven. A separate payment stream will flow from a second asset if a deposit is proven. Under the definition of natural resources discussed earlier, the sale of a right to explore is not strictly a natural resource activity. It is simply the sale of a government asset. An up-front receipt that could be characterized as the sale of an asset might be seen as a capital transaction rather than a recurrent revenue. If this is the case, it may need to be treated differently when measuring revenue
392
Searle
capacity in a fiscal equalization calculation - how are capital transactions to be handled? If an up-front payment is agreed to be in lieu of a h r e cash flow, it is less likely to be seen as a capital transaction but may again have implications for the measurement of revenue raising capacity. Should it be seen as a capacity that existed in the year in which it was received, or a capacity that should notionally be distributed over many years - the period of future cash flow to which it is equated? Rent income. There are many different and acceptable means by which annual revenues can be gained from the utilization of natural resources". In general, these payments are based on either the annual value, value added, or volume of production of the enterprise; or the profitability of the enterprise. Whatever base is used, it results in the receipt of a recurrent revenue stream over a period of years. The size and variability of the income stream will depend on the tax system, the volatility of costs and the prices for the commodity. The firm may not be profitable in some years but it must be assumed that there will be some public revenue flow for some part of the period of the economic activity. Back-end Payments. As an alternative to part of any rent income, governments sometimes receive the public infrastructure built by the firm as part-payment for the extracted material. This is more frequently economic infrastructure but can also include social infrastructure. As discussed earlier, it could be port and railway facilities that still have a value to the government but not to the firm that built them, or it could be schools and health facilities that will be used by the community long after the extractive activity is completed. To the firm that built them, these are a business expense just like taxes. To the government, they are a financial benefit just like taxes, except that it had no (or little) control in how the funds were spent.
3.4
The Allocation of the Revenue
So what right do sub national governments have to the receipt of natural resource revenues? In economic terms, it is logical that at least some of the revenue should attach itself to whatever unit(s) of government have or need to undertake expenditure as a result of the activity. Some revenue should therefore flow to sub national government if it incurs expense, which is very likely because of its usual infrastructure responsibilities. The recipient sub national government units may extend beyond the jurisdiction of 'origin' of the resource, and the payment may logically be either an up-front payment or a revenue stream depending on the mix of capital and recurrent expenditure undertaken by the sub national government(s). Recurrent expenditure by government may be limited to the period of the activity or
Revenue Sharing, Natural Resources, and Fiscal Equalization
393
may extend beyond that period. The need for up-front expenses may even dictate the structure of the revenue stream governments negotiate with the firm. There is no doubt that if a sub national government incurs expense (either up-front, during the economic activity or continuing after that activity has ceased) because of the exploitation of a natural resource, it is entitled to at least the value of that expense. And this also applies for the central government. In some ways, the income flow needed to cover public sector costs can be seen notionally as hypothecated income. There is also no doubt that the owners of a resource are entitled to some revenue from the use of their resource. If a local government owns the assets being exploited, there is therefore a strong argument that it should receive some additional compensation, but that does not mean that other jurisdictions should receive no benefit. The absolute volume of royalty revenues can, as we know, be so large and so limited in its timeliness and spatial distribution if based solely on the place of origin, that it can be detrimental to national economic efficiency and create unwanted inequalities of opportunity or differences in service delivery standards. National unity and internal security may be enhanced if the distribution is somewhat wider than a strict origin basis (whatever that means). Nevertheless, there are mechanisms that can be used within an equalization system that will allow all the revenue to go to the place(s) of origin and still not create the unwanted side effects. These are discussed later in the paper. If the nation rather than a sub national jurisdiction owns the assets, it is difficult to see that the area of origin has a greater claim to a share of the excess (non-hypothecated) royalties than any other jurisdiction in the nation. But these conclusions say nothing about what level of government should be responsible for setting the royalty rate and collecting the revenue from the
firm.
3.5
Managing the Revenue Base
There seems no reason why the management of the public sector revenue collection should attach itself to the concept of ownership. It is also clear that central governments will have a greater capacity to deal appropriately with more complex royalty schemes, more complex industrial circumstances and more sophisticated taxpayers. They will allow less opportunity for cross-border 'loss' of tax base through firms moving their profit to lower taxing jurisdictions, and thus increase the chances of the firm's revenue contribution being higher. They are also in a better position to audit the firm's revenue returns and arrange the distribution between jurisdictions on an agreed basis.
Searle
394
Natural resource companies often operate in an international environment and their cost structures, including tax imposts, have to be seen in that light. Usually, international trade is a central government rather than sub national government function and it would seem logical to have the setting of resource royalty rates done by the same level of government that handles international trade relations for the nation. The volatility of natural resource revenue and the central government's larger budget capacity to even out the impact of this volatility is a further reason why it is probably best for a higher level of government to manage natural resource revenue collection. In Australia, where natural resource revenues are the prerogative of the States, international commodities dealers and firms had the States competing against one another and kept prices artificially low until the States agreed to register and exchange the offers they were being made and thus counteracted the cartels. This conclusion to have natural resource revenues managed by the central government is not universally accepted and there are many arguments in practice why it should not be applied in an individual nation's circumstances. Bahl and Tumennasan (2002) propose the allocation of natural resource revenues to sub national governments in Indonesia. For Nigeria, where the last thirty years have seen no substantial public benefit derived from oil extraction, Sala-i-Martin and Subramanian (2003) argue for the development of a system that gives direct personal benefits to all citizens.
3.6
Natural Resource Utilization and Fiscal Equalization
We indicated earlier that fiscal equalization is about equalizing, either fully or partially, the provision of services by governments at the same level. We have also seen that the utilization of natural resources has an expenditure (service provision) aspect to it as well as a revenue perspective. We can assume that local governments will most likely have different expenditure needs associated with natural resources as well as different revenue capacities arising from the utilization of those resources. We now need to consider how these dual impacts of natural resource utilization might be handled in the calculations associated with fiscal equalization. In its most basic form, the budget equation lying behind an equalization grant formula might be expressed as:
E = R + G, where E is the level of expenditure in a jurisdiction that will enable it to provide the standard level of services; R is the own-source revenue capacity of the jurisdiction; and
Revenue Sharing, Natural Resources, and Fiscal Equalization
395
G is the grant the jurisdiction needs to equalize its capacity to provide the services. In the case of natural resource activities, there is an impact on both sides of this equalization budget. If we make equalization calculations on one side of it only, we are implicitly making an assumption about the equalization situation on the other side - we are implying there are no differences in the impact natural resources have on that part of the jurisdictions' budgets. An assessment of different capacities to raise revenue is an implicit acceptance that natural resources will also have different impacts on service provision demand and cost. Logically, therefore, we should do an equalization assessment on both sides of the budget. The revenue assessments. The revenue assessments are influenced greatly by both the pattern and type of revenue flow, as well as the way revenue bases are measured. Assuming sub national governments receive at least some of the revenue, and thus make it relevant to the equalization mechanism through the measurement of revenue capacity, the first issue is whether they collect it as an up-front payment or not. If the revenue is an up-front amount, we need to consider whether or not it is a capital receipt. In general, capital transactions on both sides of the budget are excluded from equalization assessments. Equalization has traditionally been seen as relating to recurrent expenses (services) and recurrent revenues.19 For simplicity, we might best assume that capital transactions are not included in the equalization calculations. But like any such assumption, we should make it only with a full understanding of what it is implying. Here, the assumption is either: that there are no differences in jurisdictions' capacities to raise capital revenue or need for capital expenditure; or that in every jurisdiction's case, the ability to raise capital revenue matches their need for capital expenditure. Neither of these assumptions is likely to be correct but that does not prevent governments from deciding that equalization should only apply to recurrent transactions. If the revenue is up-front but in lieu of a future cash stream, it is much more likely to be seen as recurrent revenue. The question then is whether or not to notionally spread it over the relevant period when undertaking assessment calculations. Is it a revenue raising capacity that exists only in the year in which the lump sum is received, or does part of the cash relate to capacity in future years? If the pattern of revenue flow is intended to match capital expenditure requirements associated with the start of a project, should the 'recurrent' lump sum revenue be seen as capital anyway because it is being used to fund capital infrastructure expenditure?
396
Searle
Clearly, these decisions must be taken on a case-by-case basis. Assuming a need to measure jurisdictions' revenue capacities on an annual basis, however, there are still some important issues to consider. If the revenue is seen as pertaining to an enterprise type activity of the sub national government, such as a water resource, there is a need to consider how such revenues are to be assessed, but this decision must be linked to other enterprise based revenues of the sub national government sector. If they were to be omitted, as is sometimes the case, the implicit assumption is that all sub national governments have the same per capita capacity to raise revenue from this source. Across a nation, there are usually many different natural resource minerals being utilized and many different bases on which royalties are paid. The most common bases on which payments are made are volume of production, value of production and profit, but combinations of these and others are also used. Some jurisdictions will use only one of the common bases, others might use all three. Some local government areas will have no natural resources at all. How should the assessments be handled? Should all natural resources (and in particular minerals and oil) be combined or should there be a separate assessment for each type of resource? How should the revenue base or bases be measured? If the revenue base is to determined by examining the usual practice of the jurisdictions, it will be tempting to look at how royalties are based in each industry and split the assessment of minerals and oil revenue up to match the different bases. Royalties on coal might usually be based on volume of production. Royalties on iron ore might always be based on value. This degree of disaggregation of the assessment is possible but can lead to many problems. It also creates an incentive for a jurisdiction with sole access to a particular mineral to hold the royalty rate down as it can dictate the standard and do so without any detriment to its revenue capacity assessment. But do we really need a separate assessment for each of, say, the 25 minerals found in a nation, when no one recipient government has more than three? What do we do where one sub national government uses a different base for its coal royalties, or half the jurisdictions that have coal use value rather than volume. What do we do if only one jurisdiction has a certain mineral? Can we get all the data we need to place each of the assessments on a comparable base - those jurisdictions using a volume base may have no data on value, etc? What might be found best is to combine all the minerals and oil assessments into one. Based on an assumption that it is ultimately the net income (or profit) flow to the mining industry that is being used as the revenue base, the mix of the revenue bases actually used for different minerals (or even firms) becomes irrelevant. If data are available on the
Revenue Sharing, Natural Resources, and Fiscal Equalization
397
mining industry's input to GST, these can be used as a proxy to do an overall assessment of the capacity of the sector to contribute to public revenue. The capacity of the sector to contribute to public revenue, and the capacity of the jurisdictions to impose and collect royalties, may not be the same2' but to assume so is a beneficial simplification of the process and can give acceptable results. It does not, however, overcome the problems arising from a very uneven distribution of this revenue base across jurisdictions, especially if all the revenue is provided to sub national governments. An uneven distribution of revenue from natural resources can be overcome by an equalization system if the pool of funds available for redistribution is large enough or if the differences in capacity, even if large, are not a large contributor to total differences in revenue capacity. In Australia, for example, the mining sector is only 12 per cent of States' ownsource revenue but the assessment has a larger redistributive effect2' than any other single category of either revenue or expenditure. The overall size of the equalization however, together with the fact that it is a closed system, allows all these capacity differences to be overcome. If it were much more than the 12 per cent of States' total own-source revenue, it would create more difficulties. In Canada, that is not the case because it is an open system and the central government fbnds the amount necessary to achieve equalization. There, the increase in oil production and prices over time has resulted in changes to the system to redefrne the objective and reduce the extent to which equalization is being achieved. The expenditure assessments. Based on our earlier assumption, capital expenditure by local government can be omitted from the expenditure assessments. If, however, the expenditure related to natural resource utilization is recurrent in nature, it should be included somewhere in the scope of expenditure need assessments. Such recurrent expenditure might be directly related to the mining activity and cover debt charges for publiclyowned economic infrastructure, roads maintenance and environmental management; or it might only be indirectly related to the activity and relate to such things as the running of schools and hospitals. The assessments of the sub national governments' 'needs' for all these functions of government, if they are identifiable, should take account (at least in theory) of whether differences in jurisdictions' needs arise because of natural resources activity. Where there are no expenditure assessments in the equalization system, it is clearly inequitable to include all the natural resource revenue base in the assessments if we know that some of it has to be spent on activities associated with developing the revenue base. It might be seen in this guise as an investment and the net revenue (and revenue base) be included in the assessments of recipient governments' capacities. In this way, the sub
398
Searle
national governments are hnded to cover their costs and only the surplus revenue is brought into the assessment.
3.7
Natural Resource Utilization and Governance
While not the topic of this paper, it should also be pointed out that the utilization of natural resources by any nation has considerable governance issues relating to it. Nigeria is a prime example of where these issues have been dealt with very poorly and there has been very little public benefit derived from the oil and gas utilization over the last thirty years. In postconflict Sierra Leone, it is estimated that up to 80 percent of the diamonds produced in the nation leave without royalties being paid. Natural resource utilization seems to present particularly attractive conditions for the development of graft and corruption that must be avoided if appropriate levels of public benefit are to be achieved. Governments must establish systems that maximize revenue collections rather than allow competition that reduces revenue effort. They must develop an industry ethic that sees contributions to the public good as being worthwhile and have systems in place to monitor the activities of the sector. None of these tasks are easy, particularly in less developed countries where personal wealth is still being accumulated by most of the population and systems of guardianship, sponsorship and nepotism can be culturally entrenched.
4.
CONCLUSION
Shared revenues and revenues from natural resources are important considerations in the design of intergovernmental fiscal relation systems. They are often some of the largest revenue sources allocated to sub national government and are often distributed on an 'origins' basis. Decisions regarding shared revenues and natural resource revenues often result in a very uneven distribution of fiscal capacity between jurisdictions, making the achievement of HFE objectives more difficult. This paper has shown that there are many important considerations in deciding issues of revenue sharing and what level of government should have the rights to collect natural resources revenues. It has shown the interrelationships between the policies designed to influence the size of the VFI and the decision on the extent of HFE that is to be achieved (or is achievable) by the redistribution to sub national governments. Shared revenue arrangements should be formalized in a written agreement that gives all parties at each level the same rights and obligations. The
Revenue Sharing, Natural Resources, and Fiscal Equalization
399
agreement should specify exactly what is being shared, what rights to decide tax rates and exemptions each party has, how collections are to be handled, how and when payments are to be made, and many other issues. Revenue from natural resources is often considered suitable for sharing between levels of government. To the extent that it is needed to h n d expenditures associated with the economic activity, there is some justification for this. The impact of natural resource revenue on HFE assessments is very different to other shared revenues because of the 'investment' expenses associated with natural resources. Natural resources also result in much more difficult problems when the revenue capacity is being measured. 'Full' equalization can be achieved in an environment of revenue sharing where local governments collect all the natural resource revenues. This is only possible, however, if the size of the redistributive pool is large enough andfor the differences in capacity to raise revenue from natural resources are not very large. In many cases, neither of these conditions apply and great care should be taken that decisions on natural resource revenue sharing do not prevent the achieving of equalization objectives. The over-riding consideration, however, is that governments can (and do) have different policies operating in parallel and should be aware of the repercussions of one for the other. The policy instruments used to overcome VFI have easily measurable impacts in terms of that objective. They do, however, often make the achievement of HFE more difficult. Policy makers need to be aware of this inter-relationship before any decisions are taken and, preferably, should design an overall fiscal transfer system covering both VFI and HFE, thus maximizing the chance of both objectives being better achieved.
Notes Commonwealth Grants Commission 2005, p.4. Assuming, of course, that local governments still had their own means andlor the right to vary service delivery standards from those to which they were equalized. In this sense, collections are being used as a proxy for the revenue capacity of each tier of government. It would not be appropriate, of course, to use collections as an indicator of the extent to which VFI should be overcome by equalization transfers. World Bank 2003, p 37. See, for example, McLure and Martinez-Vazquez (2002). In other words, each level of government should be provided with revenue sources that have an elasticity equal to the elasticity of demand for the services for which they are responsible. Sometimes the ability to set the rate and the responsibility for collection of the shared taxes are assigned to different levels of government. In Germany, for example, only the
Searle federal authorities can set tax rates but practically all taxes are collected at the State (Laender) level. Intergovernmental Agreement on the Reform of Commonwealth-State Financial Relations, June 1999, para 3 1. The complete documentation on this tax reform is held in Commonwealth Grants Commission, General Revenue Grant Relativities 2000 Update, Supporting Information, p 372.
At present, all eight State and Territory Governments in Australia are controlled by a different political party to the central government, making this a more likely possibility if any suggestion for change were to be made. In China before the tax sharing reforms of 1994 which created separate central and provincial government tax administrations, all taxes were collected by the provinces. During the ten years prior to that reform, central government revenues declined dramatically as provinces granted business tax preferences and exemptions at the cost of central revenues. See Bahl and Martinez-Vazquez, 2005. Never the less, it would obviously be desirable to avoid giving higher levels of revenue sharing to irresponsible governments. Republic of Indonesia Law No. 2211999 regarding Regional Governance and Law No. 2511999 regarding Fiscal Balance Between Central Government and the Regions. See, for example, Martinez-Vazquez and Timofeev (2005). In such systems, the level of equalization funding is done in absolute terms and the central government has a commitment to fund the scheme, irrespective of the fiscal impact on its budget. The Government of Canada has overcome this problem by 'tinkering' with the arrangements every time the budgetary impact became too great, thus changing the basis of the distribution over time. A more complete discussion of this issue is to be found in Brosio, G, (forthcoming). In Australia at least, the rights to the sea and what lies under it are with the State for the first xx kilometers from the shoreline, then become central government assets. This creates some interesting agreements regarding off-shore oil and gas reserves. In Columbia for example, oil royalties received by local governments must be used to fund priority projects in education, health, sewage systems and water supply. Brosio, G, 2004 discusses these matters very thoroughly. In recent times, Australia has included capital needs in its equalization assessments through a complicated procedure that is extremely intensive in its data requirements. A detailed discussion of that process is available in Searle (2002). These might differ because of structural differences within the mining industry in different localities, the laws relating to royalties and nationally accepted practices that result in some sectors of the mining industry being expected to make a lower contribution than others. In 2002-03, the ACT had no revenue raising capacity, and the other States varied between 390 per cent above the national average capacity to 95 per cent below that average. Commonwealth Grants Commission, Report on State Revenue Sharing Relativities, 2004, Review, Supporting Information, p213. The pool size in Australia is determined more by the obiective of overcominr! - VFI and is, infact, many times that required to overcome HFE.
Revenue Sharing, Natural Resources, and Fiscal Equalization
References Bahl, Roy, and Jorge Martinez-Vazquez. 2005. Fiscal decentralization and economic reform in China. In The Dynamics of Federalism: The Political Economy Reality, T.S. Srinivasan and Jessica Wallack, eds., Oxford: Oxford Economic Press. Bahl, Roy, and Tumennasan, Bayar. 2002. How Should Revenues from Natural Resources be shared in Indonesia? Georgia State University, USA. Brosio, Giorgio. 2001. Decentralization in Africa. In Managing Fiscal Decentralization, Ehtisham Ahmad and Vito Tanzi, eds., Routledge. Brosio, Giorgio. Forthcoming. The assignment of revenue from natural resources. In Handbook on Fiscal Federalism, Ehtisham Ahmad and Giorgio Brosio, eds. Commonwealth Grants Commission. 2005. Report on State Revenue Sharing Relativities, 2005 Review. Canberra: Australian Government Publishing Service. Commonwealth Grants Commission. 2000. General Revenue Grant Relativities 2000 Update, Supporting Information. Canberra: Australian Government Publishing Service. Drysdale Peter, and Shibata, Hirofumi. 1985. Federalism and Resource Development. Australia: George Allen & Unwin. Feehan, James P. Forthcoming. Equalization and the Provinces: Natural Resource Revenues: Partial Equalization can Work Better. Institute of Intergovernmental Studies, Queen's University, Kingston, Ontario. Gillis, Malcolm, Dwight H Perkins, Michael Roemer, and Donald R Snodgrass. 1996. Economics of Development. New York: W W Norton & Company. Martinez-Vazquez, Jorge, and Andrey Timofeev. 2005. Choosing between centralized and decentralized models of tax administration. International Studies Program Working Paper No. 05-02.Andrew Young School of Policy Studies, Georgia State University, Atlanta. GA. McLure Charles E., and Jorge Martinez-Vazquez. 2002. The assignment of revenues and expenditures in intergovernmental fiscal relations. Accessed March 25, 2002, Retrieved from http://www.worldbank.org/wbi/publicfinanceldocumentsMcClure%2O&%2OMartinez.pdf
Sala-i-Martin, Xavier, and Arvind Subramanian. 2003 Addressing the natural resources curse: An illustration from Nigeria. Retrieved from www.econ-upf.edu/docs/papers Searle, Bob. 2002. The Australian fiscal equalization system and capital transactions: The final step. International Centre for Economic Research Working Paper No. 0212002, Torino, Italy. World Bank. 2003. Decentralizing Indonesia, A regional public expenditure review overview report.
Chapter 13
THE NATURE AND FUNCTIONS OF TIED GRANTS BOB SEARLE* AND JORGE MARTINEZ-VAZQUEZ** Consultant on Fiscal Decentralization, Formerly the Commonwealth Grants Commission, Australian Government** Department of Economics, Andrew Young School of Policy Studies, Georgia State University
1.
INTRODUCTION
Tied (or conditional) grants are typically a key element of most intergovernmental fiscal transfer systems.' They are an effective tool for central and upper level governments to pursue a variety of objectives but in order to understand how they work, tied grants need to be seen as part of the whole system of intergovernmental fiscal relations operating in a nation. This paper discusses many aspects of the design and management of tied grants, and concludes that they need far more careful consideration than is often given to them. Before discussing specific issues relating to tied grants, however, we think it will be useful to clarify some terminology and review the place and role of tied grants in the overall context of intergovernmental fiscal relations.
2.
SOME TERMINOLOGY ISSUES
The terminology used in discussing intergovernmental fiscal relations issues, as in the case of other fields of public finance, is as yet far from standard. Different authors use different expressions to mean the same thing, as in the case of tied, conditional, or specific purpose grants. Often, terms are not clearly differentiated and in some cases the same terms are used with different meaning by different authors. In this way, it is as if readers are implicitly asked to come to grips with an author's use of terminology before they can understand what is being said. Therefore, we believe that a discussion of issues associated with tied grants should begin
404
Searle and Martinez-Vazquez
with what we in this paper assume them to be. At the same time, it is beneficial to place what we mean by tied grants in the wider context of intergovernmental fiscal relations. Grants, transfers andpayments are the first point on which clarification is needed as they are often used interchangeably. Generally, the term "transfers" is used to mean all funds that are moved from one level of government (usually the central government) to a n ~ t h e r .Grants ~ are funds given by one government to another to perform the constitutional or statutory mandates of the recipient government. Payments are transfers from one government to another for services performed by the recipient government on behalf of the transferring government - these services are usually performed under a contract or other formal agreement between the two governments. Most frequently, payments are for services that have already been performed and for which an account has been issued; while grants relate to services that have yet to be performed and are not received as a result of an account being issued. The two most general types of grants are first, those that the recipient government can spend as it wishes and, second, those that are required to be used for pre-determined functions by the recipient government. Depending on the national context, the frst type might be described as untied, unconditional or general purpose grants. As the names indicate, these transfers are made to the recipient government for it to do with as it likes.3 Untied grants represent a general shift outwards in the budget constraints of the recipient governments and therefore they may lead to tax cuts for the taxpayers of the recipient jurisdictions as well as (or instead of) more or better-quality services. We choose to use the term 'untied grants' when discussing this type of transfer. Some authors have argued that, at least in a unitary country that has decentralized government, there are really no such things as untied grants. In those circumstances, the sub national governments exist as a function of central government legislation that specified the functions for which those governments are responsible. It can therefore be argued that even where there are no conditions attached to the transferred funds, there is a strong implication that they are to be used for the functions mandated to the sub national governments.4 Depending on the structure of federal countries, this argument might also apply there. It is an interesting discussion but one we need not enter in this paper. Untied grants are most often used as the instrument through which central governments address vertical and horizontal fiscal imbalances in the system of intergovernmental fiscal relations. In the case of vertical imbalance^,^ central governments may use, for example, different forms of tax sharing with subnational jurisdiction^.^ Horizontal fiscal imbalances are typically addressed via (untied) equalization grants, which may or may not take into
The Nature and Functions of Tied Grants
405
account differences in fiscal capacity andlor expenditure needs of recipient governments, and thus may equalize to different extents across sub-national jurisdictions with different economic resources and tax bases, different costs in the provision of public services and so on. The second category of grants is described as tied, conditional, spec@ purpose, block, sectoral, functional, categorical or earmarked grants. In general, all these alternative terms have the same meaning and are often used interchangeably. The first two imply a wider set of conditions than the later four because they allow for conditions that are not related to the function for which the transfer is provided. The later four terms all imply that the conditions under which the transfer is made apply only to the function for which the transfer is provided. There is sometimes a differentiation made in the terminology based on the type of specifications that are attached to the receipt of the grant, but this is unusual. Sometimes there are also more subtle differentiations in the use of these terms, indicating different degrees of autonomy to spend the funds and so on. For example, a block grant is used for transfers that must be spent by sub-national governments on a specific sector (for example, education) but with sub-national governments maintaining full control over how the funds are spent within that sector; as opposed to specific purpose grants that usually have more narrowly defined conditions, for example that the funds must be spent for free school lunches. We choose to use the term tied grants when discussing this general type of transfers. As will be seen, this does not imply that the tying (the conditions applied to the receipt of the grants) is only about the functions on which the grants are to be spent. While governments in different countries rely on tied intergovernmental transfers for different reasons, the most generally accepted reasons for using this type of transfer include: Assuring the delivery of delegated services to sub-national governments; Compensating for the presence of (positive or negative) spillovers or "externalities" between jurisdictions in the provision of regional or local public services; Offering additional incentives for funding national priorities and "merit goods" that have been assigned as responsibilities of sub-national governments; and Simply compensating sub-national governments for the costs they have incurred in the implementation of central government programs.
In the most general sense, tied grants are therefore transfers of funds aimed at increasing the level of spending (or at least guaranteeing a minimum level) on functions that are the mandatory responsibilities of the recipient government.7 All other reasons behind tied grants (externalities,
406
Searle and Martinez-Vazquez
merit goods and so on) also have the basic purpose of stimulating the level of spending of sub-national governments in particular areas. Whether tied grants actually achieve their general objective of stimulating expenditures on specific areas of expenditures by local governments ultimately depends on the incentives and constraints the grant arrangements offer to these jurisdictions. It is important to note that all government moneys are fungible. In the case of untied grants, for example, there is nothing to stop sub-national jurisdictions from rebating taxes to their residents and in this manner indirectly not spending part or all of the untied grants they receive. In practice, sub-national governments appear to spend, and not rebate, a large portion of the grant funds they receive. In the fiscal decentralization literature, this phenomenon has received the colorful name of the "fly paper" effect, representing the image that "money sticks where it first hits".' By comparison, tied matching grants present sub-national governments with greater incentives to increase expenditures in some particular areas. In this case, the central government provides partial cost reimbursement for certain sub-national government activities. Matching grants are a special type of tied grants which require a specific matching contribution by the subnational government in a particular expenditure area in order to be eligible for the grant. In effect, this type of grant reduces the relative price of spending in the particular service or activity under the umbrella of the grant. For example, the central government may match with a dollar every dollar that a sub-national government spends on that particular activity from its own resource^.^ The most stimulative type of transfer is likely to be "payments" from the central to sub-national governments with full reimbursement for certain approved costs. For example, the central government may agree to fully reimburse sub-national governments for the costs of vaccinations made available through local governments as part of national illness prevention programs. Diagrammatically the relationship between transfers, grants and payments, and the types of grants, can be presented as shown in Figure 1.
3.
FISCAL TRANSFERS AND DECENTRALIZATION
As it is presented in the literature, decentralization is about establishing andlor giving sub-national governments power to make decisions about revenue raising and service delivery in the local area. As discussed above, most frequently intergovernmental decentralization design also results in a
The Nature and Functions of Tied Grants
Transfers
I
Grants - for functions that are the responsibility of the recipient government
Untied
I
Payments - for
functions that are the responsibility of the transferring government
Tied
I
Matching and Non-Matching
Figure I. Transfers, Grants and Payments
vertical fiscal imbalance in favor of the central government and a need for grants to the sub-national governments. What form the intergovernmental grants system takes is an important indicator of the extent to which real decentralization is being achieved. In particular, it tells us about the real power being given to the sub-national governments. If the majority of the grant funding is being given as untied funding, it can be seen that the central government is truly decentralizing decision making authority and forcing the local governments to decide what services are to be provided.'0 If, on the other hand, the grants to sub-national governments are all tied and those grants have very specific conditions attached to them in relation to service delivery, it can be implied that the central government is maintaining a lot of control over what the sub national governments are doing -decentralization is far less complete. As the central government reduces the degree of specification in the tied grant conditions, it could be said that the real level of decentralization is increasing because, at least within functional groups of services, the sub national governments have power to make decisions and influence the types and quantities, as well as the manner, in which their services are delivered. The use of different grant types and different types of conditions is ultimately a question of the power balance between the giving and the receiving governments. It is therefore a good indication of how real a nation's decentralization is. In Sierra Leone for example, which is involved in a phased devolution of many major functions to local councils established in July 2004, the legal and administrative framework is structured around a strategy that will:
Searle and Martinez- Vazquez
408
1. Commence with a separate tied grant for each function devolved to local councils; 2. Commence with very detailed tied grants for each (or most) of those functions; 3. Remove the detail within the conditions as councils are capable of accepting more authority and autonomy, until most of the funding for each function is a general tied (block) grant; 4. Maintain sufficient funding through detailed tied grants to give the central government capacity to achieve 'national7 objectives as opposed to local objectives; and 5. Eventually combine most of the general tied grants into a large untied grant, with the tied grants aimed at achieving national objectives continuing to run in parallel.
The strategy for grant design in Sierra Leone parallels to a large extent that quite successfully adopted in ~anzania." What form the grant system takes therefore can have very important implications for public service delivery in a nation. The grant system to a large extent distributes the decision-making power between the different levels of government and shapes the service delivery pattern to the achievement of either nationally identified priorities and objectives or locally identified priorities.
TYPES OF CONDITIONS There are several ways in which the types of conditions attached to tied grants might be classified. For purposes of discussion, we differentiate between conditions relating to service delivery, financial arrangements, administrative procedures, information requirements and public relations. In many countries, all five types of conditions are used in tied grant arrangements. Service delivery conditions are those that specify the function on which the funds are to be spent. They are the most common condition attached to a tied grant but can range from being very specific to being very general in nature. A grant that must be used for primary education, for example, could specify 'studentlteacher ratios, number of years schooling, number of computers per class, text books per student' etc. On the other hand, it could still be a tied grant for education but be 'for primary education purposes7. When a tied grant is simply given for a function and there are few if any specific conditions attached to it, it is sometimes referred to as a block grant. Service delivery conditions can relate to the client group as a whole or to a specific subset of that group. The primary education grant could be
The Nature and Functions of Tied Grants
409
designed to benefit all students, or could be specifically to provide better services for disadvantaged groups of students. Special services for indigenous or other minority groups of service users are often provided for through this means. Service delivery conditions can be about how services are delivered as well as what services are delivered. The grant may, for example, be conditional on what curriculum is used or what limits are placed on health diagnostic tests undertaken. They can even specify what services are not to be delivered. In the health field, they may limit the age after which no heart replacement surgery or IVF services are provided from public funds. A grant to support university education might specify that no postgraduate degree is to be funded from this source. Financial conditions most frequently show themselves in the matching requirements (sometimes also called 'counterpart funding requirements'). As discussed above, these conditions are such that the sub national government must contribute to the function from its own revenues according to a predetermined proportion. As we saw, the advantage to the central government in using these conditions is that they increase the leverage it can place on sub national expenditure patterns. It is relevant here to also point out some of the disadvantages associated with matching provisions in tied grants. First, depending on the grant arrangements, a potentially large disadvantage can be that sub-national governments with greater fiscal capacity can take more from the grants pool than those with a lower capacity to co-finance, resulting in larger inequalities of service within a nation. Sometimes, central governments attempt to address this issue by establishing different matching rates depending on the fiscal capacity of sub-national governments. Matching requirements can also lead to inefficiencies by the spending governments. The incentive to spend more in specific areas provided by matching grants, a desirable thing by design, may also have a negative side. It is common for sub-national governments, for example, to talk of matching grants as being '50 cent dollars', especially where the priorities accord with theirs anyway. Psychologically and otherwise, it is more difficult to be as concerned about higher unit costs and forms of inefficiencies when you are only spending '50 cent dollars.' The cost benefit analysis of the recipient governments is quite different to that existing when all the funds spent have been raised by the sub national government. Clearly, this issue varies in importance as the ratio of matching requirement varies. Financial matching conditions may not only vary depending on the fiscal capacity of the jurisdiction, they can also vary depending on whether the grant is for capital or recurrent purposes. And despite the incentive provided by the effectively lower price provided by the matching grant, there is no guarantee that they will actually have an impact. Where they are supporting
410
Searle and Martinez-Vazquez
a new function, matching requirements may result in additional expenditure on that function. Where they are 'new' support for an existing function, however, a matching requirement may only become a substitute for the expenditure that the sub national government was funding from its own resources. To avoid this, central governments may accompany a matching requirement with a 'maintenance of effort' requirement and then match the grant funding to additional effort. But this may not result in any additional expenditure on the hnction if the recipient government believes its budget is already satisfying the priority that function needs within its budget. It may not be prepared to spend even '50 cent dollars' on the function. The timing requirements of matching grants may also lead to inefficiencies in sub-national governments' expenditure patterns. In some developing and transitional countries, central government agencies are often late in designing and arranging funding for their matching grant programs. This leads sub-national governments to hoard liquid budgetary resources so as to be able to achieve the matching fund requirements when the funds are needed. The funds are then spent in a huny toward the end of the fiscal year. Matching grants can be a danger to a central government's hard budget constraint if they are open-ended. In these circumstances, the sub national governments dictate the ultimate level of central government expenditure by their decisions on what they want matched. It is therefore important for the central government to put an annual cap on the funds to be made available under any matching grant program. This in turn creates a problem of resource allocation because, presumably, the central government would not be happy with a simple 'first-come-first-served' basis of allocation to the sub national governments and would like some control over the locational distribution of the available funds. Recipient governments tend to dislike matching grants because they limit the flexibility they can use in determining their budget expenditure pattern. They believe matching conditions give the central government too much leverage to influence the sub national budget. This is a real problem where the increased grant funding is provided as a result of a small but politically strong minority where there might be a real case for service enhancement in some areas but the central government makes the offer nation-wide. Some jurisdictions may have correctly assessed that they need not spend any more on the function being funded, but are now 'forced' to do so or be seen by their constituents as rejecting an offer of central government assistance that other sub-national governments are benefiting from. Administrative conditions are most frequently about the environment in which services are provided. They might, for example, have some relationship to the quality of the services provided and require teachers' qualifications and salaries to be at a set level or for there to be a certain number of textbooks for each student. They might be about the environment
The Nature and Functions of Tied Grants
41 1
in which the staff work and require the government providing the services to have written policy statements and agreed practices on a variety of things. However, they might also be related to efficiency in production and service delivery and the fight against corruption, and require sub-national governments to conform to pre-determined (international) standards of procurement and asset control. Whatever their purpose, administrative conditions can either be an undesirable interference in the autonomous decision making of sub-national governments or result in an improvement in the processes and the quality of the services provided. Information conditions are important to the longer-term management of the function and can be used to develop databases beneficial to resource allocation both within and between jurisdictions. They relate to the data the recipient governments must give the central government. This information may provide not only data to evaluate the effectiveness of the tied grant program but also data for a greater understanding of the service being funded. In an education grant, for example, the information recipients are required to provide might cover not only how the money was spent, but also basic data on how many schools and teachers there are in the jurisdiction, the composition of the student body (to identify special funding groups), the quality and outcome of the education the students received, and so on. Information conditions can be very important in assisting service providers to understand their environment. Managers will often have an overall feel for what drives variations in demand for their service and what causes their costs to vary, but they do not often have the data to know just how much more use one group actually makes of their service than others, or what the differences in unit cost are for different clients or different services. An information provision requirement that covers these aspects provides both a better base for service provision planning and a better base to make inter-regional comparisons of the standard of services and which jurisdictions are operating more efficiently than others. Public relations conditions are not essential to service delivery or efficiency but are favored by politicians. These conditions are usually about politics and re-election and a perceived need of politicians to take credit for good news. An extreme example is the requirement for a sign at a roadworks site that tells the public the central government Minister for Roads (Mr X) gave the sub national government the money to do the work. A less prominent example is the discrete logo of the providing central government Ministry on the front of the brochures advertising a new program. We have previously seen how a service delivery condition can be either specific or general in its requirement, but each of the other types of conditions can also be detailed or general in their makeup. They can also
Searle and Martinez- Vazquez
412
vary for capital as opposed to recurrent grants and may have to vary depending on whether they are for short term as against long term programs. It could be claimed that very short term (catastrophe-based) grants are a separate category of grants, but we think it is better to see them as a subset of tied grants. In this way, we can be more certain that all the relevant issues are likely to be raised during the design stage of the program.
5.
WHY USE TIED GRANTS?
For the recipient governments, the answer to this question is easy. Very likely, they would prefer not to have any tied grants and accept all the grants as untied funding. Sub-national governments would argue that they are in a better position to assess the expenditure needs of the local people anyway and there should be no interference from the central government. From their perspective, all tied grants do is force the distribution of scarce resources away from what the local people desire. From the central government's perspective, tied grants are favored because they give them a capacity to:
1. Influence (control) the standard of services being provided by each of the sub national governments - in this regard, equalization of service provision is oRen cited as a reason for tied grants; 2. Influence the range and level of particular services being provided across the nation - in this regard, tied grants serve a purpose that untied grants cannot achieve as they enable the central government to force an improvement in specific services that are the responsibility of the sub national governments; 3. Ensure minority groups and the disadvantaged receive a minimum level of services; 4. Ensure national priorities and objectives are met within sub national service provision; 5. Ensure cross-border service provision externalities are catered for appropriately by adjusting the funding received by sub national governments; 6. Improve the long term management of the government function by providing a better database for future managers; and 7. Take credit for good performances of the sub national governments, while still being able to blame those governments for poor performances. Given the relative revenue raising strength of most central governments, these are strong reasons why very few systems of intergovernmental fiscal
The Nature and Functions of Tied Grants
413
relations do not have tied grants as one of their major components. Most countries have very large numbers of tied grants and tend to have specific conditions rather than general arrangements that might be considered in a block grant. Ultimately, central governments establish tied grant programs because of the level of influence they give them over the activities of the sub national governments. The programs may be dressed-up as the best way to see that community expectations of government are met, but they are, in reality, about control and power. As mentioned earlier, they can be seen as a negative influence on the general movement towards accountable and efficient decentralization that otherwise is being pursued by those same central governments.
LIMITATIONS IN THE USE OF TIED GRANTS Tied grants can be used very effectively to ensure that pre-determined levels of input, and perhaps output or outcome, are achieved for any specific function provided by the sub national governments. They can be targeted very well to ensure the needs of small groups of service users are met and can be used to achieve equalization of (minimum) service provision by all governments at the sub national level. But it is more difficult to use tied grants to adjust for differences in the revenue raising capacity of the recipient governments. For example, it would be possible to calculate each government's revenue capacity on a Representative Tax System (RTS) approach and establish a block grant program that: (i) was to h d specific services provided by sub-national governments; and (ii) was received on the sole condition that the RTS-based revenue collections were met. Such a scheme would be very difficult to manage, however, and for all practical purposes, the tied grants would remain largely unrelated to revenue capacities. However, some governments use formulas to distribute tied or block grants, often with some elements in these formulas tried to approximate the fiscal capacity of sub-national governments. This is, for example, the approach recently introduced in ~anzania." Tied grants are therefore more widely used to achieve service provision and delivery objectives than to overcome differences in revenue raising capacity. The overall effectiveness of their use therefore depends on the objectives of the central government. The importance of the inability of tied grants to adjust for revenue capacity differences depends, of course, on the importance of own-source revenue in the budgets of the sub national governments. If they have little or nor capacity, as in South Africa and many other developing countries, this inability has little significance.
414
Searle and Martinez-Vazquez
TIED GRANTS AND EFFICIENCY There is a great deal of literature on this aspect of tied grant systems.13 We need here only to briefly mention several aspects of both economic and administrative efficiency that relate to them. Economic EfJiciency. We have already discussed several ways in which the use of tied grants can influence the efficiency of public sector service provision. All of them are related in some way to the fact that they may undermine accountability of the recipient government because that government is not being responsible to the people it serves for raising revenue from those people. The mismatch between revenue collections and expenditure associated with vertical fiscal imbalances and the preponderance of grants results in reduced accountability and efficiency. Governments would operate more efficiently if each had to answer to its people for all the money it wanted to spend, and thus raise it in taxes. This is not a common experience, however. Many countries seemed to have decided for better or worse that any inefficiencies arising from the concentration of revenue collection in the hands of the central government can be justified in terms of greater revenue collection efficiency by that government andlor a more appropriate delivery of sub national government services across the nation when local revenue capacity is not the sole determinant of local service standards. Administrative EfJiciency. We will see later that a tied grant system requires an administrative structure at both the central and sub national levels of government. It is often argued that these administrations duplicate one another's efforts and that they can bring a great deal of administrative inefficiency into the public sector. The Commonwealth Grants Commission in Australia estimated in 1990 that, for 1988-89, the administrative costs associated with seven of the largest tied grant programs was just under 1 percent of the value of the programs, but this varied considerably from grant-to-grant and was much larger where detailed conditions rather than block grant approaches were taken. l4 Administration costs are likely to be much larger for small tied grant programs. Although there have been recent reforms in many OECD countries to consolidate large numbers of tied grants programs into a smaller number of block grants, in many developing and transitional countries the natural (politically driven) proliferation of small tied grant programs and sub programs remains a concern in this regard. Tied grants also cloud the lines of responsibility and reduce the community's understanding of what level of government is responsible for what services. They result in a general reduction in government responsibility by enabling the sub-national governments to blame the central government when things go wrong, and the central government to continue to blame the sub-national governments even though it may be responsible
The Nature and Functions of Tied Grants
415
for the ultimate level of funding being allocated to the h c t i o n and the particular details for how a service gets delivered. The time perspective of a tied grant program can also be an influence on efficiency. The recipient governments are more interested in a program if they know it is to be 'long-term' in its nature. They will then establish proper administrative systems and try to be as efficient as possible in its operation. If, on the other hand, they think the program is a 'one-off or will not last beyond two or three years, there is little incentive for them to take on permanent staff to manage it and incorporate it into their normal operations. One of the discussion points of any tied grant program is therefore how long the central government is prepared to provide funding for it. On the other hand, the incentive of the central government is, on most occasions, to eventually have the service funded or expanded by the tied grant incorporated into the sub-national service provision as part of their normal activities. For this reason, it will encourage the sub-national governments to assume continuity of the service fiom the start. It is unlikely however, to promise continuity of funding beyond a three or five year period.
8.
DESIGNING THE TIED GRANT SYSTEM DETERMINING THE TOTAL POOL
The ultimate decision on the size of the pool of funds to be allocated to tied grants is usually at the discretion of the central government. It presents some difficult management issues, however, because the total and the division of that total between programs (and between capital and recurrent) cannot be considered in isolation from one another. There are probably at least as many pressure points for increases in the available funds as there are tied grant programs. In most countries, there is less initial emphasis placed on the total pool of tied grants than there is on the total of each of the tied grant programs and the total pool is the result of many individual decisions. It is only after the proposed total has been arrived at in this way that there is any evaluation of it and either adjustments made to individual grant programs to make the total fit within the central government's budgetary constraint, or the total to be distributed as tied funds adjusted to meet that requirement. This is not the best way to proceed. It would be better if, before commencing the decisions leading to the size of each tied grant program, the central government decided the overall size of the grants to sub national governments, and the mix of those grants between tied and untied, capital and recurrent, as part of its budgetary considerations in the context of a Medium Term Expenditure Framework (MTEF). Grants are often a very
416
Searle and Martinez- Vazquez
large element of a central government's budget and not to have this level of control over them is detrimental to good budgeting practice. It may not be possible to estimate the need for 'catastrophe' based grants because they will be unknown, but even those might have a tentative amount placed on them as a precaution for budget management purposes. Capital grants are more likely than others to be funded from loan funds. If this is to be the case, the budgetary considerations are quite different from those associated with grants for recurrent purposes funded from central government recurrent revenue. Grants funded from loan funds will need separate consideration in the budget and separate discussions with the sub national governments.'5 The level of debt (or future debt) may be a determinant of the size of the capital grants the central government is prepared to make available. We have seen that the design of the grant system in a country indicates where the overall control by governments lies. The greater importance of tied grants in the financing of sub-national governments is a clear indication that much of the power lies with the center and little of it with the "transfers dependent" sub-national governments. However, the proliferation of tied grants also speaks of the diffusion of power at the center. To just 'let it happen' in this way, as is often the case, seems irresponsible. When there are several hundred tied grant programs and sub programs, as is common, where and how are reductions to be approached if the resulting proposed total is too high and reductions have to be made within the tied grant component of a nation's fiscal transfer system?
9.
DESIGNING THE TIED GRANT SYSTEM DETERMINING THE DISTRIBUTION OF EACH GRANT
The distribution of a tied grant between recipient governments would be expected to match the intention of the government initiating the program. It may also differ for grants intended to cover capital as compared to those for recurrent purposes. If the true objective of the central government is to put more money into electorates that favor its re-election, as discussed by Khemani (this volume), the information needed would be very different from where the intention is to distribute the available funds on a 'needs' basis. In that case, the definition of need has to be determined and a distribution system designed that is based on that definition. Basic questions of intent such as these are political decisions and will vary across grant programs. Once they have been answered, the subsidiary question is whether the calculation of the distribution is done by the
The Nature and Functions of Tied Grants
417
responsible line Ministry, the Ministry of Finance or a separate and independent agency of the central government. The most usual approach is to have line Ministries do the calculations but that can change depending on the basis of distribution required. At the extreme of political expediency, the distribution might be effectively determined by the staff of the ruling political party, based on electoral voting patterns, but nominally distributed by some government agency. The definition of need will change depending on whether the grant is to fund capital or recurrent functions. On an assumption of recurrent funding, the definition can be based either on a desire to increase the standard of a service overall, or on the relative needs so that the available resources are spread across the recipients to provide an equal level of services in all areas. In the former case, there is usually a net increase in the level of required expenditure so a 'maintenance of effort' and/or matching requirement is put in place. In the later, which is more often associated with a long-standing grant program, governments are happy with the absolute level of services being provided and the central government wishes only to influence the spatial distribution of that expenditure. Where equalization of service provision capacity is the objective, the basis of distribution may not be too dissimilar to that discussed in the paper by Vaillancourt and Bird in this volume. The distribution between jurisdictions would usually be based on variations in the inherent demand for the service, based on some objective data, and may or may not also take account of variations in unit cost. For most services, and therefore most tied grants, the data will relate to demographic differences but may also be such things as the length and type of roads, the number and size of bridges, the length of coastline, the number of farms or the total number of properties, depending on the function being funded. The important thing is that actual levels of demand and/or actual cost structures are not used to measure relative need. To do so establishes a grant design inefficiency through which the recipient governments compete with one another to change their service provision levels and/or costs to try to get a larger share of the grant. Where this occurs, the grant distribution method can lead directly to the inefficient use of resources because sub-national governments are not making optimal resource allocation decisions. Most frequently, needs-based tied grants for capital purposes are for 'approved projects' and require the submission of proposals by recipient governments. An alternative has been proposed in the paper on public infrastructure equalization by Levtchenkova and Petchey in this volume. No matter how it is approached, determining a needs-based distribution for capital grants requires a completely different administrative approach to a recurrent grants program, but it still requires many aspects to be pre-set. These include the type of projects that can be approved, what the maximum
418
Searle and Martinez- Vazquez
(and minimum) grant will be for any one project and government, and what costs are to be accepted as capital for this purpose.
10.
DESIGNING THE TIED GRANT SYSTEM DETERMINING THE CONDITIONS OF EACH GRANT
We have seen that the central government can apply a wide range of conditions to any tied grant program and that these will differ depending on the degree to which it is willing to give real autonomy to the sub-national governments. Ultimately, it is the role of the central government to determine the conditions it wants to attach to 'its' money. At the extreme, the sub-national government either accepts the results and receives the grant, or rejects the conditions and does not receive any grant. What the central government should do when deciding what conditions it wants to apply is to consider the full range of objectives it has relating to the service being funded, both in the short and the longer term. It should also think of any non-service provision objectives it might have that could be achieved by conditions attached to a grant. It should then consider in turn each type of condition discussed earlier and consolidate its requirements before presenting them to the sub-national governments. Obviously, conditions need to be realistic, achievable and manageable by both the central and sub national governments. Some are better initiated by the line Ministry and some are more likely to be initiated by the Ministry of Finance (or even the Minister representing political interests). In the final event, the set of conditions it is proposed be attached to a tied grant should be the product of discussions between at least the line Ministry and the Ministry of Finance in the central government. We would expect the line Ministry to be more aware of the need for service delivery, administrative and data provision conditions and the Ministry of Finance to be more concerned with financial conditions and budget incentives. What is clear is that no one central government agency should cover all the possible conditions that need to be considered.
11.
THE ROLE OF SUB NATIONAL GOVERNMENTS IN DESIGNING THE TIED GRANT SYSTEM
It is because a tied grant program should be realistic, achievable and manageable that the sub-national governments should be involved in the process of developing the program. In the best circumstances, the central
The Nature and Functions of Tied Grants
419
government negotiates the value of the grant, the basis of distribution and the conditions that are applied to it. The sub national governments can always be over-ruled (and it is likely that they would be over-ruled on most issues) but it benefits the process greatly if they are given their voice, on the conditions in particular, as a reality check on the feasibility of what is being proposed. The conditions of tied grants are rarely achieved without cost to the recipient governments and these costs are better estimated by those involved rather than the central government. The extent of the involvement of sub national governments will vary depending on the type and size of the grant. For minor amounts, a committee of sub national government officials may be all that is needed to discuss the issue with central government staff. For major programs involving large amounts of money, the negotiations and discussions might be at the political level and involve all the relevant Ministers fiom both the central and sub national governments. Even in those cases, however, we would expect the political representatives only to deal with the issues of principle and the details to be sorted out by the relevant civil servants from both sides. The extent of the discussion of the program details prior to it being implemented is another indicator of the degree of control maintained by the central government. If it simply tells the recipients what is going to happen, we have less decentralization than if the sum of the grant, the basis of its distribution and the conditions attached to its receipt are negotiated and agreed prior to the implementation.
12.
THE APPROPRIATION OF TIED GRANTS BY THE CENTRAL GOVERNMENT
Once the size of the tied grant has been determined, it must be approved by the central parliament as part of the budget processes before the amounts can be paid to the sub national governments. There are two issues involved here. They are whether it is only the total of a grant that is included in the budget approval processes or the split between recipients as well; and whether the appropriation for a grant is part of the relevant line Ministry's vote or done from a central location within the budget and controlled by the Ministry of Finance. Assuming Parliament has the power to change the Government's budget, the first issue is based on whether or not we want Parliament to have the power to change the grant distribution from that derived on the basis of the pre-determined method. It can only do this if the distribution to each recipient is part of the budget. As the distribution does not have budgetary implications and the Parliament would only wish to change it for political
420
Searle and Martinez-Vazquez
reasons, we suggest the only aspect of tied grants that should be part of the budget should be the total hnds allocated to the grant. The distribution of that amount between the sub national governments should be controlled by the Government andlor the Parliament (in that they must determine or approve the pre-set basis of distribution) but they must not then have the power to change the distribution simply because they do not like the results. If it wishes to change the distribution, the Government andlor the Parliament must either change the required basis of distribution or show that their required basis has not been correctly implemented. Whether the tied grants are appropriated as part of a line Ministry's vote or this is done under the control of the Ministry of Finance is a question of control within the central government budget implementation processes. What is important is that the payments are made on time and according to the pre-determined distribution, with as little administrative cost as possible. In less developed public sectors or where corruption is a greater possibility in the line Ministries, there is usually more management capacity in the Ministry of Finance and this is also a consideration. In most cases, these considerations lead to the vote and the payments being controlled by the Ministry of Finance, but this is probably not essential if the government has a well-designed and effective expenditure control system in place. A matter of lesser importance but one that can be important to the attitude sub-national governments take to the program is whether or not funding for it is included in any Medium Term Expenditure Framework (MTEF) data published by the central government in its budget documentation. If the recipients can see that the central government is assuming a continuity of the program, at least for the MTEF period, they will be happier to conform to the conditions of the program and take more effort and attention with staffing and administration to ensure it runs effectively.
13.
TRANSFERRING THE FUNDS TO SUB NATIONAL GOVERNMENTS
We have talked thus far only about conditions that are imposed on the recipient governments. One of the few conditions that fall on the central government is to make payments according to some pre-determined frequency and method. Depending on what is being funded by the grant, the frequency of payment might be weekly, monthly, quarterly or as required to pay accounts (in the case of capital developments). This will be a condition of the grant that the sub-national governments have a keen interest in and they will argue for payments in advance and as frequently as possible. Where financial conditions, particularly matching requirements, are imposed, there is a need to agree how these will influence the frequency and
The Nature and Functions of Tied Grants
42 1
rate of funds transfer. Will the sub-national governments be expected to make their contribution to the program before any tied grant is made, or will the tied grant be provided on an imprest basis and the imprest account simply not be replenished unless the sub national contribution is made? There are also matters of funds control that will be subject of the nation's accounting and financial management standards that will need consideration. Will the recipient governments be required to have a separate bank account for each grant, for example or should it be managed in a unified treasury account? In a nation where there is a national standard for financial management and accounting, these aspects might not be relevant but, in cases where the sub national governments operate to their own standards and those differ from the central government's, financial management aspects may be important elements of the financial conditions attached to a tied grant program.
14.
REPORTING, MONITORING AND EVALUATION OF TIED GRANT PROGRAMS
It must be assumed that the continuation of payments under a tied grant program is dependent on the recipient conforming with the conditions attached to the program. In reality, central governments are probably loathe to discontinue payments but, for the purposes of this paper, we must assume the conditions of the grant are to be adhered to. Checking on this adherence requires a system of reporting, monitoring and evaluation, not on the effectiveness of the program but on whether the conditions for future payments have been met. Reporting and Monitoring. Like the level of negotiations involved in developing a tied grant program, the means and level of reporting by the recipient governments will vary with the size and importance of the program. For minor grants, the central government may accept a simple statement from the CEO of the recipient government's line Ministry or Ministry of Finance saying the funds have been spent and the conditions have been adhered to. In these circumstances, and particularly where the conditions require the provision of information or data to the central government, this might be all the monitoring that is done. At the other extreme, the reporting might require a detailed audit report and statement by the recipient line Minister or Minister of Finance, as well as the provision of the required data and other information. In both cases, as indicated earlier with reference to any financial matching requirement, there may or may not be a need to report and monitor conformity on a regular basis - presumably with the same frequency as payments.
422
Searle and Martinez- Vazquez
In unitary countries, even if decentralized, the national auditor-general usually has power (and often a legal responsibility) to audit the activities of sub national governments. This is not always the case in federations and, in some, it is not legally possible without agreement between the parties. In these circumstances, difficulties can develop about what level of government does the audit associated with the financial reporting and monitoring, and what degree of acceptance the central government might give any audit report if it is done for (or by) the sub-national; government. This is more of a problem where a sub-national government has different audit standards to the central government and must be worked out before the tied grant program is finalized. Auditing of matching requirements and for grants for capital purposes have particular issues when examining for compliance. What expenses by the sub national governments qualify as matching? What expenditures are accepted as being capital in nature? It is not unknown for a recipient government to 'adjust' its accounting procedures to functionalize more overhead expenses once a matching requirement is introduced. Neither is it unknown for governments to adjust the definition of capital they are applying and include assets of lower value than before and overhead costs such as building design as capital expenses. These matters can be sorted out after the event but it is obviously better practice if they are considered and agreed before the grant program is finalized. Evaluation. Some conditions attached to the tied grant will lead themselves to evaluation rather than auditing. If service quality outcomes are a condition, for example, these will be measured in subjective as well as objective ways and require skills other than financial auditing. This is probably best done by the central government line Ministry responsible for the function (if such a Ministry exists) or an independent agency contracted to the central government.'6 The other aspect of evaluation that must form part of the program is a general program evaluation - is the public investment as measured by the financial inputs, giving the community the expected level of outputs and outcomes? This is irrelevant to the continuation of payments while the program continues to exist, but is highly relevant to the continuation of the program. It should also be an input into the continuation of specific conditions that might be currently included in the program. Ultimately, the central government must be satisfied that the conditions of the grant are being met before further payments are made. This will require some certification by a central government representative. Depending on the types of conditions that require monitoring and evaluation, this in turn might mean co-operation between the central government line Ministry and Minister of Finance. The most efficient place for the final certification to be made is in the agency from which the future payments have to be authorized.
The Nature and Functions of Tied Grants
423
This will dictate whether it is a line Ministry or a Ministry of Finance representative that makes the overall certification of compliance.
15.
TIED GRANTS AND COMMUNITY INVOLVEMENT
Because tied grants relate to a specific function of the sub national government, the area to which they apply is likely to have an active lobby group established to pressure public officials to put more resources into its area of special interest. An aged persons association will, for example, pressure governments to spend more on services its members use more frequently. A teachers' union will lobby governments to spend more on schools, and so on. These pressures will be applied at several stages of the tied grant process. They will add pressure to establish a tied grant in the hope of additional resources being allocated to their 'pet' function; they will add pressure to enlarge the share of public sector resources allocated to that function; and they will add pressure on the sub national governments to ensure the funds are spent effectively and efficiently. Overall, we can expect a higher level of community involvement in areas subject to tied grants than other activities of government.'7 Taken as a whole, we can also expect this community involvement to be an active force in correctly allocating resources to government service responsibilities and monitoring the activities of governments. Because of the monitoring fbnction performed by the community, we might also expect there to be less corruption in the provision of services where tied grants programs are operating. Some tied grant programs, particularly those relating to capital development, allow for a formal involvement of community groups in prioritizing the projects the recipient government proposes for funding. In these circumstances, the program might require the recipient government to show that the projects it proposes have been initiated by or have the agreement of the community before they can be accepted for funding. Tied grants give governments another opportunity to be transparent with the people and provide information on what they plan to do (and then what they have achieved). The program design processes result in all the documentation the people need to understand what their governments want to achieve, how they intend to achieve it and how they are going to measure what they have achieved. This documentation should be accumulated at a national level and published as part of the national budget documentation. It should provide details for all programs and sub programs and could be a valuable addition to the ability of the central government to check the acceptability of its proposals with the people and force its objectives on the
424
SearIe and Martinez- Vazquez
sub national governments. It would also be of assistance to the people in monitoring what their sub national governments are doing.
16.
ACCOUNTABILITY FOR TIED GRANTS SERVICES
Governments are accountable to the people they represent. The central government is accountable for the services for which it is responsible and the sub-national governments are accountable for the services they are mandated to provide. Tied grants bring an additional element into this framework, however, because they make the sub national governments accountable to the central government for that part of the supported service that is funded from the grant. These relationships are shown in Figure 2. On the negative side, as we discussed above, the funding of sub-national governments with grants decreases the overall accountability to its constituency v i s - h i s financing though its own taxes. The issue is whether or not the sub national governments are accountable to the people for the services funded fi-om tied grants. Maybe the answer depends on the degree of control they have in what services are provided and the closer the program gets to block funding, the more accountable they become for the services provided. A difficulty often encountered is that the people are unaware of such niceties and simply hold 'government' accountable. When something goes wrong in these circumstances, the central government correctly tells the people the function is a sub-national government responsibility, and omits any comment on the adequacy of the resources it has provided. The sub-national governments cannot (or do not) differentiate between the two components of its service provision and tell the people the central government is not giving them enough hnds to provide a better service. Thus, each level of government blames the other, the people are confused and no-one really takes responsibility. In this way, tied grants can be detrimental to governments' accountability.
17.
TIED GRANT AGREEMENTS
We have now seen that there are many aspects of a tied grant program that are better handled and clarified while the program is being developed. We have also seen that both sides have many commitments, although the
The Nature and Functions of Tied Grants
Central Government Accountability Services
b
The people
Sub National Governments , - - - - - - - - -2
hnded from tied grant programs
I
Services not hnded from tied grant programs
----------
Central Government The people
fi Accountability
I
I
Figure 2. Accountability for Government Services more obvious are for the payments to be made and the conditions of receipt to be adhered to. We believe tied grants programs benefit greatly fiom the development of a formal agreement between the central government and each of the recipients. Where possible, this should be a contract supportable under the law but that is not always possible depending on the administrative and legal regime of the country. In some nations, special legal provisions are necessary before one arm of the government can contract another arm of the same government to do something. In others, this is not legally possible under the constitution. Never-the-less, we believe formal documents are beneficial in clarifying the terms of the program and ensuring both parties have considered all aspects of their commitment before the program is finalized. A written document also gives a firmer basis for the reporting and monitoring associated with the program. The signing of a formal contract or agreement can also be a significant sign to the people of their governments' intentions and provide a basis for community monitoring of activities. Such events can be important milestones in a government's public relations and interaction with lobby groups and other community associations.
426
18.
Searle and Martinez- Vazquez
TIED GRANTS AND EQUALIZATION
We discussed earlier that the more common practice within intergovernmental fiscal transfer systems is to use untied grants to achieve any overall equalization of fiscal capacity and service provision that the nation aspires to. We have since seen that tied grant programs are the source of funding for some of the same services which might be subject to that equalization objective. There is therefore a question of how the tied grants and the equalization objective interact. Assuming for the point of discussion that equalization is about an equalization of service provision by sub-national governments (and several papers in this volume discuss the appropriateness of such an assumption), the issue is whether or not the services we are equalizing include or exclude those funded from tied grants. If all services are included in the equalization process, the grant receipts are counted as part of the revenue capacity of the recipient governments and the expenditure and services funded from the tied grants are seen as part of the standard of services currently being provided in each jurisdiction. Any inequitable distribution of services resulting from a tied grant will be equalized by an adjustment to the untied grant distribution based on equalization. If equalization is not to be applied to the services being fimded from tied grants, the grant revenue is not included in the revenue capacity measures of the recipient government, and the services and expenses resulting from the grant are not included as services being equalizing. Any inequitable distribution of services funded under the tied grant program is untouched by the equalization process. Ultimately, the approach to be taken to this question is a political one and the answer should come from the government. In some circumstances (where political considerations are applied to the distribution), the government will want the distribution of the services funded from the tied grant to be separated from the equalization process. In others, where the government wants some check on the activities of the line Ministries' distribution of tied grant funding, it may require the services to be included in the equalization process. This later approach is not popular with line Ministries which see the equalization process 'over-riding' their tied grant determinations. The Commonwealth Grants Commission in Australia, which has responsibility for recommending an equalization based distribution of an untied grant and includes both revenue and expenditure assessments in its calculations, has faced this question since the early 1970s. At that time, there was a large expansion in the tied grant programs (in both number and importance) and there were stark differences in how those grants were determined. The two largest were for schools education and health. One
The Nature and Functions of Tied Grants
427
resulted in States that spent more getting higher grants. The other resulted in States that spent more getting lower grants. Neither conformed with the principle of fiscal equalization the Commission was required to apply and it therefore developed criteria to determine how each tied grant should be handled. Appendix A holds details of that Commission's approach. Where tied grants are to be included as sub national revenue and the services are to be considered in the equalization process, any grants-in-kind will also have to be considered. In many nations, the central government chooses to become a direct provider of sub national services rather than give the sub national governments grants to expand their services. These are grants-in-kind and should normally be treated exactly as if they were actual grants. One thing that might make them different is that the sub-national government has absolutely nothing to do with setting the expenditure priorities and, if they are distributed on totally political grounds, for example, it may be inequitable to make an adjustment to the untied grant distribution because of the political decisions of the central government. The size of these grants-in-kind can also become relevant if they are very large and their treatment by 'inclusion' would imply an unreasonable degree In these of budget flexibility by the sub-national government. circumstances, part or none of the tied grant-in-kind might be treated by the inclusion approach. What is important is that the Ministry or agency charged with determining any equalization-based distribution of an untied grants pool knows what is required of it and has pre-determined and publicly known principles on which it will make its decisions of how tied and untied grants interact.
19.
HOW DO WE KNOW WHEN TIED GRANTS ACHIEVE THEIR OBJECTIVES?
In some circumstances, tied grants achieve all they were designed to achieve. In other circumstances, they achieve very little (if anything), even if they have specific conditions attached to them rather than being in the form of a block grant. If there is very little being spent on a service prior to the tied grant program being established, the tied grant is a net addition to the expenditure on the service and the conditions of the grant are adhered to, there is no doubt it will be more likely to achieve the objective of the central government, at least as measured from the financial input side. This is particularly so where the sub-national governments agree with or accept the policy priority implicit in the program and have little reason to avoid adherence with the conditions of the grant.
428
Searle and Martinez- Vazquez
If the function is for a large and existing hnction such as education or health, the tied grant is poorly designed and the monitoring that conditions are met is not well done, the tied grant can very quickly become simply a substitute for sub national government expenditure and achieve none of the objectives of the central government. This substitution effect is probably the most common reason why tied grant programs fail to achieve any specific goals in their design. It is particularly prevalent where the sub-national governments do not agree with the implied policy priority behind the grant and they therefore have an incentive to avoid conforming with the conditions. There seems to be a relationship between the size of tied grant programs and their effectiveness. This could be because small grants are usually more tightly targeted or address perceived needs for a separate and more readily identifiable subset of the population. Where grants are a large percent of the total spent on a service (that is where they are moving closer to being block grants), they are more difficult to focus on specific services or service users because they are being spent on more 'normal' parts of the function. They are therefore easier for the sub national governments to incorporate into their everyday service provision. We should note that tied grant agreements can be a negative influence on service provision change and efficiency. If the central government is forced into agreements that last too long (say, exceeding five years), the conditions written into those agreements can actually prevent service providers in the sub national governments from changing the methods of delivery andlor types of services provided to meet changing community expectations and requirements. In Australia, this is said to have happened with the largest of its tied grant programs - Health Care Grants - that are signed for five-year periods and, in effect, lock in levels of medical technology current when the agreements are signed.18
20.
TIEDIUNTIED GRANTS AND INTERNATIONAL DONORS
This paper has been written on an assumption of tied grants being paid from a central government to sub national governments. The same principles apply, however, with any tied grant from an international agency or donor nation to a developing country. In that case, grant assistance for general budget supplementation might be seen as untied grants and program grants might be seen as tied grants. In most instances, international aid will be synonymous with grants-inkind as the donor funding, particularly for program grants, is unlikely to be seen as part of the national budget.
The Nature and Functions of Tied Grants
429
Where sub national governments receive tied grant equivalent funding from international agencies of donor nations, they should be seen as the equivalent of tied grants (or grants-in-kind) and treated the same way in any equalization system of untied fhnding the nation might operate.
21.
CONCLUSION
This paper covers many issues associated with tied grants and has been structured to help those involved in the design of a tied grant program. Tied grants are beneficial for several reasons and it is most unusual not to have at least some tied grants in a fiscal transfer system. They are valuable tools through which to achieve national objectives, improve services to specific groups of people and ensure at least a minimum level of service is provided in all parts of the nation. They are not the preferred transfer mechanism of local government because they can be seen as limiting the recipient's budget flexibility but this negative aspect is oRen overstated. We have shown how the structure and type of conditions attached to a grant are strong indicators of the true level of autonomy the central government wants to pass to local government, and how this can vary over time by varying the conditions. Tied grant programs are thus powerfid tools in managing the relationship between the different tiers of government in a nation. We have also shown the relationship between tied grant conditions and governance and efficiency issues. There are arguments on both the affirmative and the negative sides of these aspects and it would be expected that different parties would have different views, depending on whether they are in favor of the conditions imposed or not. Because tied grants are such an important part of the governance structure within a nation, we are strongly of the view that it is necessary to involve the recipient governments in the design of a tied grant program. They will give a 'reality check' to the proposed arrangements and ensure the program is practical and effective. At the end of the program design, it is essential that there be a formal written agreement detailing all aspects of the program's administration and funding flows. Remember in this regard that the conditions must specify the frequency and size of payments as well as the obligations placed on the recipients. Agreements are better if they have legal status but should, in any case, have a dispute settlement clause allowing an independent court to arbitrate. Conditions attached to tied grants can be wide-ranging in their aims and are not limited to service delivery aspects of the program. A well designed agreement will avoid many disputes once the funds commence to flow and the resultant services are provided.
430
Searle and Martinez- Vazquez
Tied grants will usually be distributed between jurisdictions on the basis of need but this may not be the case. There is an important question regarding the relationship between tied and untied equalization grants and we argue that this is a matter on which governments must give direction. Ultimately, it is a question of the extent to which equalization is an objective for the services fwnded by the tied grant.
The Nature and Functions of Tied Grants
Appendix A. Equalization and the Treatment of Tied Grants The Australian Commonwealth Grants Commission Approach
There are three ways in which tied grants (SPPs or Specific Purpose Payments in Australian terminology) are treated in the Commonwealth Grants Commission's modela. In summary, the three approaches can be explained as follows: (i) The inclusion method, where expenditures financed from the tied grant and from State sources are both included in the standards used in the expenditure assessments. The tied grant itself is treated as part of the Commonwealth revenue payments available to finance part of the total financial assistance requirement of each State. (ii) The exclusion method, where the tied grant payment and expenditure funded from it are removed from the revenue and expenditure standards used in the assessments. (iii)The absorption method, where expenditures financed from the tied grant and from State sources are included in the expenditure standards. The tied grant revenue is treated as though it were part of the general revenue pool distributed among the States. The treatment given to each tied grant fiom the central government (the Commonwealth) to the States is arrived at separately after the Commission has asked itself the questions illustrated in Figure A-1 . Notes: a
Details of the calculation model are available in the Commonwealth Grants Commission Report on State Revenue Sharing Relativities, 2004 Review, Supporting Information, Attachment B, Canberra, 2004, available at www.cgc.gov.au
Searle and Martinez- Vazquez
Does the SPP provide assistance for a standard budget function?
Is the payment a reimbursement for services provided to or for the commonwealth? Should the payment be excluded for simplicity because it is distributed consistently with Commission assessments of State needs? Is the payment a minor payment for which needs are not assessed and the introduction of additional disability factors into the assessments would not be warranted? Is there a constraint on State budgets not allowed for in the assessment?
No, out of scope
EXCLUSION Yes (to any)
Yes
I
d Should the interstate distribution of the SPP influence the relativities?
1
INCLUSION
H
H No
ABSORPTION
Figure A-1. Treatment of Tied Grants
I
The Nature and Functions of Tied Grants
Notes This paper will use the terms "tied grant" and "conditional grant" to mean the same thing. As further discussed below, this type of grants also receive other names, such as "specific purpose grants." In hierarchical systems of intergovernmental relations, intermediate-level governments (states, provinces, regions, and so on) usually make downward transfers of funds to lower-level governments (municipalities, cities, counties, and so on). Less often, in some countries, lower-level governments also make transfers to upper level governments. This is the case, for example, when shared taxes are collected by the lower-level governments (as in Germany, or some provinces in Canada and some regions in Spain.) To emphasize this point, these grants are sometimes called "lump sum" grants. This is more clearly the case when central governments accompany the use of untied grants for financing sub-national governments with minimum expenditure requirements or standards (or norms) for service delivery in a number of sub-national functions. Vertical imbalances arise when the different levels of government do not have an adequate level of self-financing (mostly own taxes and other revenues) to cover the expenditure responsibilities assigned to them in the system. Tax sharing is one of the most common forms of untied vertical transfers and they are often computed on a derivation or origination basis. See the discussion in Searle, and in Bahl and Wallace (this volume). Some countries use tied grants also with equalization purposes by incorporating in the distribution formula elements of expenditure needs and/or fiscal capacity. See, for example, the discussion in Boex and Martinez-Vazquez (forthcoming ) for the case of Tanzania. See Courant, Gramlich and Rubinfeld (1979), Oates (1979) and Filimon, Romer and Rosenthal(l982). There is a large variety of matching tied grants. For example, the central government may use different matching rates for different categories of sub-national governments. Depending on the funds availability, matching grants can be open-ended or close-ended, with caps for each sub-national government or for the entire program. These issues are discussed further below. This has been, for example, largely the case in Indonesia that decentralized many services from 1 January 2001. However, Indonesia is already reassessing its 'big bang' approach to decentralization and there is now a move to include more tied grants in the system. See Boex and Martinez-Vazquez (forthcoming). In Tanzania, until now there has been no untied equalization grant program. Currently, the central government is exploring the introduction of an equalization grant but with limited funding vis-A-vis the block grants program. See, for example, Oates (1972), Gramlich (1997) and Bird and Smart (2002). Commonwealth Grants Commission (1990, p 167). This does not imply, of course, that the opportunity cost of funds for capital and recurrent purposes is not considered together and balanced against each other. For example, in the United States the Government Accountability Office (GAO), that responds to Congress, routinely investigates and evaluates tied grants and other federal .mom-ams. " This simply follows from the concentration of benefits in a relatively small group, as predicted by Mancur Olson in The Logic of Collective Action (1971). Discussions between the Ministry of Finance, Indonesia and the Australian Department e of Health, 1 June 2004 and the New South Wales State Health Department, 3 ~ u n 2004.
Searle and Martinez- Vazquez
References Ahmad, Ehtisham, and Martinez, Leo. Forthcoming. On the design and effectiveness of targeted expenditure programs. Bahl, Roy, and Sally Wallace. 2004. Intergovernmental transfers: The vertical sharing dimension. this volume. Bird, Richard M., and Michael Smart. 2002. Intergovernmental fiscal transfers: International lessons for developing countries. World Development, 30(6): 899-912. Boex, Jameson, and Jorge Martinez-Vazquez. Forthcoming. Fiscal decentralization in Tanzania. Palgrave. Commonwealth Grants Commission. 1990. Report on Issues in Fiscal Equalisation, Volume I1 - Appendixes and Consultant's Report. Canberra: Australian Government Publishing Service. Commonwealth Grants Commission. 1990. Report on Issues in Fiscal Equalisation, Volume I1 - Appendixes and Consultant's Report. Canberra: Australian Government Publishing Service. Commonwealth Grants Commission. 2004. Report on State Revenue Sharing Relativities, 2004 Review. Canberra: Australian Government Publishing Service. Also available at www.cgc.gov.au Commonwealth of Australia. 2004. Federal Financial Relations, Budget P a p a No. 3, 2004-05. Canberra: Australian Government Publishing Service. Courant, Paul N., Edward M. Gramlich and Daniel L. Rubinfield. 1979. The stimulative effects of intergovernmental grants: Or why money sticks where it hits. In Fiscal federalism and Grants-in-aid, Peter M. Mieszkowski and William H. Oakland, eds., Washington D.C.: The Urban Institute. Filimon, Radu, Thomas Romer, and Howard Rosenthal. 1982. Asymmetric information and agenda control: The bases of monopoly power and public spending. Journal of Public Economics, 17: 5 1-70. Gramlich, Edward M. 1997. Intergovernmental grants: A review of the empirical literature. In Financing Federal System: The Selected Essays of Edward M. Gramlich, E.M. Gramlich, ed., Cheltenham, UK. Edward Elgar. Khemani, Stuti. 2006. The political economy of equalization transfers. This volume. Levtchenkova, Sophia, and Petchey, Jeff. 2006. A model of public infrastructure equalization in transitional economies. this volume. Oates, Wallace E. 1972. Fiscal Federalism. New York: Harcourt Brace Jovanovich. Oates, Wallace E. 1979. Lump-sum intergovernmental grants have price effects. In Fiscal federalism and Grants-in-aid, Peter M. Mieszkowski and William H. Oakland, eds., Washington D.C.: The Urban Institute. Olson, Mancur. 1971. The Logic of Collective Action: Public Goods and the Theory of Groups. Cambridge, Mass.: Harvard University Press. Spahn, Paul Bernd. 2004. Intergovernmental transfers: The funding rule and mechanisms. This volume. Spahn, Paul Bernd. Forthcoming. Contract federalism. In Handbook on Fiscal Federalism, Ehtisham Ahmad and Brosio, Giorgio, eds. Vaillancourt, Francois, and Richard Bird. 2006. Expenditure-based equalization transfers. This volume.
Chapter 14
INTERGOVERNMENTAL LOANS: THEIR FIT INTO A TRANSFER SYSTEM DANA WEIST' The World Bank
1.
INTRODUCTION
Governments decentralize for various political and economic reasons. It can be a means to move decision making closer to people, to enhance the efficiency and responsiveness of service delivery, and to make tax systems more productive. In some countries, it may also promote national cohesion (e.g., Indonesia). Done well, decentralization can lead to all of the benefits promised by a multi-tiered intergovernmental system: better public services, enhanced local accountability, and a potential tool for poverty alleviation. But if decentralization is done badly, it can lead to macroeconomic instability, deterioration in service delivery, corruption and collapse of the safety net.2 Decentralization outcomes depend critically on the soundness of it's the system design - which spans political, fiscal, and administrative policies and institutions - and thoroughness of implementation (Litvack, et al. 1999). The key challenges are to balance responsibilities with accountability and resources, and to create incentives for the implementation of decentralization to match formal decentralization arrangements. These arrangements involve a combination of intergovernmental fiscal relations (expenditure and tax assignments, intergovernmental transferslgrants, and regulations for subnational borrowing), accountability mechanisms (electoral rules, transparency and information dissemination, channels for participation), channels for administrative responsibility and oversight (civil service arrangements, monitoring capacity), and programs to build local capacity. They should be thought of as a system, with the component parts being internally consistent and designed to minimize unintended consequences (Bahl, 2000).
436
Weist
This paper focuses on the importance in developing countries of integrating capital financing systems (intergovernmental loans and transfers) with intergovernmental transfer systems. These two systems are often considered separately, with unintended outcomes. Most recognize that without predictable transfers from higher levels of government, local governments are often unable to finance basic service delivery or infrastructure investment. However, linkages between the two systems affect other areas as well. To pursue equalization objectives, promote an efficient use of resources, and develop municipal credit systems, intergovernmental loans and transfers should be integrated within a consistent financing framework. Without such a framework, poorer local governments are unlikely to invest in needed infrastructure, and intergovernmental transfers may impede the development of a local credit market, especially when grants are available for the more creditworthy jurisdictions for infrastructure investments that could be financed through borrowing. The paper focuses on the capital financing systems for local governments in three East Asian countries. It reviews the impacts of weak integration of those systems, and the challenges of establishing a consistent financing framework.
2.
OVERVIEW OF DECENTRALIZATION IN EAST ASIA
Decentralization is occurring around the world, and fundamentally changing the role that subnational governments play in delivering and financing service^.^ Unlike Latin America and Central and Eastern Europe, which went through substantial periods of decentralization in the 1980s and 1990s, decentralization is a more recent phenomenon in East Asia. The Philippines first embarked on intergovernmental reform in 1991, with the enactment of its far-reaching Local Government Code. In 1994, as a response to the pressures of globalization and to manage the transition from a command to a market-oriented economy, China initiated a comprehensive fiscal reform program (the tax sharing system), which has profoundly affected the balance of powers between central and local governments. The 1997 Thai Constitution mandates decentralization to local administrations. In January 2001, Indonesia launched a "big bang" approach to decentralization that has substantially changed the way public services are financed and delivered. Figure I below shows the subnational share of total revenues collected in Indonesia, the Philippines, and Thailand. By comparison, subnational governments in the OECD account, on average, for only 20 percent of total revenue collections.
Intergovernmental Loans: Their Fit into a Tran.$er System
I
OECD
Thailand
Indonesia
Phihpines
Figure I . Subnational Revenues as a Share of Total Revenues, Selected Countries
Working at the subnational level in East Asia is complicated by the large number of local governments, and the limited role of the provinces. As shown below in Table 1, substantial service delivery responsibility has been devolved to 410 cities and districts in Indonesia, about 1,500 cities and municipalities in the Philippines, and almost 7,000 local governments in Thailand. In addition to the complexity of dealing with numerous local governments, many of them have limited capacity to discharge their newly delegated responsibilities.
Table]. Structure of Local Governments, Selected East Asian countries4 Tier 1
Tier 2
Tier 3
Indonesia
32 Provinces
4 10 Cities and Districts,
4,038 subdistricts 68,783 Villages
The Philippines
80 Provinces
114 Cities 1496 Municipalities
41,945 Barangays
Thailand
75 Provinces, 8 1 1 Districts
1129 Municipalities, Bangkok, Pattaya
6746 Tnmbon Administrative Org
Weist
Table 2 below summarizes some of the key elements of a well designed intergovernmental fiscal framework, namely:' Clearly defined and reasonable expenditure responsibilities; Local revenue autonomy; A stable, predictable and formula-based transfer system; Framework for responsible borrowing; and Consistent financing framework, that promotes the integration of grant and loan financing. Table 2 shows that the Philippines has the strongest intergovernmental fiscal arrangements (though with ample scope for improvement), perhaps because of its longer experience with decentralization. Many of these elements are weak in Indonesia and Thailand. It's notable that none of the three countries has a well developed system of capital transfers for infrastructure investment - such transfers are relatively small in the Philippines, the DAK is not fully operational in Indonesia, and most capital transfers in Thailand are allocated based on political decisions.
2.1
Financing Local Investments
Capital transfers from higher levels of government are used by many nations to finance basic infrastructure investments by local governments. In developing countries, investment loans or grants for financing local infrastructure often flow either through a central ministry (e.g., Ministry of Finance, Ministry of Local Government), sectoral ministries (e.g., Ministry of Health), or a financial intermediary (e.g., a Municipal Development Fund). In countries with more highly developed capital markets and more established intergovernmental fiscal systems (i.e. Hungary), a third source of financing is a municipal capital market.6 Capital transfers that flow through ministries are often designed to be matching (i.e., require a contribution, even if modest, by local governments.) Such transfers recognize the potential disincentive effects on local operation and maintenance of total (i.e., 100 percent) central financing of local investments, and have the advantage that they require some local revenue effort and demonstrated ownership. Matching transfers to finance specific investments are often also justified on the basis of the existence of spillover benefits or central government interest in priority programs such as health or environmental protection (Bird and Smart, 2001). The amount of central government transfers for a particular investment should be determined by balancing the priority of the investment from the central government's perspective with the affordability of long-term transfers at the central level,
Intergovernmental Loans: Their Fit into a Transfer System
439
and with the extent to which such transfers affect local government behavior. However, determining the proper matching share to achieve the above objectives can be quite complicated. Some argue that matching grants favor better-off local governments who can afford to match larger amounts, or conversely, penalize poorer local governments who cannot afford the matching requirement and may therefore forego the investment. Others argue that the larger investments needed in wealthier local governments are proportional to the matching funds received and that equity is not a concern. To pursue equalization objectives, the income level of the local government could be used to define the extent of matching, with wealthier local governments either receiving a smaller matching share than poorer local governments or a ceiling being placed on the amount of matching grant that can be received. As shown below in the cases of the Philippines and Indonesia, such equity concerns were an important determinant of a "loan-grant financing mix" for determining the matching that would apply to different local governments and their types of infrastructure investment. A second source of financing local infrastructure investment is through the use of a financial intermediary. The experience of financial intermediaries has been mixed.7 The intermediaries that have been the most successful have used transparent eligibility criteria for defining local governments and types of projects, financed economically and financially viable projects, had high repayment rates, used market-based interest rates, and encouraged the development of private credit markets for subnational lending. Many intermediaries do not meet these characteristics, with credit directed for political purposes, high loan repayment rates, subsidized interest rates, and impediments to the development of private credit markets. Figure 2 below shows that five key elements are needed to establish a municipal credit system: (i) creditworthy local governments, (ii) viable projects, (iii) financing through a financial intermediary or credit market, (iv) a supportive intergovernmental fiscal system, and (v) an overall regulatory framework. The weaknesses in intergovernmental fiscal systems noted in Table 2 introduce several impediments to the development of a municipal credit system, which are discussed in each country context below.
3.
THE PHILIPPINES EXPERIENCE
In the Philippines, the Local Government Code enacted in 1991 shifted many functions and revenues to Local Government Units (LGUs), which include provinces, cities and municipalities. It provides LGUs with a formula-based transfer to support their responsibilities. The Code defines the total Internal Revenue Allotment (IRA) in any one year as 40 percent of the gross national (internal) government tax revenues averaged over the
440
Weist
Figure 2. Building Blocks of a Municipal Credit System
three years prior to the current year. The IRA provides all LGUs with direct, unrestricted transfers. It is allocated first among subnational levels -- 23 percent to provinces and cities, 34 percent to municipalities, and 20 percent to barangays - and then within level by a formula based on population (50 percent), land area (25 percent) and equal share (25 percent). The IRA accounts for 94 percent of total transfers (World Bank, 2003b). Categorical grants to LGUs are relatively small, and come from various sources: (i) lump sum allocations under the General Appropriations Act (GAA) of various years; (ii) allocations made by national government sector agencies from their own budgets; and (iii) lump sum and/or line item appropriations at the direction of legislators. The Code does not mention matching grants. Transfers account for more than three-quarters of provincial and municipal revenues. Cities, however, have more financial autonomy and rely on the IRA for about 40 percent of their revenue.
intergovernmental Loans: Their Fit into a Transfer System
44 1
Prior to 1991, LGUs were required to borrow from Government Financial Institutions (GFIs) for capital financing, including government banks and the Municipal Development Fund (MDF). Local borrowing was never a substantial source of LGU financing compared to revenues from the IRA, grants and local taxes.8 In the mid-1980s, GFIs stopped lending to LGUs because of mounting LGU loan arrears and defaults. As a result, the National Government instituted a first debt-relief program for LGUs, and a second after the enactment of the Code which granted LGUs a clean slate vis-i-vis government lenders. After the Code was introduced, LGUs were no longer confined to seeking credit from GFIs: they could float their own bonds; enter into build-operate-transfer (BOT) arrangements; and in general tap various sources of credit financing. In support of the Code's decentralization objectives, the Department of Finance (DOF) prepared an LGU Financing Framework in 1996.~ The Framework defined a vision and an action program for moving to marketbased financing of LGU capital investments. The vision was anchored on two premises. "First, LGUs have varying levels and records of creditworthiness and bankability. Second, their financing needs are huge. Therefore, the private sector (BOT investors, bondholders, commercial banks), the GFIs and MDF all have a role to play in meeting LGU financing needs.. ..The ultimate objective is to graduate LGUs to private sources of capital which are vast and promising, but remain largely untapped." The Framework recommended that steps be taken to achieve seven reform objectives: (i) increase LGU use of BOT arrangements; (ii) develop the LGU bond market; (iii) promote LGU access to private banks; (iv) optimize the involvement of GFIs in LGU financing; (v) restructure and reorient the Municipal Development Fund; (vi) improve LGU capacity to raise their own revenues; and (vii) tap technical assistance and financing from official development assistance. As part of this Framework, the Inter-government Coordinating Council (ICC) and the National Economic Development Authority (NEDA) developed a matrix specifying the loan-grant-equity mix that would apply to LGU projects that required loan or grant financing, whether the project was supported by multilateral or national funds. The matrix attempted to enhance the transparency of subsidies for LGU programs; clarify the eligibility of certain investments for matching grants; and specify the matching grant formula that would be applied to certain types of projects. The matrix classified types of projects (i.e., environmental, social, incomegenerating), based on an assessment of externalities and public goods, and LGU classes were classified based on local ability to pay.
Table 2. Assessment of Intergovernmental Fiscal Relations in Selected East Asian Countries
Clearly defined and reasonable expenditure responsibilities?
Local revenue autonomy? Local revenue sources Ability to set tax rates Administrative capacity Stable, predictable and formula-based transfer system? Equalization Public
Indonesia Somewhat. Central government retains responsibilityonly for judiciary, religious affairs, national defense and security, fiscal and monetary affairs, and international diplomacy; provincial role very limited; inconsistencies among laws, regulations, and decrees; confusion about minimum service standards Limited. Limited local tax bases, some discretion in setting tax rates; weak administrative capacity, much scope for nuisance taxes
The Philippines Mostly. Many functions devo1;ed; undlear devolution of environment, natural resources and infrastructure. LGUs can join cooperative units in delivering services. Central government imposes unfunded mandates and wields undue influence in some areas.
Thailand Somewhat. 245 functions are being devolved in phased manner. Central government still actively providing "local" services; sectoral decentralization of health and education distinct from devolution to local governments.
Yes, though ample scope for expansion of Real Property Tax, local business tax; and user charges. Limited authority to increase tax rates.
No. Very limited local taxation; revisions in local tax laws under discussion for past 10 years; weak capacity
Somewhat. Significant reliance on shared taxes (huge disparities in natural resources); DAU (equalization) is formula-based; DAK (tied grants) relatively unused.
Yes. IRA formula stable and generally predictable; equalization could be improved; ODA for cost sharing; augmentation funds confuse responsibility.
Somewhat. Formula-based transfer system is slowly being implemented; other transfers (especially for capital projects) distributed on an ad hoc basis.
Table 2. Assessment of Intergovernmental Fiscal Relations in Selected East Asian Countries (Continued)
Framework for responsible borrowing? Legal basis for borrowing Hard budget constraint
Consistent financing framework? Economic and financially viable projects High loan repayment rates Integration of financing and transfer systems Promote private c a ~ i t a markets l
Indonesia Somewhat. Legally permitted, but currently on hold. Local borrowing limited by rules. Need for prudential regulation of subnational borrowing, work-out arrangements. Small local tax base undermines hard budget constraint
The Philippines Yes. Most borrowing through MDF or GFIs. Debt service limited to 20 percent of income. Need for prudential regulation of subnational borrowing; work-out arrangements
Thailand Yes. RUDF established with small market. Working on municipal bankruptcy code. Need for prudential regulation of subnational borrowing; work-out arrangements
No. Onlending complicated by KMK35 and RDAlRDI flawed, with many local government defaults
Yes, but. IRA intercept provision assures high repayment for MDF and GFIst; DOFINEDA financing framework never implemented; impediments to competitive local capital markets
No. RUDF promotes viable projects and loan repayment. Lack of integration between transfer system and onlending system. MOF is discussing establishment of Credit Guarantee Corporation.
444
Weist
The loan-grant-equity matrix has been implemented inconsistently. Support for it has declined with changes in government administration, and the matrix and resulting formulae has been used only for projects implemented by the Municipal Development Fund Office (MDFO), which must adhere to World Bank conditions on the use of its loan funds. Though recent cost-sharing schemes reflect significant progress in rationalizing LGU grant policies, much more remains to be done, including the development of a transparent formula for matching grants for all eligible LGU projects, regardless of funding source; and clarifying the grant-making mechanism for LGUS." Successful implementation of the policy would further require that the Philippines Government consistently apply the policy, and that multilateral and bilateral donors "play by the rules" established. Little progress has been achieved in enabling LGUs to access private capital based on creditworthiness: the MDFO, the Development Bank of the Philippines (DBP) and the Land Bank of the Philippines (LBP) dominate the LGU credit market (World Bank, 2003b). Private financial institutions have no market presence - their inability to become depository banks for LGUs impedes their entry into this market. Further, only government financial institutions can intercept the IRA transfer to guarantee local repayments, which disadvantages private banks, and prevents the creation of a "level playing field." With almost no mobilization of domestic capital, almost all long-term credit to LGUs is provided through Official Development Assistance. The most significant development has been the creation of an LGU Guarantee Corporation (LGUGC) to guarantee debt issues financed from private sources. The LGUGC is the first privately-managed local government guarantee corporation established in a developing country in Asia. Since the inception of the LGUGC, six LGU bond issues have been floated (totaling PhP1.35 billion), all with the guarantee of the LGUGC." The principles and strategy embodied in the framework are still valid and are important to the Philippines' decentralization process. Yet the framework has only been weakly implemented. As a result, LGU borrowing for critical investments occurs in a non-competitive environment, and funds from multilateral sources are channeled in inefficient ways. For example, the World Bank's projects do not always reflect the same credit conditions, which leads to instances where LGUs invest in certain types of infrastructure based on more lenient credit terms, rather than because of the soundness of the investment or its contribution to the community.
lntergovernmental Loans: Their Fit into a Transfers System
THAILAND'S EXPERIENCE Prior to FYOl, over 70 percent of the intergovernmental grants in Thailand were allocated for specific investment projects. The Ministry of Interior allocated these grants in an ad hoc and highly politicized manner. The amounts allocated varied greatly from year-to-year, and actual allocations were not known until well after the fiscal year began. Hence, the basic requirement of a decentralized system having transparent and stable intergovernmental transfers was violated. Nor did the grants reflect the broader intergovernmental framework and the vertical fiscal imbalance between the central and local governments. Since the decentralization reforms have begun, the intergovernmental transfer system has been in a process of transition. The National Decentralization Committee determines the amount of shared taxes and grants, as well as how they are allocated to local governments. The Department of Local Government Promotion in the Ministry of Interior allocates grants to local governments. From FYOO to FY03, the total amount of grants (the "distributable pool") and the allocation criteria differed every year. In FY02, for example, good governance and tax effort grants were initiated, but these were discontinued in FY03. In FY04, a portion of the general purpose grant was allocated to enhance local revenue mobilization. Specific grants are declining in importance (from 60 percent in FYOl to 20 percent in FY04), while general grants are increasing as an incentive to encourage the performance of the mandatory functions specified in the Decentralization Action Plan. In FY04, about 40 percent of general grants were allocated to municipalities, about 54 percent to tambon administrative organizations, and about 7 percent to provincial administrative organizations, to implement the decentralized functions defined by the Decentralization Action Plan. Equalization transfers are among the largest general transfers (44 percent of total general transfers in FY04), and are allocated by a formula based on the following factors: (i) an equal share (25 percent), (ii) population (30 percent), (iii) area (5 percent), (iv) inverse of local own revenue mobilization (20 percent), and (v) inverse of specific grants (20 percent). While these equalization transfers are a welcome addition to the intergovernmental fiscal system, and the formula promotes predictability, it is not yet known whether they are equalizing differences in fiscal capacity or expenditure. The decentralization objectives espoused in the new Constitution are being implemented (albeit slowly), and by the end of 2001, local governments -- including provincial administrative organizations, municipalities and tambon administrative organizations -- provided about 20 percent of government goods and services. Many of the 980 new
446
Weist
municipalities will need to invest in infrastructure to fulfill their new responsibilities. In support of its decentralization objectives, and as a means of building local capacity and developing a municipal credit market, in 1997 the Thai Government began preparing a financial intermediary called the Regional Urban Development Fund (RUDF.) When the RUDF was initially designed, the Thai economy was growing. The situation changed drastically during the East Asian economic crisis and many unanticipated events delayed the RUDF's implementation, including volatile financial markets, declining local government revenues, and various Government programs that competed with the RUDF in providing "cheaper" funds for local infrastructure. For example, few other funds required full feasibility studies before local governments became eligible for loan funding. The initial design also underestimated the effort and new learning required by municipalities to participate in the RUDF program; and municipalities' knowledge and skills in subproject preparation and analysis (including feasibility studies), managing revenue-generating assets, and engaging the community in municipal decision making were rudimentary. Nonetheless, the RUDF has successfully: Established a new financial intermediary and its requisite policies and procedures, Developed analytical tools for measuring municipal creditworthiness, Provided technical assistance in determining subprojects, conducting feasibility studies, and organizing civic fora, Promoted community participation in local decision making through civic fora, Built a portfolio of economically feasible and financially viable subprojects,'2 and Financed an estimated Baht 620 million (approximately $15 million) in local infrastructure. Despite these accomplishments, the RUDF financed only about half of its original projected pipeline of subproject investments ($30 million). In retrospect, the delays associated with establishing a financial intermediary during Thailand's economic crisis were substantially underestimated, and many design aspects were overly complex and/or inappropriate. Notably, numerous obstacles within the intergovernmental framework impeded the functioning of the RUDF. Competing funds were available from other sources (e.g. Miyazawa funds for economic stimulus during the crisis, Senatorial allocations) or below-market loans with easier terms were available from other government agencies (e.g., Department of Local
intergovernmental Loans: Their Fit into a Transfers System
447
Administration's Municipal Development Fund, Ministry of Interior's Cooperative Fund, Krung Thai ~ a n k . ) ' ~ With regard to establishing a municipal credit system, Thailand has made progress in establishing a financial intermediary, strengthening the creditworthiness of selected municipalities, and developing a limited number of viable projects. These are important achievements, but they do not constitute a municipal credit system and additional financing from a credit corporation would not establish a viable municipal credit system. By 2006, local governments are expected to collect 35 percent of government revenues and will have substantially more responsibility for providing and financing local services. In developing local capacity and promoting the development of a municipal credit market, the RUDF continues to be a key instrument to implement the government's decentralization objectives. However, unless the overall intergovernmental framework under which the RUDF operates is rationalized, its potential impact on strengthening the capacity of local governments and establishing a viable municipal credit system will be limited.
INDONESIA'S EXPERIENCE14 Indonesia's law on fiscal relations (Law 2511999) introduced significant changes in its transfer system (Hofman, et al. 2002, World Bank, 2003a). It established a general allocation grant -- the Dana Alokasi Umum (DAU) which is financed by a minimum of 25 percent of central of domestic revenues. The DAU consolidated previous transfers for local civil servants' salaries and the Inpres program. It is allocated among the provinces (2.5 percent), and kota and kabupaten (cities and districts, 22.5 percent.) Allocations within levels of regional government are based on a formula which depends on measures of expenditure needs and revenue capacity. Law 2511999 also contains provisions for special earmarked grants, the Dana Alokasi Khusus (DAK). These are mainly sectoral, conditional grants designed in consultation with line ministries, but to date have been limited in use. Regional governments rely on transfers (exclusive of shared taxes) for over two-thirds of their revenues. Indonesia's legislative framework for borrowing is defined in Law 2511999, Government Regulation 10712000 and a decree from the Ministry of Finance (KMK 3512003.) Laws 2211999 and 2511999 allow the regions to borrow, and allow borrowing from foreign sources through the central government.'5 Law 2511999 states that borrowing is permitted only for projects generating a financial return, but PP10712000 clarifies that financial returns can be direct or indirect. It further states that regions cannot have outstanding debt that exceeds 75 percent of the previous year's general
revenue, and the debt-service coverage ratio must not exceed 40 percent. Regions are also permitted to borrow funds for short-term periods for cash flow management. The legislative framework reflects a largely market-based credit system. The central government does not guarantee local government loans, and all creditor conditions are the responsibility of the local borrower. When lending to the regions, the central government can intercept the DAU if the region fails to meet its debt service obligations - a common problem in the previous credit system where local defaults and arrears in the RDNRDI accounts were prevalent. KMK 3512003 permits on-lending for cost-recoverylrevenue-generating projects and defines on-granting (transfers) for non-cost recoverylnonrevenue generating projects. The share of required counterpart funds depends on region's fiscal capacity, where fiscal capacity is defined as: General Revenues - Employee Expenditures
Based on this value, regions are categorized into high, medium and low fiscal capacity, with different matching rates based on their category. Lack of clarity in the details of the on-granting framework and delays in finalizing it led to a period where local investment spending declined significantly, especially for donor funded investments, which account for a significant share of Indonesian investment. The most recent revisions to Law 25 relax some of the restrictions related to local borrowing, and the provisions of KMK35 are less relevant. A provision in the revised law requires that regions report their borrowing and debt positions to the central government twice a year. If such reports are not provided, or payments on central government loans are not made on a timely basis, regions face the threat of delayed transfer payments for revenue sharing or the DAU.
6.
THE CHALLENGES OF DEVELOPING AN INTEGRATED FINANCING SYSTEM
In the Philippines, the lack of integration between transfers and capital financing resulted in noncompetitive LGU borrowing, inefficient use of funds from multilateral sources, and little progress in enabling LGUs to access private capital based on creditworthiness. In Thailand, the Government's objective of establishing a financial intermediary to support the decentralization process, build local capacity and establish a municipal credit market was impeded by inconsistencies in the intergovernmental fiscal framework. In Indonesia, the lack of a tied grant system (i.e. DAK) and
Intergovernmental Loans: Their Fit into a Transfers System
449
delays in implementing the on-granting framework led to a period of declining infrastructure investment. In all three countries, developing a transparent formula for matching grants to fund local government investment projects, regardless of funding source, would substantially strengthen the intergovernmental fiscal system. None of the countries reviewed above has an integrated financing system. It could be argued that the decentralization frameworks in Indonesia and Thailand are relatively recent, and that such a framework will be fully developed once the other aspects of the complex intergovernmental fiscal systems are defined further. As borrowing restrictions are lessened in Indonesia through the revision to Law 25, and local governments become more active borrowers in Thailand, however, the importance of such a framework is attenuated. The Philippines proposed its framework in 1996 but has not yet been able to implement it. What explains this poor performance? In all three countries, a substantial portion of funds from central agencies (or deconcentrated spending) for current and capital spending flows through a system that is often unmonitored, and could undermine the consistency of a financing framework. Until such flows are measured and their incentive effects understood, it will be difficult to implement a consistent financing framework. A second problem is the multiplicity of entities involved in financing local investments - Ministries of Finance and Planning, sectoral ministries, Congress/Parliament, donor agencies. Coordination among such entities may be difficult without an overarching entity or authority responsible for overseeing the entire intergovernmental fiscal system.16 Finally, the lack of recognition of the disincentive effects of separate approaches may account for the lack of attention. The need for considering expenditure needs and fiscal capacity is well acknowledged in designing transfers aimed at meeting equalization objectives. Having sufficient local revenues is generally a prerequisite for permitting local borrowing. But linking the transfer system with the borrowing framework has not generally occurred. Broadening the intergovernmental fiscal framework to reflect transfers for infrastructure investment and their impacts on borrowing, may result in more systematic approaches that avoid the unintended effects of separate systems.
Notes 1. 2.
Lead Public Sector Specialist, the World Bank, [email protected]. This paper reflects the views of the author and not the World Bank. See for example, Bardhan and Mookherjee (2000), Crook and Manor (1998), Kahkonen and Lahni (2001), and Prud'hornme (1994).
3. 4. 5. 6.
7. 8.
9. 10. 1I. 12. 13.
14. 15.
16.
Of the 75 developing and transition countries with population greater than 5 million, all but 12 claim to be embarking on some form of fiscal decentralization. See Burki et al. (1999). See World Bank (2003a, 2003b). Mechanisms for local accountability are a critical element of an intergovernmental fiscal framework, but they are not discussed in this section. See, for example, Mihaly Kopanyi, Deborah Wetzel and Samir El Daher, intergovernmental Finance in Hungary: A Decade of Experience: 1990-2000 (The World Bank, 2004.) See for example, George Peterson, Building Local Credit Systems, Urban Management Program Policy Paper, August 1997. On average, LGU borrowing accounted for only 1.25 percent of their total income for the period 1981-1993 (2.0 percent for cities, 0.5 percent for municipalities and 1.35 percent for provinces). See Llanto et al. (1996). The Framework was published in 1996 and presented at a Consultative Group meeting in Tokyo in 1997. The cost-sharing arrangement is dependent on the sector of the grant, and typically follows a certain percentage sharing between the national government and the LGU. See World Bank (2003b). Note that the total amount guaranteed by the LGUGC to date is roughly equivalent to the outstanding portfolio of the Development Bank of the Philippines but is substantially less than the outstanding LGU portfolio of the Land Bank. Typical local projects include: municipal office buildings, municipal markets, childrens' centers, schools, pawnshops, a water gate, and a public library. This lending offered municipalities lower interest rates without the conditions associated with the RUDF for prudent operation (i.e., requiring a civic forum and feasibility study, regulating local procurement activities, etc.). Some of the sub-projects prepared with technical assistance from the RUDF were subsequently financed by the Municipal Development Fund because of its lower interest rates. This section draws heavily on Jasmin Chakeri, KMK35: Implementation and Potential Problems, Presentation to the World Bank Indonesia Country Team, World Bank Indonesia website (undated). As in the area of financial management, which is also treated in both laws, borrowing is treated a slightly different way in both laws. This overlap, and partial conflict is the result of a turf battle between two ministries which is yet to be resolved. The South Africa Finance and Fiscal Commission is one of the few authorities with responsibility for intergovernmental transfers and borrowing.
References Bahl, Roy. 2000. How to design a fiscal decentralization program. In Local Dynamics in an Era of Globalization, S. Yusuf, W. Wu and S. Evenett, eds., 41-100. New York: Oxford University Press. , and Paul Smoke. 2003. Restructuring local government finance in developing countries; Lessons From South Africa. Cheltenham, U.K.: Edward Elgar. Bardhan, Pranab, and Dilip Mookherjee. 2000. Capture and governance at local and national level. American Economic Review, 90(2): 135-39.
Intergovernmental Loans: Their Fit into a Transfers System
45 1
Bird, Richard M., and Michael Smart. 2001. Intergovernmental fiscal transfers: Some lessons from international experience. Paper present at Symposium on Intergovernmental Transfers in Asian Countries: Issues and Practices, Asian Tax and Public Policy Program, Hitotsubashi University, Tokyo, Japan, Burki, Shahid Javed, Guillermo E. Perry, and William R. Dillinger. 1999. Decentralizing the State. Washington, D.C.: World Bank. Crook, R., and James Manor. 1998. Democracy and Decentralisation in South Asia and West Africa. Cambridge: Cambridge University Press. Ebel, Robert D., and Serdar Yilmaz. 2002. On the measurement and impact of fiscal decentralization. Washington, D.C.: The World Bank Institute. Faguet, Jean-Paul. 1997. Decentralization and local government performance. Washington, D.C.: The World Bank. Hofman, Bert, Kadjatmiko, and Kai Kaiser. 2002. The making of the big bang and its aftermath: A political economy perspective. Paper present at Georgia State University Conference: Can Decentralization Help Rebuild Indonesia? Kahkonen, Satu, and Anthony Lanyi. 2001. Decentralization and governance: Does decentralization improve public service delivery PREM Note. The World Bank, Washington, D.C. Kopanyi, Mihaly, Deborah Wetzel, and Samir El Daher. 2004. Intergovernmental Finance in Hungary: A Decade of Experience: 1990-2000. Washington, D.C.: The World Bank. Lewis, Blane D., and Jasmin Chakeri. 2004. Decentralized Local Government Budgets in Indonesia. Jakarta: The World Bank Llanto, Gilbert et al. 1996. Local government units' access to the private capital markets, Philippines Institute for Development Studies. Martinez-Vazquez, Jorge, and Jameson Boex. 2001. Russia's transition to a new federalism. 97. World Bank Institute: Washington, DC, WBI Learning Resources Series. Peterson, George. 1997. Building local credit systems, Urban Management Program Policy Paper, The World Bank. Prud'homme, Remy. 1994. On the dangers of decentralization, World Bank Research Working Paper, No. WPS 1252. The World Bank, Washington, D.C. Weist, Dana. 2004. Thailand's decentralization: Progress and potential. Mimeo prepared for the EAP Decentralization Flagship Study. World Bank 2002. Implementing Decentralization in Thailand: The Way Forward. Washington, D.C.: The World Bank. World Bank 2003a. Decentralization Indonesia: A Regional Public Expenditure Review Overview Report. Washington, D.C.: The World Bank. World Bank 2003b. Philippines: Decentralization and Service Delivery-From Promise to Performance. Washington, D.C.: The World Bank. World Bank 2003c. Philippines: Improving Government Performance: Discipline, Efficiency and Equity in Managing Public Resources. Washington, D.C.: The World Bank. World Bank (2000). Thailand Public Finance Review: Public Finance in Transition. Washington, D.C.: The World Bank.
DISCUSSANT COMMENTS BEYOND EQUALIZATION GRANTS STRICT0 SENSU
&MY PRUD'HOMME Professor Emeritus, University of Paris XI[ France
SEARLE AND WEIST
1.
INTRODUCTION
The common thread in the papers presented at this session is that too narrow a view of fiscal equalization grants can be misleading or dangerous. It is necessary to look at fiscal equalization in a broader setting and to integrate a number of additional mechanisms, objectives, and instruments in the design of an equalization grant scheme. This message is reinforced by the fact that one of the presenters who made a career -and a bright one- in the equalization grants business chose to speak not on equalization grants strict0 sensu, but on related issues. The narrow standard approach, illustrated in Table 1, is as follows. Let us assume two sub-national jurisdictions, called Rich and Poor, of equal population and area, with different "fiscal capacities" and different "expenditures needs". The equalization grant problem is: what grant system will "equalize,' that is, eliminate or reduce or compensate for, these disparities? Line 3 of table 1 provides a very simple, not to say simplistic, answer. This analytic framework can -and should- be expanded in at least four directions. The first two are discussed in the papers by Bob Searle and Dana Weist, but the others are also worth some attention.
Prud'homme
Table I. The Fiscal Equalization Problem Rich Fiscal capacity Expenditure needs Fiscal equalization grant Infrastructure endowment Contribution to natl. budget Benefits from natl. budget Net gain from natl. budget
2.
70 80
10
700 700 450 -250
Poor 20
100 80 150 200 450 +250
LOOK AT WHAT LIES BEYOND "FISCAL CAPACITY"
The notion of fiscal capacity is not as straightforward as it might seem. The paper by Bob Searle is an exploration of this idea. What sub-national government financial resources should be included in the definition of "fiscal capacity"? This is discussed under two headings: revenue-sharing and natural resources. It is difficult to disagree with the useful and subtle analysis of the various forms of revenue-sharing presented, and even more with the conclusion that what comes under the name of revenue sharing should be included in fiscal capacity. One could, however, question the usefulness of the very concept or expression of "revenue-sharing". The word is used to describe two completely different things. It can refer to tax-yield sharing. A particular national tax, or a part of it, is set aside and distributed to sub-national governments, generally on the basis of what has been collected in each jurisdiction. Such a system is very common in Latin America under the name of co-participacion. But it should be clear that the share received by the sub-national governments is in no way a tax. It is a transfer from the central government. Saying that this transfer is "financed" by the shared tax makes no sense. It is obviously financed by all the taxes collected by the central government. The shared tax is just a way of calculating the total amount of the transfer, and sub-national collection is simply a criteria --one of the worst possible criteria- to allocate this total amount to the various jurisdictions. Tax-yield sharing is not at all equalizing. It is on the contrary disparities-increasing, as correctly noted by Bob Searle. An equalization grant may therefore be necessary to undo what tax-yield sharing has done. Tax sharing can also refer to tax-base sharing. In this case, several levels of government have access to the same tax base, but each of them is
Discussant Comments
455
responsible for setting its own tax rate. It is this freedom to increase or decrease tax rates -and to reap the political costs and benefits associated with it- that defines this type of "shared tax" as a local tax. As such, it must obviously be included in the calculation of fiscal capacity. The question of who actually collects the tax is in general not important. A local tax can be collected by the central government, on behalf of the local governments, and at the rate decided by them. This is what is done in France, for instance, against a reasonable fee, and it is not considered to affect the "localness" of such taxes. Central government assessment and collection even has one important side benefit. It makes it much easier to compensate for differences in tax bases since local tax bases are assessed in a uniform fashion for all local governments and are known to the central government. This general rule of the non-importance of the assessing and collecting level of government has, however, one important exception. In some developing countries, assessment ratios and collection ratios (the ratios of what is actually assessed and collected to what should, in principle, be assessed and collected) are far below one, and vary from one local government to another. Increasing or decreasing these informal ratios functions just like increasing and decreasing formal tax rates. A tax benefiting local governments that has a given constant formal tax rate but varying tax assessment and tax collection ratios can be considered as a true local tax. Bob Searle offers an interesting, although rather inconclusive, discussion of natural resources revenues. It shows how unclear "natural resource" is as a concept. Some are publicly owned, other are privately-owned. Some are a gift of God, other are largely man-made. Some are exhaustible, other are replenishable. Some produce a modest income flow and are like any other sector, others produce an enormous rent. Some require public investment, others don't. The key issue is probably that of rent, the difference between sales values and costs (including external costs and a fair return on capital). First, it is not clear, at least from an economic viewpoint, why such rents should be locally rather than nationally appropriated. Second, if they are, it is clear that they should be treated as part of local fiscal capacity.
3.
LOOK AT INFRASTRUCTURE DISPARITIES
The main message of the paper by Dana Weist (and also the paper by Sofia Levtchenkova and Jeffrey Petchey presented in the preceding session) is that one cannot neglect disparities in infrastructure endowment between sub-national governments. It is presented in the context of local government borrowing, on the case of three East Asian countries, but the thrust of the
456
Prud'homme
argument is that looking at differences in fiscal capacity or expenditure needs (the two first lines of Table 1) would miss a substantial dimension of existing disparities. Differences in infrastructure endowment (as indicated in line 4 of table 1) are as important and arguably more important than the other types of differences. There is an ambiguity about how they should be treated. It has been argued that the more infrastructures, the greater the expenditure needs (in terms of heating, maintenance, servicing, etc.). Indeed, in the early 1990s in the Czech Republic and Poland, the amount of infrastructure was a criterion in grants allocation formulae. Those local governments that were well endowed asked for greater grants to operate their infrastructure. Those who were poorly endowed also asked for greater grants - to create their infrastructure. There is also a time dimension to the problem. Infrastructure is long lasting, whereas equalization grants are yearly. This can create a mismatch. Consider an area that is less endowed that the average (on a per capita basis), but with a rapidly declining population. Is it better to build infrastructure in that region in the name of equity, even though it will be underutilized after a few years, or not to build this infrastructure in the name of efficiency, even though it means discriminating against that region now? Infrastructure endowment could be introduced as an element of fiscal capacity. Public service provision, like any other good or service, is produced by a combination of capital and labor (and other operating expenditure), with some possibilities of substitution at the margin. Fiscal capacity is the ability to pay for labor and other operating expenses. It could be "enlarged" to incorporate infi-astructure capital, or the yearly costs associated with it (depreciation plus the opportunity cost of capital). In the language of Table 1, one would take, let us say 10% of line 4 and add it to line 1, to get an enlarged, but perhaps more meaningful, notion of "fiscal capacity". This procedure would substantially change the magnitude of the equalization grant. Compensating for differences in infrastructure endowment is often considered more important than compensating for differences in the capacity to provide services. The argument offered is that infrastructure is more "strategic", that it contributes to growth more effectively than mere services consumption. This is implicitly the position taken by the European Union in its regional policies. Grants from the (small) EU budget to countries and regions most frequently take the form of infrastructure grants; and they are based on GDPIcapita, not tax capacity, in the sense that only regions with a GDPIper capita below 75% of the European average (and also regions with particular difficulties) are eligible. Borrowing to reduce infrastructure endowment differentials is a doubleedged sword, as Dana Weiss is well aware. Facilitating borrowing across the
Discussant Comments
457
board is likely to benefit primarily the richer sub-national governments those that are already better endowed than the average, and at the same time can more easily obtain the trust of lenders. An improved borrowing system (the well functioning bond market dreamt of by many) can therefore increase, rather than reduce, existing disparities. To avoid this unwelcome effect, one must devise complex systems that facilitate infrastructure investment in low-income sub-national governments. This can be done either by larger capital grants or lower borrowing rates. But this in turn runs the risk of pushing these low-income sub-national governments towards over capital-intensive modes of service provision.
4.
LOOK AT FEED-BACK EFFECTS
Grants and taxes are treated as if they were independent. It can be argued that they are not, and that too high a level of grants will have important feedback effects on both tax rates and tax bases. In the language of Table 1, line 3 has an impact upon line 1 that should be taken into account. The standard theory has it that sub-national governments will consider the utility of an additional dollar of public expenditure and compare it with the disutility of an additional dollar of taxes. This comparison, it is claimed, is independent of the amount of grants received. This view is very doubtll, for at least two reasons. One is that the marginal utility of public expenditures declines. The after-grant marginal utility is much lower than the before-grant marginal utility. If the latter is equal to the marginal utility of taxes, the former will not be. A second reason is that median voters react in relative, not absolute, terms. They compare a percentage increase in taxes with a percentage increase in expenditures. Look at Table 1. In region Rich, where grants account for a small share (12.5%) of resources, a 10% increase in expenditures1 will require an 11% increase in taxes2: this is a politically reasonable proposition. In region Poor, however, where grants represent a large share (80%) of resources, a similar 10% increase in expenditures3will require a 50% increase in taxes4: this is a politically suicidal proposition, and politicians would laugh at it. In short, grant levels that are too high tend to kill local tax rates. We verified this on a sample of Brazilian municipalities (Prud'homme 1998). Controlling for tax bases and size, it could be shown that the higher the transfers received, the lower the tax rates. Introducing transfers per capita in a logistical form, it appeared that there was a threshold in transfers per capita above which tax effort practically vanished5. A similar feed-back mechanism is probably at play with tax bases. If equalizing grants fully compensate for differences in tax bases, sub-national governments are deprived of an incentive to increase their tax bases.
Prud'homme
458
Increasing tax bases, in the form of attracting industries for instance, often entails costs, in the form of noise, pollution, increases in real estate prices, etc. What makes some sub-national governments accept such costs is often the hope that attracting industries will also increase their tax bases, and this, in turn, will either make it possible to increase public expenditures or to lower tax rates -and to reap the associated political benefits. Breaking this virtuous mechanism is dangerous. Differences in tax capacities are double-faced. On one side (the most important side, admittedly) they are innate or God given; and as such they should be compensated. On the other side, they are acquired, or man-made; and as such they should be encouraged.
5.
LOOK AT OTHER GOALS AND OTHER INSTRUMENTS
Equalization is not the only goal of transfers, and transfers not the only instrument of equalization. An excessive focus on equalization grants could obscure those rather evident truths. Compensating for fiscal disparities and expenditure needs is an important function of transfers, but by no means the only important one. Four other important h c t i o n s can be briefly mentioned. First, transfers are necessary to correct vertical imbalances (as well as horizontal ones). Even if all the sub-national governments of a country were identical -thus making equalization unnecessarytransfers would nevertheless be very much needed. The expenditures that can be optimally decentralized are much larger than the taxes that can be optimally decentralized; hence the need for transfers. Second, transfers can be very useful to reconcile or articulate the views of different levels of government. It is not true that a perfect division of labor between levels of government can and should be arrived at. On many issues, such as education, health and safety, for instance, the problem is not so much: "who will do what?', but rather: "how can equally legitimate viewpoints be reconciled?". There are many answers to that question (negotiations, information, mandates, etc.) but tied grants and differential transfers are one of them. You, local government, can chose as you want between roads and sport, but because we, central government, believe that roads are more important for the country, we will subsidize your roads at 50% and your stadiums at 20%. Third, a standard justification of transfers, valid even in the absence of horizontal and vertical balance objectives, is to compensate for spillovers. Because the benefits of some local government expenditures extend
Discussant Comments
459
geographically beyond their jurisdictions, there is a danger of sub-optimal provision in these cases. This requires compensating transfers. Fourth, a major objective of transfer systems is to promote local economic development and to reduce disparities in economic development between jurisdictions, in terms of income and not only in terms of public service provision. It can be thought that this objective is well served by decentralization, and that sub-national governments know better (than central government) how to achieve it. It can also be argued that equalization of services is not the same thing as equalization of income or growth potential. The example of the Economic Union, as mentioned above, suggests this. The EU transfer system has not tried to equalize service provision in member countries and regions (the EU budget, which is only 1.2% of European GDP, would not have made it possible anyway), but has explicitly focused on subsidies to infrastructure investments in lagging regions. This strategy seems to have met with some success. Disparities between regions and countries did decline (this is called increased cohesion in Brussels jargon); but whether this is the result of the strategy followed or of broader market forces is anybody's guess. Similarly, even if one focuses on the very worthwhile goal of public services equalization, transfers are only one instrument amongst several. In particular, central government taxes and expenditures should not be forgotten. In most countries they continue to play a major role. The last lines of Table 1 are intended to illustrate this. Let us assume that national taxes are 10 times as important as sub-national taxes and proportional to them; let us assume further that central government expenditures are equally divided between region Rich and region Poor. It follows that the central government budget redistributes enormous amounts from region Rich to region Poor. This is an unintended (and often unrealized) equalization effect as important (or even more important) than most equalization grants system. In conclusion, it is important not to take too narrow a view of equalization grants. The focus should not be on equalization grants strict0 sensu, but rather on the equalization dimension of grants which is itself only one dimension of grants. Grants, in turn, are also only one dimension of intergovernmental fiscal relations
Notes 1. 2. 3. 4. 5.
From 80 to 88 From 70 to 78 From100to110 From 20 to 30 To be fair, a similar study on France failed to show this effect, for unidentified reasons.
Prud'homme
References Prud'homme, Rkmy. 1998. State and Local Public Finance in Parana: Structure and Issues. For The Government of Parana.
6 THE BIGGER PICTURE
Chapter 15
THE POLITICAL ECONOMY OF EQUALIZATION TRANSFERS STUTI KHEMANI' The WorldBank
1.
INTRODUCTION
Normative theories of fiscal federalism postulate that intergovernmental transfers should be determined by equity and efficiency considerations, to support local governments in providing differentiated public goods to heterogeneous populations, while ensuring an even distribution of basic services across all regions (Musgrave, 1959, 1983; Oates, 1972; Gramlich, 1977). However, a recent surge of empirical evidence shows that the distribution of transfers across local jurisdictions, and what local governments do with these transfers is heavily influenced by political incentives facing both national and local policy-makers. Although the notion that intergovernmental transfers, like most other public policy issues, are influenced by and interact with political institutions and processes, is far from earth-shattering, until recently there had been little attempt to internalize the political implications when analyzing the role of transfers in equalizing access to basic services for all citizens. However, the new evidence shows that politics has such substantial impact on both the distribution and use of resources for local service delivery that it cannot be avoided nor side-stepped when developing intergovernmental fiscal policies for equalization. This chapter reviews the evidence on the impact of politics on the distribution of national resources across regions, and on the performance of local jurisdictions in using these resources for service delivery. This evidence is largely "reduced-form" in that it measures political impact on distribution and service delivery for a given system of intergovernmental sharing of resources and responsibilities. The impact of political institutions is of course more fundamental, in that it helps shape the intergovernmental institutions themselves. But the chapter abstracts from this more
464
Khemani
fundamental interaction, and focuses on the implications of political analysis for technocratic solutions to the problem of equalization transfers, given currently operational intergovernmental institutions. Section 1 shows that there is remarkably consistent and robust international evidence that politics impacts the distribution of transfers across local jurisdictions, with particular forms of political decision-making leading to particular forms of political distortions in the distribution of national resources across regions. The impact of political considerations on distribution may not necessarily be regressive, if for instance citizens in lowincome regions are particularly active and informed participants in political processes, but it is more likely to lead to deviations from the benchmark outcomes assuming social welfare maximization or equalization of basic services across regions. Fiscal institutions such as formula-based transfers or delegation of distribution decisions to independent agencies have limited success in curbing political influencealthough they do change the terms of political bargaining, they do not cancel completely the impact of political considerations on the distribution of resources across regions. This is both because the formula itself is determined by a political process, or is not binding and leaves room for political discretion, and because even when it is binding national governments can turn to other fiscal instruments at their disposal, such as direct spending or financing deficits of sub-national governments, to place f k d s where there is political advantage. This analysis suggests that the objective of equalization of resources across local jurisdictions might be best served through equalization of political representation, that is through political institutions, in the absence of which even the most sophisticated fiscal arrangements are likely to fail. Section 2 argues that the design and functioning of the system of intergovernmental sharing of resources and responsibilities has political implications, in that it influences relative accountability of different tiers of government to citizens. This is an important issue particularly in the context of newly decentralizing countries in the developing world, where national resources are often concentrated at the center, for historical and sometimes geographic reasons, and where nascent systems of intergovernmental transfers leaves room for uncertainty and confusion about the resources and responsibilities of local governments. This undermines local government accountability and creates perverse incentives for misallocation of public resources transferred fiom higher tiers of government. More generally, the impact of transfers on actual delivery of basic services by local governments depends upon their political incentives. Thus, even if national political agents were to get resource distribution equalized, according to the needs and production possibilities of different jurisdictions, whether these transfers convert to equalization in access to basic services
The Political Economy of Equalization Transfers
465
depends upon how local politics works across jurisdictions. It is likely that the poorer regions of a country are also those where the poor find it harder to influence public policies and hold their representatives accountable for policy outcomes, because of entrenched inequalities and imperfections in political competition. This section explores the role of conditional grants and matching grants in addressing this problem of political accountability. The broad conclusion is that unless accompanied by reforms in political institutions, or informed by the nature of political market imperfections, such transfer design strategies are likely to be blunt instruments to solve the problem of political accountability. Just as national governments can circumvent formula-based design to target transfers to politically favored constituencies, local governments are also likely to circumvent conditionalities in transfers to continue to pursue their political objectives. Section 3 provides a conclusion to this analysis by suggesting directions for future research on how political incentives can be explicitly internalized into the design of intergovernmental institutions to promote the objective of equalization.
2.
HOW DOES POLITICS IMPACT THE DISTRIBUTION OF TRANSFERS?
Central decisions about the regional distribution of resources actually take place within a political economy context where national legislators are elected from regional constituencies, and political bargaining within the legislature determines outcomes (Weingast, 1979; Weingast, Shepsle, and Johnsen, 1981; Baron and Ferejohn, 1989; Becker, 1983). For transfer instruments where the legislature has the greatest power in determining allocations to states or counties, political variables measuring legislative bargaining influences the distribution of intergovernmental transfers. Inman (1988) argues that the pattern of distribution of central grants to the states in the United States does not seem consistent with policies designed to correct inefficiencies of a decentralized tax system, but rather reflects decisions taken by bargaining within the central legislature. Indeed, in the new literature on the importance of legislatures, "equalization" has come to mean something quite different than what is assumed in this volume-equal votes in the legislature produce equal distribution of resources, with states and counties having greater per capita representation in the central legislatures receiving greater per capita resources (Ansolabehere, Gerber, and Snyder, 2003; Porto and Sanguinetti, 2003). The first evidence of the importance of political factors came from the examination of the distribution of federal welfare spending during the Great Depression in the US under President Roosevelt. Wright (1974) found a
466
Khemani
strong positive correlation between New Deal spending per capita and electoral votes per capita across states. However, careful scrutiny of these New Deal spending results by Wallis (1998) highlights the sensitivity of these results to empirical specification. More recently, definitive evidence on the importance of legislative representation is provided by Ansolabehere et a1 (2003), using a "natural experiment9'-in the rnid-1960s, the U.S. Supreme Court ordered the redistricting of U.S. state legislative districts to equalize legislative seats per person across counties. They find that counties with relatively more representation per person prior to this court-ordered redistricting received relatively more transfers from the state per person. But after redistricting, counties that lost legislative seats received a smaller share of transfers, with approximately $7 billion diverted annually from formerly overrepresented to formerly underrepresented counties. Porto and Sanguinetti (2003) provide evidence from a developing country, Argentina, that overrepresented provinces, both at the senate and at the lower chamber, have received more intergovernmental transfers than more populous and less represented provinces. For transfer instruments that are amenable to direct control by the central executive, such as the President, Prime Minister, or central cabinet, political targeting can be more fine-tuned, with regional and even district-level voting patterns in previous elections influencing the distribution of transfers. At the regional level, the partisan identity and influence of regional governments has emerged as empirically significant in explaining the distribution of national resources. Grossman (1994) models grants to the U.S. states as being determined by the "political capital" of state politicians and interest groups in the national legislature, and finds that empirical measures of this are positively correlated with per capita grants. Levitt and Snyder (1995) find that the percent voting Democratic in a congressional district is a significant factor explaining the distribution of overall federal assistance spending across districts in the U.S, when Democrats control the US Congress and the Presidency. Khemani (2003) finds that state governments in India that belong to the same political party as the national government receive greater transfers, and interprets this within a model where the political objectives of state and national policymakers are linked through the party system. Three separate features of voting at district levels have been found to be empirically significant in influencing the distribution of national resources, whether as transfers to local governments or through direct spending in national programs: one, the degree to which voters are "core" supporters of the party in power at the center in that they vote largely on ideological grounds; two, the degree to which voters are "swing", that is, they have weak ideological leaning and are largely influenced by policy actions; and three, the degree to which the electorate participates in elections and is
The Political Economy of Equalization Transfers
467
informed about policies. (Schady, 2000; Case, 2001; Johansson, 2003; Besley and Burgess, 2002; Stromberg, 2004). The extent to which "core supporters" of the ruling national political party receive more or less than voters that are "swing" is a matter of prolific debate in the literature, both theoretically and empirically, with evidence on both sides of the issue, depending upon the local context of electoral competition. Dixit and Londregan (1996, 1998) have stressed the importance of "swing" voters for office-motivated politicians that use spending programs in order to win votes. However, Cox and McCubbins (1986) argue that vote maximizing candidates that are risk averse will over-invest in their closest supporters, while only more risk accepting candidates will pursue "swing" voters. Lindbeck and Weibull (1987) and Snyder (1989) contrast two different objectives-maximizing the votes a party obtains, and maximizing the probability of winning a majority of seats. If the policymaker's objective is to maximize the number of votes received more resources should be allocated to those districts where races are tight, or where more voters are likely to be "swing". On the other hand, if the objective is to maximize the probability of winning a majority of seats in the legislature--as is required to form the executive government in some electoral systems-resources should also be targeted to those core support districts without which the party would be hard pressed to win a majority. Case (2001) finds that block grants from the federal government in Albania to commune-level governments are higher both to those communes where a higher percentage of voters voted for the President's party in the past election, and also to those districts where the race was tight, that is where the absolute difference between the commune's vote for the party and the 50% mark (required to be crossed for victory) was small. Schady (2000) similarly finds that spending under a Presidential poverty alleviation program in Peru was targeted both to provinces that voted in larger numbers for the President, and to those where the election race was tight. Johansson (2003) estimates the number of swing voters directly using election survey data in Sweden, and finds that municipalities with many swing voters receive larger grants than other municipalities. There is emerging evidence that areas where voters are more likely to participate in elections (by turning out to vote), and are better informed about public policies, receive not only more government resources, but also better government performance in service delivery. Stromberg (2004) finds that between 1933 and 1935 in the United States federal assistance to lowincome households was greater in counties where more households had radios and were thus more likely to be informed about government policies and programs. The spread of radio particularly improved information access for rural voters, accounting for as much as 20 percent more in social assistance funds to a rural county than an identical urban county. Besley and
468
Khemani
Burgess (2002) show that state governments in India respond more to declines in food production and crop flood damage through higher public food distribution and calamity relief spending where newspaper circulation, particularly in local languages, is greater. Although the impact of politics on transfers may not necessarily be regressive, if poor districts are the ones with "swing" voters (or "core" supporters) and informed voters, or if poor regions are the ones politically affiliated with the national government. But more often that not, it is the poor that lack political power, that are less likely to participate in political processes, or more likely to live in areas where the reach of information media is limited, and hence are less critical a constituency for political opportunists at national levels. Boex and Martinez-Vazquez (2004) provide a survey of international evidence on the impact of political and economic factors on the distribution of intergovernmental transfers, and conclude that politically better represented and wealthier sub-national jurisdictions are consistently the ones receiving greater transfers.
2.1
Can Institutional Mechanisms Curb Political Influence?
There are two institutional mechanisms that countries have used to curb the impact of political influence-establishing formulae for transfer distribution based on measurable economic characteristics of sub-national jurisdictions, such as local income, population, revenues; and delegating the distribution of transfers to independent agencies, outside the direct line of control of the political executive.' However, until recently (Khemani, 2003), there was no evidence in the literature that explicitly tested whether these institutional mechanisms indeed make a difference. For instance, in Australia, intergovernmental transfers are determined by an independent Commonwealth Grants Commission, which is supposed to be "free from political and bureaucratic bias" (Matthews, 1994, p. 16). Worthington and Dollery (1998) find evidence that some transfers that are not subject to strict fiscal equalization formula that govern other fiscal assistance grants in Australia, are distributed across states in a manner that is consistent with a Grossman-style story of states with greater "political capital" receiving greater transfers. However, they do not provide any evidence to show whether formula-driven financial assistance grants, on the other hand, are indeed impervious to political control, as suggested by the different institutional framework within which they are determined. The Indian federation provides a valuable laboratory for the purpose of examining the impact of both formula design and delegation to an independent agency because of the existence of two major channels of formula-based, general purpose federal transfers to state governments: one
The Political Economy of Equalization Transfers
469
that is determined by an explicitly political body made up of the executive heads of the central and state governments, while the other is determined by a quasi-judicial body with constitutional authority. Using disaggregated data on transfers and political variables for the major Indian states from 19721995, Khemani (2003) finds a pattern of evidence that shows that while the transfers that are determined by political agents are indeed distributed to serve political objectives, the distribution of transfers by an independent agency curbs political influence and is consistent with promoting equity. These transfers are by far the largest source of central assistance to the states, together constituting 30 percent on average of state revenues, and over 50 percent of state borrowing. Transfers in India that are determined at the discretion of political agents are significantly greater to those states whose governments belong to the same political party as that of the national government. Furthermore, amongst partisan states, these discretionary transfers are greater to those states where the party controls a smaller proportion of seats allotted to the state in the national legislature, and therefore has more to gain. Politically affiliated states where the ruling party controls less than half to a quarter of the seats assigned to the state in the national legislature receive discretionary transfers that are greater by 10 to 30 percent of the sample average. In contrast, transfers that are determined by the independent agency with constitutional authority serve to counter these partisan effects on resources available to state governments. Constitutional transfers are also effected by the same political variables, but in exactly the opposite direction than those predicted by the model of electoral competition-politically affiliated states receive lower constitutional transfers than non-affiliated states. Such an outcome is predicted as a Nash equilibrium of a simultaneous-move game between the two central agencies-one having political objectives, and the other equity objectives-that determine resource transfers to state governments. Khemani (2003) argues that these results suggest that constitutional rules indeed act as a check on politically motivated distribution of resources by the national executive. The mandate of the independent agency is to provide equalizing transfers, with greater resources allocated to disadvantaged states. If non-affiliated states are politically disadvantaged, and likely to have fewer national resources directed towards them, whether through intergovernmental fiscal transfers or direct spending by the central government, then greater constitutional transfers would be directed to them not because of political motives but because they happen to be the resource-poor states. If the two sets of transfers are pooled, then the partisan effect on discretionary transfers dominates, that is, total general-purpose transfers from the center are greater when a state government is politically affiliated with the center. Affiliated states whose ruling parties control less than half of
470
Khemani
the state's seats in the national legislature receive total transfers that are greater by 4 to 18 percent of the sample average. This suggests that even when delegation to an independent agency makes a difference, it is difficult to completely reverse the impact of political influence. Furthermore, the contrast between the different types of Indian transfers suggests that the difference is due to the effect of constitutional rules on the general decision-making process rather than the difference between formuladriven versus non-formulaic, discretionary transfers. Although both statutory tax sharing and plan grants are formula-driven, we find a partisan effect on each of them albeit in meaningfully opposite directions. These findings highlight the significance of the political incentive environment within which policy decisions are made, and the limitations of technical formulae in neutralizing or blocking the impact of political imperatives. Although the evidence for the impact of delegation to the independent agency in India is so striking, and so substantially constrains political influence on intergovernmental transfers, the effectiveness of such delegation will depend heavily on the local context. Delegation cannot in general be expected to be successfid. For instance, even within India, State Finance Commissions that are supposed to advise state governments on intergovernmental transfers to locally elected village governments are viewed as largely ineffectual. In South Africa, the Financial and Fiscal Commission, while playing an important role in the initial years of the new democracy, has progressively lost its influence as the country made its transition from conflict years. These examples suggest that the politics that influence the distribution of resources between different tiers of governments may inevitably determine the design and effectiveness of independent commissions aimed at insulating intergovernmental finances from political capture. Even if institutionalized formulae are effective in determining the distribution of those transfer instruments that they are supposed to govern, national political agents typically have other fiscal instruments at their disposal that they can use to target resources to those regions where they stand to gain political advantage. Khemani (2003) discusses how just as the national political agents in India allowed for the creation of an independent Finance Commission, they almost simultaneously created the Planning Commission which allowed them access to other transfer instruments that would be amenable to political manipulation. Khemani (2007) provides alternate evidence on bypassing transfers that are formula-driven andlor delegated to independent agencies, namely, that of deficit financing-fiscal deficits of affiliated state governments are higher, that is, the center uses both direct (central loans) and indirect financing (guarantees, bail-outs) of deficits as fbrther instruments of resource transfer to state governments since
The Political Economy of Equalization Transfers
47 1
these are not subject to the same institutional constraints as formula-based transfers and deliberation by an independent agency. In summary, getting around political considerations is difficult-although institutional mechanisms can change the nature of political bargaining over resources, and even constrain political opportunism to some extent, they cannot hope to entirely do away with it. An important element in trying to equalize access to basic services across regions, conditional on regional needs and own revenue potential, is political representation and participation of sub-national jurisdictions in political processes. The recent empirical evidence suggests that at least as important as technical fiscal design of intergovernmental transfers is the design of political institutions that determine representation and participation. Therefore, perhaps the way to think about equalization policies is not in terms of getting around politics, but indeed making politics central to the policy-making process, with a quest for designing political institutions that promote equalization of representation and participation in political processes across sub-national jurisdictions.
3.
HOW DOES POLITICS IMPACT THE USE OF TRANSFERS FOR SERVICE DELIVERY?
An implicit assumption in the literature on intergovernmental transfers is that the recipient sub-national governments would use these transferred resources to serve the interests of their local citizens, and that the design problem is really one of solving the bargaining game between local jurisdictions of who gets what from the national pie, or simply of equalizing fiscal capacities across local jurisdictions. However, the nature of intergovernmental sharing of resources and responsibilities also has implications for local political accountability. In many developing countries that are newly decentralizing, and where local jurisdictions have little ownrevenue potential, over-dependence on national transfers undermines the accountability of local governments to their citizens. Rodden (2002) finds that in a large cross-section of countries over time dependence of subnational govemments on transfers is positively associated with higher subnational fiscal deficits, evidence that is consistent with the idea that greater dependence on transfers creates conditions for reduced accountability of local governments to their local electorate. The World Development Report 2004 (The World Bank, 2003) and Ahmad et a1 (2005) discuss more generally how incomplete institutions of decentralization, and poor transfer arrangements, can lead to breakdowns in service delivery by local governments. The essence of their analytical framework is that the delivery of services requires strong relationships of
472
Khemani
accountability between the actors in the service delivery chain. In contrast to the delivery of goods or services in a private, competitive market, where the service provider is directly accountable to the client or consumer of the service, the delivery of public services involves at least two relationshps of accountability. First, clients as citizens have to hold policymakers or politicians accountable for allocating resources towards these services. Second, policymakers in turn need to hold the service providers accountable for delivering the service (Figure 1). Weaknesses in service-delivery outcomes can be attributed to a breakdown in any of these links along the route of accountability.
Policymakers
Poor People
Providers
Figure 1. The Framework of Accountability Relationships
Devolving responsibility for public services to lower tiers of government means that the politician who is responsible is now a locally-elected one. The hope is that this would make him more accountable to the citizens, as they can monitor him more closely and attribute changes in service quality to him more easily. That is, decentralization will strengthen the citizen-local politician relationship of accountability, and thereby the other relationships of accountability for service delivery. However, as shown in Figure 2, decentralization introduces one more relationship of accountability, namely between the central and local policymaker. The rules and practices governing fiscal transfers, regulation and expenditures between central and local policymakers affect service delivery through their effect on the accountability relationship between local policymakers and providers. Ahrnad et a1 (2005) provide a striking example of this in the area of administrative decentralization, or lack thereof, when fiscal and political
The Political Economy of Equalization Transfers
473
decentralization proceeds without being accompanied by explicit staffing strategies or public administration r e f o n n l o c a l service providers continue to be full employees of upper-tier government, or local governments have only limited ability to hire and fire providers. Misalignment between the structure of the government bureaucracy and the assignment of service responsibilities to different tiers confuses incentives, weakens accountability for service delivery, and creates conflicts of interest instead of checks and balances. The problem of over-dependence on intergovernmental transfers for local government accountability for service delivery was mentioned earlier. Micro evidence on how local service delivery might suffer when financed overwhelmingly by transfers from above is available from Nigeria. Nigeria is one of the few countries in the developing world to have significantly decentralized both resources and responsibilities for the delivery of basic health and education services to locally elected governments. Bulk of the revenues of local governments (above 90 percent) in Nigeria are obtained from their constitutional entitlement to a share of federal revenues, which in recent years of oil price booms has implied substantial resource flows to local governments, distributed in accordance to a formula that includes equity elements. Khemani (2006) presents the following evidence that overdependence of local governments on federal transfers has undermined local accountability and created perverse incentives at the local level to rnisallocate public resources.
I
NATIONAL POLICYMAKERS
Figure 2. The framework of accountability relationships under decentralization
474
Khemani
A survey of local governments, public primary health facilities, and health care providers in the state of Kogi revealed that 42% of staff respondents report not receiving any salary for 6 months or more in the past year at the time of the survey. Variation in non-payment of salaries cannot be explained by variation in resources available to local governments, nor in actual spending on salaries reported by local governments, and is therefore evidence suggestive of a problem of general accountability of local governments in managing substantial resource transfers from taxpayers outside their jurisdiction. Non-payment of salaries of health staff by local governments is reminiscent of a similar problem of non-payment of teacher salaries in primary schools in the 1990s, when primary education was decentralized to local governments. Following nationwide agitations by teacher unions a policy of deducting primary school teacher salaries from the revenue share of local governments in the Federation Account was adopted (termed "deductions at source"), with the salaries being directly passed-on to the teachers. This "solution" of essentially converting a portion of an untied federal transfer into a specific purpose grant for teacher salaries, although successful in ensuring that teachers get paid, has unintended pernicious effects of undermining overall accountability of local governments. Local governments claim that deductions at source in essence lead to "zero allocations", thereby preventing them from carrying out any of their responsibilities for service delivery. Such uncertainty about resources actually available to local governments facilitates local evasion of responsibility under the guise of fiscal powerlessness. What local governments do receive as transfers is therefore sometimes treated as the personal fief of local politicians
3.1
Taking Politics Seriously - What Factors Shape Political Incentives for Service Delivery by any Tier of Government?
Even if resources and responsibilities between the central and local governments can be effectively decentralized, there remains the question whether locally-elected governments will have better incentives for service delivery. This section will seemingly digress from the issue of intergovernmental transfers, and focus seriously on the conditions under which local governments, or any elected government for that matter, will have the right incentives to effectively provide basic services. I argue that this digression is warranted because the extent to which equalizing transfers policies might effectively equalize access to basic services across local jurisdictions depends upon local political incentives. If the true underlying
The Political Economy of Equalization Transfers
475
objective of equalization transfers is to ensure access to basic services for all citizens, then transfer design should be considered in the context of political market conditions that determine the incentives of political agents at all tiers of government for service delivery. It is a common observation across countries, rich and poor alike, that substantial public expenditures are systematically misallocated, for example to wage bills for bulky state administrations, to farm subsidies that impose distortionary costs on the economy and fail to benefit the poor, and to large infrastructure projects that allow political rent extraction without creating sustainable assets, all at the expense of quality public services. These misallocations have a disproportionateimpact on the poor, who are known to benefit from increased access to public services. Even resources allocated to broad public services such as basic education and health might be ineffective in actually delivering those services, if, for example, the posts of teachers and doctors are used to extend the patronage of government jobs, rather than being held accountable for actual service delivery. Misallocation has persisted despite a sea change in the way in which governments are selected and remain in office. From 1990 to 2000, the number of countries governed by officials elected in competitive elections rose from 60 to 100.~Democratization might be expected to benefit most the "median" or average voter, who in most developing countries is "poor". Yet, public policy in emerging democracies does not seem to have benefited poor voters. Why do policy-makers that depend upon political support from the poor not effectively deliver basic services to the poor? Keefer and Khemani (2005) argue that imperfections in political markets explain this puzzle. Political market failures in this analysis are reduced to three broad features of electoral competition-one, lack of information among voters about politician performance; two, social and ideological fragmentation among voters that leads to identity based voting and lower weight placed on the quality of public services; and three, lack of credibility of political promises to citizens. Informed voting is costly, and voters may have difficulty in coordinating information to reward (or punish) particular politicians or political parties for specific actions that improve (or worsen) the quality of public services. Similarly, socially andfor ideologically fragmented societies are less able to provide the incentives to their political agents to improve broad public services, because voting is more likely to occur along the dimension of narrowly-defined identities. Even if voters are informed and coordinated in focusing on specific policies, if political competitors cannot make credible promises prior to elections, incumbents are more secure from challenge and have fewer incentives to be responsive to citizens. If politicians are credible only to a few voters, with whom they can maintain clientelist relations, then public resources are allocated to targeted benefits for these "clients", instead of to broad public services.
476
Khemani
A strong conclusion of the analysis here, which is difficult for development policy, is that the most adverse effects of political market imperfections are felt in the area of broad public services, such as health and education. It is especially difficult for voters to assess the quality and efficiency of service provision in these areas and to evaluate the responsibility of specific political actors for service breakdowns or poor outcomes. By the same token, political competitors find it especially difficult to make credible promises about service provision. Voters cannot easily collect information that would verify that politicians have fulfilled their promises. Moreover, even if they could, politicians in many countries can only make credible promises to narrow groups of voters. For these voters, it may be politically more efficient to promise narrow targetable goods-such as a farm subsidy, a contract for an infrastructure project (especially if it doesn't need to get built or if the contractor can get away with using poor quality materials), a government job as a teacher or a doctor (especially if they won't be held accountable to show-up to work in schools and clinics). The large mass of unorganized voters would suffer from the resultant poor quality of services-bankrupt state utilities that bear the subsidy burden, dilapidated roads and undrinkable water, and empty schools and clinics where children don't learn and the sick don't get treated-but would find it difficult to coordinate action to improve political incentives. Social fragmentation in the electorate exacerbates these problems of voter coordination in determining reward and punishment based upon political actions towards the quality of public services. Within this framework of analysis, decentralization to locally elected governments will improve political incentives and service delivery outcomes if voters are better informed and likely to use information about local public goods in their voting decisions for electing local governments, if there is greater social homogeneity and coordination of preferences for local public goods, and if political promises are more credible at local levels. Confronting the political economy of service delivery at local levels brings out a dilemma-if local governments in poor areas are also the ones that suffer most fiom lack of political accountability, because poor voters are more likely to vote in uninformed ways, being susceptible to campaign slogans, or polarized along non-economic, ideological dimensions such as religion or ethnic identity, and political promises are particularly lacking in credibility or prone to clientelism in environments of social inequality, then even if equalization design leads to greater resources being transferred to poor regions, the local governments might not convert these resources into actual services. In such environments of low political accountability, tinkering with the design of intergovernmental transfers, such as moving from untied transfers
The Political Economy of Equalization Transfers
477
to grants conditional upon actual service delivery outcomes can only be a solution if political incentives at the national level are conducive to actually enforcing these conditionalities, and punishing errant local governments. But often, these local governments are locally powerhl, and control the local vote bank which is of value to national political leaders, who would therefore be loathe to enforce conditionalities and antagonize local political leaders. Quite apart from the possibility of such a political nexus between national and local governments, there might be simple administrative obstacles to perfect monitoring and control by national governments over local political representatives. Just as national governments can circumvent fiscal rules to target resources to regions where they stand to gain political advantage, local governments also might be able to circumvent transfer conditionalities. Understanding the nature of political market imperfections can help to optimally design the institutions of decentralization, or to analyze the impact of available and operational institutions. If political markets function better at more centralized levels of government, then decentralization might not need to be political, but rather designed to help central politicians with agency problems, such as gathering information, monitoring performance, and enforcing norms for locally produced and consumed services.
3.2
What can Policymakers do in the Face of Political Market Imperfection?
What can a policymaker at the national level, with the right political incentives for equalizing access to basic services for all citizens, do when inheriting an existing decentralized system where locally elected governments, with varying degrees of accountability, are responsible for basic services, financed out of intergovernmental transfers? One of the "political market imperfections" that might be particularly tractable to a policy solution is that of information asymmetries. Greater information dissemination about the roles and responsibilities of government, and the outcomes of public resource allocation might help improve the problem of political accountability. Based on this political economy view of public accountability this paper proposes a specific type of policy intervention to promote the availability of minimum basic services in all regions, namely, providing citizens with greater information about the actual outcomes of basic services and which agents are responsible for these. As mentioned earlier, Besley and Burgess (2002) show that state governments in India are more responsive to declines in food production and crop flood damage via public food distribution and calamity relief expenditure when newspaper circulation, particularly in local languages, is greater. They explain this impact by appealing to the ability of informed
478
Khemani
citizens to use political institutions such as elections to reward or punish public agents. In Uganda, the national education ministry began publishing information about school grants provided to districts in newspapers and other media, after survey evidence revealed that district governments were not transferring these resources to schools. A follow-up survey in Uganda showed that this information dissemination had a substantial impact in preventing leakage of public funds away from purposes intended in public budgets (Reinikka and Svensson, 2004). Such evidence suggests that when voters have access to the media, and the media publishes information about particular policies, they are able to extract both greater resources and better performance from political agents with regard to these publicized policies. But such evidence does not tell us whether an increase in media access improves government incentives either to target the neediest citizens with transfers or to invest in public goods benefiting large segments of the population. It could be that the mass media make it easier for politicians to take credit for targeted payoffs to particular constituencies, leading them to reduce expenditures on public goods or on broad-based social programs. In countries where the reach of information media is limited and difficult transportation and communications systems divide the electorate, it cannot be relied upon to promote equalization efforts, in the sense of promoting equal access to basic services for all. The information base of poor rural citizens of developing countries might particularly be skewed in a way that diminishes their ability to hold elected officials accountable for the quality of public services. Therefore, although reforms to solve information problems should include loosening restrictions on the media, including dependency on government advertising, and reducing barriers to entry for media firms, promoting competitive media markets may not be enough to ensure adequate coverage of government performance. Competitive information media might provide inadequate information if the markets they serve consist largely of better-off groups that do not consume public services. External interventions, such as journalist training, subsidized information transmission, or information campaigns by civic society organizations, may be useful in promoting the diffusion of the information needed for political accountability, particularly in poorer countries. There are numerous examples of successful information campaigns on public services, but no rigorous evidence for what kinds of information dissemination strategies systematically alter the nature of political a c c ~ u n t a b i l i tWe ~ . ~can, however, speculate on some features of the type of information and dissemination strategy that would make a difference in terms of improving political incentives. Recent work on the psychological underpinnings of social communication indicates that getting information dissemination to have the desired impact on actual outcomes is a particularly
The Political Economy of Equalization Transfers
479
difficult mechanism design problem (Lupia, 2003). The problem is as follows-suppose there is a target audience that lacks sufficient information to accomplish certain tasks; what kind of information and dissemination strategy is necessary and sufficient to give the target audience the requisite skill to accomplish those tasks? Lupia (2003) provides a framework for thinking about the content and nature of information dissemination that is likely to make a difference. He identifies three necessary conditions that information dissemination strategies must satisfy in order to be effectiveone, it should be brief enough so that citizens actually pay attention and register the information (the battle for attention and working memory);two, it should be pertinent enough that people actually retain the information for long enough periods of time as required to evaluate the actions of their agents (the battle for elaboration and long-term memory); and finally, the information must make a difference to the choice citizens make, that is, must be actually used by them to evaluate their agents (the battle at the precipice of choice). Keeping these three principles of cognitive science in mind, we may begin to characterize the types of information that are likely to make a difference for political competition. Information about broad public sector performance aggregates, whether based on surveys, budget studies, or report-cards on sector-wide service provision, is likely to be politically relevant only if it provides individual voters with a sense of how their specific representatives in government hurt or help them. It is even more likely to promote reform when the information providers show evidence of being able to coordinate voter responses to poor service provision. This suggests that they are more effective to the extent that they transcend the role of simple information collectors or survey firms and take a more aggressive stance themselves with regard to political accountability. Finally, information campaigns are most helpful when they tell citizens not just how bad services are, and which are worse than others, information that citizens tend to have already, but also how bad the services are in their neighborhood relative to others, who or what processes are responsible for this, and whether services have improved as a result of specific policy reforms pursued by political representatives. The process of collecting such information might, in and of itself, trigger improvement if the organizations collecting it could credibly threaten to mobilize voters around public service issues. Politicians in all countries respect interests that can bring voters to the polls. To the extent that the process of information collection by organizations lowers their costs of mobilizing voters, service improvements are likely to follow. Independent agencies with no direct political ambitions and with local credibility--civic society organizations, NGOs, or even constitutionally created advisory agencies, such as fiscal commissions-might be the
480
Khemani
implementers of such information campaigns. They would present a latent political challenge to incumbents without in fact having any political mandate or ambitions, other than mobilizing citizens and holding political agents accountable around particular policy issues. Their power would be derived from their credibility on public service issues, and their ability to disseminate information widely amongst voters on the performance of public services. Similar ideas on the role of evaluating spending programs from the perspective of actual service delivery, and disseminating information about such evaluations have been expressed much earlier (see Shah, 1997; and Picciotto and Weisner, 1998), but rigorous analysis of which basic outcome indicators to use for evaluation, and how to disseminate the information so that it indeed makes a difference for government accountability and performance is still lacking. To address this gap, experimental studies are planned or underway in a number of countries-Benin, India, Nicaragua, Pakistan, Peru, Uganda-and coordinated by the Development Research Group at the World Bank. These studies focus on outcomes in two sectors-education and health. Survey instruments are developed to measure particular outcomes, that are likely to be the appropriate ones within the individual country context. For example, in India, political economy analysis suggests that information about basic learning outcomes in education service delivery-how many children can actually read and write simple sentences-might be particularly effective in mobilizing citizens to demand better performance along this specified dimension from government agencies. The dissemination strategy is a grassroots campaign at the village level through village meetings and pamphlet distribution by a local NGO. In Peru, the politically salient issue is non-transparent distribution of schooling inputs across public school districts, and the experiment will focus on training school management and communities to use web-based tools to access information about the pattern of distribution of inputs. Each of these strategies will be implemented using a randomized, experimental methodology so that impact on what changed in the process of accountability, and whether it actually improved outcomes can be scientifically evaluated. The overall objective is to develop a set of tools to improve the effectiveness of public spending programs-1) what outcomes, 2) how to measure them, 3) how to disseminate information about them, and 4) what institutional characteristics are needed for an appropriate implementing agency. The results are expected to emerge by 2007, and reported in a compiled volume.
The Political Economy of Equalization Transfers
CONCLUSION This chapter has argued that analyzing intergovernmental transfers from a political economy perspective is essential in order to promote equalization objectives of transfers, interpreted broadly as promoting equal access to basic services for all citizens. Political incentives of government agents impact how transfers are distributed across regions and sub-national jurisdictions, and how these local jurisdictions use the transferred resources to service their citizens. On the question of distribution, recent evidence suggests that the political characteristics of voters in a region-whether they participate in political processes, whether they are informed about public policies, whether they are core supporters of the national political party or "swing" voters-have a large effect on resources transferred to a region. Instruments such as formula-based transfers and delegation of distribution criteria to independent agencies can have the power to curb political influence, but cannot undo it, because more flexible fiscal instruments are usually available to national policymakers to bypass the control of formulae and independent agencies. Poor regions are more likely to have voters that are less active in political processes, less informed about public policies, and less critical a constituency for national political leaders, and hence political distortions in intergovernmental resource allocation is more likely than not to be nonequalizing. Given this evidence, moving beyond fiscal design of intergovernmental transfers and investing in understanding what kinds of political institutions promote broad-based representation and participation would therefore be critical for equalization policies to be effective. On the issue of how transfers are used to service citizens, political economy analysis emphasizes the impact of decentralization institutions in general on political incentives of local governments. The design of transfers, and related institutions of fiscal and administrative decentralization, directly affect accountability of local governments to their citizens. Accountability for actual delivery of basic services is likely to be hampered by three broad "political market imperfections9'--one, information constraints amongst citizens about the resources and responsibilities of their local governments, and actual outcomes of public policies; two, social cleavages that prevent citizens from coordinated action to extract good performance fiom their governments; and three, credibility problems that limit the platform of political competition to patronage policies or clientelism that benefit only a few, politically-connected citizens, at the expense of the many. Again, the dilemma of equalization policies is that local governments representing poor citizens are more likely to suffer from these political market imperfections, so that even if substantial resources were transferred to these jurisdictions, they might not translate into actual services.
Khemani
482
This political economy analysis suggests information campaigns as a particular, tractable, instrument that might be used to address political constraints to the provision of minimum basic services across all regions. Information campaigns might also be particularly amenable to be combined with policy instruments of intergovernmental transfers, such as transfers conditional upon service delivery outcomes. Thus far the literature on intergovernmental transfers has focused on the informational needs of the agencies determining transfers, and some amount of information sharing, although largely with other units of government, and not directly with citizens. The political economy analysis suggests a role for carefully designed information dissemination by these agencies that are traditionally only in the business of determining grants, to mobilize and concentrate citizen attention on issues of service delivery. These are clearly preliminary ideas, but a start in the direction of moving beyond technocratic solutions based solely on normative fiscal analysis. If the objective of true equalization in the sense of equal access to basic services for all citizens is to be realistically pursued, the technocratic tool-kit would benefit from the development and addition of new tools to assess the nature of political institutions, political market imperfections, and strategies to address these imperfections.
Notes 1. 2.
3. 4.
The findings, interpretations, and conclusions expressed in this paper are those of the author, and do not necessarily represent the views of the World Bank, its Executive Directors, or the countries they represent. The Commonwealth Grants Commission in Australia is the best example from amongst the older federations of the world. However, it is in the newer federations in Asia and Africa that decision-making over intergovernmental transfers are increasingly being delegated to an independent agency, such as the National Finance Council in Malaysia, the Revenue Mobilization Allocation and Fiscal Commission in Nigeria, and the Finance Commission in India (which is studied here). According to the number of countries reported in the Database of Political Institutions (Beck et al, 2001) as having competitive elections for executive and legislative office (EIEC and LZEC equal to seven). A celebrated example is the public service report cards developed by the Public Affairs Centre in Bangalore, India (www.pacindia.org).
The Political Economy of Equalization Transfers
References Ahmad, J., S. Devarajan, S. Khemani, and S. Shah, 2005. Decentralization and service delivery. In Handbook of Fiscal Federalism, E. Ahmed and G. Brosio, eds., Forthcoming, Surrey, U.K.: Edward Elgar Publishing. Ansolabehere, S., A. Gerber, and J. Snyder. 2002. Equal votes, equal money: Court-ordered redistricting and public expenditures in the American states. American Political Science Review, 96: 767-777. Baron, D., and J. Ferejohn. J. 1989. Bargaining in legislatures. American Political Science Review, 83(4): 1181-1206. Becker, G. 1983. A theory of competition among pressure groups for political influence. Quarterly Journal of Economics, 98: 371-400. Beasley, T. and R. Burgress. 2002. The political economy of government responsiveness: Theory and evidence from India. Quarterly Jounal of Economics, 117(4): 1415-1452. Boex, J., and J. Martinez-Vazquez. 2004. The determinants of the incidence of intergovernmental grants: A survey of the international experience. Public Finance and Management. Forthcoming. Case, A. 2001. Election goals and income redistribution: recent evidence from Albania. European Economic Review, 45: 405-23. Cox, G., and M. McCubbins. 1986. Electoral politics as a redistributive game. The Journal of Politics, 48: 370-89. Dixit, A., and J. Londregan. 1996. The determinants of success of special interests in redistributive politics. The Journal of Politics, 58: 1132-55. Dixit, A., and J. Londregan. 1998. Fiscal federalism and redistributive politics. Journal of Public Economics, 68: 153-180. Gramlich, E. M. 1977. Intergovernmental grants: a review of the empirical literature. In The Political Economy of Fiscal Federalism, W. E. Oates, ed., Kentucky: Lexington Books. Grossman, P. 1994. A political theory of intergovernmental grants. Public Choice, 78: 295303. Inman, R. 1988. Federal assistance and local services in the United States: the evolution of a new federalist fiscal order. In Fiscal Federalism, H. Rosen, ed., Chicago: University of Chicago Press. Johansson, E. 2003. Intergovernmental grants as a tactical instrument: empirical evidence from Swedish municipalities. Journal of Public Economics, 87: 883-915. Keefer, P., and S. Khemani. 2005. Democracy, Public Expenditures and the Poor. World Bank Research Observer, 20: 1-27 Khemani, S. 2007. Party politics and fiscal discipline in a federation: Evidence from the state of India. Comparative Politcal Studies, 40(6). Khemani, S . 2006. Local government accountability for health service delivery in Nigeria. Journal of African Economies, 15(2): 285-3 12. Khemani, S. 2003. Partisan Politics and Intergovernmental Transfers in India, Working Paper No. 3016, Development Research Group, The World Bank, Washington, D.C. Levitt, S., and J. Snyder. 1995. Political parties and the distribution of federal outlays. American Journal of Political Science, 39: 958-80. Lindbeck, A., and J. Weibull. 1987. Balanced budget redistribution as the outcome of political competition. Public Choice, 52: 273-97. Lupia, A. 2003. Necessary conditions for improving civic competence: a scientific perspective. Mimeo. University of Michigan, Department of Political Science, Ann Arbor.
484
Khemani
Mathews, R. 1994. Fiscal Equalization: Political, Social, and Economic Linchpin of Federation. Canberra: Centre for Research on Federal Financial Relations. Australian National University. Musgrave, R. 1959. The Theory of Public Finance: A Study in Public Economics. New York: McGraw Hill. Musgrave, R. 1983. Who should tax, where, and what? In Tax Assignment in Federal Countries, C. McLure, Jr., ed., Canberra: Centre for Research on Federal Financial Relations. Australian National University. Oates, W. 1972. Fiscal Federalism. New York: Harcourt-Brace-Jovanovich. Pereyra, P. 1996. A politico-economic approach to intergovernmental lump-sum grants. Public Choice, 88: 185-201. Picciotto, R. and E. Weisner. eds. 1998. Evaluation and Development: The Institutional Dimension. New Brunswick, U.S.A. and London, U.K.: Transaction Publishers Porto, A., and P. Sanguinetti. 2001. Political determinants of intergovernmental grants: evidence from Argentina. Economics and Politics, 13: 237-56. Rao, G. 1998. India: intergovernmental fiscal relations in a planned economy. In Fiscal Decentralization in Developing Countries, Richard M. Bird and Francois Vaillancourt, eds., Cambridge: Cambridge University Press. Reinikka, R., and J. Svensson. 2004. The Power of Information-Evidence from a Newspaper Campaign to Reduce Capture. Policy Research Working Paper No. 3239, The World Bank, Washington, D.C. Rodden, J. 2002. The dilemma of fiscal federalism: intergovernmental grants and fiscal performance around the world. American Journal of Political Science, 46(3): 670-687 Schady, N. 2000. The political economy of expenditures by the Peruvian Social Fund (FONCODES), 1991-1995. American Political Science Review, 94: 289-304. Shah, A. 1997. Balance, Accountability, and Responsiveness: Lessons about decentralization. Mimeo. The World Bank: Washington, D.C. Stromberg, D. 2004. Radio's impact on public spending. Quarterly Journal of Economics , 119(1): 189-222 Weingast, B., K. Shepsle, and C. Johnsen. 1981. The political economy of benefits and costs: A neoclassical approach to distributive politics. Journal of Political Eocnomy, 89: 642-64 Weingast, B. 1979. A rational perspective on congressional norms. American Journal of Political Science, 23: 245-62. The World Bank, 2003. World Development Report 2004: Making Services Workfor Poor People. Washington, D.C.: The World Bank. Worthington, A., and B. Dollery. 1998. The political determination of intergovernmental grants in Australia. Public Choice, 94: 299-3 15. Wright, G., 1974. The political economy of New Deal spending, an econometric analysis. The Review of Economics and Statistics, 56: 30-38.
DISCUSSANT COMMENTS A TANGO BETWEEN POLITICS AND TECHNICAL SOLUTIONS CAN WORK FOR EQUALIZATION
MUSHARRAF R. CYAN Asian Development Bank
The paper, 'the Political Economy of Transfers' by Stuti Khemani, has brought up an important but often underemphasized aspect of intergovernmental transfers. Politics, as the author has discussed, determines the outcomes of transfers in an important way. The paper has brought up an important aspect for the success of equalization schemes by discussing this dimension of transfers. It has also proposed that intergovernmental transfers are likely to be a blunt instrument for resource allocation if implemented without consideration of the political economy surrounding them. The author has argued for greater dissemination of information to increase debate in support of equalization schemes. The discussion is valid from the point of view of ensuring success of those schemes. The paper proposes that any formula does not of itself automatically determine the transfers that result from an intergovernmental fiscal transfer system. Politics play a role. Beyond this conclusion, however, there can be two ways to look at the issue. One looks at politics as a spoiler; that is, something that does not allow the formula to adequately determine the desired outcomes for transfers and shares. This viewpoint is vividly expressed in the paper, which adequately demonstrates that politics do affect the outcomes of transfers and that formula results remain only a partial determinant of transfers. The second view sees politics as a determinant that is just as significant as technical considerations, maybe even legitimately so. Recognition of this limitation of technical solutions will lead to appropriate action to prevent equalization schemes from dwindling in their impact. For some forms of
486
Cyan
political input into equalization transfer schemes, we can argue that politics should be channeled to work with the technical determinants. If treated well, politics can become a force for implementation of formula-based equalization schemes. It is the latter viewpoint that is being expressed by Khemani. The underlying point of seeing politics as a spoiler is that politically determined transfers are discretionary and may not be preferentially in favor of the weaker capacity regions. Where politics is adding to the resources made available to such regions, it is not a matter of concern and is not being considered in the paper. In a functioning democracy, allocation of resources remains at the center of politics and the working of the state. Equalization transfers are, therefore, a matter which would be debated, hotly contested and ultimately defined in the policy arena according to the legitimately dominant political choice that the electorate has appointed as its representative government. This principle is not greatly contested. The prerogative of the legislature to carry out appropriations is a generally accepted norm. Federal systems however are complex arrangements for allowing articulation of different interests and resolution of competing claims. As part of a federal arrangement, often as a component of a federal scheme itself, there are broad societal agreements to allocate resources to some communities or regions in preference to others. It is this notion that creates room for equalization transfers outside the functional domain of the legislatures. Once the principle for such transfers is established, the formula based determination of shares provides a rational, transparent and debatable means to achieve the policy objectives. The ends are not contested but the means is seldom an agreed and acclaimed criterion. Even in the developed world, formula for sharing fiscal resources remain a contested terrain'. A closer review of the choices of formula, how they are arrived at, and how they are implemented shows that transfer formula is not a policy instrument that can be easily crafted to incorporate the political choices of a government. Unlike an ordinary appropriations bill, which has many concomitant rigidities and apron strings, an equalization formula has historical, social and political connotations: the most powerful being a level of acceptance of an existing arrangement amongst the various contestants2. It is rare that devising a formula takes place on a clean slate, and changing a formula or a resource sharing arrangement is more difficult than devising a new one. In some situations, in Japan for example, simply contemplating changes to an existing transfer system leads to much political debate3. Nation building objectives of transfers are palpable in developed4 as well as developing countries. This is probably more so in developing countries where access to development is often seen as being part of nationhood. This is potent stuff for politics and any technical considerations will be very weak in the face of such considerations. ~ i ~ e r and i a ~India have both used
Discussant Comments
487
preferential transfers to rein in centrifugal politics. South Africa had a more sanguine scenario at the end of apartheid and did not create political pressures for immediate development6. In the case of Russia, it seems that politics of coalition-building have taken over an earlier similar trend7. This at times sets up different dynamics for equalization where meeting the politically voiced concerns is more important than a technically determined equalization system. Recently, the same has been happening in Pakistan where natural resource revenues are likely to be part of a larger political settlement with some provinces. It is important to recognize that politics is not a constraint to equalization. It could only be considered a constraint if equalization were to be designed and implemented in a hypothetical model. Politics is a determinant of transfers and equalization schemes. Recognition of this fact can help the mechanism and institutional design of transfer schemes to meet the objective of equalization. Recognizing the importance of politics and its inevitability and legitimacy in setting policy choices will help design and implement better equalization schemes. In fact, working with the politics and at times adding components of politics might be a way to achieve greater success. The key is to set up a relationship between politics and technical solutions. Due to the specific nature of politics in each country this will need to vary from country to country and even within countries. Pakistan's four Province Finance Commissions each set up different transfer schemes to suit their intergovernmental politics8. To make equalization successful, politics and technical advice have to move ahead with a synergistic harmony, one contributing to the other. This can only happen where technical advice is provided with an understanding of politics. Design and implementation of equalization schemes requires the setting up of a tango of synergistic advancement of equalization and politics. The requirements for this are that: a. Technical advice should come in early in the political process and help build the political case for preferred equalization choices; b. Constitutional rules and institutional arrangements for equalization should be in place; and c. Technical considerations should be formulated clearly to create a robust play of politics around them. Where technical support fails, even well crafted equalization schemes can generate perverse incentives for sub-national governments. These steps are elaborated in the following paragraphs. Transfers and equalization are political choices in any country. Whether technical considerations or politics determine the grant shares of
488
Cyan
jurisdictions is a concern all around the world, even in developed countries with considerable experience with designing and managing equalization schemes9. Arguments in favor of decentralization strengthening government accountability recognize the role of political processes in ensuring appropriate taxation and financing of serviceslO.The nature of equalization scheme has implications for the commitment to following a formula. In some federations, formula-driven outcomes are used to give an agreed (if only temporary) arrangement to an ever-contested field. A formula provides a manageable ruse for politicians to settle contested claims to shares of national resources. In this sense, it reduces ongoing debates and unpredictability, and probably lowers transaction costs. As a result, politics in such situations will be pro-formula. However, where a formula is at best an interim settlement in a field of counter forces, attempts at seeking what has not been agreed in a formula is likely to remain an option for some of the political parties. These attempts often begin soon after the results of the formula become known. Satisfying such attempts provides avenues for discretionary transfers. In this case, discretionary transfers might be a workable support to retain a formula-based distribution as the mainstay of transfers. Pakistan's National Finance Commission formulae have been stabilized in this manner. The best case scenario for the full implementation of a formula-driven outcome comes from a wider political acceptance of the formula. Political campaigns and electoral choice should determine resource allocation criteria and this in turn should be translated into a formula. This kind of dispensation has a high chance of faithful implementation; but for this to happen, technical advice needs to be available to the politicians much earlier in the process. Currently, the technical advice usually only comes into play after the basic policy is in place. In other cases, technical debates may remain at a distance from politics. Politicians need to be aware of the technical aspects of the issue before they determine the resource allocation criteria that are to be followed in the formula. Any absence of constitutional rules creates incentives for violation of any formula. Where rules are weak and politics is a series of evolving situations to suit the government, chances of a formula-based equalization dispensation surviving remain weak. There will be many violations as time goes by. This can be seen in the case of Russia where weak rules and evolving political agreements created a milieu that can only suite a discretionary support to regions1'. But by no means is this a lone example. Institutions can also buttress formula-based arrangements. In countries where no autonomous agency exists to manage the formula based distribution, the only solution available is to mount a political crisis to try to force the government to stick to a formula. In coalition governments where interests can be cobbled together through patronage and transfers to specific
Discussant Comments
489
regions and constituencies, any likelihood of pressuring governments in this way becomes farfetched. An important step is to set up, through the process of grant design, competitive politics between equalization and counter-equalization forces. This can happen where an autonomous commission provides membership or right of audience to competing interests and regions. This has been the operational procedure in some countries. Pakistan's National Finance Commission brings together the two richer provinces with the two poorer ones. The Commission is less of a technical body and more of a political bargaining house. Constitutionally, it is above the parliament and has the status of a mechanism within the federal constitutional scheme. Competitive politics within the Commission's proceedings can be long-drawn and at times lead to unsatisfactory solutions, but the formula is always adhered to. Violations provide an immediate recourse to parliament, other constitutional fora and the Supreme Court. The extent to which equalization schemes can be effectively included in grant systems depends in large measure on the constitutional rules. The National Finance Commission in Pakistan is mandated to determine the transfers from a pool of shared taxes. Provinces jealously guard this as a general purpose grant. They also strongly defend their right to transfers emerging from natural resource revenues. In this sense, their established prerogative ensures their shares but militates against the federal government laying down any conditions for further inter-jurisdictional distribution in favor of weaker local governments. In some other cases, institutional mechanisms are a way to provide voice to the weaker jurisdictions. It is here that the technical advice works best because it provides technical arguments in favor of the claims of the weak. This contributes to political solutions. Work on sharing of natural resource royalties and providing arguments in favor of the claims of politically weak regions rich in natural resource is a case in point. Institutions are generally needed to adequately administer a sophisticated equalization scheme. Robust formula would be a necessary accompaniment of the scheme and institutions will exert pressure to lay down such a formula. However, the importance of a formula for politics should also be recognized. It does not obviate the play of politics but redirects it by creating a template for its operation. An equalization formula approach can contribute to positive outcomes for politics. It is different from the politics of discretionary grants in that:
1. It addresses to some extent the question of unpredictability; 2. A formula determines a basis for contestation and some definition of rights; it can become sophisticated over time; 3. The existence of a formula renders discretion less legitimate;
Cyan
4. An opaque formula sometimes serves the purposes of politics well by taking the load off the politicians; 5. A formula becomes untouchable and lets the sleeping dogs lie: in such situations, changing the formula requires a play of politics; and 6. Any formula requires periodic calibration to suit the objectives of the equalization policy; an existing formula, even if difficult to change, provides grounds for debate.
The extent to which institutions become arrangements to 'combat political influence' is complex. The purpose of institutional arrangements is to reduce transaction costs but not to bypass politics. In fact, they provide a forum for the play of politics and set limits on the politicians. If this is an effective feature of the institutional arrangements, they can help politics achieve equalization. Pakistan's National Finance Commission decisions are above the fray of parliamentary allocations -- this is a strong constitutional principle and the basis of Pakistani federation, where one province with only 5 percent of the national population has a veto power over resource allocation. As a result, transfers to provinces from the Commission have been relatively stable and in accordance with the f ~ r m u l a ' ~This . does not mean that it leads to less politics. The province of Sindh uses more resources per capita in providing poorer standards of services but cites the low standards in asking for more resources. This brings up another important question. Can equalization work through inputs alone and, even where performance based grants are not equalizing, can they be totally excluded? In Pakistan, because of the status of the National Finance Commission, these questions are above executive discretion. The formula effectively limits the discretion of the government and creates a recognizable criterion against which to measure the performance of the executive. Most importantly, the formula creates new entitlements and, once those entitlements are in place, politics is necessary to change them. Federations work on the basis of norms and institutions. The extent to which equalization schemes are designed around the fknctionality of institutions and norms determines their success. Institutional arrangements can also provide a basis for technical debate. There is more likelihood of policy makers seeking technical solutions through institutions than through open-ended parliamentary processes. Pakistan's Province Finance Commissions are much more effective than the National Finance Commission in directing equalization in accordance with technically sound advice. Through them, the Provinces are seeking to work with formulae which target weaker local areas. Some technical issues have an impact on the politics of transfers. Looking at these issues, politics can become more than a constraint or force undermining the carefully crafted technical solutions in the domains of:
Discussant Comments
a. b. c. d. e. f.
Choice of variables; Data systems; Incentives for performance; Nature of revenue assignments amongst the tiers of governments; Patronage politics; and Planning transition.
If these are not attended to, equalization schemes can set up perverse incentives for sub-national governments. Developing countries grappling with nascent decentralization find this a difficult area. Working with indicators like low enrolments or literacy, and lags in infrastructure development, can set up situations where past and present inaction is rewarded. In this case, politics will often try to use formula distributions to hide previous inefficiencies or poor performance. Some variables used in a formula can make policy choices clearer but, in other cases, they draw veil of opacity over the calculations. In Pakistan, a debate on the distribution of national fiscal resources has been underway for the last three years. Each province argues for a different set of variable to suit its position, thus preventing a consensus. This is a healthy debate that focuses technical energies on political policy choices. Technical studies and discussions bring out the relative merits of various positions. The bargaining stance of each province in the National Finance Commission deliberations is a charged political position, aided by technical points. At the same time, it helps the provincial governments communicate with the electorate as well as attempt to raise their share of revenue. The political nature of the Commission, as opposed to the technical Commissions in Australia and India, allows a debate of this nature to take place. Where calculations are complex, their translucency does not allow for political debate to engage with technical work. On the other hand, over simplification or ignoring to draw up necessary technical options might not be an answer. For developing countries attempting to set up formula-based transfer systems, a major constraint remains undeveloped data systems. Due to a narrow range of variables that can be used in formula design in those circumstances, not all aspects of policy or regional choices can be addressed. In such situations, sub-national governments will always be vying for additional resources since some of their special needs will not have been addressed through the formula. Credibility of data systems can be another issue. Where formula uses data to capture underdevelopment, for example, indices devised to calculate its relative measures are often hotly contested. Unreliability of data (either real or perceived) opens doors for some disaffected jurisdictions to stake claims outside the formula. In this manner, relative unreliability of data weakens the credibility of formula-based shares and creates space for negotiated
492
Cyan
(political) settlements. A central government's position will usually be to strictly adhere to the formula but this is weakened once flaws in data come to light. Transfers generate the need to judge performance. In a legal sense, Pakistan's local governments are seen are subservient to the provinces. Their major dependence on transfers to finance local service deliveryI3 does not help their situation; provinces show a lot of concern for performance and whether local governments are working toward achieving provincial policy priorities in education, health, water supply and roads. (All these sectors are nearly fully financed from provincial transfers.) As a result, the situation has moved towards delineation of performance criteria. That these should be clear, manageable and measurable to provide a norm-based intergovernmental relationship is now accepted in Pakistan. The formula for transfers when using service delivery indicators provides a mechanism for playing out this upward accountability of local governments to the province: where h d s are well spent the indicators should improve. It also ensures that local government performance is rewarded in a transparent manner without the play of partisan politics. Few improvements have been achieved, however, and continuing use of performance indicators will probably lead to enunciation of mechanisms to ensure that politics does not set up perverse incentives for local government performance. Whether an equalization scheme is part of a comprehensive grant system or is an isolated component also attracts politics. If there are negotiated transfers, or transfers operating under non-equalization principles, and these are channeling major resources to some jurisdictions, the political energies are likely to go into those arrangements rather than equalization transfers. In other cases, equalization-based formula can lead to countervailing forces. The recent Canadian experience of provinces under-pricing natural resources and failing to develop new ones is related to the equalization schemeI4. Similarly, ascendancy of oil revenue in Nigeria led to a decreasing importance of 'area of derivation' as a principle for revenue sharing15. This kind of situation results in politics preserving revenue sharing arrangements at the expense of equalization. Arrangements for central taxation and local spending have an inherent instability. This certainly applies in case of Pakistan. The central government earns the odium for tax collection the local government receives the kudos for carrying out works. Most of the tax revenue comes from the two richer provinces. Even a per capita sharing of these resources would provide much higher than baseline revenues to the weaker provinces. Planning equalization schemes in such situations has its limitations because richer jurisdictions start demanding higher shares and argue for the return of revenue. In Pakistan, the provinces have little incentive to levy higher taxes since the rate and collection of taxes are controlled by the federal government.
Discussant Comments
493
Instead, they follow the course of politically demanding higher shares of federal resources. The negative attributes earned by the federal government due to provincial jostling for more resources weakens its nerve for further tax effort. In this way, the nature of equalization, seen in the context of the general taxation scheme, does not set up positive politics and places limitations on revenue generation. The way politics works in a society also influences the determination of equalization schemes. Patronage-based politics creates pressures for all politicians. This is a major issue in Pakistan today16 where there are a large number of representatives in the three tier local government system as well as the provincial and federal legislatures. All of them are expected to deliver tangible developments and help meet the infrastructure backlog. This has set up a kind of competition17 only manageable by instituting simple sharing arrangements. The provincial representatives take up a large portion of the pool of resources, leaving little fiscal space for sophisticated equalization schemes1*. The scope of sound equalization schemes is therefore circumscribed in many cases. Past rigidities and a transition towards an equalization scheme are also subjects of political interplay. Past expenditure patterns are tied up in salaries in some recent decentralization cases like Indonesia and Pakistan. There is a reliance on public sector jobs in many jurisdictions. Reduction in the salary bill has a high political cost. As a result, pursuit of equalization in such a situation usually has to produce a result under which all jurisdictions are winners in the short run and allocations are changed in the medium term. Transition has to be planned in a way that equalization creates sufficient support in the poorer jurisdictions. It needs to leave the existing (excessive) allocations as they are and address imbalances through additions to the pool over time. Thirdly, if the richer jurisdictions can be given access to credit and other incentives from the capital market, they can allow equalization schemes to work for recurrent budgetary activities. In conclusion, technical considerations in an equalization scheme have to play out policy choices in more than one ways. It can be remarkably useful to recognize the importance as well as nature of politics while designing and implementing equalization schemes. Chances of success of an equalization scheme are proportional to the extent of politics getting integrated into that scheme. The exact solution will vary both from country to country and over time.
Cyan
Notes 1. 2. 3.
4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.
Bogdanor, 200 1 Ibid. Sato, 2002 Guihery, 2001 Edevbie, 2000 Bahl, 2001 Popov, 2004; Martinez-Vazquez & Boex, 2001 ADB/WB/DfID, 2004 McKay, 2004 Lockwood, 2005 Popov, 2004 Ahmad et al, 2003 ADB/WB/DfID, 2004 McLean, 2003 Uche & Uche, 2004 Department of Finance, SIndh 2004 ICG, 2004 ADB/WB/DfID 2004
References ADBIWBlDfID. 2004. Devolution in Pakistan, Asian Development BanklWorld BankJDepartment for International Development, UK, Islamabad, Pakistan Ahmad, Nuzhat, and Syed Ashraf Wasti. 2003. Pakistan. In ZntergovernmentalJiscaltransfers in Asia: Currents practice and challenges for the&ture (Asian Development Bank, TA 5902), Paul Smoke and Kin Yun-Hwan, eds., 176-218. Retrieved from ht~://www.adb.org/Documents/Books/Intergovernmental~Fiscal~Transfers/Intergov~m ental-Fiscal-Transfempdf
Bahl, Roy. 2001. Equitable vertical sharing and decentralizing government finance in South Africa. Andrew Young School of Policy Studies Working Paper, Georgia State University, Atlanta, GA. Bogdanor, Vernon. 2001. Devolution in the United Kingdom. Oxford, UK: Oxford University Press. Department of Finance, Sindh. 2004. Report on Implementation of Sindh Province Finance Commission Awards During FY 2002-03 & 2003-04, Karachi, Pakistan. Edevbie, David. 2000. The politics of 13 percent derivation principle, remarks of commissioner for finance and economic planning, Delta state Nigeria, quoted on website of Urhobo Historical Society. Retrieved from h t t p : / / w w w . w a a d o . o r g / E n v i r o n m e n t / F e d G o 3PercentA ~ 1location.htm Guihery, Laurent. 2001. An economic assessment of German fiscal equalization schemes since 1970-What prospects for a unified Germany. Public Finance and Management, l(4): 393-419. Lockwood, Ben. 2005. Fiscal Decentralization-A Political Economy Perspective. Coventry, UK: University of Wanvick and CEPR.
Discussant Comments
495
Martinez-Vazquez, Jorge, and Boex, Jameson. 2001. Russia's transition to a new federalism. World Bank Institute Working Paper, Washington, DC. McKay, John. 2004. Questions Posed to the Parliamentary Secretary to the Minister of Finance during debate on Bill C-24-Federal-Provincial Fiscal Arrangements. McLean, Iain. 2003. Fiscal Federalism in Canada. Oxford, UK: Nuffield College, University of Oxford. Popov, Vladimir. 2004. Fiscal federalism in Russia: Rules versus electoral politics. Comparative Economic Studies, 44 (4): 5 15-4 1 Sato, Motohiro. 2002. Political Economy of Fiscal Decentralization. Tokyo, Japan: Hitsubashi University. Uche, Chibuike & Uche, Ogbonnaya. 2004. Oil and the politics of revenue allocation in Nigeria. ASC Working Paper 5412004, African Studies Centre, Leiden, The Netherlands.
Index
absolute equality 299 absorptive capacities 120 ACIR 343,365,367 administrative capacity 179,442 administrative efficiency 4 14 administrative grants 235 Advisory Commission on Intergovernmental Fiscal Relations 365 agency costs 153, 156,158 aggregate external finance 3 10 Argentina 127, 171, 172,218,228,232, 249,263,466,484 asymmetrical transfer systems 176 Australia 13-20, 28,30,58,68,73,79, 82, 93, 129, 137, 143, 149, 158, 160-168, 172, 175, 179, 195,208,223,236, 242,244,248,252,255,263,268, 269,270,274,281-288,294,305, 3 17,340,345,349,355,371,377, 382,385, 394,397,400,414,426, 434,468,482,491 average effective tax rates 3 17 Back-end Payments 392 big bang approach 436 budget affordability 3 12 budget autonomy 168,338 budget resource envelope 3 11 build-operate-transfer (BOT) 441 buoyancy 123,171,241,246 burden index 28 1 Canada 7, 13, 130-137, 143,158,160, 162, 175, 176,224,237,248,254,263, 264,279,285-288,294,317,340, 371,385,397,400,433,495 capacity equalization 4, 13, 15,28,62, 82,159,289,293
capacity indicators 160, 187, 190,267, 294 capital expenditures 8,234,309,346, 347,350,351,352,353,355,356, 358,362 capital financing systems 436 categorical grants 167,440 China 32-40,43,50,58, 143,215,216, 222,246,247,270-277,283,286, 289,294,311,320,339,340,343, 400,436 civil society forums 6, 141 civil society monitoring 153 closed funding 7, l7O,2O 1 Colombia 171,232,287,320,340 Commonwealth Grants Commission 19 composite needs index 33 1 comprehensive equalization 62,82. 268, 394 conditional grants 40,42,53, 151, 168, 221,227,230,238,239,263,269, 270,281,284,348,447,465 co-participation 454 cost correction factors 193 cost disabilities 13 cost reimbursement approach 226,227 cost reimbursement grants 227,245 county-based local government areas 322 Czech Republic 309,342,456 Dana Alokasi Khusus (DAK) 447 Dana Alokasi Umum (DAU) 37,447 debureaucratization 101, 102 Decentralization Action Plan 445 deconcentrated spending 449 deconcentration 101 deductions at source 474 degrees of equalization 176,284 Department of Local Administration 447 Department of Local Government Promotion 445
Index depreciation expenses 77,79 Development Bank of the Philippines (DBP) 444 disabled pension scheme (AI) 281 discretionary transfers 228, 230,469, 470,488 disincentive effects 438 distortionary costs 475 distortionary effects 195 distributable pool 226,241,245,366,445 earmarked grants 36,37,46,52,276, 405,447 earmarked transfers 232,234 Ecuador 217,218,230,248,294,320 efficient central planner 132 elasticity of intergovernmental transfers 24 1 equalization entitlements 134,269 equalization formula 7,9, 16,22,23,28, 42,53, 169, 188,260,298,301,315, 320,338,340,341,468,486,489 equalization grants 3,4,5,6,7,9, 10, 13, 15,36,42,52, 53,57, 146, 172,249, 251,254,255,264,277,288,291, 292,297,298,300,318,404,453, 456,458,459 equitable growth 6, 100, 103 European Union 456 expenditure needs 3, 13, 16,25,36,37, 38,40,45,51,53,96,98, 104, 130, 137, 149, 160, 175, 177, 180, 181, 184,206,2 14,225,240,246,25 1, 255,259,260,266,291,298,300, 301-309,311,321-329,334-339,341, 363,366,383,394,405,412,433, 447,456,458 expenditurenorms 42,247,296,308, 309,3lO,3Il, 339,365 expenditureresponsibilities 3,31,34,98, 225,240,296,297,309,372,433, 438,442 expense effort 66 extra budgetary accounts 2 16 family allowance in agriculture (AFA) 28 1 Federation Account 220,294,301,338, 474
FFSR (Fund for Financial Support of the Regions) 226 Financial and Fiscal Commission (FFC) 355 financial capacity index 28 1 financial input side 427 financial intermediary 438 fiscal decentralization effect 2 10,212 Fiscal dependency 2 10 fiscal equalization principles 130 fiscal externalities 14, 16, 18,27,28,30, 247,263 fiscal gap 18,293,329,330,333,334, 336,364,365,366 fiscal liability 385 flypaper effect 208,406 formula-based equalization 37,291,292, 486,488 formula-based normative grants 366 formula-based transfer system 438 France 208,252,455,459 Fundo de Participacao dos Estados (FPE) 276 Fundo de Participacao dos Municipios (FPM) 276 General Appropriations Act (GAA) 440 general grants system 37 general revenue grants 148, 172, 174, 177,202 Georgia 242 Germany 145,160,255,278,494 Ghana 143,242,244,320 Government Financial Institutions (GFIs) 44 1 grant pool 19,20,21,29,33 1,341,354 grant volatility 385 grants dependency 169 grants-in-aid 149 health care grants 63, 148,428 horizontal disparities 7, 132 horizontal fiscal balances 97 horizontal fiscal equalization (HFE) 61 horizontal fiscal imbalance 9, 104,373 horizontal scheme 262 horizontal shares 205,226 Horizontal spillovers 167 housing gross expenses 75
Index Human Development Indicator (HDI) 46 incentive-neutral transfer 164 India3,58, 104, 127, 143, 147, 158, 159, 160,219,229,233,242,249,271-
280,281,289,294,317,320,338, 340,466,469,477,480,482-486,491 Indonesia 3 1-39,40,46,49,50,54,59, 104, 126, 127, 143, 162,210,214, 218,231,240,248,288,293,294, 305,311,319,338,342,371,381, 394,400,401,433-439,442,449, 450,493 Inpres program 447 interagency externalities 167 interagency reallocation of resources 165 Inter-government Coordinating Council (ICC) 441 inter-jurisdictionalexternalities 3 18 inter-jurisdictionalredistribution 367 interjurisdictional spillovers 105, 106, 107 inter-municipal equalization 165,202 Internal Revenue Allotment (IRA) 42, 108,235,439 inter-regional disparities 52 inter-regional transfer 16,29 Japan 13, 143,223,234,246,294,3 11, 339,45 1,486,495 lagged tax sharing scheme 171 Land Bank of the Philippines (LBP) 444 land rates 62 Latvia 215,222,242,244,294,302,339, 342 level of service provision 66,68,71,77, 281,372,375 LGU Guarantee Corporation (LGUGC) 444 load shedding 100 Local Government Units (LGUs) 439 macroeconomic fiscal planning 225 macroeconomic forecasting models 347 macroeconomic risks 170, 171 Malawi 23 1,237,242,244,301,305,341 matching grants 40, 167, 170,340,406, 409,410,433,439,440,444,449,465
Medium Term Expenditure Framework (MTEF) 4 15 merit goods 96, 134,405,406 Mexico 49, 127,219,229,234,248,287, 320 moral hazard 135, 169, 171, 173, 174, 20 1 multi-tiered intergovernmental system 435 Municipal Development Fund (MDF) 44 1 Municipal Development Fund Ofice (MDFO) 444 Nash equilibrium 16,469 National Economic Development Authority (NEDA) 44 1 needs-resources gap 206 neo-institutional economics framework 141 net fiscal benefit 64,260,265,363 Nicaragua 320,32 1,480 Nigeria 143,146,171,220,229,242, 244,249,293,294,301,320,338, 341,394,398,401,473,482,492, 494,495 non-equalizing transfers 230 non-tax revenues 108, 197, 199,22 1,264 normative grants 238 OECD 259,286,289,295,414,436 old-age and survivors insurance (AVS) 28 1 open-end financing 165 own-revenue mobilization 32 own-source revenue 20,2 1,98, 123,327, 376,394,397,413 Pakistan 104, 126, 143, l46,2 18,230, 234,243,244,480,487,490,491,494 Pareto optimal 26 patronage-based politics 493 payroll tax 20,62 per capita expenses 65,66,83 per capita needs 29,65,66,330 Poland 143,456 political accountability 465 political capital 466 political decentralization 102,473
Index poverty alleviation program 467 poverty reduction 6 , 3 1,57,98, 100, 101, 103,114, 124, 125,126,324 preference revealing mechanisms 169 pre-tax sharing system grants 277,339 progressive tax structure 32,241 pure equalization transfers 165 rate tax-back 135 recurrent funding 4 17 Regional Urban Development Fund (RUDF.) 446 Representative Expenditure System (RES) 267,302 Representative Revenue System (RRS) 317 Representative Tax System (RTS) 261, 264,383,413 revenue assignments 3,491 revenue autonomy 130,216,297,322, 340,438,440,442 revenue capacity 7,36-39,45,52, 104, 137, 175,305,317,364,383,392, 395,399,413,414,426,447 revenue mobilization effect 2 10,2 12,213 revenue raising capacity 75,92, 177, 178, 320,392,395,400,413 revenue raising efforts 61,62, 79 revenue sharing 8,36,42,53,64,93, 109, 142, 147, 149, 150, 158,214,253256,281,314,320,338,340,371, 376,378,381,385,398,399,400, 448,454,492 revenue-raising capacity 104, 105, 107, 108,122,265,281,316,373 revenue-raising powers 98 Russia 45, 128, 173,216,220,226,230, 234,248,293,297,305,310,312, 314,339,341,382,451,487,488,495 services tax revenue 63, 148 shared tax approach 2 14,2 17 Sierra Leone 225,247,398,407,408 South Africa 8,52, 126, 143, 150, 162, 175,203,220,240,246,249,252,
294,312,317,341-349,353,357, 358,362,413,450,470,487,494 spatial allocation 4, 16,28 special grants 64, 147, 149, 169, 174,279
specific purpose payments (SPPs) 63, 269,43 1 Sri Lanka 243,244,320 Staffing Grants 235 subnational borrowing 435 substitution effect 167, 174,202,428 Support Grants 235 swing voters 467 Switzerland 13, 143, 167, 175,220,238, 249,270,272,273,274,280,281, 284,287,288,317,342,382 Tanzania 126, 143,230,247,3 11,341, 408,413,433,434 tax avoidance 2 16 tax competition 196 tax reimbursement grants 172 tax return grants 42 Tax Sharing System (TSS) 37,277 taxes as countercyclical policy instruments 170 Thailand 34,40,44,49,54,222,23 1, 234,249,436,438,442-449,451 the Philippines3 1,36,42,47,48,54,57, 95,96, 104, 111, 117, 122, 127, 132, 214,240,248,436,439,444,448 tied grants 3,8,9,62,75,82,403,409, 412,419,423,424,427,429,431, 433,442,458 transaction costs 6, 141, 153,488,490 transfer dependency 15,32, 174, 197, 210,212 transfer-allocation formula 37 transfer-as-you-go system 172 transition countries 7,56,208,210,211, 245,246,283,286,291,325,333,450 Uganda 147,15 1, 162,243,244,294, 301,311,317,343,478,480 Uganda LGFC 3 11 Ukraine 143,309,3 l4,3 15,327,339, 340,342 uncertainty costs 153, 158 unconditional grants 164, 172,201,215, 338 unfunded mandates 110, 125,442 user charges 62,390, 391,442 Venezuela 320
Index vertical fiscal balance 165, 166, 197,293, 338 vertical fiscal imbalances (VFI) 8, 106, 166,372,414 vertical share 7,205,208,210,211,217, 225,226,240,242,243,245,253 vertical spillovers 167 vertically asymmetric grants 165 Village Development Committee (VDC) 298 zero allocations 474