34401 EAST ASIA Decentralizes MAKING LOCAL GOVERNMENT WORK THE WORLD BANK EAST ASIA DECENTRALIZES Making Local Government Work EAST ASIA DECENTRALIZES Making Local Government Work THE WORLD BANK Washington, D.C. © 2005 The International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org All rights reserved 1 2 3 4 08 07 06 05 The findings, interpretations, and conclusions expressed herein are those of the author(s) and do not necessarily reflect the views of the Executive Directors of the International Bank for Reconstruction and Development / The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgement on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development / The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Internet: www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org. ISBN-10: 0-8213-6059-0 ISBN-13: 978-0-8213-6059-0 eISBN-10: 0-8213-6060-4 eISBN-13: 978-0-8213-6060-6 Cover photos (left to right): Tran Thi Hoa/World Bank (2002); Hugh Sitton/Getty Images; and Ray Witlin/World Bank (1996). Library of Congress Cataloging-in-Publication Data East Asia decentralizes. p. cm. Includes bibliographical references and index. ISBN 0-8213-6059-0 (pbk.) 1. Decentralization in government--East Asia. JQ1499.A56D424 2005 352.2'83--dc22 2005047796 Contents Preface ix 1 EAST ASIA DECENTRALIZES 1 Roland White and Paul Smoke 2 THE RULES OF THE INTERGOVERNMENTAL GAME IN EAST ASIA: DECENTRALIZATION FRAMEWORKS AND PROCESSES 25 Paul Smoke 3 ACHIEVING FISCAL SUSTAINABILITY 53 Sudarshan Gooptu 4 FISCAL DISPARITIES IN EAST ASIA: HOW LARGE AND DO THEY MATTER? 67 Bert Hofman and Susana Cordeiro Guerra 5 PUBLIC EXPENDITURE ON THE FRONTLINE: TOWARD EFFECTIVE MANAGEMENT BY SUBNATIONAL GOVERNMENTS 85 Edward Mountfield and Christine P. W. Wong 6 SUBNATIONAL OWN-SOURCE REVENUE: GETTING POLICY AND ADMINISTRATION RIGHT 107 Robert R. Taliercio 7 MANAGING HUMAN RESOURCES IN A DECENTRALIZED CONTEXT 129 Amanda E. Green 8 DECENTRALIZING HEALTH: LESSONS FROM INDONESIA, THE PHILIPPINES, AND VIETNAM 155 Samuel S. Lieberman, Joseph J. Capuno, and Hoang Van Minh 9 EDUCATION REFORMS IN EAST ASIA: POLICY, PROCESS, AND IMPACT 179 Elizabeth M. King and Susana Cordeiro Guerra 10 DECENTRALIZING BASIC INFRASTRUCTURE SERVICES 209 George E. Peterson and Elisa Muzzini 11 GOVERNANCE GONE LOCAL: DOES DECENTRALIZATION IMPROVE ACCOUNTABILITY? 237 Jose Edgardo Campos and Joel S. Hellman 12 COMMUNITY-DRIVEN DEVELOPMENT: DECENTRALIZATION'S ACCOUNTABILITY CHALLENGE 253 Susan Wong and Scott Guggenheim v vi Contents LIST OF BOXES, FIGURES, AND TABLES BOXES 1.1 Variants of Decentralization: Pros and Cons 6 1.2 Hard Budget Constraints 16 5.1 The Backdoor Route to Decentralization in China 98 5.2 China: Devolution by Default Rather Than by Design 99 5.3 The Block Grant Experiment in Ho Chi Minh City 100 6.1 Revenue Assignment across Subnational Governments in China's Gansu Province 118 6.2 Improving Tax Administration in Quezon City, Philippines 123 6.3 Local Revenue Administration in Action in East Asia 125 7.1 Personnel Exchange in India and Japan: Whose Capacity Is Being Strengthened? 144 7.2 Lessons from Uganda's District Service Commissions 150 8.1 Local Innovations in Health Service Delivery in the Philippines 163 9.1 School Report Cards in Paraná: A New Incentive System 199 10.1 Vietnam's Red River Delta: Efficiency Gains from Decentralizing Irrigation 218 10.2 Vietnam: Watershed Management 227 10.3 Indonesia: When Transport Regulations Distort Trade 228 10.4 The Philippines: Toll Road Management 231 12.1 CDD Projects in Three East Asian Countries 257 12.2 Funds Arriving in Cambodian Schools 260 FIGURES 1.1 Subnational Expenditures as a Share of Total Public Spending 2 4.1 Percentage of Population Living below PPP$2 a Day in 2002, East Asia and Pacific 69 4.2 Disparity in Provincial per Capita Revenues Before and After Transfers 75 7.1 Subnational Employees as a Share of Total Government Employment 134 7.2 Personnel Spending as a Share of Total Subnational Government Expenditures 134 9.1 Net Enrollment Rates in East Asia, 2000 181 9.2 Central and Local Government Education Spending in the Philippines 191 9.3 Literacy and Enrollment Rates in China, by Province 193 9.4 Literacy and Enrollment Rates in the Philippines, by Region 194 9.5 Enrollment Rates in Indonesia, by Province 194 11.1 The Accountability Framework 238 11.2 Decentralization and the Accountability Framework 239 11.3 The Negative Impact of Decentralization, as Cited by Firms in Indonesia 244 11.4 Obstacles to Business, as Cited by Firms in Indonesia 244 11.5 Informal Payments after Decentralization, as Cited by Firms in Indonesia 245 12.1 Project Disbursement in Indonesia by Ratio, Fiscal Years 2000­2 260 12A.1 Bankwide CDD Lending for Fiscal Years 2000­3 264 12A.2 Regional Trends--CDD Investments by Fiscal Year 264 12A.3 CDD Project Breakdown by Country 265 12A.4 CDD Project Breakdown by Sector 265 12A.5 Sectoral Breakdown of CDD Projects by Country 265 TABLES 1.1 Economic and Demographic Trends in East Asia, 1975­2002 5 1.2 Expenditure Shares and Functional Allocations in East Asia 10 1.3 Vertical Organization of Intergovernmental Systems in East Asia 11 1.4 Subnational Fiscal Structure of Selected East Asian Countries 13 Contents vii 2.1 Levels of Government Administration 27 2.2 Decentralization Policy 28 2.3 Decentralization Frameworks 30 2.4 Basic Features of Political Systems 32 2.5 Subnational Assemblies and Elections 33 2.6 Subnational Budgetary and Personnel Autonomy 34 2.7 Subnational Transparency 35 2.8 Subnational Civic Participation and Civil Society 36 2.9 Subnational Functional Assignments and Expenditure Shares 39 2.10 Subnational Revenues 40 2.11 Intergovernmental Transfers 42 2.12 Responsibility for Managing Decentralization 44 2.13 Decentralization Strategies 45 2.14 Building Decentralization Capacity 47 3.1 Subnational Borrowing Practices 57 3.2 Subnational Fiscal Risks 61 4.1 Disparities in Provincial Revenues before Grants, per Capita 70 4.2 Disparities in China's Provincial Fiscal Outcomes, per Capita, 1979­2002 70 4.3 Variation in Revenues across Provinces in Indonesia, 1994­2002 71 4.4 Equalization in Intergovernmental Transfer Systems 72 4.5 Disparities in Province-Level Revenues After Grants, per Capita 75 4.6 Evidence of Equalization 75 4.7 Disparities in Province-Level HDI Indices, 1997­2002 76 4.8 Disparities in Province-Level Persons per Hospital Bed, 1997­2002 77 4.9 Disparities in Province-Level Life Expectancy, 1996­2002 77 4.10 Disparities in Province-Level Literacy Rates, 1994­2002 77 4.11 Expenditure Disparities and Service Outputs and Outcomes at the Subnational Level 78 4.12 Fiscal Disparities and Public Services Outputs and Outcomes in China, 1985­2000 79 5.1 Expenditure Decentralization in East Asia in Comparative Perspective 86 5.2 Expenditure Assignments in East Asian and Comparison Countries 88 5.3 Strength of Core Expenditure Management Processes at Subnational Levels 90 5.4 Do EA5 Countries Meet Conditions for Effective Decentralization of Expenditure Management? 95 5.5 Distribution of Expenditures and Revenues in China, by Administrative Level 96 6.1 Estimated Own-Source Revenue of Subnational Governments 108 6.2 Classification of Subnational Taxes by Degree of Central vs. Local Control 109 6.3 Tax Assignment by Local Government in the Philippines 110 6.4 Own-Source Revenue Assignment 113 6.5 Highest-Yield Fees and Sources of Operating Revenues of Subnational Governments in the Philippines, 2000 115 6.6 Most Significant Charges by Local and Provincial Governments in Indonesia, 2002 115 6.7 Own-Source Revenue Composition by Type of Local Government in the Philippines, 2001 119 6.8 Composition of Regional Revenue in Indonesia, 2002 119 6.9 Annual Nominal Growth in Local Own-Source Revenue in Indonesia, 1996­2002 120 6.10 Tax Administration Models in East Asia 121 7.1 Key Features of Administrative Decentralization 130 7.2 Central Authority over the Subnational Civil Service 133 7.3 Staffing Authority among Subnational Governments 135 viii Contents 8.1 Selected Health Status Indicators 160 8.2 Health Status of the Poorest and Richest Population Groups in Selected East Asian Countries 162 9.1 Student Performance on Mathematics and Science Tests 181 9.2 Percentage of Decisions Related to Lower Secondary Education at Each Level of Government, 1998 and 2003 184 9.3 Locus and Mode of Key Decisions in Lower Secondary Education, 1998 and 2003 185 9.4 China: Education as a Share of Total Fiscal Expenditures 190 9.5 China: Per-Student Educational Expenditure, 1989, 1997, and 2000 190 9.6 Philippines: Median Values of SEF Resources per Pupil 191 9.7 Indonesia: Per Capita Education Spending, 2001­2 192 9A.1 Cross-Study Comparison of Functions 202 10.1 Road and Highway Investment in China 211 10.2 Road and Highway Investment in China, by Type 212 10.3 Piped Water Coverage within the Urban Population, Vietnam 212 10.4 Capital Spending as a Share of Local Government Expenditure, Philippines 214 10.5 Citizen Perceptions of Corruption in Different Layers of Government in Indonesia 216 10A.1 Water Supply Coverage Rates 234 10A.2 Sanitation Coverage Rates 234 10A.3 Median Investment in Water Supply and Sanitation, 1990­2000 235 10A.4 Median Urban Tariff Rate 235 11.1 Regulatory Burdens on Firms in Indonesia 245 11.2 Perceptions of Corruption in the Philippines 247 11.3 Sources of Information on Politics, as Cited by Households in the Philippines 248 11.4 Correlation between Funding Priorities of Public Officials and Households in the Philippines 248 11.5 Responses from Facilities on Administrative Delays in the Philippines 249 12.1 Points of Local Government Engagement in CDD Projects 262 12A.1 World Bank Lending in Support of CDD, Fiscal Years 2000­3 264 12A.2 World Bank­Funded CDD Projects Open in the East Asia Pacific Region, as of June 30 2003 266 Preface Decentralization has gained momentum across The report received useful advice from a team of East Asia, and it is time to take stock of the experi- peer reviewers comprising Richard Bird, Roy Bahl, ence so far. This study reviews intergovernmental James Hicks, Stuti Khemani, Keith McClean, reform in the region, distills key messages, high- Ranjana Mukherjee, David Rosenblatt, Dana Weist, lights positive experiences, and points out areas and Don Winkler. where policy makers will need to take priority Many people both inside and outside the World action to avoid going down the wrong path. Where Bank provided comments on the report during draft- relevant, the study draws on international experi- ing. Valuable contributions were received from ence while recognizing the economic, political, and Madga Adriani, Mats Anderssen, Naomi Aoki, cultural factors unique to East Asia. Philippe Auffret, IIhem Baghdadli, Jitendra Bajpai, The report has been written primarily for policy Aldo Baietti, Joven Balbosa, Halsey Beemer, Simeth makers who are currently making decisions on Beng, Kirida Bhaopichitr, Jasmin Chakeri, Songsu these issues, but academics, business people, and Choi, Luis Crouch, Soren Davidsen, Lynnette de la development practitioners should also find it use- Cruz Perez, Robert Ebel, Wolfgang Fengler, Cyprian ful. It does not attempt to provide an exhaustive Fisiy,James Ford,David Gomez-Alvarez,Philip Gray, analysis or a detailed practical manual on decen- Arvind Gupta, Alejandro N. Herrín, Dingyong Hou, tralization in East Asia. Space limitations do not Boun Oum Inthaxoum, Emmanuel Jimenez, Kai permit coverage of all facets of decentralization Kaiser, Wolfgang Koehling, Philip Lam, Qing Lani that may be important. Nor is every country in the Wu, Jennica Larrison, Blaine Lewis, Xiaofan Liu, Puti region included. Instead, the volume focuses on six Marzoeki,StephenMink,AmitabhaMukherjee,Tariq East Asian countries where decentralization is a Niazi, Antonio M. Ollero, Demetrios Papathanasiou, major issue: Cambodia, China, Indonesia, the Juan Antonio Perez, Tanaporn Poshyanand, Sjamsu Philippines, Thailand, and Vietnam. The report is Rahardjs, Merwin Salazar, Rubi Sugana , Bambang intended to be helpful in tracking progress on Suharnoko, Hioraki Suzuki, Luiz Tavares, Thanh Thi decentralization in East Asia today and in guiding Mai, Jessica Tisch, Roy van der Weide, Minh Van reforms as they go forward. Nguyen, Khuankaew Varakornharn, Emiliana Vegas, Thisreportwasproducedbyamultisectorteamled Jonathan Walters, Liping Xiao, Bastian Zaini. Despite by Duvvuri Subbarao and Roland White. Chapter efforts to compile a comprehensive list, some who authors are listed on the Contents page. The authors contributed may have been inadvertently omitted. benefited from background papers and input from The team apologizes for any oversights and reiterates various specialists, namely Alex Brillantes, Hana its gratitude to all who contributed to this report. Brixi, Jasmin Chakeri, Robert Ebel, James Ford, Kai The team undertook a wide range of consulta- Kaiser, Bui Duong Ngheim, Barbara Nunberg, tions including workshops in Cambodia, China, Duvvuri Subbarao, Helen Sutch, Charas Suwan- Indonesia, the Philippines, Vietnam, and the Lao mala, Jorge Martinez Vazquez, and Christine P. W. People's Democratic Republic. The participants in Wong. Team assistance was provided by Muriel these workshops included policy makers, academ- Greaves, Gloria Elmore, and Walter Meza-Cuadra. ics and nongovernmental organizations. The work was carried out under the overall direc- Book design, editing, and production were coor- tion of Homi Kharas. dinated by the World Bank's Office of the Publisher. ix 1 EAST ASIA DECENTRALIZES Roland White and Paul Smoke A fundamental transformation in the structure of delivery functions reasonably effectively. Where government has been taking place across East Asia. decentralization has proceeded more gradually, as Before 1990 most East Asian countries were highly in Cambodia and Vietnam, it has produced some centralized; today subnational governments have gains in service delivery and public participation at emerged as the fulcrum for much of the region's the local level. Decentralization has also unleashed development. They deliver many critical services local initiative and energy. New service-delivery and account for a significant fraction of total public models have begun to emerge, and the potential expenditures (see figure 1.1). Though East Asia's for continued improvement and innovation has decentralization has come later than in some other become tangible. parts of the world, it is now here to stay. But there have also been problems, and there is a From China to Cambodia to Thailand, countries real danger that programs could stall. While East of varying sizes, income levels, and political systems Asian decentralization does not pose the macro- are moving government down this path. Reform economic risks confronted in other regions, it does processes are under way in almost every country. face three distinctive challenges, which play out Demands for accountable government whose serv- differently in each country setting. The first chal- ices reach the grassroots effectively are on the rise. lenge relates to the design of sound intergovern- Thus, a wholesale recentralization of government is mental organizational arrangements. For example, unlikely. With all that is invested in successful unclear assignments of functions among levels of development of these decentralized public sectors, governments threaten to sidetrack decentralization the key question is no longer whether to decentral- reforms in some countries. The second challenge ize. It is how best to design intergovernmental concerns the development of robust financial structures and manage the implementation process mechanisms for channeling money to subnational to achieve optimal results. governments. In some countries, the failure to In many countries initial progress is encourag- allocate sufficient own-source revenues to local ing. Where decentralization "leaps" have been governments could hamper their ability to deliver attempted, as in Indonesia and the Philippines, services, for example. The third challenge relates they have gone fairly smoothly. Intergovernmental to the accountability of local governments and fiscal systems have been institutionalized. Workers the capacity of their management systems. Attenu- have been transferred from central ministries to ated accountability and weak management--of local governments without significant disruptions. both financial and human resources--could con- And local authorities have taken up their service- strain effective implementation of decentralized 1 2 East Asia Decentralizes FIGURE 1.1 Subnational Expenditures as a early view of the record provided. The chapter Share of Total Public Spending proceeds to examine the three main intergovern- 80 mental challenges outlined above, looking at the 69 organizational structures, financial mechanisms, 70 and subnational government management and 60 accountability systems that have emerged in East 50 48 Asian countries engaged in the decentralization 40 process. This discussion touches on themes that percent 32 will be taken up again in later chapters. Finally, the 30 26 chapter distills some key messages for East Asian 20 17 policy makers grappling with decentralization 10 10 issues. 0 The ensuing chapters are organized as follows. Chapter 2 delves into greater detail on the frame- China works, structures, and processes of decentralization Cambodia Indonesia Thailand Vietnam Philippines across East Asian countries. Chapters 3 and 4 dis- Source: Refer to chapter 5, table 5.1. cuss the macro-fiscal issues of maintaining fiscal sustainability and ensuring interregional equity. Chapters 5 through 7 explore the nuts and bolts of functions by subnational entities. In sum, the decentralizing public management systems, includ- potential benefits of harnessing the power of sub- ing public expenditures, revenues, and human national government in East Asia are substantial. resources. Chapters 8 through 10 examine the But national leadership across the region will be nexus between decentralization and service deliv- critical in responding consistently and systemati- ery in key sectors: health, education, and infra- cally to these three broad challenges if decentraliza- structure. Chapters 11 and 12 cover governance tion is to live up to its promise. and citizen empowerment. This Study Why Does Decentralization Matter Clearly, governments in East Asia have an impor- for East Asia? tant stake in getting decentralization right. They Intergovernmental reforms adopted in the process will need to develop strategic approaches that fit of decentralization are important because they country conditions but that also benefit from inevitably affect broader country performance regional and global lessons of decentralization. in three critical, interrelated areas: the economy, This study seeks to provide guidance to country service delivery, and governance. reformers by reviewing the East Asian experience so far, and by furnishing an analytic perspective on the design and implementation challenges that face The Economy policy makers in the region. This chapter sets the stage for more detailed dis- East Asia's remarkable achievements in economic cussions in subsequent chapters. It provides an growth and poverty reduction over the past 30 overview of East Asian decentralization. First, it years can be attributed largely to significant public shows why decentralization is an important issue investment in human capital formation and infra- in the East Asian context by highlighting its structure, and to the establishment of a regulatory broad implications for economic, governance, and environment conducive to private enterprise.1 With service delivery outcomes. The discussion then decentralization, subnational governments are now examines the origins of decentralization in East at the heart of a range of investment, fiscal, and reg- Asia, pointing to structural and political factors ulatory activities that affect both the pace and qual- that have driven the process. The approaches taken ity of economic growth. For example, they are now by East Asian governments are discussed, and an responsible for planning and financing economic East Asia Decentralizes 3 infrastructure, such as local roads and irrigation the richest province is 16 times that of the poorest schemes, and for regulating and taxing businesses. region. In Indonesia, the Philippines, and Vietnam, In some East Asian countries, such as Cambodia, own-source revenues are even more skewed. Such the role of local and regional authorities in these disparities can be mitigated by fairly designed sys- areas is still limited. But in most, including China tems of intergovernmental transfers. In the absence and Indonesia, it has become crucial. of such mechanisms, decentralization may rein- There are fiscal dangers associated with decen- force inequitable distribution of services across tralization. To the extent that newly empowered jurisdictions, potentially posing serious concern for subnational governments can transfer significant both subnational and national political leaders in financial liabilities to the center, the expansion of the region. subnational activities could generate macroeco- nomic risks for national governments. With the Governance possible exception of China, this has yet to emerge as a problem in East Asia. But local off-budget fiscal While the successes that came to be known as the activity, subnational borrowing, and inadequate Asian Miracle were partly attributed to well- local control of contingent liabilities are common performing public sectors, serious deficiencies in throughout the region. These weaknesses, com- public sector governance were evident in many bined with deepening autonomy and expanding countries by the time the 1997 financial crisis hit access to domestic and international capital mar- East Asia (World Bank 2000). Low levels of gov- kets, could aggravate longer-term risk unless insti- ernment accountability, transparency, and probity tutional, regulatory, and monitoring systems contributed to the calamity. Despite postcrisis improve. efforts to reform these governance weaknesses, problems persist. In particular, institutions of public financial accountability often diverge Service Delivery strongly from internationally accepted standards, While many East Asian countries have strength- especially with respect to procurement, account- ened the delivery of public services in recent years ing and auditing, and performance management (chapters 8, 9, and 10), profound problems and information systems in government (Mount- remain. It is precisely in the areas where these dif- field 2001). ficulties are concentrated--such as primary Perception surveys suggesting that corruption health, education, and potable water supply-- runs deep and wide across East Asia are particu- that subnational governments in East Asia have larly vexing. Transparency International's 2003 assumed most of their responsibilities. As dis- Corruption Perceptions Index ranks the six coun- cussed in later chapters, performance in these tries covered in this report in the bottom half of sectors has been mixed. In addition, subnational the 133 countries surveyed, ranging from China at governments have been required to assume regu- 66 to Vietnam at 100 (Transparency International latory roles in areas (such as environmental pro- 2003). The World Bank's Governance Research tection) for which they are poorly structured and Indicators for 2002 show East Asia's control of cor- have little technical capacity. The concern is that ruption in a steady decline since 1996 (Kaufmann if local and intergovernmental systems do not et al. 2003). Country-specific diagnostic surveys in function well, countries are unlikely to sustain Cambodia, Indonesia, and Thailand also point to positive service-delivery trends, or to reverse neg- the high costs of corruption for households and ative ones. enterprises alike.2 Recent Investment Climate The equity of service delivery is also a concern. Assessments in Cambodia and the Philippines pro- Wide divergences in natural resource endowments vide further evidence that graft--pervasive in the and economic concentration, combined with the justice sector and revenue administration--is a design of frameworks for own-source revenues at major obstacle to doing business.3 the subnational level in most countries, have pro- Decentralization magnifies these governance duced pronounced horizontal fiscal inequalities in challenges for subnational governments. Opportu- East Asia. In China, the per capita own-revenue of nities to strengthen grassroots accountability are 4 East Asia Decentralizes paired with risks that, absent tight central controls, up to 2015. The interplay between economic corrupt practices could proliferate (Bardhan and growth, urbanization, and intergovernmental Mookherjee 2000). Chapter 11 discusses these reform in East Asia is not clearly understood. But, issues at length, emphasizing the importance of as in other parts of the world, the basic economic intergovernmental reform design to ensuring good and demographic transformations in East Asia governance in decentralized contexts. seem to have created an environment conducive to decentralization, raising the stakes for intergovern- mental systems to function effectively (World Bank 2000). What's Driving Decentralization? Political factors. While these structural changes Decentralization is a process,not an event,and stipu- have generated pressures for decentralization, lating a starting point in any country is a difficult and political factors appear to have been the more prox- sometimes controversial exercise. But there is broad imate and powerful drivers of the process.4 Demo- consensus that,for most countries in EastAsia,inter- cratization appears sometimes to have been an im- governmental reform gained significant momentum portant trigger. In the Philippines and Indonesia, in the 1990s. The evolution of intergovernmental decentralization was adopted after the sudden reform has been distinctive for each country, but collapse of authoritarian regimes (Marcos in 1986 common dynamics can be identified. For the most and Soeharto in 1997, respectively) fueled demand part, long-run structural transformations--mainly for legitimate, local representation. Thailand's economic and demographic--have created an envi- prodemocracy movement in the 1990s paired ronment conducive to decentralization, while pow- protests against the inordinate role of the military erful political imperatives have precipitated and in politics with the development of an initial decen- shaped it. tralization framework. Explicit moves toward plu- Structural factors. In most East Asian countries, ralism have not figured in China and Vietnam. But decentralization has been preceded by significant in both cases, more audible demand for citizen par- and continuous periods of economic growth and ticipation and good government at the local level urbanization, broadly suggesting a relationship has driven decentralization. Certainly, the impedi- among these three phenomena (see table 1.1). ments that China's sheer physical size poses to As nations develop and urbanize, there is effective central control has reinforced these ten- growing pressure to provide services to rapidly dencies through the refinement of an intergovern- expanding and increasingly concentrated popula- mental framework that has long had quasi-federal tions (World Bank 2000). The challenges facing features (World Bank 2002a). national governments in effectively administering Political dynamics have also shaped the nature services create demands to empower subnational of decentralization arrangements. In Indonesia, the governments to meet the burgeoning needs. need to minimize the probability of provincial sep- Indeed, for developing and developed countries aratism in the fragile post-Soeharto era produced a alike, evidence shows that growth in per capita system that favored subprovincial governments. income is associated with a declining central Central political elites have also sought to prevent government share in public investment (World the emergence of regional provincial rivals by Bank 2000). pushing authority and resources down to less pow- Urbanization rates in the region are particularly erful city and district levels of government. Faced informative. With the exception of Thailand in with calls for grassroots "People's Power," post- recent years, urban growth rates over the last Marcos regimes in the Philippines created multilay- quarter-century have been high for the six coun- ered systems that diffused subnational power tries under consideration here, particularly during among different jurisdictions, for example. In con- earlier periods in China, Indonesia, the Philippines, trast, low levels of party competition in China and and Thailand. Although Cambodia and Vietnam Vietnam may explain the relative administrative began urbanizing later, their urbanization rates power of the provinces, as these don't act as plat- have accelerated and are projected to be substantial forms for rival political interests. TABLE 1.1 Economic and Demographic Trends in East Asia, 1975­2002 GDP per Urban capita, Compound annual population, PPP, rates of growth, % of total Population 2001 current/constant PPP population Urbanization rate 2002 (in 1995 (millions) US$) 1975­1983 1984­1994 1995­2002 1980 2000 1975­1980 1985­1990 1995­2000 World 6,199 6,837 7.6/1.3 4.1/1.3 3.3/1.8 39.2 47.1 2.72 2.72 2.22 Cambodia 12 1,747 -- -- 5.4/-- 12.4 16.9 2.25 3.65 6.25 China 1,281 3,796 11.2/5.1 10.9/7.9 7.8/6.1 25.7 35.8 3.39 4.09 2.84 Indonesia 212 2,768 11.1/4.5 7.3/4.4 1.9/0.2 22.1 42.0 4.90 5.00 4.74 Philippines 80 3,672 8.5/2.1 2.1/ 0.7 2.4/0.9 37.5 58.5 3.75 4.87 3.64 Thailand 62 5,932 10.8/4.3 9.7/6.8 2.1/0.0 26.8 31.1 4.71 2.36 1.61 Vietnam 81 1,956 -- 7.3/4.8 6.2/4.7 19.4 24.3 2.56 2.85 3.22 Sources: World Bank (SIMA); United Nations 2004, tables A.2 and A.6. Note: PPP purchasing power parity. 5 6 East Asia Decentralizes BOX 1.1 Variants of Decentralization: Pros and Cons Deconcentration refers to decentralization of cen- governments are responsible for delivering a set tral government ministries and arrangements of public services and for imposing fees and whereby subnational governments act as agents taxes to finance those services. Subnational of the center. Sometimes regional branches of governments have considerable flexibility in central offices and agent governments have selecting the mix and level of services they pro- some authority to make independent decisions, vide. Other levels of government may provide usually within central guidelines. Often, though, intergovernmental transfers. For devolution deconcentrated local government lacks author- to work, the central and local governments ity over the scope or quality of local services and must act as partners, with the former keeping its how they are provided. commitment to devolve functions, and local offi- Under delegation, subnational governments cials agreeing to make difficult choices and rather than branches of central government are develop the capacity to exercise their powers responsible for delivering certain services, sub- effectively.a ject to some supervision by the central govern- Intergovernmental systems usually have ment. Delegation may improve efficiency when some elements of each of these variants. In prin- subnational governments can better administer ciple, devolution should improve efficiency by programs of national interest--including certain giving citizens more influence over the mix and aspects of education, water, and health--in level of services, and by giving local govern- ways that better reflect local circumstances. The ments greater incentives to mobilize resources. center, or sometimes intermediate government, The associated efficiency gains--combined with determines what should be spent, and may also the ability to mobilize untapped revenues at the set minimum service standards, while subna- subnational level (from the formal and informal tional governments define the details. The sectors)--may prove particularly significant in design of intergovernmental fiscal transfers, and countries with diverse economies, cultures, the degree and nature of central monitoring, geography, or tastes for services. Indeed, assign- influence the balance between central and local ing some local control over expenditure and tax decision making under delegation. decisions can be key to nation building. Devolution is the most complete form of decentralization. Independent or semi- a. Bahl 1999. independent and, typically, elected subnational How Are Governments and through a "Big Bang" decentralization in the Approaching Decentralization? aftermath of Soeharto's fall and the 1997 financial In response to these structural and political drivers, crisis in Indonesia. These fast starters introduced East Asian governments have taken different the basic elements of a decentralization framework, approaches to decentralization, combining standard subnational democratic elections, and substantial elements of delegation, deconcentration, and devo- resource sharing swiftly. Considerable follow-up lution found in many intergovernmental reforms policy and legislative work to create a fully coherent around the world (see box 1.1). Country approaches and functional system remains. can be divided into three broad categories: fast The incrementalists (China and Vietnam) have starters, incrementalists, and cautious movers. taken a more piecemeal approach to decentraliza- The fast starters (the Philippines and Indonesia) tion. In China, the government made ad hoc policy have rapidly introduced major structural, institu- decisions and passed legislation directly affecting tional, and fiscal reforms in response to a sudden subnational governments (some decentralizing, and far-reaching political stimulus. Sweeping some recentralizing), but decentralization is not an decentralization reforms were introduced in the officially documented policy. Instead, it is anchored late 1980s after the fall of Marcos in the Philippines, in historical realities and broader political and East Asia Decentralizes 7 economic reforms. In Vietnam, decentralization China lacks even the incomplete policy framework policy has been more formal than in China, involv- of the Indonesian case but has gradually developed ing a sequence of specific legislative measures. But one of the most decentralized intergovernmental with the exception of a few bolder, asymmetric structures in the region (World Bank 2002a). In experiments in some of the major cities, decentral- contrast, countries with well-developed, more com- ization in Vietnam has been limited and incremen- prehensive policy frameworks have yet to imple- tal. In both cases, decentralization has focused on ment decentralization reforms. Thailand's decen- administrative and fiscal reform, with modest tralization framework is elaborated in great detail political change and the retention of considerable in the Constitution, in law, and through parliamen- central control over subnational governments (in tary and cabinet decisions, but the country has been law and policy if not always in practice). slow in carrying out reforms.5 Cautious movers (Cambodia and Thailand) have DecentralizationinEastAsiaisacomplexblendof established significant elements of decentralization characteristics in other respects as well. Decentral- at the formal policy and legislative levels, but there ized subnational governments have, for the most has been limited progress with implementation. In part, been assigned substantial functional mandates Thailand,the ambitious decentralization framework and responsibility over large fractions of total public developed in the wake of the 1997 Constitution has spending. And even where limited recentralization been only partly implemented. In Cambodia, elected has taken place (such as in China in 1994), this was a commune councils have limited functions and temporary move to improve system performance.So receive only modest resources. The provincial sys- the trend has been mainly toward greater local tem is stronger but heavily managed by national line autonomy--albeit at varying rates. The difficulty, ministries and centrally appointed governors. Only however,isthatthistrendcoexistswithweakintergov- recently has the government undertaken further ernmentalinstitutionalenvironmentsandarecordof work on devolution policy. In both countries, the generally poor (while in some instances improving) commitment to extensive decentralization appears performance of core mandates by most subna- to be limited. tional governments. In some countries (Indonesia, the Philippines) centralized, top-down systems for economic investment and service delivery have Cross-Country Trends weakened. Oversight and regulation of the emerging The broad categories outlined above describe a intergovernmental arrangements are undeveloped. cluster of characteristics related to the pace and And local systems--lacking appropriate bottom-up timing with which different countries have intro- incentives and human resource capacity--are not in duced decentralization. Other features of decen- place to take up the slack. tralization in the region defy straightforward classi- The result is a kind of "institutional limbo"which fication. There is an ongoing unevenness in the way typifies decentralizing environments internation- in which decentralization is being planned and ally. In East Asia, it has two main features. First, implemented across countries. whether by design or as a result of slippages in the For example, the degree to which countries uti- implementation process, intergovernmental struc- lize consistent and intentional policy frameworks to tures have substantial internal inconsistency. The guide reform varies considerably and appears unre- functions of different levels of government overlap. lated to the pace or appetite for reform. Indonesia Bottom-up accountability of locally elected bodies is decentralized faster than any other country in the dampened by top-down methods for appointing key region, but the Big Bang took place in the absence officials.And the discretion given to local authorities of a comprehensive policy framework. While recent in spending unconditional fiscal transfers is effec- revisions of the key Indonesian framework laws tively curtailed by central government control over (Laws 22 and 25 of 1999, now Laws 32 and 33 of human resources. The aforementioned substantial 2004) have clarified outstanding issues in the gov- capacity constraints--both local and central-- ernmental transfer system, many other policy compound these policy and design deficiencies. elements--such as the regulatory frameworks for Second, the decentralization process itself has subnational borrowing--have yet to be defined. slowed, even among countries that had been on the 8 East Asia Decentralizes more accelerated, "fast starter" path described leadership and direction of decentralization are above. For example, repeated attempts to revise widespread: aspects of the Local Government Code in the Philippines have borne little fruit, and there are few · In Indonesia, the recently completed modifica- signs of significant advances in implementing other tion of Laws 22 and 25 (now Laws 32 and 33 of key policy initiatives (e.g., the reform of the Internal 2004) was complicated, and hence delayed, by Revenue Allotment grant system). In China, recent policy disagreements between the ministries of intergovernmental reform has become more mod- Finance and Home Affairs. In fact, these laws est, restricting its focus to introducing equalizing now provide for simultaneous regulatory juris- adjustments to the existing central-provincial fiscal diction of both ministries over a number of transfer system. Reforms that would address more subnational fiscal and financial matters. fundamental problems, such as those of subna- · In Cambodia, responsibility for policy develop- tional debt, do not seem to be on the immediate ment and implementation of decentralization is horizon. In Indonesia, aspects of Law 32 of 2004 divided across a number of key interministerial represent a tightening of central control over local agencies, and coordinated progress has proven budgeting and own-source revenue decisions, alter- difficult. Only recently has the central govern- ing the more deliberate move toward subnational ment attempted to institute a single process of autonomy present in the initial phases of the Big policy development on issues related to decon- Bang reforms. centration and decentralization, backed by a Of course, policy and design inconsistencies, program of joint donor support. along with stop-and-start progress, characterize · In the Philippines, it appears that opposition institutional reform processes in developing coun- from major government banks has played a role tries around the world. Progress is rarely monoto- in ensuring that the legal and regulatory envi- nic, and policy reconsideration and reformulation ronment for local borrowing continues to place can even be sensible and salutary. To the extent private investors at a competitive disadvantage that such fits and starts reflect problematic aspects (World Bank and Asian Development Bank of the decentralization process in East Asia, how- 2003; Pellegrini and Ma 2002). ever, they are mainly explained by political and · In contrast to other regions, finance ministries bureaucratic conflicts, which break down into in East Asia are less active in managing decen- three broad types. tralization than are ministries of Interior, Home First, conflicts between national agencies and Affairs, or Planning, which have strong incen- subnational governments--usually over the pace of tives to limit local autonomy. These central min- decentralization and resource questions (such as istries often prevail, as those who would cham- revenue allocations)--have been gaining impor- pion decentralization tend to be less powerful. tance. Some conflicts are explicit and formal, such as public disputes between central and local govern- Third, in some countries, conflicts over the ment officials over jurisdictional issues. Some are scope of decentralization have become an issue informal and implicit, such as disputes between between elected political leaders and administra- local and central authorities over local tax bases and tive bureaucracies. For example, the Indonesian the regulatory powers of subnational governments. Parliament developed its own bill to reform Law 22, The latter are particularly prevalent in countries as an alternative to the legislative drafting efforts of where the law remains unclear. In Indonesia, local the ministries of Finance and Home Affairs. Similar authorities have increasingly begun to assert them- dynamics appear to have slowed attempts to reform selves in competition with central and provincial aspects of the intergovernmental transfer system in government in local business regulation. In China, the Philippines. local governments have disputed centrally imposed While these conflicts may be no more pro- fiscal constraints and raised local revenues outside nounced in East Asia than in other decentralizing the regular budget in response. regions, it does appear that they have seriously Second, in most East Asian countries, con- impeded progress on key reforms in some coun- flicts among central agencies themselves over the tries (Cambodia, Indonesia, and the Philippines). East Asia Decentralizes 9 Even where agreement has been reached among 2001, while total health spending remained in line central agencies on intergovernmental frameworks, with other East Asian countries. In education, the the arrangements that have been struck tend to fix positive trend is more pronounced. In China, edu- in place confusion over functional mandates that cation spending nearly tripled as a share of total perpetuate interagency squabbles which, in turn, expenditure from 1978 to 2001 (from 6.7 to 18 per- slow implementation. Moreover, in a number of cent). In the Philippines, total government spend- countries, the absence of clear intergovernmental ing on education grew from just over 2.5 percent to reform policies, combined with weak management well over 4 percent of GDP between 1991 and 1998. capacity at the center, has generated incentives and In Indonesia, the share of spending on education opportunities for local authorities to work outside has risen to 4 percent of GNP from only 1.4 percent the formal system. Indeed, almost half of Chinese before decentralization. The subnational share subnational government revenue is, by some of education resources has also increased, partly estimates, not captured in formal local budgets because funds and functions have been transferred (Ahmad et al. 2000). With decentralization, local to the local level, and partly because local govern- payrolls for temporary workers have expanded ments are expected to mobilize additional resources in the Indonesia, the Philippines, Thailand, and to perform these functions. Vietnam. Indonesian businesses are increasingly The record on infrastructure is mixed. Local subject to regulation and taxation imposed by local infrastructure investment in China has recently ordinances that have not been harmonized with been sustained at high levels, but an initial postde- existing central government legislation. These prac- centralization spending surge in the Philippines has tices are largely the unintended result of inade- given way to a slow but steady decrease in spending quately designed reform programs, and illustrate since 1995. There are some indications that hori- the difficulties countries face in trying to system- zontal disparities in spending in some sectors have atize decentralization policies. widened during decentralization (education in China). But this has not been widely found and Early Impacts is not entirely consistent with the evidence on service-delivery outcomes. Despite initial anxiety in many countries, the effects of decentralization in East Asia appear to have been largely benign so far. In particular, there seems not Outputs and Outcomes to have been a systematic deterioration in the deliv- The positive trends in health outcomes in East Asia ery of key services, a major preoccupation in many that began in the 1980s have been broadly sus- countries. Indeed, while subnational governments tained through the period of decentralization have had difficulties with service delivery, so have reform. During the period of intensive intergovern- some central governments. These are still early mental reform in Indonesia, the Philippines, and impressions. Comprehensive assessment of the Vietnam, infant mortality rates (IMRs) fell and life effects of decentralization is yet to be done. But ini- expectancy grew steadily, albeit slowly. In some tial trends can be detected in three broad areas: countries, gaps in health status between rich and resources (particularly fiscal resources), outputs poor local governments have narrowed markedly. and outcomes, and propoor service delivery.6 In the Philippines, the difference in IMRs between the richest and poorest regions was further reduced Spending Envelopes in the 1990s, and by 2000 the IMR of the poorest Overall, decentralization seems to have had a posi- area was lower (Eastern Visayas, at 10.7 per 1,000 tive effect on aggregate fiscal spending. In health, live births) than that of the richest (Manila, 19.4 average annual expenditure in Indonesia and per 1,000) (chapter 8). This favorable trend was Vietnam remained constant as a proportion of due, in part, to improved health outputs and serv- gross domestic product (GDP) during the most ice coverage that extended its subnational reach: intense decentralization reforms, from 1997 to the proportion of births attended by trained health 2001. In the Philippines, health spending at the personnel, and the share of the population with local level exceeded that at the national level by access to water or sanitation, rose discernibly over 10 East Asia Decentralizes the postreform period. Similar trends were evident areas touched upon earlier in this chapter. These in education during periods of intergovernmental include the organizational arrangements that con- reform. As China decentralized, literacy rates rose stitute a country's basic intergovernmental struc- from 1982 to 1999, most dramatically in the poorer ture; the financial resources that subnational regions.(Further examples of service-delivery trends governments can mobilize and the way they are can be found in chapters 8, 9, and 10.) distributed; and the management and accountabil- ity systems that govern resource allocation and control at the subnational level. The following dis- Service Delivery to the Poor cussion highlights the key characteristics of these The impact of decentralization on the poor is not challenges. It draws on international experience to well understood. National decisions related to stimulate thinking about pitfalls to avert and inno- decentralization (fiscal equalization, nationally vative approaches that might be adapted to each mandated service levels) affect the poor. So do local country's circumstances. processes (budgetary allocations, planning deci- sions, and service mix and quality). In 2001, prede- Organizing the Intergovernmental System centralization survey data on perceptions among Indonesian households of service delivery sug- The first broad challenge faced by decentralizing gested that the poor knew less about programs and East Asian governments is how to design the policies affecting them than did other groups. A organizational structures of the intergovernmental postdecentralization follow-on survey is under way system. This section examines how functions and to understand these issues more deeply. Better expenditure responsibilities are assigned, explores monitoring and evaluation--and more data--are issues related to coordination among intergovern- needed to get a fuller picture of service delivery mental structures and subnational entities, and performance in poor communities throughout the examines questions of special configuration and region. subnational boundaries. Functional and Expenditure Assignments. Sub- stantial functions have been assigned to subna- Intergovernmental Systems: tional governments in East Asia (see table 1.2). But What Challenges Do East Asian in most countries, specific responsibilities are Countries Face? unclear, and overlaps among the different levels of As countries in the region continue to decentralize, government are common. In Indonesia, despite they will need to design and implement reforms of recent legal revisions (Law 32 of 2004), a lack of their intergovernmental systems in the three broad clarity regarding central and subnational functions TABLE 1.2 Expenditure Shares and Functional Allocations in East Asia Subnational Functional allocationsb expenditure Country (% of total)a Education Health Social welfare Cambodia 17 Provincial Provincial Provincial China 69 Local Local Local Indonesia 32 Local Local Local Philippines 26 Central, Central, Central, provincial, local provincial, local provincial, local Thailand 10 Central, provincial Central, provincial Central, provincial Vietnam 48 Provincial, local Provincial, local Provincial, local Sources: Various country case studies prepared for this report and World Bank statistics. a. 2001, except for China, Indonesia, and Vietnam (2002). b. "Provincial" refers to the level immediately below the central or national level; "local" refers to all levels below the provincial level. Classifications vary across countries; municipalities, for example, can be at either the intermediate level (as in Cambodia and Vietnam) or the local level (as in the Philippines and Thailand). East Asia Decentralizes 11 TABLE 1.3 Vertical Organization of Intergovernmental Systems in East Asia Average population of Levels of Number of first-tier first-tier subnational subnational subnational governments Country governmenta governments (millions, 2002) Cambodia 2 24 0.5 China 4 32 40.0 Indonesia 3 32 7.0 Philippines 4 149 0.5 Thailand 3 76 0.8 Vietnam 3 61 1.3 Sources: Various country case studies prepared for this report and World Bank statistics. a. "Level" refers to an organ of government with some degree of formal budget (expenditure) authority. In some cases (such as Indonesia), this can be highly circumscribed, particularly at the lowest levels. remains, as do a number of inconsistencies between coordination, both among subnational govern- Law 32 and various sectoral laws (such as Law 31 of ments and between subnational governments and 2004, which assigns all fisheries functions to the the center. central government). In the Philippines, some All East Asian countries have multiple levels of clauses in the 1991 Local Government Code appear subnational administration or government, but the to distribute functions unambiguously across levels structures vary widely (table 1.3 and chapter 2). of government, but this clarity is diluted by other These variations make simple cross-country com- clauses allowing both national government agen- parisons difficult, but a rough pattern of vertical cies and local government units to initiate devolved organization is evident. Intergovernmental struc- activities. In China, there is no national law that tures in Indonesia and the Philippines are oriented clarifies the functions of each level of local govern- toward the subprovincial level, though interprovin- ment; expenditure assignments are decided by the cial variations in size, wealth, and influence are provinces and thus differ across regions (World wide. In both, however, countries now have limited Bank 2002a). powers and functions relative to subprovincial Achieving greater de jure clarity on functional governments--perhaps to the point that being mandates will not entirely eliminate vertical intera- precluded from oversight, coordination, and regula- gency conflict, and even mature intergovernmental tion of city and district governments has become systems are characterized by some dynamic tension dysfunctional for the intergovernmental system over jurisdictional issues. But East Asia may be able as a whole. Subnational structures in China and to learn from countries that have faced these ques- Vietnam focus more on the intermediate level. tions in other parts of the world. As with many These are "nested hierarchies," in which central gov- other issues, solving these problems is usually ernment determines the overall character of the sys- about overcoming political obstacles more than tem and deals directly only with provincial adminis- technical challenges, and an ongoing dialogue trations; provinces oversee subprovincial levels; and involving various stakeholders will be necessary to so on (World Bank 2002). This has allowed the resolve these challenges in most countries. development of significant variation in fiscal and Organization of Intergovernmental Structures institutional arrangements below the provinces. and Coordination among Subnational Entities. The Cambodia and Thailand are at the other end of character of decentralization varies considerably the spectrum, with the most centralized structures across East Asian countries, ranging from decon- in the region. In both, there are elected subprovin- centration to devolution, and encompassing dif- cial governments, but they have not been assigned ferent levels of subnational government. The inter- significant resources, authority, or functional man- governmental structures in turn have varying dates. In Cambodia, provinces operate fully under implications for the scope of intergovernmental the control of the center. In Thailand, provinces 12 East Asia Decentralizes have some independent functions in principle, but aim primarily to improve tax compliance and their realization in practice has been limited. enforcement, including the exchange of personal When planning and budgetary authority is dis- and corporate information between federal and persed laterally and devolved downward, there is a state levels and across states. This exchange may be need for horizontal coordination across subna- accomplished either through a voluntary uniform tional entities and vertical coordination among exchange of information or through separate bilat- levels of government. As the responsibilities of local eral exchange agreements. At present, 45 states use authorities expand, the effects of their activities can the uniform exchange of information agreement as spill over into neighboring jurisdictions, intensify- a framework for exchanging this kind of taxpayer ing the need for horizontal cooperation across information--and for cooperative activities to boundaries. This area has not yet been well studied, combat tax avoidance (Duncan and McLure 2005; but initial indications are that horizontal coopera- Ebel and Taliercio 2004).7 tion has generally been weak in the region. There In Estonia, localities have the authority to levy are some examples of progress in some sectors, local gross receipts taxes and then verify taxpayer such as the Joint Health Councils in Yogyakarta, reports by checking reported gross receipts against Indonesia (chapter 8), but these tend to be excep- the central government's VAT declarations. To en- tions. Some donor projects are attempting to intro- force the tax payment, localities may choose to duce incentives to stimulate such cooperation--in deny operating licenses to businesses that have not China, for example--but systematic initiatives are paid the tax and/or they may choose to contract not common in the region. with the National Tax Board for collection of the Vertical coordination between central and sub- tax (Sootla et al. 2000). Mexico presents a special national governments is also critical for major case of cooperative administration whereby the planning and investment activities. In Indonesia, central government may enter into agreement with provinces are reportedly encountering difficulties states to audit and otherwise verify compliance in facilitating cooperation across local govern- with federal laws in exchange for a portion of the ments, leading to underperformance in tasks with federal revenues. States thus reduce costs of admin- large spillover effects and significant economies of istration by using their knowledge of the local scale, such as ocean management. An ongoing economic activity, which the central government tug-of-war between the center and the regions on might not possess (Mikesell 2003). investment approval and land has cooled local Spatial configuration and boundaries of subna- investment (World Bank 2003b). These difficulties tional entities. The wide variation in the structure are related in part to the underlying lack of clarity of subnational governments across East Asia is in the division of functions and powers among the consistent with the political and geographic diver- different levels of government--and in part to the sity in the region. As decentralization has pro- changing scope and diminishing authority of gressed, two important concerns have emerged for national planning agencies under decentralization. the spatial organization of local authorities. First, To maintain efficient planning, budgeting, and there is the question of the administrative effi- regulation of infrastructure investment, govern- ciency and capacities of subnational units. In ments need to make conscious efforts to develop Cambodia, studies have raised issues about the abil- models of vertical and horizontal cooperation that ity of communes to operate effectively given their can function in a more complex institutional very small sizes (1,600 communes serve 12 million landscape. people) (Rusten et al. 2004). In Indonesia, where There are many international examples of suc- around 400 kabupaten/kota8 governments serve cessful intergovernmental cooperation. A number well over 200 million people, the limited political of these lie in the area of tax administration. In the connectivity between decision makers and con- United States, the sovereignty accorded the states stituents in large districts is often considered by the constitution has created a highly independ- problematic. ent system of tax administration and generated the Second, there is more fragmentation of subna- need for cooperative arrangements between state tional jurisdictions. In Indonesia, the number of and federal tax authorities. These arrangements kabupaten and kota increased from 292 to 434 East Asia Decentralizes 13 between 1998 and 2003; in Vietnam, the number of system of more than 800 local authorities into 284 provinces expanded from 61 to 64 during 2003. municipalities and districts by introducing legisla- This fragmentation raises questions about whether tion and establishing an independent, nonpolitical the process for creating new subnational units cre- body to scrutinize local government boundaries on ates incentives for increasingly inefficient local gov- the basis of detailed, technical criteria. While the ernment structures. Recent research on Indonesia, final boundary determinations were not without for example, indicates that "perverse fiscal and controversy, the overall streamlining of the system bureaucratic rent-seeking initiatives also may be at is widely regarded as having laid the platform for a work in the creation of new regions" (Filtrani et al. major improvement in local government efficiency 2004, p. 3). Proliferation often introduces substan- and operations (Glasser and White 2004). tial instability in local institutional frameworks, discouraging economic investment and possibly Accessing and Distributing Financial Resources undermining the institutional consolidation needed for effective government administration. The second common challenge in East Asia's inter- East Asia has much in common with other governmental reform relates to access by subna- regions here. Decentralization in a number of tional governments to financial resources. This African, Central European, and Latin American discussion covers fiscal resources such as own- countries has been characterized by a proliferation source revenues and intergovernmental transfers, of local governments, sometimes to the point as well as local government borrowing activities where the average unit has become inefficiently and the extent to which they are subject to hard small (Bird et al. 1995). In some of these instances, budget constraints. fragmentation has been followed by a later phase of Own-source revenues. East Asian decentraliza- consolidation, in others not. These dynamics are to tion has been characterized by a relative lack of some extent intrinsic to decentralization. But inter- own-source revenue autonomy, in both the range national experience provides examples of transpar- of local revenue sources assigned to subnational ent processes that force a more objective examina- governments and their authority to determine the tion of technical factors--such as the impact of tax base and rate. As in other areas, comprehensive boundary changes on the fiscal position and data on fiscal arrangements are not available for administrative capacities of local authorities--and most countries in the region, but table 1.4 presents that appear to lead to more rational outcomes. an informed impressionistic overview of key fea- South Africa is worth examining in this regard. In tures of the intergovernmental fiscal structure, 2000 it rationalized a wasteful and dysfunctional including own-source revenues (see chapter 6). TABLE 1.4 Subnational Fiscal Structure of Selected East Asian Countries Own-source Unconditional Conditional Informal Country revenues Shared taxes transfers transfers revenues Cambodia Low Low Higha n.a.b High China Moderate High Low High High Indonesia Low Moderate High Low Moderate Philippines Moderate Moderate High Low Moderate Thailand Low High Moderate Moderate Low Vietnam None High Low High Moderate Sources: Subsequent chapters and unpublished country case studies prepared by World Bank staff for this report. Note: "Low," "moderate," and "high" refer to the rough proportions of total subnational revenues attributable to each revenue source relative to international experience. a. Refers only to the commune level. b. Most "provincial" agencies are deconcentrated arms of central ministries, so the term "transfer" does not apply. 14 East Asia Decentralizes A brief survey of typical own-source revenues in is significant scope--and an imperative--for most East Asia demonstrates the overall lack of subna- East Asian countries to empower and encourage tional autonomy. Among the six countries consid- subnational governments to generate revenues. ered in this study, none permits personal income Intergovernmental transfers. Subnational gov- taxes or general sales taxes at the subnational level. ernments are invariably in a better position to The real property tax on land or on land and build- spend resources than to raise them, so intergovern- ing improvements is effectively a national tax in mental transfers form a substantial part of the pub- Indonesia, Vietnam, and Thailand. Real estate lic sector fiscal structure in most countries. In (property) taxes are a potentially robust source of OECD countries, subnational governments depend local revenue, but they are either permitted only on transfers to finance about 40 percent of their on a very narrow base, as with unused land in expenditures, on average, with the numbers for Cambodia, or subject to maximum rates set by the Eastern Europe and Central Asia, and for Latin center, as in China and the Philippines. Business America and the Caribbean, slightly lower, and receipts taxes are also underused, except in the those for Africa closer to 60 percent (Ebel and Philippines and China, where they constitute about Yilmaz 2002).9 a third and a fifth of local revenues, respectively. East Asia follows this pattern, with transfers Some decentralization laws allow the imposition of accounting for significant but widely varying modest excises and fees, including those on motor proportions of total local revenue, ranging from fuel in Indonesia, vehicle and vessel use in China, nearly 100 percent in Cambodia to about 34 percent markets in Cambodia and Indonesia, and public in Thailand. Local fiscal dependence on transfers is transportation in Thailand. about 70 to 80 percent in Indonesia and the Philip- One result of this limited authority over subna- pines and 50 percent in Vietnam. In addition, in tional tax policy is increasing informality, through countries such as China, Thailand, and Vietnam, which subnational governments seek "unofficial" subnational governments obtain much of their sources of revenue. Allowing subnational govern- income from shared taxes, a form of intergovern- ments to enact new taxes and fees has generated mental transfer. These vary widely across the region, some productive entrepreneurial behavior in sev- and have important additional implications for the eral countries, but it has simultaneously created dependence of local governments on non­own- problems. Chinese subnational governments have revenue sources of funding (see chapter 2). developed an abundance of "illegal" extrabudgetary Many East Asian countries have recently been fees, with distortionary effects. In Indonesia, a moving toward internationally accepted "best prac- proliferation of "nuisance" taxes yields limited tice" norms by simplifying complex intergovern- revenues, imposes high administrative costs on mental transfer systems, improving the trans- subnational governments and compliance costs on parency and predictability of allocations, and taxpayers, and, in some cases, impedes inter- increasing subnational government discretion over regional trade. Subnational governments in the the use of these resources (Schroeder and Smoke Philippines have created a complex variety of taxes, 2003). This is particularly true in Cambodia, fees, and charges, many of which are uncollected or Indonesia, and the Philippines, where transfers undercollected. are formula-driven and largely not earmarked. Weak local revenue authority creates depen- Thailand and Vietnam have also begun to move in dency on higher levels of government, restricts this direction, though central control over the use subnational autonomy, and undermines the link of transferred resources remains tighter than in between services and finance needed for strong other countries. China's transfer system is less local accountability--a key factor in successful transparent, having developed piecemeal over the decentralization. Combined with the incentives for years; recent efforts have focused on improving poor collection of allowable taxes and fees and the horizontal equalization. proliferation of "illegal" taxes, this contributes to a These trends are positive, but the increase in culture of noncompliance and undermines the local discretion over how transfers are spent often integrity of the fiscal system. Although political masks important constraints in local decision dynamics pose serious challenges in this area, there making. There is a degree of internal inconsistency East Asia Decentralizes 15 in the design of intergovernmental transfer well as the public sector­dominated character of systems. While transfer policies seem to favor local the banking industry and the regulatory advantages expenditure discretion, human resource policy typically conferred on state-owned banks. As a tightly restricts it. For example, the Internal result, lending to subnational authorities, and the Revenue Allotment in the Philippines and the dana associated credit risk, tends to concentrate in the alokasi umum in Indonesia are not earmarked public sector even in countries such as the Philip- transfers. But because they are used primarily to pines where greater private participation in the cover local wage costs, and because the central gov- subnational debt market is an explicit policy goal.10 ernment retains significant control over local civil Third, there are questions about the extent to service staffing and employment conditions, local which subnational governments are subject to hard discretion over the use of these grants is limited. budget constraints (see box 1.2). China is the only The actual amount of intergovernmental transfer East Asian country in which soft local budget con- funds that can be spent on truly local priorities in straints may pose a major problem at the macro- the region is much smaller than first appears. economic level, largely because of the significant Subnational borrowing and hard budget con- off-budget expenditures of local governments and straints. Subnational borrowing varies widely across their widespread--albeit indirect--borrowing. East Asia (see chapter 3). Such borrowing is virtually Cambodia is at the other extreme, as communes are nonexistent in Cambodia, highly restricted in responsible for only about 2 percent of total public Indonesia, moderate in the Philippines and Vietnam, spending and cannot borrow. While Cambodian and extensive but indirect in China. The frameworks provinces account for more spending (17 percent), that regulate local borrowing and the actual practices 80 percent of that is under the direct control of the of subnational governments are complex. In China, central government. Provincial off-budget activities for example, subnational borrowing is illegal except are minor, and provinces may not borrow, leaving when financed by higher levels of government. How- little scope for accumulating contingent liabilities. ever, local authorities effectively raise credit (and The other countries lie between these two extremes, hence incur liabilities) through special-purpose although data weaknesses, the complexity of regu- vehicles created for investment and financing. latory frameworks, and the opacity of local Therefore, while local borrowing is legally more con- accounting systems make the situation unclear. In strained in China than in many other countries, it is Indonesia, where subnational governments are usu- in reality both widespread and unregulated. ally entitled to borrow, but where these powers are The emergence of sustainable subnational bor- currently restricted,11 regulatory weaknesses and a rowing in East Asia is hampered by a number of poor record of loan repayment indicate potential serious problems. First, while the economic and concern in the medium and long term. Whether demographic growth of cities is propelling signifi- local borrowing will become a problem will depend cant latent demand for financing for subnational on whether these restrictions are relaxed, as well as loans, policy and regulatory frameworks are poorly on reform of the subnational borrowing framework developed. Thus, while the larger cities and towns and the evolution of the intergovernmental transfer in East Asia are in need of increased financing flows system, among other factors. (primarily to fund infrastructure investment) and The degree to which subnational governments are expanding their capacity to service debt, the are subject to hard budget constraints depends on lack of reliable financial data, constrained borrow- several aspects of the intergovernmental system, ing authority, and the absence of rules in the event especially the regulatory environment surrounding of default interact to undermine the de facto credit- subnational borrowing. Ultimately, if East Asian worthiness of subnational governments and make countries wish to expand sustainable private invest- them unattractive investment propositions. ment at the subnational level without misalign- Second, private lending to subnational govern- ing incentives and risking undesirable macroeco- ments remains low throughout the region, even nomic impacts, they need to develop much more where central governments have attempted to thorough policies and regulations and devote stimulate it. This situation arises from the weak significant efforts to implementing them. This creditworthiness of subnational governments, as will require a fundamental reassessment of the 16 East Asia Decentralizes BOX 1.2 Hard Budget Constraints The ability of subnational entities to borrow debts by borrowing from their own banks, or funds is fundamental to the concept and prac- from a central financial authority that provides tice of fiscal decentralization and local fiscal additional funding, without accompanying--and autonomy. Subnational borrowing can be an severe--consequences for default. effective tool for local development, if limited to The problem with a subnational entity bor- financing capital investment expenditure. When rowing from itself is clear. A pattern of central structured well, it can improve both economic government bailouts may also be costly. Not only efficiency, in which the cost of debt repayment will this undermine the center's own credit stand- matches the flow of benefits over time, and ing, but it can also lead to selective allocation of intergenerational equity, so that future genera- credit across subnational governments according tions who benefit from investments also share to political, ethnic, or religious favoritism. Capital the responsibility for payment. investment can become inefficient, and national The key to making subnational borrowing cohesiveness may be jeopardized. work is the presence of hard budget constraints. The second strategy is to establish a fiscal These ensure that subnational governments can- framework that ensures subnational budget dis- not transfer the liabilities they accrue to higher cipline. Central or higher-level governments levels of government, either by shifting debt may limit local borrowing to capital expendi- service obligations upward or other means such tures only, as in Argentina, Latvia, and Poland. as expenditure deferrals or the accumulation of They may place statutory limitations on local contingent or implicit liabilities. The best way to debt, as in Italy, Portugal, and Canadian enforce a hard budget constraint is to establish provinces. And they may provide for a municipal procedures that clearly signal that local govern- bankruptcy law, as have Hungary, South Africa, ments will bear the costs (and accrue the bene- and many member countries of the Organisa- fits) of their fiscal decisions. tion for Economic Co-operation and Develop- When this message is conveyed to creditors, ment. Other rule-based approaches include pro- asset owners, and voters, the market mechanism hibitions on foreign borrowing in Ethiopia and comes into play. First, creditors will demand Ireland, and the establishment of intergovern- sound fiscal decisions by withholding credit. mental coordinating committees. For example, Second, knowing that local fiscal decisions can the Australian Loan Council provides informa- directly affect property values and rents, asset tion to the financial markets on local govern- owners will have a strong incentive to lobby ment fiscal positions, and the French Regional against imprudent borrowing and debt buildup. Chambers of Accounts may establish remedial Third, political factors may also be important in fiscal measures if a deficit appears. countries with local elections, where voters can A range of sanctions may also be established. oust underperforming local officials. In Austria, an explicit Domestic Stability Pact sets In well-established intergovernmental sys- overall fiscal targets for municipalities as a tems, the fiscal discipline of the capital market group, and allows the transfer of surplus/deficit suffices. Yet markets work only when good infor- obligations and rights across subnational gov- mation is available on local fiscal performance ernments if the group target is not met. In Brazil, and risk, when the central government makes a partly in reaction to sharp increases in subna- credible commitment that it will not provide tional debt levels (up to 17 percent of GDP in bailouts, and when accountability systems are 1996), the central government not only restricts robust. Where these conditions are not in place, debt service levels but also penalizes governors as in many developing countries, two reinforcing and mayors for irresponsible performance strategies can help. First, it is important to ensure through impeachment or imprisonment. that subnational entities do not "pay down" objectives of subnational borrowing, close study dissemination of better credit data. East Asia could of the local borrowing environment, a broad draw on the experiences of other countries that framework conducive to disciplined subnational have begun--and in some cases successfully credit activity, and the central capacity to monitor completed--such exercises, including Hungary, local credit activities, including collection and Mexico, Brazil, and South Africa. East Asia Decentralizes 17 Some East Asian governments are beginning to governments have often relied on nontransparent tackle at least some aspects of this agenda. For allowances to boost civil service pay. In China, example, the Department of Finance in the Philip- where subnational governments return budget sur- pines has recently begun developing mechanisms pluses to the central government, local civil service to measure and monitor subnational liabilities, and employment has swelled in the absence of incen- the department intends to create an early warning tives to local managers to keep staffs lean. system to identify impending local debt defaults. In The design of decentralization has also had the rest of the region, however, efforts appear to important impacts on the accountability, incen- address much narrower concerns. Indonesia has tives, and capacity of local civil servants. Unclear focused on developing a regulatory system for the policies have sometimes created confusion con- onlending of donor-sourced investment funds cerning accountability relationships. In Indonesia, within the public sector (KMK 35).Notwithstanding for example, conflicting laws have produced ineffi- the positive steps mentioned above, the Philippines cient overlaps in authority. In China, the Philip- has invested significant effort in restructuring the pines, and Vietnam,"double subordination" to cen- Municipal Development Fund Office into a nonbank tral authorities and local assemblies conflates the financial institution that--like two government accountability of local staff. banks (the Land Bank and the Development Though some East Asian countries have estab- Bank)--will cater to the subnational market. Gov- lished formal mechanisms for merit-based recruit- ernments have devoted much less attention to ment and performance management, these tend to broader policy and regulatory issues surrounding lack teeth in reality, and civil servants are rarely subnational borrowing. held accountable for their behavior. Seniority and party loyalty have frequently trumped performance in promotion decisions. In China, the awarding of Creating Sound Management across-the-board performance bonuses has under- and Accountability Systems cut their ability to leverage civil service quality. In Management and accountability systems often the Philippines, where staff performance has not, determine the performance of intergovernmental until recently, been rigorously evaluated, patronage mechanisms. Designing and implementing systems and nepotism have figured prominently in recruit- that work on the ground are not easy tasks. This is ment and promotion. In countries where wages are where decentralization often falters. The following low and absenteeism is endemic, service delivery section examines three key elements of this third quality cannot be maintained. Throughout East common challenge: human resource management; Asia, less well-resourced local governments are planning, financial management, and information often unable to attract enough staff to fill the at local and national levels; and local accountability required number of spots or to fund centrally set systems. wages. For local governments that manage to staff Human resources. East Asian countries have their civil services despite poor incentives, the lack started down the path of decentralizing human of strong systems for career management also resources (see chapter 7). In most cases, subna- dampens the incentives of local functionaries to tional governments have the authority to hire and perform. assign junior staff, but the central government International experience shows that countries retains control over the aggregate number and pay often neglect the details of administrative decen- levels of local staff.12 Under these arrangements, the tralization. In Central and Eastern Europe, confu- overall management of the civil service effectively sion about the role of the postcommunist state remains largely centralized. prompted many governments to decentralize tasks To compensate for this limited local formal without providing adequate resources (Verheijen autonomy, subnational practices in some countries 2002). In Latin America, local authority to control have tended to circumvent formal systems. In the personnel costs was critical in enhancing macro- Philippines and Thailand, local managers have economic management under decentralization avoided central guidelines by hiring contractual (Burki et al. 1999). Yet even when far-reaching workers. In Indonesia and the Philippines, local on paper, decentralization of human resource 18 East Asia Decentralizes management has usually been quite limited in do not inform the budget in a meaningful way. In practice. Governments tend to decentralize the civil China and Vietnam, insufficient coordination service in similar ways, giving local governments between planning and budgeting means that plans autonomy to hire and fire staff and supplement typically far outstrip resources. centrally set salaries, while the center retains con- The budgeting process is often inefficient and trol over wage rates and interregional mobility only loosely scheduled. In China, the budget (Evans and Manning 2004). Concerns about local process often starts only two months before the fis- capacity, interregional equity, and national unity cal year begins, creating delays throughout the sys- can spur recentralization of the civil service and tem as each level of government must wait for threaten the decentralization agenda. information from the previous tier before creating Despite the relative inattention to administrative its own expenditure and transfer plans. In Indone- decentralization, a few interesting experiences in sia and the Philippines, poor estimates and unpre- local civil service management provide lessons dictable release of transfers over the budget year for East Asian countries. In Uganda, independent distort budget execution. District Service Commissions manage the recruit- Difficulties with local financial management ment of local civil servants. The commissions have systems are also common. In China, Indonesia, the experienced some growing pains, such as limited Philippines, and Vietnam, programs to modernize ability to discipline and dismiss staff, but they can financial management systems are in their infancy, still serve as a useful model. In Pakistan, reforms particularly at the subnational level. Though envision organizing lower-echelon staff into cadres management information systems in the Philip- to facilitate their mobility across districts, while pines and Vietnam produce general reports, they provinces will officially employ more senior staff to cannot do so in a timely or accurate way. Year-end allow vertical movement into central government accounts are thus not available in Vietnam until service. In this case, the challenge has been to ensure 18 months after the end of the fiscal year. In the that district governments rather than provinces Philippines, local financial management is still hold civil servants accountable for performance. based mainly on manual systems. Personnel exchange programs in Japan enhance the Internal and external auditing of expenditures at career prospects of local civil servants by transfer- the subnational level are of extremely low quality, ring them between subnational governments and suffering from low capacity and confused man- the center. Broadly, international experience sug- dates. In Indonesia, the legal framework is unclear gests that some central controls on wages and civil with regard to the authority of various auditors in service mobility, combined with local autonomy examining local budgets. In the Philippines, inter- over hiring and promotion of more junior staff, nal auditing is nonexistent among most local gov- have been the preferred policies for most countries ernments, and the regional offices of the Commis- moving from centralized to decentralized human sion on Audit, which have the power to audit local resource management systems. These experiences governments, are overstretched. In Thailand, only a show how important human resource arrangements few large-scale local authorities have internal audi- can be to making intergovernmental systems work. tors, and two external auditing units--the Office Planning and financial mangement. As is com- of the Auditor General and provincial auditing mon in newly decentralizing environments, local teams--have considerable credibility but limited planning in East Asia is fairly weak (see chapter 5). capacity to provide services to all local authorities. Though the Local Government Code of the Philip- Decentralization implies a shift rather than a pines prescribes a participatory approach to plan- weakening of systems and capacity at the center. In ning, less than a third of subnational governments fact, as the intergovernmental structure becomes accounted for external inputs in creating develop- more complex, central ministries need to develop ment plans, according to one survey (Azfar et al. new systems to monitor and manage it effectively. 2000). Indonesia has also begun to introduce a par- Two challenges are typical of East Asian countries. ticipatory planning process at the kabupaten level, First, development of the systems connecting dif- but project affordability is not a factor in expendi- ferent layers of government often lags far behind ture choices, so this process creates "wish lists" that the policy decisions and institutional arrangements East Asia Decentralizes 19 they are meant to support. Cambodia is an extreme trol of expenditures to performance-based account- case: grant disbursements to provinces are so ability, but these are in the early stages. consistently late and so divergent from initially Low capacity within oversight bodies challenges budgeted amounts that budgets themselves have horizontal accountability at the subnational level. become virtually meaningless. But similar difficul- Local legislatures and judicial institutions often ties with the timely release of intergovernmental lack the financial and human resources to hold transfers are fairly widespread in the region. local administrations accountable. In some cases, Second, central governments face growing prob- political corruption prevents local politicians from lems in monitoring the activities of local govern- exercising oversight of local bureaucrats. In May ments. China's deficiencies in managing subna- 2004, 43 of 55 members of Indonesia's West Suma- tional liabilities have already been mentioned. The tra legislative council were found guilty of corrup- Philippines and Vietnam have a limited ability to tion. Though this experience highlights widespread monitor and control the addition of "temporary" corruption in local administrations in Indonesia, it or "informal" workers to local establishments. In also provides a successful example of judicial over- Indonesia, decentralization has undermined data sight and prosecutorial effectiveness. systems in sectors such as education, for which Civil society groups,including the media,can also local authorities are now responsible. As a result, raise awareness of the extent and costs of weak gover- the central government is less able to measure ser- nance, in terms of monitoring government per- vice delivery in these sectors than before decentral- formance and giving citizens a voice to demand ization (see chapter 9). accountability from their leaders (see chapter 12). Accountability systems. A number of interlocking The importance of civil society grows as decentral- factors constrain local accountability in East Asia. ization expands citizens' access to government Upward accountability--or central oversight of actors.InthePhilippines,the SocialWeather Stations local administrations--tends to be stronger than produced a Report Card of Pro-Poor Services to share other forms of accountability but suffers from a citizen feedback on public services. Vietnam is lack of clarity in functional allocations, incomplete launching a similar initiative. In China, some locali- flow of information, and inadequate monitoring. ties have introduced Citizens' Charters through Horizontal accountability--the monitoring of local which service users rate how government agencies bureaucrats by local politicians, and of local gov- and public utilities perform against their stated com- ernment by local legislatures and courts--is diluted mitments. However, although civil society organiza- by clientelist politics and a serious lack of capacity tions are growing rapidly in most East Asian coun- among local countervailing powers. Downward tries, they tend to concentrate in national capitals accountability, or responsiveness to citizens, is and are therefore less influential at subnational lev- often interrupted by a tendency to focus on compli- els.Where they do exist, local civil society groups are ance rather than performance. constrained by insufficient financial and human East Asian countries have historically empha- resources, political interference, and weak links to sized top-down, ex ante control of subnational central organizations. expenditures. As decentralization evolves and local International experience shows that accountabil- governments gain greater financial autonomy, both ity stems from numerous processes and relation- ex post review processes (such as external audits) ships, including those outside the formal public and horizontal and bottom-up systems inevitably sector. Enhancing access to information is a critical become more important in deepening accountabil- step in bringing external pressure to bear on govern- ity. For any of these to work, local data must be pro- ments. In Mexico, the government has established a duced, shared, and maintained. In Vietnamese System of Evaluation of Municipality Transparency communes, poor record keeping makes monitoring to measure how much information municipal the use of funds extremely difficult. In China, local governments are offering their citizens and to reliance on extrabudgetary funds has reduced encourage greater transparency. In Uganda, after a transparency, and thus accountability, in public Public Expenditure Tracking Survey revealed that spending. Thailand and Vietnam have experi- schools had received only 13 percent of per-student mented with mechanisms to shift from central con- grants, the central government launched a massive 20 East Asia Decentralizes information campaign, spreading the word through gests that progress in moving decentralization for- the media and on notice boards outside schools and ward is rarely monotonic; programs proceed in fits district offices. Four years later, a repeat survey and starts. But there is cause for concern in East showed that over 90 percent of resources were Asia that a low-level equilibrium has set in, with reaching schools. Community report cards have only minimal reforms proceeding at a glacial pace. produced similar successes, notably in Bangalore, Recognizing the importance of national require- India, where citizens rated public services and the ments in developing decentralization approaches, media widely disseminated results. Workshops this report does not offer specific recommen- allowed providers and clients to interact, and some dations on the way forward. These must be devel- agencies took steps to improve their service delivery oped at the country level. But the paramount need systems. to make decentralization work broadly across Finally, citizens need opportunities to hold local the region may warrant general guidance to governments accountable. In the Brazilian city of policymakers as they think about the strategies Porto Alegre, an experiment in participatory bud- needed to advance the range of intergovernmental geting improved public service delivery considerably reforms appropriate to their particular contexts. and has since expanded to over 80 cities. The gram Drawing on the earlier sections of this chapter, sabha public forum in Indian local governments and and previewing the themes that will recur in later community oversight committees in Bolivia allow chapters of this report, the following three broad citizens to monitor government expenditures and substantive imperatives merit priority attention in help make decisions regarding future activities. decentralization: Across the spectrum, the most successful mecha- Improving the organization of the intergov- nisms tend to be those that link information, capac- ernmental system will usually require focused ity, and opportunities for participation. attention on policy and, often, legislative develop- ment in three core areas: Conclusion: Key Messages and Challenges · Clarifying the expenditure assignments and func- tional roles of the various levels of government As the preceding discussion has shown, East Asia's with a view to eliminating (or greatly narrow- decentralization process--like the region itself--is ing) jurisdictional overlaps and reducing the characterized by heterogeneity as well as a set of gaps between mandates and funding. shared features, many of which are common to the · Rationalizing the vertical and horizontal organi- decentralization experience of other countries zation of the intergovernmental system to ensure around the world. As elsewhere, the thrust of an appropriate balance in the roles of central, decentralization in the region has been determined intermediate, and local authorities, and to inject largely by structural and political factors. These fac- greater transparency and objectivity into the tors suggest that decentralization is likely to be a processes that determine the size and geographic pivotal fact of East Asian life for the foreseeable boundaries of subnational entities. future. It is already affecting prospects for eco- · Developing systems to facilitate lateral and verti- nomic development, possibilities for "good gover- cal cooperation between subnational govern- nance" in country institutions, and the quality of ments and central government agencies, partic- service delivery, especially for the poor. Some early ularly in the areas of planning and investment. indicators give reason to hope that outcomes in these areas are trending in the right direction--that Strengthening local fiscal and financial struc- is, that the benefits of decentralization in improv- tures involves expanding the financial resources ing access and voice for local constituencies will available to subnational governments while improv- outweigh the costs of disruption and inefficiency ing their incentives to use these resources efficiently. that overhauling formerly centralized institutions Three focus areas are evident: own-source revenues, inevitably bring. But the decentralization process subnational borrowing, and intergovernmental trans- has been uneven and, in some countries, may fers. Substantial progress has been made in East Asia actually have stalled. International experience sug- with respect to intergovernmental transfer policy East Asia Decentralizes 21 and systems, but this dynamic area will require While capacity limitations have surfaced as an ongoing attention. In the other two instances, the important issue in decentralization processes picture is more problematic.Throughout the region, throughout the developing world, two aspects policy environments are not conducive to the emer- require particular attention in the East Asian gence of enhanced own-source revenue or vibrant environment. First, insufficient capacity at the and disciplined borrowing systems. Little serious local level is not the only issue; central govern- policy work is under way, and current solutions tend ments also require improvements in their ability to be at odds with emerging international best prac- to administer the intergovernmental system and tice. The problems that arise in these areas, including drive the decentralization process effectively. the lack of hard budget constraints at the local level, Second, capacity-building efforts that focus often stem more from enforcement failures than purely--or even predominantly--on the supply from poor policy. Rules need to be well designed and side are likely to bring limited success. For effectively enforced. And local authorities need to capacity to expand and endure, reform efforts have incentives to raise revenues and control expen- need to generate effective and ongoing demand ditures. Serious efforts are needed in this area. for enhanced capacity at the local level, and to create systems for responding to that demand. Developing the functional systems that under- Reforming subnational human resource systems pin the effectiveness of intergovernmental struc- will be critical to this endeavor. tures has three important dimensions: · Deepening and enhancing accountability. Effective Finally, in addition to these substantive chal- accountability systems required to reap the lenges, three messages on the nature of the decen- potential developmental benefits of decentralized tralization process are important: government are consistently weak across the region. Substantial improvements are needed in · Given varying conditions within countries, inter- formal and informal bottom-up processes of governmental frameworks will need to be crafted accountability at the local level, and in top-down with enough flexibility to allow for appropriate systems for generating information and effec- asymmetries in the design and implementation tivelymonitoringlocalperformancebythecenter. of decentralization structures and processes. National and local governments alike need better · The inevitably uneven pace of reforms and the data on all aspects of the local and intergovern- need to capitalize on political opportunities as mental system, particularly subnational finance they arise call for a reform process that is more and local government performance in delivering piecemeal than comprehensive. International services. Systems must therefore be developed to experience suggests that focusing on a few key ensure regular and accurate production and dis- areas where change is possible and getting these tribution of basic local government data. right is likely to yield greater success than trying · Improving the performance of intergovernmental to accomplish too much simultaneously on too functional and management systems. As intergov- many reform fronts. ernmentalstructuresevolve,systemic"plumbing" · While the inevitably long-term nature of the needs to be built to allow these structures to func- decentralization process must be understood, tion properly. Intergovernmental transfers need each country's policymakers should take stock of to be disbursed, local financial statements need to progress on intergovernmental reforms and be produced and audited, and budgets need to be assess whether the pace and energy of reforms are produced in a timely manner. In many East Asian sufficient to meet the important challenges dis- countries, these systems have not kept pace with cussed in this report. Mindful of the magnitude the evolution of the intergovernmental structure, of problems that are likely to accumulate if local and local performance has suffered. and intergovernmental structures remain weak · Building capacity. The best-designed intergov- and incentives are not in place, governments need ernmental system in the world will not func- to combine political capital, strategic focus, and tion effectively without sufficient capacity, technical effort to ensure consistent progress on particularly in the area of human resources. decentralization in the region. 22 East Asia Decentralizes Endnotes Monentary Fund conference on fiscal decentralization, Washington, DC, November 2000. 1. In 2001, central government spending on capital in East Azfar, Omar, Tugrul Gurgur, Satu Kähkönen, Anthony Lanyi, Asia and the Pacific was 3.6 percent of GDP, compared and Patrick Meagher. 2000. "Decentralization and Gover- with 3 percent in Eastern Europe and Central Asia and nance: An Empirical Investigation of Public Service Delivery 1.6 percent in South Asia. (Figures are derived from the in the Philippines."Washington, DC: University of Maryland World Bank's World Development Indicators and World IRIS Center and the World Bank. Bank 2004b.) Bahl, Roy. 1990. "Implementation Rules for Fiscal Decentraliza- 2. In Indonesia, households perceived corruption as the top tion." Working Paper 99-1. International Studies Program, national problem. In Cambodia, urban citizens and foreign School of Policy Studies, Georgia State University, Atlanta, firms called corruption the leading problem, while rural citi- Georgia. zens called it the second most serious problem (after the high Bardhan, Pranab, and Dilip Mookherjee. 2000. "Capture and cost of living), as did domestic firms (after street crime). In Governance at Local and National Levels." American Eco- Thailand, respondents called it the third most serious nomic Review 90 (2):135­9. national problem, after the poor economy and high cost of Bird, R., R. Ebel, and C. Wallich, eds. 1995. "Decentralization of living. See World Bank, Partnership for Governance Reform the Socialist State." Washington, DC: World Bank, Intergov- in Indonesia 2002b; World Bank 2000; and Phongpaichit ernmental Finance in Transition Economies. et al. 2000. Burki, S. J., G. Perry, and W. Dillinger. 1999. "Beyond the 3. For example,in Cambodia,82 percent of firms reported some Center: Decentralizing the State." Washington, DC: World level of bribe payments, and 71 percent of large firms sug- Bank. gested that such payments are frequent (World Bank 2004a). Duncan, H., and C. E. McLure, Jr. 2005. "Tax Administration 4. This view is consistent with earlier regional analyses of Mechanisms for Intergovernmental Cooperation." In decentralization by the World Bank in Latin America and Cordes, Joseph J. Cordes, Robert D. Ebel, and Jane G. Grav- East and Central Europe (Burki al. 1999; World Bank 2001). elle, The Encyclopedia of Taxation and Tax Policy. Washing- 5. To date, only 172 of the 245 functions specified for devolu- ton, DC: Urban Institute Press. tion in the Master Plan have been or are in the process of Ebel, R. D., and R. Taliercio. 2004. "Subnational Tax Policy and being devolved, and the rate of devolution has declined over Administration in Developing Economies." Tax Notes Inter- the past two years. The Nine Policy Measures to improve national March: 919­936. local revenue mobilization, approved by the Cabinet in Ebel, R., and S. Yilmaz. 2002. "Globalization and Localization: 1994, have made slow progress toward legal enactment Decentralization Trends and Outcomes." Washington, DC: (chapter 2). World Bank, Regional and Sectoral Studies. 6. These assessments are based on initial data that will need to ------. 2003. "On the Measurement and Impact of Fiscal be reviewed over time. Attributing impact to intergovern- Decentralization." In Public Finance in Developing and mental reform is also problematic, as decentralization has Transitional Countries: Essays in Honor of Richard M. Bird, occurred in parallel with other policy changes. eds. Jorge Martinez-Vazquez and James Alm. Cheltenham, 7. A review of intergovernmental tax administration in E. UK: Edward Elgar Press. Asia is provided in Ebel and Taliercio 2004. Evans, A., and N. Manning. 2004."Decentralization: A Review of 8. Kabupaten are districts; kota (or kotamadya) are cities. Staffing Practices in Eight Countries." Manuscript. 9. These figures are based on a limited sample of countries in Filtrani, F., B. Hofman, and K. Kaiser. 2004. Unity in Diversity? each region. The Creation of New Local Government in a Decentralizing 10. Borrowing from private sector sources typically constitutes Indonesia. less than 1.5 percent of total annual subnational revenues Freire, M., and J. Petersen. 2004. Subnational Capital Markets in in the Philippines, despite the 1996 Local Government Unit Developing Countries: From Theory to Practice." Washington, Financing Framework, which stresses private lending to DC: World Bank. local governments as a core objective. Glasser, M., and R. White. 2004. "South Africa." In Subnational 11. The Finance Ministry has imposed a temporary freeze on Capital Markets in Developing Countries: From Theory to subnational borrowing, but indications are that this will Practice,"ed. Mila Freire and John Petersen.Washington, DC: expire in 2005. World Bank. 12. There are many legitimate reasons for the center to retain Kaufmann D., A. Kraay, and M. Mastruzzi. 2003. "Governance some authority over civil service management, such as to Matters III: Governance Indicators for 1996­2002." Policy maintain minimum standards for the working conditions Research Working Paper No. 3106. Washington, DC: World and professional qualifications of public servants, to Bank. broaden their career paths, and to unify a fragmented Mountfield, E. 2001. "Special Focus: Public Financial Account- nation. On the other hand, as shown by Vietnam's experi- ability in East Asia." Washington, DC: World Bank. ment with block grants to districts and departments in Ho OECD (Organisation for Economic Co-operation and Develop- Chi Minh City, service delivery and operational efficiency ment). 2002a. "The Fiscal Decentralization Initiative for can be improved by extending autonomy over budget allo- Central and Eastern Europe (Budapest) and the Organiza- cations to civil service managers and allowing them to keep tion for Economic Cooperation and Development." Paris: the savings. OECD. ------. 2002b. "Fiscal Design Surveys across Levels of Govern- Bibliography ment." Tax Policy Study #7. Paris: OECD. Pellegrini, A., and Cecilia Soriano Ma. 2002. "A Study to Revisit Ahmad, E., Li Keping, and Thomas Richardson. 2000. "Recen- the LGU Financing Framework and Its Implementation." tralization in China?" Paper presented at International Manila: Department of Finance. East Asia Decentralizes 23 Phongpaichit, P., N. Treerat, Y. Chaiyapong, and C. Baker. 2000. World Bank. 2000. "Cambodia Governance and Corruption "Corruption in the Public Sector in Thailand: Perceptions Diagnostic: Evidence from Citizen, Enterprise, and Public and Experiences of Households." Official Surveys, 2000." World Development Report: Entering Rusten, C., K. Sedara, E. Netra, and P. Kimchoeun. 2004. "The the 21st Century. Washington, DC: World Bank; and New Challenges of the Decentralisation Design in Cambodia." York: Oxford University Press. Monograph. ------. 2001. "Decentralization in the Transition Economies: Schroeder, L., and Paul Smoke. 2003. Intergovernmental Fiscal Challenges and the Road Ahead." Washington, DC: World Transfers: Concepts, International Practice and Policy Issues. Bank. In Intergovernmental Transfers in Asia: Current Practices and ------. 2002a. "China National Development and Subnational Challenges for the Future. Ed. P. Smoke and Y. H. Kim. Manila: Finance: A Review of Provincial Expenditure." Washington, Asian Development Bank. DC: World Bank. Sootla, G., et al. 2000. "Local and Regional Tax Administration in ------. 2003a. "Decentralization and Poverty Indonesia." Estonia."In Mihaly Hogye, ed, Local and Regional Tax Adminis- Washington, DC: World Bank. tration in Transitional Countries. Budapest: Local Government ------. 2003b. "Decentralizing Indonesia." Washington, DC: and Public Service Reform Initiative/Open Society Institute. World Bank. Steffensen, J., E. Sewankambo, and P. Tidemand. 2004. "A ------. 2004a. "Seizing the Global Opportunity: Investment Comparative Analysis of Decentralization in Kenya, Climate Assessment and Reform Strategy for Cambodia." Tanzania, and Uganda." Draft Country Study. World Bank, Washington, DC: World Bank. Washington, DC. ------. 2004b. World Development Report 2003­2004: Making Transparency International. 2003. "Corruption Perceptions Services Work for Poor People. Washington, DC: World Bank; Index 2003." and New York: Oxford University Press. United Nations. 2004. World Urbanization Prospects 2003. World Bank and Asian Development Bank. 2003. "Philippines: New York: United Nations. Decentralization and Service Delivery." Washington, DC: Verheijen, A. J. G. 2002. "Removing Obstacles to Effective World Bank. Decentralization: Reflecting on the Role of Central State World Bank, Partnership for Governance Reform in Indonesia. Authorities." In Mastering Decentralization and Public 2002b. "A National Survey of Corruption in Indonesia". Administration Reforms in Central and Eastern Europe. Washington, DC: World Bank. 2 The Rules of the Intergovernmental Game in East Asia: Decentralization Frameworks and Processes Paul Smoke Although political forces have largely driven decen- The Basics of Decentralization tralization in East Asia and most countries face in East Asia similar reform challenges, their decentralization This section briefly examines the origins and paths experiences are far from uniform. Countries have of decentralization in the region. It also outlines the adopted different intergovernmental structures, levels and structures of government, compares the proceeded at uneven paces, and adopted a wide thrust of decentralization policy in various coun- range of implementation strategies. This diversity is tries, and describes the enabling frameworks. not surprising, as East Asian countries vary greatly in geographical size, population, history, economic structure, and political and institutional dynamics, The Origins and Evolution all of which influence the form that decentraliza- of Decentralization Reforms tion can and should take. This chapter provides expanded context for the SomeEastAsiancountries,suchasChina,havealong analysis presented in chapter 1 and lays a foundation tradition of limited decentralization, while the con- for later chapters. After reviewing the origins of cept is more recent in countries such as Cambodia. decentralization, it compares the basic intergov- Crisis sparked decentralization in Indonesia and the ernmental frameworks, structures, and processes Philippines, while in China and Vietnam it is part evolving in Cambodia, China, Indonesia, the of a gradual process of market and public sector Philippines, Thailand, and Vietnam.1 The chapter reform. In a few cases, decentralization is essentially focuses, in turn, on enabling frameworks, the gover- complete (in structure if not in practice), such as in nance environment, fiscal decentralization, and the the Philippines, or heavily in process, such as in management and implementation of decentraliza- Indonesia. In other cases reforms are less advanced, tion reforms. either with limited policy development, as in 25 26 East Asia Decentralizes Cambodia, or a substantial but only modestly Weak attempts to decentralize in the 1970s and implemented framework, as in Thailand. 1980s did not gain political momentum. The East The Philippines has the strongest history of Asian economic crisis in 1997 hastened the fall of democratic decentralization in the region. The the Soeharto regime. Around that time, a successful country's colonial heritage established limited dem- independence referendum in the former province of ocratic roots, and a series of presidential decrees East Timor and growing complaints from resource- enacted under the autocratic Marcos government-- rich provinces about insufficient revenue autonomy including the Local Tax Code, the Real Property Tax increased pressure for reform. The 1999 decentral- Code, and the Local Fiscal Administration Code-- ization legislation was a direct response to this polit- laid the institutional foundation for decentraliza- ical crisis and a perceived need to hold the diverse tion. The country reestablished democracy after the and tenuously unified country together. The reform fall of Marcos in 1986, and decentralization and devolved power primarily to subprovincial govern- local autonomy were among the fundamental prin- ments, largely because of fear among national lead- ciples embodied in the 1987 Constitution. The ers that empowered provinces could fuel regional Aquino administration launched a pilot decentral- ethnic and political conflicts, leading to further sep- ization project and established autonomous regions aratism or federalism. The government revised the in Muslim Mindanao and the Cordilleras. But not initial decentralization framework legislation in until 1991 did constitutional provisions begin to 2004. This increased higher-level control but left take robust shape through the Local Government unresolved important aspects of the intergovern- Code, which mandated significant devolution to mental system. Local capacity is deficient in many local governments. Today a formal decentralization areas, and citizens and government officials, both framework is essentially complete, but much effort elected and appointed, are still learning to function remains to realize full implementation. in the evolving democratic environment. Although China's lack of democratic institutions Vietnam, which became a centrally planned may be seen as an impediment to decentralization, communist state after the Vietnam War, has increas- the country has some history of subnational author- ingly formalized the subnational government ity because its size has made central control difficult. framework since the mid-1990s. As in China, eco- The provinces, in particular, have long enjoyed nomic reforms (doi moi) spurred initial progress on a degree of administrative and fiscal autonomy. intergovernmental reform. The center still exerts The market transition that began in the late 1970s substantial control, but subnational governments enhanced the subnational role. As economic reform have some discretion. Provinces have greater pow- progressed, changes in the composition of expendi- ers, including considerable authority over lower lev- tures and relative prices moved the burden of public els. Popular participation and grassroots demand spending to subnational governments. These adjust- for political voice have grown, but Vietnam remains ments led to further modifications of government a one-party state and a fairly centrally driven sys- operations, but the country never adopted a formal tem. The country has moved forward with its decentralization policy. Still, by 1993, the central decentralization framework and conducted some government was collecting only 20 percent of public successful policy experiments, but implementation revenues. This prompted the substantially recentral- is uneven and additional reforms are required. izing Tax Sharing System reforms in 1994. Public Thailand has been modestly enhancing the role demand for more responsive government and greater of subnational entities for some time, but decentral- entrepreneurial freedom has also shaped central- ization has been a priority only since the Seventh subnational relations, but formal intergovernmental National Economic and Social Development Plan political reforms have been limited. (1991­96). The plan emphasized developing local Indonesia--also a large country and spread over infrastructure, providing credit to expand and thousands of islands--had elements of decentral- improve local services, and helping local authorities ization during its Dutch colonial period. As in many mobilize capital and pursue development projects. ethnically diverse countries colonized by European The May Five democracy movement emerged in the powers, building national unity through greater mid-1990s to demand stronger democratic institu- centralization was the goal after independence. tions more insulated from the military, which has The Rules of the Intergovernmental Game in East Asia: Decentralization Frameworks and Processes 27 long played a pivotal role in Thai politics. The Eighth Reforms adopted in 2001 led to the election of com- Plan (1997­2002) advocated stronger local institu- mune councils and provided them small intergov- tions, the 1997 Constitution formally enshrined ernmental transfers without formal service respon- decentralization, and later legislation detailed it. The sibilities or own-source revenues. This approach country has formally adopted many reforms but focuses on meeting immediate community needs implemented few of them, and political consensus and developing trust between citizens and the gov- on further progress remains unclear. ernment as a first step in decentralization. Provincial Cambodia's decentralization is relatively unique. reforms have been limited, except for the adaptation Following elections brokered by the United Nations of Seila mechanisms to provide provincial support (UN) in 1993, the center reclaimed power from to communes and a few ad hoc reforms by individ- provincial governors--who had previously ruled ual sectoral ministries. A program to build capacity with a free hand--in order to impose discipline on is under way and the country is planning further the intergovernmental system. The UN-funded reforms, but it is unclear how the system will evolve. Cambodia Resettlement and Reintegration (CARERE) Project of the early 1990s, particularly its Structures of Subnational Government second stage known as Seila, experimented with sweeping local institutional and governance reforms The structure of subnational governments in the in many areas. Seila's success in delivering develop- region varies considerably (see table 2.1). Most ment projects made its formal integration into the countries have three or four levels of administra- government system attractive to the ruling party. tion. In China, the Philippines, and Vietnam, each TABLE 2.1 Levels of Government Administration Country Subnational levels of government Cambodia Two levels in two parallel systems: · Provincial administrations (20) and municipalities (4) with provincial status divided into districts and khans · Elected commune and sangkat (urban commune) governments (1,621) divided into villages China Four levels: · Provinces (22), autonomous regions (5), and large cities (4) · Prefectures and cities (300) · Counties (2,100) · Townships (44,000 ) Indonesia Three levels (de jure): · Provinces (33), special regions (2), and capital city (1) · Local governments: kotamadya (cities) and kabupaten (districts) (440) · Desa (villages) Philippines Four levels: · Provinces (79) · Cities (112) · Municipalities (1,496) · Barangays/villages (41,944) Thailand Four levels with top three formally empowered: · Provinces (75) · Districts and municipalities (811) · Tambons (subdistricts) (6,744) · Villages (67,000 ) Vietnam Three levels: · Provinces (58) and municipalities (3) · Districts (600) · Communes (10,000 ) Source: Compiled by the author from multiple sources. 28 East Asia Decentralizes level is an active player. In Thailand, all but the low- Decentralization Policy est level have formal authority. In Cambodia and East Asian countries also vary in the extent to Indonesia, two levels have independent powers, which their decentralization policy emphasizes while other levels perform mostly subsidiary deconcentration, delegation, or devolution of administrative and political functions.In Indonesia, functions and revenue authority (see box 1.1). the lower tiers have no formal functions or inde- Some countries such as China and Vietnam have pendent budgets, but centrally managed--often seen legal or de facto deconcentration of functions donor-funded--community development schemes to subnational entities that remain substantially have channeled substantial resources to them (see accountable to the center, though elements of dele- chapter 12). In Cambodia, the provinces (with dis- gation and devolution have emerged. Thailand is trict subdivisions) and communes (with village gradually shifting its focus from deconcentration subdivisions) have functional mandates, although to devolution, but reform remains at a relatively with a different relationship to the center. early stage. Indonesia, the Philippines, and, to a A few countries also have ad hoc subnational certain extent, Cambodia have emphasized devo- institutional arrangements such as special status for lution of responsibilities to more autonomous the capital and other major cities. These include subprovincial entities, but specific approaches dif- Jakarta in Indonesia and the Bangkok Metropolitan fer (see table 2.2). Administration and Pattaya City in Thailand. China's decentralization is unusual in that eco- China, Indonesia, and the Philippines have created nomic reform rather than specific deconcentration a number of autonomous regions, often in areas of or devolution policies has shaped its evolution. special political, historical, or ethnic significance. A During the last two decades China has transitioned few countries have interjurisdictional structures from a largely deconcentrated system to one designed to meet specific needs. The Philippines, that incorporates elements of delegation and for example, has designated 16 planning regions, devolution. Subnational governments have become each with a Regional Development Coordinating more responsible for financing their expanding Council composed of provincial governors, city functions from their own revenue, both formal and mayors, and representatives from national agencies informal, giving them more autonomy except in and the private sector. TABLE 2.2 Decentralization Policy Country Policy orientation Cambodia Hybrid case, with deconcentration to provinces and devolution to communes; commune system new and given greater emphasis, but provinces are more significant in terms of public expenditures. China Main focus on deconcentration to provinces and larger cities, although lower levels have larger public expenditure role and elements of de facto devolution have emerged in some areas; provinces have considerable regulatory control over lower levels. Indonesia Focus on substantial devolution to cities and districts, which replaced earlier emphasis on deconcentration to provinces; limited formal role at lowest levels; 2004 reforms increased the role of higher levels. Philippines Focus on devolution to subprovincial units, but provinces still play a significant role. Thailand Historical focus on deconcentration to provinces and districts, but 1997 framework shifts toward devolution to municipalities, districts, and subdistricts; implementation has been limited. Vietnam Focus mainly on deconcentration with stronger role for provinces, including regulatory control over subprovincial levels; subnational governments have been allocated rights over specific functions, approaching devolution. Source: Compiled by the author from multiple sources. The Rules of the Intergovernmental Game in East Asia: Decentralization Frameworks and Processes 29 sectors with mandated service standards (see chap- resources are subject to central control. Provincial ters 1, 6, and 10). departments continue to be major providers of In Vietnam, decentralization policy blends a national services, though the country has not deconcentration of service responsibilities with an emphasized formally deconcentrating more respon- allocation of rights that resembles devolution. The sibilities to them. latter, however, is much less developed than the for- Both devolution and deconcentration reforms mer, although provinces have considerably more are occurring in Cambodia. The decentralization to power and autonomy than subprovincial entities. elected commune councils is a limited form of devo- Even provinces are subject to minimum expendi- lution, while the central government is planning to ture requirements in some sectors, and the central deconcentrate responsibilities to provinces and government still sets rates on major sources of municipalities. Progress with devolution--albeit revenue. modest--has been more rapid than with deconcen- Thailand has long-established deconcentration tration, owing largely to the Seila Program's signifi- policies but enacted a decentralization law in 1999. cant influence over institutional reform since the The country has drafted action plans for devolving mid-1990s. Deconcentration reform is at an earlier specific functions to subnational governments, but stage and more fragmented, although a few central has assigned relatively few functions thus far. The ministries, such as Agriculture, Education, and most significant devolution has occurred in infra- Health, have experimented with limited functional structure, quality of life, and natural resources and deconcentration. Communes have relative auton- the environment. Decentralization of health and omy in pursuing small-scale local priorities with the education has not yet occurred, though the min- modest resources they receive, but their fiscal roles istries of Public Health and Education have pro- are limited. posed deconcentrating some responsibilities to area health boards and local education authorities until The Formal Basis for Decentralization local governments can meet "readiness" criteria for full devolution. The country plans to enhance The formal basis for decentralization varies widely decentralization of revenues, which are now under throughout East Asia (see table 2.3). A few coun- substantial central control. tries have a constitutional basis for subnational Decentralization reforms in Indonesia include governments, and most have at least a law or set of both devolution of authority and, to a lesser laws that defines the decentralization framework. extent, deconcentration of functions. Deconcen- The formal basis and extent of this framework do tration to provincial authorities was the dominant not seem to be closely associated with decentraliza- form of decentralization before 1999, when the tion performance, as highlighted in chapter 1. Still, emphasis shifted to devolution to city and district the nature and depth of enabling provisions could governments. Local governments have broad func- become more important as challenges to decentral- tions and receive substantial intergovernmental ization arise. transfers, but have limited revenue authority. The Thailand and the Philippines have both a con- country has increasingly developed the legal stitutional and a legal basis for decentralization. framework (most recently through Laws 32 and Thailand's 1997 Constitution clearly specifies prin- 33 of 2004), but functional responsibilities and ciples of local autonomy and elected representa- subnational revenues require further elaboration tion, and establishes specific intergovernmental and regulation. reform objectives. A National Commission on In the Philippines, deconcentration was histori- Public Sector Reform includes a subcommittee to cally important. The Integrated Reorganization implement decentralization policy. The cabinet Plan of 1972 divided the country into 11 (later 16) approved a Local Fiscal Master Plan in 1997, which regions, each with administrative authority. In 1991 defined the framework for Decentralization Act of the center devolved many responsibilities, person- 1999 reforms. Yet the country needs further legal nel, and resources to local governments. These gov- and regulatory instruments to define the sub- ernments have some revenue authority, but most national system more fully. The constitutional and 30 East Asia Decentralizes TABLE 2.3 Decentralization Frameworks Country Formal basis for decentralization Cambodia Legal and administrative basis: Provincial Budget Law (1997) gives limited functions to provinces; Law on Commune/ Sangkat Administrative Management (2001) and Election Law (2001) establish commune system; all legislation clarified in numerous administrative decrees. China No constitutional or dedicated legal basis: Comprehensive Fiscal Reform (1994), Budget Law (1995), and Tax Sharing System (1994) relevant for roles of subnational governments. Indonesia Legal basis and constitutional amendment: Law 22 on Regional Government (1999) amended as Law 32 (2004), Law 25 on Fiscal Balance between Central Government and Regions (1999) amended as Law 33 (2004), and Law 34 on Regional Taxes/Levies (2000) (to be amended) provide a framework for decentralization; constitutional amendment (2000) strengthens basis for decentralization. Philippines Constitutional and legal basis: Constitution (1987) provides for local government autonomy; Local Government Code (1991) and various Marcos-era and post-Marcos laws define aspects of the intergovernmental system. Thailand Constitutional and legal basis: Constitution (1997) specifies principles of local autonomy and elected local government; Provincial Administration Act (1997) codifies deconcentration policies; Decentralization Act (1999) defines functions and decentralization process. Vietnam Legal and administrative basis: Law on Organization (1994), Ordinance on Concrete Tasks (1996), Budget Law (1998), and revised Budget Law (2002) assign functions and resources to subnational governments. Source: Compiled by the author from multiple sources. legal basis for local government is stronger and Two main laws established decentralization in more specific in the Philippines. Articles II and X of Indonesia. Law 22 on Regional Government of 1999 the 1987 Constitution establish the autonomy of eliminated hierarchical relationships between cities local governments and give them the power to cre- and districts and higher levels of government, ate their own sources of revenue. The Local Govern- granting the former autonomy and broad responsi- ment Code of 1991 codifies existing laws on local bilities. This legislation has been revised as Law 32 government, provides for substantial devolution of of 2004, which allows for the direct election of sub- services, and creates local institutions, such as national leaders beginning in 2005, reestablishes school boards, development councils, health boards, central control over the hiring and firing of civil ser- and peace and order councils. vants, and requires ex ante approval of subnational Most other East Asian countries have a legal budgets. Law 25 on Fiscal Balance of 1999 modified but not a constitutional basis for decentralization. the intergovernmental transfer system and provided Cambodia does not have a unified decentralization for limited local revenue authority. This law has framework. However, the Provincial Budget Law of been updated as Law 33 of 2004, which further 1997 provides for modest provincial fiscal powers, defines aspects of the intergovernmental fiscal sys- and the Commune/Sangkat Administrative Man- tem. Law 34 on Regional Taxes and Levies of 2000 agement Law and the Commune Election Law of modestly enhances local revenue authority and the 2001 broadly define the functions and structures government has plans to expand these powers in of commune councils and procedures for electing future legislation. Constitutional amendments them. No legislation deconcentrates powers to passed in 2000 consolidated certain decentraliza- provincial and district governments, but the coun- tion reforms and make it more difficult for the try is developing such a law within the framework of National Assembly and the president to substan- the National Program for Administration Reform. tially reverse them. The Rules of the Intergovernmental Game in East Asia: Decentralization Frameworks and Processes 31 Vietnam has no constitutional basis for decen- role in the political process. The executive branch tralization, but a 1994 Law on Organization and a encompasses a state council, which includes the 1996 Ordinance on Concrete Tasks assign functions prime minister. The president serves as head of to provinces and districts. Decentralization is an state. The legislative body--the National People's important part of the Public Administration Congress--is elected by representatives of lower- Reform Program launched in 1995. Budget laws in level legislatures and designates the president and 1996 and 1998 also formalized fiscal arrangements prime minister. The Chinese Communist Party among levels and assigned budgeting responsibili- plays a pivotal role through its power to designate ties to subnational governments, particularly senior officials throughout the governmental provinces. More recent legislation, including the system. Vietnam is also a one-party state, with the revised Budget Law of 2002, provides more details Communist Party the leading force. Party organiza- on subnational functions and revenue sources. tions at all levels must operate within the constitu- China has the weakest formal basis for decen- tional and legal framework, but they have great tralization. Because intergovernmental changes in power in determining who can run for elected China have occurred mostly through economic office. The main legislative body is the National reform, they have no constitutional or dedicated Assembly, which localities elect directly. As in legal framework. Intergovernmental fiscal relations China, National Assembly delegates elect the presi- rest largely on a complex system of bargaining dent and prime minister. between higher-level and lower-level authorities. Cambodia is a constitutional monarchy under a Since 1994 a number of reforms and legislative democratic regime established in the 1991 peace changes have clarified the fiscal responsibilities of accord. The executive branch includes the king, as different levels of government somewhat, although head of state, and the prime minister, who holds the important areas remain undefined. The most rele- real power as head of government. The legislative vant reforms include the 1994 Tax Sharing System, branch includes a National Assembly and a Senate. the 1994 Comprehensive Fiscal Reform, and the The Cambodia People's Party has dominated recent 1995 Budget Law. elections, but other parties have won national and subnational seats. Opposition parties did well enough in July 2003 elections that negotiations The Governance Environment to form a new government took a full year. Like This section reviews key aspects of the governance Cambodia, Thailand is a constitutional monarchy environment in East Asian countries; subnational with a unitary democratic government. A directly elections, autonomy, and transparency; and the role elected Parliament selects a prime minister. The of civic participation mechanisms and civil society country has a multiparty system with a history of organizations. These factors, among others, help unstable coalition governments and military inter- determine governance quality in a decentralizing vention. Under the 1997 Constitution, however, the environment (chapters 5 and 11). country is evolving into a two-party system domi- nated by the incumbent Thai Rak Thai Party and the opposition Democrat Party, with the latter advocat- The National Political Environment ing decentralization. East Asian countries exhibit a broad spectrum of Indonesia and the Philippines are both demo- political environments. China and Vietnam are cratic republics. Indonesia's governmental structure single-party states. In Cambodia a single party is particularly complex. The main legislative body is dominates, while Indonesia, the Philippines, and the largely elected People's Assembly (DPR). In 1999 Thailand have multiple competitive parties. All of the electoral system included hybrid proportional these countries have some type of national and and district elements based on closed party lists subnational assemblies. Table 2.4 summarizes key (voters could not vote for individuals). Most of the features of their political systems. 500 DPR members were elected, but 38 were ap- China is a popular republic with a single official pointed by military and police factions. The People's political party, the Chinese Communist Party. Consultative Assembly (MPR), which included Minority parties are extremely small and play no the DPR plus 135 members selected by provincial 32 East Asia Decentralizes TABLE 2.4 Basic Features of Political Systems Governmental Political Country system competition Legislative branch Executive branch Cambodia Constitutional Multiparty; National Assembly and King (head of state); monarchy Cambodia Senate with direct prime minister (head of People's elections government) designated Party by National Assembly dominates China Popular Single party: National People's Congress President, vice president, republic Chinese elected by lower-level and state council Communist congresses (15 members, Party including prime minister) all designated by National People's Congress Indonesia Democratic Competitive People's Assembly (DPR) President elected by the republic multiparty directly elected; largely People's Consultative system consultative Regional Assembly until direct Representative Council election in 2004 (DPD) created in 2004; People's Consultative Assembly (MPR) composed of DPR and DPD manages constitutional reform Philippines Democratic Competitive House of Representatives President elected directly republic multiparty and Senate largely directly by the people system elected Thailand Constitutional Multiparty: Parliament with direct King (head of state); monarchy two elections prime minister (head dominate of government) designated by Parliament Vietnam Popular Single party: National Assembly elected President and state council republic Vietnamese by lower-level assemblies (including prime Communist minister) designated by Party National Assembly Source: Compiled by the author from multiple sources. legislatures and 65 members representing other but other parties, particularly the nationalist Demo- groups, met only as needed to elect the presi- cratic Party and the United Development Party, an dent and reform the Constitution. The country Islamic party, have become more powerful. The adopted major changes for 2004 elections. Political resounding defeat of former President Megawati representation by the military, police, and special Sukarnoputri by former army chief Susilo Bambang interest groups ceased, and Indonesians directly Yudhoyono in the 2004 direct elections signals a elected the president. A new, territorially based, and new era in Indonesia's democratization, and has largely consultative body, the Regional Representa- raised expectations of the national leadership. tive Council (DPD), was also elected, and the DPR The Philippines also has a multiparty system,and and the DPD now together constitute the MPR, competition typically requires parties to form a which retains only its as-needed constitutional coalition government. The country relies on direct reform function. Until the 1998 collapse of the elections to fill all elective offices,including the pres- Suharto regime, Indonesia was effectively a one- ident and members of the House of Representatives party state run by the Golkar Party. Golkar still plays and the Senate. The exception is the provision for a a major role and did well in the 2004 elections, limited number of special party-list representatives The Rules of the Intergovernmental Game in East Asia: Decentralization Frameworks and Processes 33 TABLE 2.5 Subnational Assemblies and Elections Country Subnational assemblies and elections Cambodia Subnational representative bodies elected through universal suffrage only at the commune level. China People's Congresses in China exist at all levels of government, but only the village level is directly elected. Indonesia Regional People's Assemblies elected at local and provincial levels. Philippines Directly elected bodies exist at all subnational levels of government. Thailand Different types of subnational governments have directly elected councils of different sizes. Vietnam People's councils at all levels of government are directly elected and ratified by the immediately superior council. Source: Compiled by the author from multiple sources. of marginalized sectors, where people vote for par- won seats on many councils. Indonesia directly ties rather than individuals. elects the Regional People's Assemblies (DPRD) at local and provincial levels. Under Law 32 of 2004, subnational leaders (governors and mayors) will be Subnational Elections directly elected and can be removed with cause by All East Asian countries reviewed here have subna- the DPRD, subject to higher-level approval. tional government assemblies, but they vary consid- Thailand holds subnational elections every four erably in whether and how they elect those assem- years at all but the lowest (village) tier. The Local blies (see table 2.5). The number of levels, the size of Election Act of 2002 shifted responsibility for con- jurisdictions, whether the elections are direct or ducting local elections from the Ministry of the Inte- indirect, the degree of political competition, and the rior to the Election Commission, a new independent relationship between elected councils and subna- constitutional agency. Various levels of subnational tional executives all influence the degree of genuine government have councils of differing sizes that are subnational representation and accountability. directly elected. Council members have elected the At one extreme are China andVietnam,where the chair of subnational councils, except in the Bangkok Communist Party heavily influences subnational Metropolitan Administration, where the governor is elections, reinforcing upward accountability. In popularly elected. Broader direct election of subna- China, People's Congresses exist at all levels of gov- tional executives (although not provincial gover- ernment, but voters elect delegates only at the village nors) is intended for the future. Among countries level, which is not a formal unit of local government. with multiple political parties, only the Philippines Subordinate congresses elect delegates to higher- conducts elections at all levels. Per the 1991 Local level congresses from party lists.InVietnam,People's Government Code, the country holds subnational Councils are elected through universal suffrage at all elections every three years, except at the barangay levels, but leaders are elected by council members level, where they occur every five years. The Local and ratified by the People's Council at the next level. Government Code also created special-purpose rep- Other countries hold more freely contested elec- resentative bodies such as Local Development Coun- tions, but not at all levels. Cambodia holds subna- cils,which formulate and ratify development plans. tional elections with universal suffrage only for com- mune councils. These are elected with a five-year Subnational Autonomy in Budgetary mandate on a proportional basis, such that the and Personnel Decisions councils can include representatives of more than one political party. The council chief is the individ- The autonomy of subnational governments varies ual receiving the most votes on the majority-party considerably across East Asian countries (see list. The Cambodia People's Party dominated the table 2.6). This section characterizes their indepen- first local elections in 2002, but other parties also dence in making budgeting and personnel decisions 34 East Asia Decentralizes TABLE 2.6 Subnational Budgetary and Personnel Autonomy Country Degree of subnational autonomy Cambodia Commune governments have their own budgets, whereas provincial budgets are linked to the national budget; strong central civil service control. China Subnational governments have their own budgets but are hierarchically integrated with higher levels and subject to central civil service regulations; control is weaker in practice and off-budget activity is considerable. Indonesia Subnational governments initially had complete budget autonomy, with next-higher level having legality review, and national civil service regulations allowed a reasonable degree of subnational discretion; Law 32 of 2004 significantly expanded higher-level control over budgeting and the civil service. Philippines Subnational governments prepare budgets with legality review by next-higher level; national civil service regulations allow subnational discretion. Thailand Local governments prepare budgets subject to certain central mandates and follow civil service regulations; major reforms planned. Vietnam Subnational governments have their own budgets, but these are hierarchically integrated and approved by higher levels; this is being phased out, and major cities have been permitted to experiment with greater autonomy. Source: Compiled by the author from multiple sources. (see also chapters 1, 5, and 7). Subnational govern- staffing of the People's Councils, but each level of ments in some countries are subject to significant subnational government now has limited discretion control by higher levels, although such control is in managing local personnel. Pilot schemes in not always exercised effectively. In other cases, sub- Hanoi and Ho Chi Minh City allow even more local national governments are more independent. discretion in managing budgets and personnel. Official autonomy is generally weak at the sub- The Philippines has established fairly strong national level in China and Vietnam. In China, sub- local autonomy, in principle. Local governments national budgets are approved by People's Con- prepare their own budgets, which are reviewed at gresses at the same level, but hierarchical linking of the national level in the case of provinces and cities, budgets, a lack of local tax autonomy, higher-level and by provinces in the case of municipalities and directives, and earmarked funding offset this dis- barangays. This review is intended to ensure that cretion somewhat. Most subnational officials are budgets meet regulatory requirements, not to inter- also appointed by People's Congresses at the same fere in composing the budgets. Civil service regula- level, but higher levels appoint top officials. These tions, particularly regarding salary, are national, but senior officials increasingly come from lower-level local chief executives exercise some discretion. ranks rather than the central bureaucracy, which Salaries of local officials may vary widely, as some may improve local accountability. Management of local governments lack the funds to meet national the subnational civil service closely follows national standards. regulations, although local leaders exercise some Indonesia, Thailand, and Cambodia fall in the discretion. middle of the autonomy spectrum. Indonesia's Although subnational People's Councils in Laws 22 and 25 of 1999 originally provided for Vietnam have their own budgets, they are integrated strong regional autonomy in principle. Local gov- into a hierarchical system that requires higher-level ernments had authority over their budgets, subject approval. Provinces have more expenditure auton- to national legality review, and technically con- omy than subprovincial levels. As of January 2004, trolled their staff subject to national regulations. provincial budgets no longer require approval from On the other hand, local governments did not the National Assembly, and Provincial People's select many of their staff, who were transferred Councils have some authority to assign expenditure under the 1999 decentralization, along with sub- and revenue functions among subordinate levels of stantial intergovernmental transfers to support government. Central rules and regulations govern them. As noted, recent revisions to Law 22 (Law 32 The Rules of the Intergovernmental Game in East Asia: Decentralization Frameworks and Processes 35 of 2004) expanded central control over budget and Subnational Transparency civil service decisions. Subnational budgets require East Asian countries have all made some attempts to formal approval rather than legality review, and improve transparency and expand access to infor- authority to hire and fire subnational civil servants mation at the subnational level, but intent has often has been significantly recentralized. been more substantial than practice (see table 2.7). In Thailand, local governments prepare and exe- Some countries, such as the Philippines, provide cute their own budgets, but they are subject to cen- extensive public documentation of and access to tral direction. A significant share of local expendi- subnational government budgets and other infor- tures is centrally mandated, with the largest portion mation, while other countries, including China, do devoted to personnel expenses (representing 30 per- not. Audits do generally occur--internally in some cent of local budgets, on average). Central directives cases, both internally and externally in others--but govern staff numbers, salaries, and benefits. Major reforms, however, are intended to eventually move countriesusuallydonotmonitorsubnationalbudget this highly centralized civil service to one where performance comprehensively. local governments have considerable authority over Transparency in China and Cambodia is low. In personnel management. China, a finance director for each subnational gov- Cambodia is a more unusual case, as its system is ernment provides an annual report to the People's new and the gap between provincial and local pro- Congress on budget implementation and the main cedures is significant. Centrally appointed provin- features of the current budget. This is the only docu- cial governors have some power and influence in ment on subnational budgets and includes only coordinating budgets, but provincial line depart- highly aggregated data. Substantial off-budget fund- ments are primarily accountable to parent min- ing also limits transparency. Internal audits are rela- istries. Commune councils have greater autonomy, tively strong in the Chinese system. Local audit in principle: they have some discretion in preparing bureaus conduct external audits, but these fall under plans and budgets if they follow basic guidelines. the direct authority of the subnational government. Under nascent decentralization, however, centrally External auditing by higher levels is infrequent. appointed key staff limit local autonomy. For There is no system for evaluating budgets, although example, the Ministry of Interior appoints the the central government does focus on meeting tar- commune clerk (though council members can gets in priority areas such as family planning and tax request a replacement if they show cause), and the collection. Cambodia has a legal framework for commune treasurer is a member of the Provincial budget review, including the 2000 Audit Law. How- Treasury (though officially required to follow the ever, the capacity to implement this system is not in instructions of the commune council). place, and public access to documentation is limited. TABLE 2.7 Subnational Transparency Country Mechanisms for subnational transparency Cambodia Evolving commune reporting requirements provide public information in some areas; National Audit Authority weak. China Limited publishing of official subnational government documents; strong internal audit; external audit weak and not independent from executive. Indonesia Various public reporting requirements but weaker in practice; Commission on Audit has mandate to review subnational governments but limited in practice. Philippines Several subnational public reporting requirements; Audit Commission review of budgets; internal audit generally weak. Thailand Subnational governments required to generate significant public financial reports; external audit hampered by capacity limitations; internal audit generally weak. Vietnam Well-defined system of reporting but weaker in practice; State Audit reviews subnational governments, but not independent from executive; weak internal audit. Source: Compiled by the author from multiple sources. 36 East Asia Decentralizes The Philippines, Indonesia, Thailand, and and Web sites. Internal auditing is weak except in Vietnam have adopted transparency frameworks, large cities. The auditor general and provincial but they do not always function well. In the audit units are credible but do not have the capacity Philippines, budget documents--including reports to audit all subnational governments. Vietnam has on implementing the previous budget and an a well-defined system of reporting from lower to annual financial report--are made public. The higher levels, and governments must make certain Department of Budget and Management and the budget data public. Internal auditing is seriously Commission on Audit require regular accountabil- deficient, primarily because of capacity constraints. ity reports, and the financial transactions of local The State Audit of Vietnam must audit all sub- governments are subject to ex post review by the national governments, but the agency is not Commission on Audit. In Indonesia, budget docu- independent of the executive. Subnational budgets ments are supposed to be public and external include no performance measures. audits are conducted, although not always on schedule because of capacity constraints in higher- The Role of Civic Participation and Civil Society level departments. The revised Constitution guar- antees freedom of information, and a new anti- Most East Asian countries officially require civic corruption law requires access to information. participation, and civil society organizations gener- However, bureaucratic barriers make exercising ally exist, but their development and influence these civic rights difficult. vary substantially (see table 2.8). This section Thailand has been improving transparency. The focuses on how central governments engage citi- 1997 Constitution guarantees freedom of informa- zens in decision making and support and regulate tion, although the country has not yet passed laws civil society organizations. In countries with lim- implementing that freedom. Local governments ited democratic institutions, such as China and must submit budget plans, financial reports, and Vietnam, mechanisms for civic participation tend procurement reports to the Department of Local to be weak, and civil society less independent of the Authority Promotion, the Bureau of the Budget, state. The roles of nongovernmental organizations and the Auditor General. All are publicly disclosed, (NGOs) also differ greatly across East Asia, with and some local authorities prepare publications some taking direct responsibility for providing TABLE 2.8 Subnational Civic Participation and Civil Society Country Subnational civic participation Subnational civil society Cambodia Civic participation introduced in Some active civil society groups the context of emerging commune emerged from the period of system, but weak in many areas. civil war. China No formal government mechanisms Civil society organizations permitted for civic participation. but heavily regulated, creating disincentives. Indonesia Civic participation encouraged and Emerging civil society, but complex and sometimes required by donors, but limited in some respects. no official mechanism. Philippines Civic participation framework in Local Relatively active civil society. Government Code (1991), but no formal mechanism. Thailand Strong constitutional and legal basis Civil society groups limited but growing. for civic participation; much weaker in practice. Vietnam Civic participation mechanisms encouraged Many civil society organizations, but under Grassroots Democracy Decree major groups are under state control. (1998), but new and unfamiliar. Source: Compiled by the author from multiple sources. The Rules of the Intergovernmental Game in East Asia: Decentralization Frameworks and Processes 37 services. Cambodia, Indonesia, and Thailand, for but those that attained formal status have moved example, boast many active NGOs, but they tend into advocacy on key policy issues. Examples not to interact extensively with local governments. include attempts by the Forum for Popular Partici- Civic participation mechanisms are most pation to push amendments to Law 22 of 1999, and developed, and civil society the most organized and the support of citizen forums by the Indonesian dynamic, in the Philippines. Civil society grew Partnership in Local Governance Initiatives, a net- during the Marcos dictatorship, when antigovern- work of local NGOs. ment sentiment was high and focused on "people's China and Vietnam do not emphasize civic par- empowerment." After the democratic transition, ticipation and do not have vibrant civil societies. civil society organizations became more active. Formal NGOs have a shorter history in China than The 1987 Constitution and Bill of Rights ensure in other East Asian countries, and social welfare is the rights of independent NGOs and facilitate considered the responsibility of central govern- popular consultation. The 1991 Local Government ment. In the late 1970s and 1980s, government Code requires public participation in barangay departments at all levels approved and managed development plans and certain local functions. social organizations. As these groups proliferated, A Barangay-Bayan Network assists barangays in the Ministry of Civil Affairs took control of this developing plans and projects, and the Local process in 1988. In 1989, after the Tiananmen inci- Government Code Network supports governance. dent, new regulations were applied retroactively Despite these positive features, the country has and became even stricter and more extensive in the room for improvement. The country has not late 1990s. Social organizations--official, semioffi- implemented key constitutional provisions on the cial, and popular--must register and win sponsor- representation of marginalized groups and local ship of a government agency. Only a minority of referenda, and civil society includes armed left- grassroots organizations has been able to register wing groups that threaten national stability. legally. Many are financed by international agen- The official framework for civic participation cies, but support is scarce in less-developed areas. and civil society is relatively strong in Thailand and In the late 1990s, a government campaign to regu- Indonesia. Thai military regimes discouraged larize NGOs required reregistration at the Ministry NGOs, but civil society is now officially considered of Civil Affairs. The number of NGOs fell from important for good governance. Thailand's 1997 180,000 in 1995 to 160,000 in 2000. Constitution requires the state to promote popular According to Vietnam's 1992 Constitution, the participation in preparing policies and plans, mak- Communist Party is the leading organ of the state, ing public decisions, and monitoring the exercise of which includes civil society and mass organiza- state power. The Constitution also enshrines the tions. Economic reform, however, has encouraged right to petition and receive a response from the the development of civil society. The country state, and to peacefully resist unconstitutional enacted a Law on Co-operatives in 1997 and issued attempts to acquire power.The Eighth Development a Grassroots Democracy Decree a year later, estab- Plan (1997­2001) supported the emergence of local lishing a legal framework for citizen participation civil society, but slow progress on decentralization at the commune level. The Law on Complaints and has constrained the development of civil society Denunciations is now under revision to expand organizations. opportunities for citizens to register complaints Neither Indonesia's Constitution nor its recent against the civil service. State-sponsored mass Bill of Rights mentions popular participation. organizations, however, are still the major form of However, the country does have some local tradi- social organization. Representatives of the Women's tion of community consultation. For decades, Union, Youth Union, Farmer's Union, and General repression and state-organized unions weakened Confederation of Labor--whose memberships social participation. Thousands of NGOs have include a large proportion of citizens--sit on sprung up since the late 1980s, but the Internal national and local committees that discuss policies Security Law limited their development. The move- affecting their constituencies. Strictly speaking, ment that overthrew Suharto dissipated without state-sponsored mass organizations are not part of developing into strong civil society organizations, civil society, although they have become somewhat 38 East Asia Decentralizes more like NGOs.A growing number of community- tures and act primarily as agents of the center.Elected based organizations, such as water users' associa- communes have few mandatory functions and tions, medical volunteers, and village development account for only about 2 percent of public expendi- committees, have formed and are enhancing tures, although enabling legislation provides for the Vietnamese civil society. A 2003 decree on NGOs eventual formal transfer of specific functions to recognizes their importance, but some of its provi- them. Other East Asian countries have assigned, sions raise concerns about how freely they can at least in broad legal terms, relatively signifi- function. cant responsibilities to subnational governments, Cambodia's decentralization legislation requires although legal provisions are not always opera- participatory planning at the commune level, tionally defined and implemented. An interesting although the extent to which this is genuine and contrast to Cambodia is Thailand, where the 1999 inclusive varies considerably. Though weak in Decentralization Act calls for the transfer of six many parts of the country, civil society groups major functions to local administrations. Because played an important role in providing community the country has implemented these legal provi- services in the absence of local government. sions only partially, Thai subnational governments Initially hostile to government-related local institu- account for only about 10 percent of public expen- tions, NGOs in Cambodia have since offered ditures, although that figure is expected to grow important expertise and capacity building under sharply. Seila, and some are partnering with new commune The Philippines and Indonesia have gone fur- councils. With little tradition of popular participa- ther in defining and implementing functional tion in local governance beyond religious-based assignments. The 1991 Local Government Code in community development, effectively incorporating the Philippines devolved substantial responsibili- civil society participation in local public sector ties to the various types of local governments, decision making will remain a challenge for the which currently account for about 20 percent of foreseeable future. public spending. They also have some regulatory powers, including land reclassification. Indonesia's Law 32 of 2004 reserves only national defense, for- Fiscal Decentralization eign policy, security, justice, monetary and fiscal This section outlines the fiscal functions of sub- policy, and religion for the center. Local govern- national governments in East Asia, focusing on ments must perform a wide range of obligatory assigned roles and own-source and intergovern- functions under Law 22 of 1999 and Law 32 of mental revenues. (See chapter 6 for more detail on 2004. The province played a smaller role in many own-source revenues, and chapters 1 and 3 for functions under the 1999 legislation, focusing information on subnational borrowing.) mainly on regional coordination and the backstop- ping of underperforming local governments. Many of the provincial functions assigned under Law 32 Distributing Functions among Levels of 2004 are similar to those assigned to local of Government governments, raising concerns about clarity and The distribution of functions among levels of redundancy. Subnational governments account for governments is far from uniform in East Asia, just over 30 percent of total spending, and that fig- with subnational roles ranging from modest to ure is expected to continue growing. dominant (see table 2.9). Although subnational China and Vietnam emphasize the sharing of governments have substantial functions in most responsibilities. China's Budget Law defines a broad countries, incomplete implementation of legal division of functions between central and local authority has resulted in low subnational expendi- governments, but does not disaggregate local cate- ture shares in some cases. The pattern of assign- gories. The result is concurrent assignment and ments also varies across government levels and significant variation across regions. Subnational sectors, and the magnitude of subnational expendi- governments have heavy safety net responsibilities, tures is not clearly related to autonomy. including pensions, unemployment, and social wel- At one extreme is Cambodia, where provinces fare, which are unusual subnational responsibili- account for less than 20 percent of public expendi- ties. The center sets broad expenditure guidelines, The Rules of the Intergovernmental Game in East Asia: Decentralization Frameworks and Processes 39 TABLE 2.9 Subnational Functional Assignments and Expenditure Shares Country Subnational functions (see chapter 5) Subnational share of expenditures Cambodia Provinces dominate subnational service delivery; Around 20% overall; 2% at communes have few mandatory functions, commune level, the rest at but legal provision for eventual transfer of provincial level (2001). more functions. China Broad legal division of responsibility between Around 70% overall; 40% at the levels without disaggregation; in practice, county level (2002). multiple levels perform many functions concurrently. Indonesia Obligatory local functions include health, Around 32% for all levels; expected education, environment, and infrastructure, to increase (2002). among numerous others; provinces were originally assigned mainly coordination and gap-filling roles, but Law 32/2004 increases their role and raises concern about lack of functional clarity. Philippines Substantial functions devolved to subnational Around 20% at subprovincial level governments, particularly health, social (2002). services, environment, agriculture, public works, education, tourism, telecommunications, and housing. Thailand Six broad functions to be devolved to local Around 10% for all levels; expected governments: infrastructure, quality of life, to increase (2001). community and social order, planning and investment and promotion of trade and tourism, management of natural resources and the environment, and culture, values, and local wisdom; slow progress on implementation. Vietnam Main functions remain centralized but different Around 50% for all levels (2003). levels share responsibilities in practice; subnational governments dominate in agriculture, forestry, irrigation, fisheries, power, water, education, and health. Source: Compiled by the author from multiple sources. at least in principle. Subnational governments high, most are centrally defined and/or managed account for around 70 percent of public spending, taxes over which subnational governments have with the county level accounting for more than little control, with the proceeds fully assigned or 40 percent. In Vietnam, intergovernmental respon- shared locally. These are in fact intergovernmental sibilities are more specific on paper, but the center transfers, but are included here with own-source and subnational levels share functions in practice. revenues because of the lack of disaggregated data Still, subnational governments have been playing a to clearly distinguish between the two in some more dominant role in some sectors, including countries. Informal, off-budget revenue is a major agriculture, forestry, irrigation, fisheries, power, issue in some countries (see chapters 1 and 6). water, education, and health. Their share of public China, Vietnam, and Thailand rely primarily on expenditures stands at around 50 percent. shared taxes. As economic reforms proceeded and subnational governments came to dominate public sector revenues, China introduced the recentraliz- Subnational Revenues: Own-Source ing 1994 Tax Sharing System noted above and and Shared further refinements in 2002. Subnational revenue Most East Asian countries have few productive own- includes shared taxes--the relative proportions sources of local revenue (see table 2.10 and chap- of which are sometimes negotiated--and several ter 6). Even where local revenue shares are relatively exclusive subnational taxes. Provinces have nearly 40 East Asia Decentralizes TABLE 2.10 Subnational Revenues Country Own-source revenues (see chapter 6) Shared sources Cambodia Subject to strong central control. Most revenue sharing occurs through Provincial sources: taxes on transportation, line-ministry budget allocations to unused land, markets, business licenses, provinces, and intergovernmental parking, slaughter; fees and charges. transfers to communes (see Commune sources: administrative fees and table 2.11). contributions required for transfer-funded development projects (current); land and property tax and user fees/charges (authority not implemented). China No formal subnational own-source revenues, Value added tax (25% share). except for a limited set of user fees/charges. Income tax on enterprises (40% share). Some national revenues fully shared with Taxes on personal income, natural subnational governments (see next column). resources, nonplan construction, salt, security and exchange (50% share). Taxes on non­VAT-sector businesses, urban maintenance and construction, urban land use, rural markets, vehicle use, property, entertainment; also various business-related taxes (100% share). Taxes shared with provinces, which control sharing to lower levels. Indonesia Subject to some central control. Main revenue sharing is through Provincial sources (substantially shared with formula transfers (see table 2.11) local level): taxes on motor vehicles, fuel, rather than shared taxes. groundwater extraction and use. Selected taxes and state-owned City/district sources (modestly shared with enterprise revenues shared with lower level): taxes on hotels and both provinces and cities and restaurants, entertainment, districts: property, natural resources, advertisement, street lighting, limited and personal income tax. mineral exploitation, parking; limited 32% of provincial and 20% of locally designed taxes under Law No. 34 subprovincial revenue (2002). (2000). Revenue sharing, especially for User fees and charges at both levels. natural resources, expanded under 35% of provincial and 6% of subprovincial Law 34 of 2004 and is not reflected revenue (2002). in above percentages. Philippines Subject to some central control. Central revenue sharing occurs Main sources: taxes on real property, mostly through intergovernmental proceeds from public enterprises, local transfers (see table 2.11). business turnover. National wealth composite (based Other sources: taxes on transfer of real on a set of national revenues property, quarries, amusement; many fees derived from related bases) and and charges. the tobacco excise tax are shared Cities can impose full set of taxes; fewer in with subnational governments. provinces/municipalities. Cities and provinces must share portions of revenues with municipalities and barangays. 30% of subnational revenue (2002). Thailand Largely centrally defined. Value added tax (30% share). Provincial: petroleum sales tax; tobacco sales Natural resources (60% share). tax; hotel tax. Sales, special business, excise taxes Subprovincial: taxes on vehicles, houses and (10% share). land, land development, signboards, slaughter. 54% of subnational revenue (2001). Various permits, licenses, and fees at all levels. 12% of subnational revenue (2001). The Rules of the Intergovernmental Game in East Asia: Decentralization Frameworks and Processes 41 Country Own-source revenues (see chapter 6) Shared sources Vietnam No formal subnational own-source revenues, Taxes on natural resources (except except for a limited set of user fees. petroleum), transfer of land-use rights, Some national revenues are fully shared with agriculture, land and housing, subnational governments (see next column). licenses, state dwelling leases, lottery revenues (100%). Value added tax, taxes on enterprise and personal income, special consumption, remittances, gas and oil fees (partial). Taxes shared with provinces, which control sharing with lower levels. 46% of subnational revenue (2003). Source: Compiled by the author from multiple sources. complete freedom to assign revenue to lower levels, lower levels. Shared taxes account for about 32 per- resulting in a variety of practices across the country. cent of provincial and 20 percent of local income, Strictly speaking, Vietnam has no subnational but Law 33 of 2004 increases tax sharing, particu- taxes. The central government controls tax bases larly on selected natural resource bases. Local and rates completely, and the Department of Tax governments exercise control--within national Administration collects all nontrade revenue. ceilings--only over a limited set of taxes, many of Subnational taxes are either assigned 100 percent to which are holdovers from the prereform era. Both the local level or shared among levels. Under 2002 provinces and local governments collect user reforms, provinces formally receive the proceeds of charges. Law 34 of 2000 allows new local sources, all shared taxes and assign portions to districts and but it led to the adoption of some problematic local communes subject to central standards. Fully and tax and nontax revenues and is slated for revision. partially shared taxes have recently provided around Overall, own-source revenues account for about 46 percent of subnational revenues, in roughly equal 35 percent of provincial and about 6 percent of proportions. local income, although the latter average masks In Thailand, subnational revenues include own- stronger performance in larger urban areas. In the collected taxes and nontax revenues, centrally col- Philippines, only cities may impose the full set of lected taxes, and shared taxes. In 2001, locally col- local taxes, while provinces and municipalities have lected revenues accounted for only 11­12 percent less taxing power. Cities and provinces must often of subnational revenues, while shared revenues share portions of their tax revenues with munici- accounted for about 54 percent, including about palities and barangays. 18 percent from the value added tax. The recent As a newly decentralizing country,Cambodia has Property Tax Act--which combines the land and established few official subnational own-source rev- building tax and the land development tax--could enues. As noted above, the government centralized provide subnational governments with more local revenues after the 1993 elections to impose disci- revenue. pline on provinces. The 1998 Provincial Budget Law Indonesia and the Philippines take a different reinstated limited provincial revenues, but most approach. Both have tax sharing, but they pool a provincial resources continue to flow through cen- high proportion of shared resources into a consoli- tral sectoral budgets. The 2001 Law on Commune/ dated fund allocated by formula as an intergovern- Sangkat Administrative Management grants rev- mental transfer (see below). Indonesia also assigns enue sources to communes, including a land and to provincial and district and city governments a property tax, but the law requires follow-up legisla- share of revenues from selected taxes. Provinces tion. The central government has formally assigned have uniform tax rates and share the revenues with only a few nonproductive fees for civil registration 42 East Asia Decentralizes to communes. Some communes also collect minor which represent a large share of subnational user fees, but these are extralegal as no enabling resources. regulation exists. The new Department of Local The intergovernmental transfer system in China Finance in the Ministry of Economy and Finance is the most complex and least transparent among has made developing commune own-source rev- the countries considered here. During the past two enues a priority. decades, the government has added components designed to address newly recognized problems without removing or altering existing elements. Intergovernmental Transfers Beyond shared taxes (discussed above), which Intergovernmental transfer systems in East Asia account for some 40 percent of transfers, there are a range from substantial to modest in terms of both variety of specific-purpose grants. These include central and subnational government budgets, quota subsidies (left over from an earlier scheme from complex to simple in structure, and from rel- that subsidized deficits on approved expenditures), atively transparent to highly nontransparent (see transfers to offset the impacts of the 1994 Tax table 2.11). In most cases the central government Sharing System, final account subsidies, and a few provides significant intergovernmental transfers, minor programs. Subnational governments rely on TABLE 2.11 Intergovernmental Transfers Country Unconditional transfers Conditional transfers Cambodia Communes receive largely unconditional Provinces receive line-ministry allocations, transfers via formula allocation from not transfers; decentralization law allows Commune/Sangkat Fund. for conditional transfers to communes. China Tax Sharing System (1994) assigns shares Complex accumulation of old and new of certain taxes (table 2.10) to systems; conditional grants account subnational governments' general for more than half of all transfers; revenue, but they are officially subject dominated by social security, wage to some expenditure guidelines. increase, and fiscal stimulus grants. Indonesia Certain taxes shared with lower levels Minor; special-purpose transfers--dana (table 2.10); formula-driven dana alokasi alokasi khusus--under development; umum revenue sharing accounts for 10 percent subnational matching at least 26% of domestic revenues; requirement under Law 33 of 2004. provincial/subprovincial shares based on responsibilities (Law 33 of 2004). Philippines Internal Revenue Allotment shares by Modest categorical grants, including the formula account for 40% of internal Municipal Development Fund, the Local revenues; 23% each to provinces and Government Empowerment Fund, and cities, 34% to municipalities, 20% to the Calamity Fund. barangays; the IRA accounts for 94% of transfers. Thailand Substantial shared tax revenues Specific grants are mostly for capital (table 2.10); "general" transfers for fiscal expenditures, with one type earmarked equalization and other purposes; some, for education and other types less such as the transfer for devolution of restricted, so not heavily conditional; compulsory functions, are not truly some "general" transfers (see left unconditional. column) subject to conditions. Vietnam Certain taxes fully assigned to or shared Before 2002, no conditional grants, only with subnational governments national program budget allocations; (table 2.10); equalization transfer Budget Law (2002) recast these as distributed by formula to jurisdictions specific transfers and provides for more where approved expenditure budgets types of conditional transfers. (based on minimum standards) exceed the sum of shared taxes. Source: Compiled by the author from multiple sources. The Rules of the Intergovernmental Game in East Asia: Decentralization Frameworks and Processes 43 transfers to finance nearly half their budgetary depend on functions. The DAU is formula-driven, expenditures in the aggregate. so the allocation of transfers is more transparent Thailand and Vietnam also have complex trans- than in the past, and the formula attempts to con- fer systems, but they are generally more transparent. sider expenditure needs and revenue capacity. Law Thailand has two main forms of intergovernmental 25 of 1999 and Law 33 of 2004 also provide for transfers besides shared tax revenues. The central special-purpose transfers: the dana alokasi khusus government devotes the bulk of seven types of (DAK). These are expected to be mainly sectoral so-called "general" grants to fiscal equalization, conditional (matching) transfers designed in con- devolution of compulsory functions, and tax sultation with line ministries, but the Ministry of promotion. Grants fulfilling the first goal are allo- Finance has only begun to experiment with DAK on cated by formula, those fulfilling the second goal a small scale. Given the weak devolution of revenue are based on the number of beneficiaries, and those powers, subnational governments rely on transfers fulfilling the third goal are based on past tax per- (exclusive of shared taxes) for more than 65 percent formance. Specific grants--largely discretionary-- of their revenues, with provinces averaging 34 per- are mostly for capital expenditures; one program cent and local governments averaging 74 percent. is earmarked for education, and larger programs The main intergovernmental transfer program in are broader. Subnational governments depend on the Philippines is the Internal Revenue Allotment transfers for about 34 percent of their revenues, (IRA). The Local Government Code requires that not including the 54 percent derived from shared this program share 40 percent of gross national taxes. internal revenues (with a three-year lag), and the Vietnam provides two types of intergov- program accounts for 94 percent of total transfers. ernmental transfers: equalization transfers and Subnational levels share the IRA pool, with 23 per- specific-purpose transfers. In the past, the central cent going to provinces and cities, 34 percent to government negotiated transfers with subnational municipalities, and 20 percent to barangays. A sim- governments mostly to fill budget gaps. As of 2003, ple formula based on population, equal share, and the government distributes the equalization trans- land area allocates the appropriate pool share among fer by formula to jurisdictions whose approved units at each level. Two modest revenue-sharing budgets (based on minimum standards) exceed the schemes--the national wealth share (national rev- sum of "100 percent shared" and partially shared enues derived from certain taxes) and the tobacco taxes. The formula must remain in place for three excise share--and a few small categorical grants also to five years. These reforms have improved the exist. Subnational governments rely heavily on trans- transparency and stability of intergovernmental fers, which account for over 80 percent of provincial transfers. Line ministries have also long used trans- budgetsandaround 70percentof municipalbudgets. fers to support national priority programs, and the Cities are more financially independent, relying on 2002 Budget Law formalizes these resources as con- the IRA for just over 40 percent of their income. ditional transfer programs. As noted above, just Cambodia's intergovernmental transfer pro- under half (46 percent) of the revenues of subna- gram is modest, reflecting its early stage of tional governments come from 100 percent and decentralization. The country launched the Com- partially shared taxes; the other 54 percent takes the mune/Sangkat Fund (CSF) in 2002 to coincide with form of intergovernmental transfers. the first election of commune councils. The CSF Indonesia's Law 25 of 1999 significantly altered relies on both domestic and external sources. The the transfer system. The dana alokasi umum (DAU) central budget contributed 1.2 percent of recurrent combined substantial transfers for local civil service domestic revenue in 2002, and that share grew to wages and the fragmented general program Inpres 2.5 percent in 2004. The Commune/Sangkat Law into a revenue-sharing fund financed by at least requires that the Cambodian government devote a 25 percent of central domestic net revenues, with share of its budget to the CSF, but how to deter- 2.5 percent assigned to provinces and 22.5 percent mine this percentage and ensure that it will grow is to cities and districts. Under Law 33 of 2004, the unclear. CSF transfers are divided into general pool increased to at least 26 percent of domestic administration and local development compo- revenues, and provincial/subprovincial shares now nents, with no less than 70 percent allocated to the 44 East Asia Decentralizes TABLE 2.12 Responsibility for Managing Decentralization Country Institutional responsibility Cambodia The interministerial National Committee to Support the Communes (chaired by the Ministry of Interior) in charge of developing and implementing decentralization involving communes; Council for Administrative Reform in charge of deconcentration involving provinces; generally weak coordination between the two. Single integrated, interministerial process created in 2004 to develop consistent decentralization and deconcentration strategies. China No formal decentralization policy, so intergovernmental issues managed through regular government institutions. Indonesia High-level Regional Autonomy Review Board initially in charge of decentralization policy but process now dominated by the Ministry of Home Affairs, with specific matters formally under the Ministry of Finance or other ministries; generally weak interministerial coordination and some problematic competition. Philippines Interagency oversight committee to monitor implementation of Local Government Code; National Economic and Development Authority, Department of Budget and Management, and Department of Interior and Local Government also play important roles. Thailand National Decentralization Committee, with broad representation from national and subnational governments as well as the nongovernmental arena, charged with making, implementing, and monitoring decentralization policy. Vietnam No formal decentralization-specific body; regular government institutions such as Ministry of Finance and Ministry of Planning and Investment manage reform. Source: Compiled by the author from multiple sources. latter. The central government allocates the general composition and role. All countries experience ten- administration share to communes and sangkats in sions between reformists and defenders of the status proportion to the number of councilors, and the quo, and various political parties and institutions local development share based on a formula with may hold different visions of decentralization. three components: equal share, a share propor- Table 2.12 summarizes arrangements for designing tional to population, and a share proportional to and managing decentralization in the region. relative poverty. Given the weak development of Because decentralization is not a formal policy in local resources, Cambodian communes depend on China, central ministries"manage"decentralization the center for almost 100 percent of their funds. through routine interactions with subnational gov- ernments. Existing institutions similarly manage the decentralization process in Vietnam. The Min- Managing Decentralization Reforms istry of Finance and Ministry of Planning and Chapter 1 highlighted the importance of moving Investment, and their provincial counterparts, are beyond the decentralization structure to consider particularly important. The Philippines has also the decentralization process. This section provides relied on existing national institutions, supported a more in-depth, country-specific look at key by an interagency committee responsible for moni- aspects of this process, focusing on responsibility toring implementation of the Local Government for designing and managing decentralization, the Code. Key players include the National Economic strategy for implementing it, and the capacity- and Development Authority, the Department of building activities designed to support it. Budget and Management, and the Department of Interior and Local Government. Besides cities and a few provinces, associations of local governments Responsibility for the Decentralization Process also play important roles in the Philippines' decen- In some East Asian countries, regular government tralization process. institutions manage decentralization. In others, spe- Indonesia, Thailand, and Cambodia all estab- cial bodies manage the process, but these differ in lished special bodies to guide decentralization. The Rules of the Intergovernmental Game in East Asia: Decentralization Frameworks and Processes 45 Indonesia's Regional Autonomy Review Board Interior, the Ministry of Economy and Finance, and (DPOD), composed of minister-level members, the Ministry of Planning particularly involved. The played a significant role in setting the initial direc- government established the National Committee to tion of decentralization policy. With a basic legal Support the Communes, chaired by the Ministry of and institutional framework for decentralization Interior, to develop and implement decentraliza- now in place, key national ministries have assumed tion to the communes. The Seila task force and principal responsibility for detailing and imple- its secretariat are also helping integrate the donor- menting broad policy parameters. The Ministry of initiated Seila Program with the deconcentrated Home Affairs plays the strongest official role, and and decentralized systems as they develop. The the Ministry of Finance and the national planning Council for Administrative Reform, attached to the agency (Bappenas) provide key inputs in specific Council of Ministers and responsible for overall areas. Sectoral ministries help develop regulations public sector administrative reform, has been the for decentralizing services, but Home Affairs is main player in deconcentration, although the Min- attempting to assert leadership. Weak coordination istry of Interior has recently taken a stronger role in and interministerial competition remain signifi- developing the required legislation. In 2004 the cant problems. Although Home Affairs is techni- Cambodian government established an integrated cally in charge, it has limited authority over other process overseen by an interministerial committee ministries with important decentralization roles. to develop decentralization and deconcentration In Thailand, the National Decentralization policy in an integrated way. Committee serves as the strategic unit for decen- tralization policy. Its members include local gov- Strategies for Implementing Decentralization ernmental and nongovernmental officials as well as representatives of central government. This com- East Asian countries have generally not imple- mittee, like the Indonesian Regional Autonomy mented decentralization strategically or systemati- Review Board, was instrumental in designing cally (see table 2.13). The tendency toward ad hoc decentralization, and is now also charged with approaches is not surprising, given the variety of monitoring and implementing reforms and pro- political rationales for decentralization and the dif- viding policy recommendations to the cabinet. fering nature of regimes in the region. Decentraliza- National agencies, particularly the Ministry of Inte- tion is mostly occurring within highly centralized rior, also engage in the day-to-day management of systems. National agencies often lack serious com- decentralization. mitment to reform, slowing progress even in coun- Responsibility for decentralization policy in tries with reasonable frameworks. The overall Cambodia is fragmented, with the Ministry of the environment is not conducive to well-planned and TABLE 2.13 Decentralization Strategies Country Nature of strategy Cambodia Limited elements of a strategy for commune system, but weakly developed and short term. China No formal strategy; some asymmetric treatment of subnational governments. Indonesia No formal strategy; some attention to key reforms after "Big Bang," such as defining functional assignments more clearly, but approach largely fragmented. Philippines Broad three-stage strategy for implementing Local Government Code, now in last phase; unclear how carefully the country followed the strategy. Thailand Detailed master plan with three phases approved by Parliament in 2002; progress modest (phase one finished in 2004 without meeting key goals). Vietnam Ad hoc strategy in that reform has been slow and controlled; some asymmetric treatment of larger cities through pilot programs. Source: Compiled by the author from multiple sources. 46 East Asia Decentralizes carefully executed implementation, the conse- opment in weaker communes. There is no clear quences of which are highlighted in chapter 1. vision of where the overall system is headed, the Because decentralization in China was essentially plan for fully folding Seila into the formal govern- a by-product of economic reform, it lacks a real ment system is incompletely developed, and, as in strategy. The central government treats various Indonesia, coordination of the key actors is inade- types of subnational governments differently, but quate. The new interministerial effort noted above there is not a developed asymmetric decen- is intended to provide direction and facilitate coor- tralization strategy. Because of the nature of the dination, but how successful it will be is unclear. administrative hierarchy and the diversity of the Though decentralization in the Philippines also country, however, provinces make decisions about emerged from crisis, the country did attempt to subprovincial roles, perhaps in some cases reflecting develop a strategy, at least on paper. A Master Plan strategic attempts to improve subprovincial per- for the Sustained Implementation of the 1991 Local formance. The lack of an overall strategy is also Government Code (1993­98) provided the blue- manifested in ad hoc central government steps to print for reform. The plan included three phases. define the intergovernmental system, resulting in Phase one (1992­93) involved the transfer of func- disjointed revenue and expenditure assignments tions, which varied by type of local government. and an inconsistent intergovernmental transfer Phase two (1994­96) gave local governments time system. to adjust to their formal responsibilities. Phase Indonesia and Cambodia also have weak imple- three (1997 onward) expected a more stable system mentation strategies. Decentralization was adopted to focus on building local capacity, with technical quickly and with little debate in both cases. In assistance from national agencies. The interagency Indonesia this occurred in a crisis situation, while oversight committee noted above was charged with Cambodia's efforts might be characterized as a monitoring implementation. The extent to which case of political opportunism. Because Indonesia's this phasing was followed is not clear, and decen- reforms emerged from crisis, the general frame- tralization continues to face political difficulties, work was pushed urgently, without much thought instability in some regions, and limited resources. given to how to make it work. The country's decen- Some central agencies have held on to functions tralization is often referred to as a "Big Bang" they were supposed to devolve, and development of because significant resources and functions were local revenue has been slow. devolved so quickly. A substantial portion of the Thailand and Vietnam are closest to having a resources, however, are used to pay for staff who decentralization strategy, but both have imple- were transferred to local governments, suggesting a mented it slowly. In 1997, after the new Thai Con- possible conscious effort to reduce the effects of stitution mandated decentralization, a Local Fiscal major reform shocks. Operational details on many Master Plan identified 17 measures to enhance of the legally devolved functions still have to be local revenues, clarify responsibility for expendi- specified, and the above-noted weak coordination ture, reform the intergovernmental transfer system, of the national agencies involved constrains the establish mechanisms for monitoring local fiscal development of a genuine strategy. systems, promote new methods of mobilizing capi- Cambodia has limited elements of a strategy. As tal for local investment, and develop local capacity. many of the newly elected communes had little or Parliament did not approve a more comprehensive no capacity or political credibility, the early design plan to decentralize administrative power to local included initially modest functional expectations, administrations until 2002. The plan includes a simple structure and staffing of councils, and clas- general framework, objectives, and guidelines for sification of communes into two categories based decentralizing administrative power in three stages. on capacity, with differential funding awarded on During the first stage (2001­4), the country was that basis. As the system matures, communes are to supposed to transfer 245 tasks, improve local and be assigned greater responsibilities. Unfortunately, regional administrative systems, eliminate overlap- there is no strategy for making further progress. ping functions, and strengthen local capacity to The classification system has been abolished with- manage functions, personnel, revenue, and assets. out any assessment of the extent of capacity devel- Unfortunately, these goals were not fully achieved, The Rules of the Intergovernmental Game in East Asia: Decentralization Frameworks and Processes 47 the coordinating National Decentralization Com- national staff can meet their shifting obligations mittee (NDC) is understaffed and inadequately and to realign relationships in the way required. financed, and the strength of political support is Local level capacity building under decentraliza- unclear. Thus, the NDC has not been able to opera- tion is generally of two types. The first type involves tionalize what appears to be an atypically carefully training to ensure that subnational staff can per- conceived implementation strategy. form their technical functions. The second type InVietnam,the movement to give greater respon- involves building governance mechanisms that are sibility to subnational governments has progressed required for a decentralized system to work effec- in stages.Expenditure and revenue assignments have tively. Either of these two types can be supply- been changed gradually, and the transparency and driven (by the central government) or demand- stability of the intergovernmental transfer system driven (by subnational governments). The latter have improved. The central government has also type is considered good practice based on the gradually introduced autonomy measures,removing recognition that a lack of demand for reforms and the requirement that the National Assembly approve the capacity needed to make them work under- provincial budgets, and giving provinces more mines their chances of being realized and sus- authority over lower levels. Pilot programs have also tained. Most countries in the region have focused accorded some urban areas greater autonomy. How- on traditional supply-driven technical capacity ever,these elements may reflect the conservatism of a building, and most governance training has also highly centralized government in a one-party state been supply-driven (table 2.14). more than a strategic effort to decentralize. Cambodia faces the greatest challenges. Capacity is weak in many communes, which had no real functions before the first elections in 2002, even as Building Subnational Capacity part of provincial administrations. Regions that As chapter 1 notes, capacity building is an impor- participated in the Seila Program developed rea- tant part of any decentralization strategy. This is sonable capacity, but other communes generally lag true not only for subnational governments but also far behind. Even Seila communes must adjust to for central agencies, which must learn new ways of new systems and procedures under official decen- doing business and new skills in developing local tralization policy, posing significant challenges. The systems, strengthening subnational actors, and Ministry of Interior and the Ministry of Economy monitoring the implementation of decentraliza- and Finance have conducted most commune-level tion. Although most East Asian countries have training, and technical staff at the provincial level relied on technical assistance and provided training help communes implement new procedures. This to central employees involved in decentralization, training, however, has barely laid a foundation, and they have paid limited attention to ensuring that much work remains at both the commune and TABLE 2.14 Building Decentralization Capacity Country Capacity-building provisions Cambodia Massive, basic program run by central government for developing commune system. China No specific decentralization-related training; most is organized at subnational level; some temporary posting of higher-level staff to lower levels. Indonesia Significant transfer of staff to lower levels; much capacity building driven by central government, although some demand from lower levels. Philippines Significant transfer of staff to lower levels; subnational governments responsible for training, which typically focuses on councilors rather than civil servants. Thailand Some transfer of staff to lower levels; subnational capacity building driven by national agencies. Vietnam Training programs for subnational staff driven by central government. Source: Compiled by the author from multiple sources. 48 East Asia Decentralizes higher levels. Deepening and institutionalizing diversity in the paths it has taken, the enabling capacity takes time, and the prominent role of frameworks that define it, and the ways in which it is donors in providing and financing decentralization structured. In Indonesia and the Philippines, the capacity-building efforts in Cambodia raises con- focus quickly shifted from deconcentration to cerns about sustainability. decentralization, and this is happening in Thailand Subnational governments elsewhere in the as well. The focus remains on deconcentration in region have greater capacity, but the levels and China and Vietnam, but there are emerging ele- mechanisms for building it vary. In China, where ments of delegation and devolution. Cambodia decentralization is not official policy, capacity makes a stark division between deconcentration and building has focused on improving overall govern- devolution. Within their basic policy thrust, all ment performance. All ministries and departments countries have multiple tiers of government or receive an annual training budget, and most train- administration, but exact forms and responsibilities ing is organized locally. The central government vary. In Thailand, most levels have a role, while the also posts its own mid-career staff to subnational focus has been on provinces and urban governments administrations for six months to a year, which in China and Vietnam. In Cambodia, Indonesia, and may boost the capacity of those governments. the Philippines, subprovincial units have been the Although Vietnam has also long been heavily cen- main targets of decentralization reforms. tralized, its decentralization program is more for- Decentralization enabling frameworks differ sig- mal. Subnational capacity is generally strongest in nificantly. The Philippines and Thailand have robust provinces and larger cities. Staff from higher levels constitutional and legal foundations, and Indonesia of government provide most training of subna- adopted constitutional reforms to institutionalize tional staff. Neither China nor Vietnam has made decentralization. Indonesia and Vietnam have legal training of citizens to interact with local govern- frameworks, although not fully developed. Weaker ments a priority. laws underpin the system in Cambodia, but addi- In Indonesia, the Philippines, and Thailand, local tional legislation is in process. China has the weakest governments have benefited from the transfer of framework, with only a few laws that refer to subna- higher-level staff. Indonesia has decentralized more tional roles. The nature and degree of development than 2 million officials since 1999. The Philippines of the enabling framework do not seem to affect transferred some 70,000 employees from central decentralization progress or quality, but all coun- ministries to local governments following the Local tries eventually need to define the roles of relevant Government Code of 1991. Given its more incre- actors in a framework that protects their rights and mental approach to decentralization, Thailand has provides a basis for accountability. made more limited transfers involving about 4,000 Although far from complete, governance is central employees so far. In all these cases, the help improving in the region through political, institu- local governments receive is not always the type they tional, and fiscal decentralization reforms. All need. Capacity levels vary widely within these coun- countries considered here have subnational elec- tries and are generally higher in urban areas, with tions, ranging from Vietnam at all levels to Cambo- significant gaps in smaller urban and rural areas. In dia at one level. Subnational councils usually have Indonesia and Thailand, higher-level agencies drive an element of direct election, but in some cases local capacity building. In the Philippines, local there are higher-level appointments or party list governments are formally responsible for building voting. In the Philippines, council leadership is capacity, but training focuses on councilors rather directly elected, and this will soon occur in Indone- than staff, and the central government, international sia and Thailand. Political competition varies from agencies, and NGOs often assist with the provision of one-party states in China and Vietnam to the such training. nearly chaotic multiparty system of Indonesia. Cambodia has multiple parties, but one dominates, while political competition is somewhat more Summary and Conclusions robust in the Philippines and Thailand. The various historical and political roots of decen- Autonomy differs in complex ways across coun- tralization in East Asia are reflected in the observed tries. Thailand and Vietnam have higher-level The Rules of the Intergovernmental Game in East Asia: Decentralization Frameworks and Processes 49 controls over subnational budgeting and personnel both design and implementation. In no case are decisions, although they are not always exercised coordination or enforcement of decentralization and are being relaxed. Philippine subnational gov- activities adequate. ernments are more independent in principle, if not Many decentralization problems result from the always in practice, while new laws curtail autonomy lack of an implementation strategy. Given China's in Indonesia. On transparency, the Philippines and context, the conspicuous lack of strategy is under- Thailand provide extensive public access to subna- standable. Both Indonesia and Cambodia are tional budgets and minimally acceptable auditing, struggling to deal with the consequences of poor while China and Vietnam do not. Incorporating planning prior to rapid, politically driven decentral- citizen input into local decisions is a priority only ization. The Philippines, Thailand, and Vietnam in a few cases, and civil society varies, from rela- have more considered implementation strategies, tively free and robust in the Philippines to heavily with articulated phases and timelines. In the controlled in China and Vietnam. Philippines, the final phase of the 1991 decentraliza- Expenditure and revenue assignments vary con- tion effort is in process, but the phasing does not siderably. In most cases there is some formal seem to have been systematically followed,and prob- assignment, but details typically need clarification. lems persist. Vietnam, out of an abundance of cau- In the Philippines and Indonesia, the process is tion expected in a centralized one-party system, has substantially or moderately advanced. Cambodia moved in an atypically structured way. Thailand has a very long way to go, and China has no clear has a well-articulated strategy, but implementation interest in formalizing assignments. Own-source has been slow. None of these countries has a clear revenues are generally weak, so subnational govern- strategy in the sense of using graduated, asymmetric ments rely heavily on shared taxes and intergovern- functional assignments and capacity-building sup- mental transfers. In China, Thailand, and Vietnam, port consistent with the performance of individual national taxes are shared on a tax-by-tax basis, local governments. often by origin. In Indonesia and the Philippines, Political factors elaborated in chapter 1 compli- most shared revenues are pooled into a consoli- cate the development of decentralization frame- dated fund for distribution as a formula-based works and strategies, and the pitfalls of trying to be transfer. Cambodia has little formal tax sharing; too normatively comprehensive in designing and provinces are funded primarily through national implementing decentralization are well known. It is sectoral budgets, and the commune transfer pool is also clear, however, that the costs of ignoring the determined on an ad hoc basis. Transfers vary problems that have often emerged as decentraliza- widely in significance, structure, complexity, and tion has unfolded in East Asia are potentially very transparency. China's system is highly complex and high. With a basic decentralization vision and a nontransparent. Other countries have somewhat degree of leadership, East Asian countries should simpler and more transparent systems, especially find it possible to accommodate political realities for general revenue sharing, but there are com- and strategically use opportunities to build more monly less transparent conditional transfers, often effective decentralization frameworks, structures, for capital expenditures. processes, and outcomes. This is the significant Although often neglected, the institutional challenge facing all countries in the region. structures and processes for defining and managing decentralization are critical for success. In most East Asian countries, decentralization is primarily Endnote overseen by regular government agencies. Special 1. The information in this chapter is largely derived from decentralization bodies exist in some countries, but country reviews prepared as background papers for this their composition and role differs. Indonesia has volume. These and other key country-specific citations are included in the references at the end of the chapter. David a high-level policy body that influences major Gomez Alvarez supplied considerable research assistance design decisions. An interministerial body in the for this chapter. Some data and clarifications on particular Philippines has primary responsibility only for countries were provided by Robert Ebel, Amanda Green, Bert Hofman, Kai Kaiser, Blane Lewis, Ed Mountfield, monitoring implementation. Interministerial bod- Amitabha Mukherjee, Duvvuri Subbarao, Rob Taliercio, ies in Cambodia and Thailand play broader roles in Dana Weist, Roland White, and Christine Wong. 50 East Asia Decentralizes Bibliography McLure, Charles E., Jr., and Jorge Martinez-Vazquez. 1998. "Intergovernmental Fiscal Relations in Vietnam." Working Alm, James, and Roy Bahl. 1999. "Decentralization in Indonesia: Paper 98­2. Atlanta, GA: Georgia State University, Andrew Prospects and Problems." Atlanta, GA: Georgia State Young School of Policy Studies. isp-ysps.gsu.edu/papers/ University, Andrew Young School of Policy Studies. ispwp9802.pdf. www.gsu.edu/~wwwsps/publications/1999/990601_ Nghieu, B.D. 2003."EAP Flagship Country Review of Decentral- IndonesiaReport.pdf. ization in Vietnam." Washington, DC: World Bank. Alm, James, Jorge Martínez-Vázquez, and Sri Mulyani Rasyid, M. Ryaas. 2004. "The Policy of Decentralization in Indrawati, eds. 2004. Reforming Intergovernmental Fiscal Indonesia." In Reforming Intergovernmental Fiscal Relations Relations and the Rebuilding of Indonesia. Cheltenham, UK, and the Rebuilding of Indonesia. Ed. James Alm, Jorge and Northampton, MA: Edward Elgar Press. Martínez-Vázquez, and Sri Mulyani, Indrawati. Cheltenham, Bird, Richard, Jennie Litvack, and M. Govinda Rao. 1995."Inter- UK, and Northampton, MA: Edward Elgar Press. governmental Fiscal Relations and Poverty Alleviation in Rocamora,Joel.2003."Legal and Policy Framework for`Participa- Vietnam." Washington, DC: World Bank. tion' in Southeast Asia." www.ids.ac.uk/logolink/resources/ Brilliantes, Alex. 2003."EAP Flagship Country Review of Decen- downloads/regionalreports/RegionalReportSouthEast tralization in the Philippines."Washington, DC: World Bank. Asia%20final.pdf. Hofman, Bert, and Kai Kaiser. 2004. "The Making of the Big Sidik, Machfud, and Kadjatmiko. 2004."Combining Expenditure Bang and Its Aftermath: A Political Economy Perspective." In Assignment, Revenue Assignment and Grant Design in Reforming Intergovernmental Fiscal Relations and the Indonesia's Fiscal Decentralization." In Reforming Intergov- Rebuilding of Indonesia. Ed. James Alm, Jorge Martínez- ernmental Fiscal Relations and the Rebuilding of Indonesia. Vázquez, and Sri Mulyani, Indrawati. Cheltenham, UK, and Ed. James Alm, Jorge Martínez-Vázquez, and Sri Mulyani, Northampton, MA: Edward Elgar Press. Indrawati. Cheltenham, UK, and Northampton, MA: Hofman, Bert, and Kai Kaiser. Forthcoming. "Decentralization, Edward Elgar Press. Democratic Transition, and Local Governance in Indonesia." Smoke, Paul. 2002. "Decentralization and Deconcentration in In Decentralization in Developing Countries: A Comparative Cambodia." Background paper prepared for Cambodia Perspective, eds. D. Mookkerjee and P. Bardham. Integrated Fiduciary Assessment and Public Expenditure Hutchcroft, Paul D. 2003. "State Formation, State Reformation: Review. Washington, DC: World Bank. Deciphering Decentralization in the Philippines and Smoke, Paul. 2004. "Expenditure Assignment under Indonesia's Thailand." Working Paper. Madison, WI: University of Decentralization: A Review of Progress and Issues for the Wisconsin, Department of Political Science. Future." In Reforming Intergovernmental Fiscal Relations and Iszatt, Nina T. n.d."Legislating for Citizens' Participation in Local the Rebuilding of Indonesia. Ed. James Alm, Jorge Martínez- Governance: The Philippines Experience." www.ipd.ph/ Vázquez, and Sri Mulyani, Indrawati. Cheltenham, UK, and logolinksea/resources/Frameworks%20Phil%20experience. Northampton, MA: Edward Elgar Press. pdf. Smoke, Paul, and Yun-Hwan Kim. 2002. Intergovernmental King, Dwight Y. 2004. "Political Reforms, Decentralization, and Fiscal Transfers in Asia: Current Practice and Challenges for Democratic Consolidation in Indonesia." In Reforming Inter- the Future. Manila: Asian Development Bank. governmental Fiscal Relations and the Rebuilding of Indonesia. Suwanmala, Charas. 2003. "EAP Flagship Country Review of Ed. James Alm, Jorge Martínez-Vázquez, and Sri Mulyani, Decentralization in Thailand."Washington, DC: World Bank. Indrawati. Cheltenham, UK, and Northampton, MA: Wong, Christine. 2003. "EAP Flagship Country Review of Edward Elgar Press. Decentralization in the People's Republic of China." Lewis, Blane D. 2002. "Indonesia." In Intergovernmental Fiscal Washington, DC: World Bank. Transfers in Asia: Current Practice and Challenges for the Wong, Susan, and Scott Guggenheim. 2003. "Community- Future, eds. Paul Smoke and Yun-Hwan Kim. Manila: Asian Driven Development and Decentralization in the East Asia Development Bank. Region." Washington, DC: World Bank. Lewis, Blane D. 2003a."Property Taxation in Indonesia: Measur- World Bank. 2000a. "Thailand Public Finance Review: Public ing and Explaining Administrative (Under-) Performance." Finance in Transition." Washington, DC: World Bank. Public Administration and Development 23 (3): 227­39. ------. 2000b. "Vietnam Public Expenditure Review: Manag- ------. 2003b. "Tax and Charge Creation by Regional Gov- ing Public Resources Better." Washington, DC: World ernments under Fiscal Decentralization: Estimates and Bank. Explanations." Bulletin of Indonesian Economic Studies ------. 2001. "The Filipino Report Card on Pro-Poor Services." 39 (2): 177­97. Process case studies. www.worldbank.org/participation/ Lewis, Blane D. 2004. "Indonesian Local Government Spending, Filipino%20Reportcard.htm. Taxing and Saving: An Explanation of Pre- and Post- ------. 2002a. "China Sub-National Expenditure Review." Decentralization Outcomes." Jakarta: World Bank. Washington, DC: World Bank. Lewis, Blane D., and Jasmin Chakeri. 2004. "Decentralized Local ------. 2002b. "Implementing Decentralization in Thailand: Government Budgets in Indonesia." Jakarta: World Bank. The Way Forward." Washington, DC: World Bank. Liang, Sharon. 2003."Walking the Tightrope: Civil Society Orga- ------. 2002c. "Vietnam: Delivering on Its Promise." nizations in China." China Rights Forum 3. iso.hrichina.org/ Washington, DC: World Bank. www.worldbank.org.vn/ download_repository/2/Sharon%20Liang.pdf. topic/vdr2003.htm. Logo Link Southeast Asia. n.d. "Learning Initiative on Citizen ------. 2003a. "Cambodia Integrated Fiduciary Assessment Participation and Local Governance." www.ipd.ph/ and Public Expenditure Review." Washington, DC: World logolinksea/index.html. Bank. Martínez-Vázquez, Jorge, and James Alm, eds. Forthcoming. ------. 2003b. "Cambodia Poverty Reduction Strategy Paper." Reforming Intergovernmental Fiscal Relations and the Washington, DC: World Bank. Rebuilding of Indonesia. Cheltenham, UK, and ------. 2003c. "Cambodia Rural Investment and Local Northampton, MA: Edward Elgar Press. Governance Project." Washington, DC: World Bank. The Rules of the Intergovernmental Game in East Asia: Decentralization Frameworks and Processes 51 ------. 2003d. "Decentralizing Indonesia: Regional Public ------. n.d. "Participation and Civic Engagement in Poverty Expenditure Review." Washington, DC: World Bank. Reduction Strategy. Washington, DC: World Bank. ------. 2003e. "Indonesia Public Expenditure Review: Fiscal www.worldbank.org/participation/PRSP/alp.htm. Challenges in a New Era." Washington, DC: World Bank. ------. n.d. "Vietnam Process Case Study." Washington, DC: ------. 2003f. "Philippines: Decentralization and Service World Bank. www.worldbank.org/participation/web/ Delivery: From Promise to Performance." Washington, DC: webfiles/vietnam.htm. World Bank. Ye, Zhang. 2003. "China's Emerging Civil Society." Washington, ------. 2003g. "Philippines: Improving Government DC: Brookings Institution. www.brook.edu/fp/cnaps/papers/ Performance--Discipline, Efficiency, and Equity in Manag- ye2003.htm. ing Public Resources." Washington, DC: World Bank. 3 Achieving Fiscal Sustainability Sudarshan Gooptu A well-designed system for fiscal decentralization and subnational levels will yield to uncertain inter- promises many benefits, including enhanced eco- governmental flows, capricious revenue policies, nomic growth and macroeconomic stability. If, and a lack of medium-term fiscal planning. however, decentralization is designed badly or Much is at stake in getting this decentralization implemented and monitored improperly, such a sys- system right. East Asian countries have recovered tem can foster "soft" budget constraints--loosened rapidly from the 1997 financial crisis, posting high controls on the fiscal activities of subnational growth rates. Regional output grew by 6.7 percent governments--that may undercut the spectacular in 2002 and 7.9 percent in 2003 and an estimated economic growth occurring in East Asia today. 7.8 percent in 2004. Growth is expected to reach This chapter highlights the challenges of decen- 7.1 percent in 2005--the strongest record since the tralization in the context of macroeconomic man- start of the global and regional recession in late agement, especially in achieving fiscal sustainability 2000. The number of people living below US$2 per and providing a medium-term environment for day is estimated to have fallen to around 34 percent sustained growth. The focus is twofold. First, the in 2004, amounting to some 636 million people-- chapter highlights the critical need for central gov- down from 50 percent as recently as 1999, repre- ernments to monitor subnational fiscal and quasi- senting 890 million people (World Bank 2004, 2005 fiscal activities through well-designed reporting, table 1.1). auditing, and financial management institutions, as The policy challenge for each country is to well as information databases. Second, the chapter maintain its high growth rate while strengthening addresses the need to institutionalize incentives for its fiscal health through intergovernmental coordi- subnational governments to publicly and regularly nation and monitoring. Unfortunately, this task report relevant financial information. is far from straightforward. For a start, whether Most East Asian countries are at the early stages developing countries experience a causal relation- of this twofold process. Unless the process keeps ship between decentralization, macroeconomic sta- pace with decentralization, the result will be subna- bility, and economic growth is unclear. The reasons tional arrears and pressures for central bailouts--the are twofold. First, the data for measuring the extent latter occurring through ad hoc financial transfers, to which a country has fiscally decentralized are or accounting"adjustments"that take significant tax simply not available--a worldwide problem that and spending activity off-budget. Once countries also exists in East Asia. Second, the line is not abandon the discipline of hard budget constraints, always clear between formal and informal institu- systematic budget planning and execution at central tional arrangements for enforcing hard budget 53 54 East Asia Decentralizes constraints. Even if the right data and definition time frame for assessing the impact of different were available to measure decentralization, track- types of public spending is also important. Analysts ing fiscal risks can be difficult when subnational see health and education spending, for example, as governments have hidden, off-budget expenses and an investment in human capital with a long-term funds, as in some East Asian countries. payoff, resembling physical investment.4 Still, the elements of responsible fiscal manage- Ebel and Yilmaz (2003) reproduced this analysis ment, and the criteria for maintaining fiscal disci- by weighing degrees of revenue autonomy. They pline in a decentralizing environment, are well found that tax autonomy and nontax autonomy have known. This chapter therefore focuses on empirical a positive correlation with economic growth, while conclusions regarding the links between fiscal tax sharing has a negative relationship (see chap- decentralization and growth, the role of budget ter 6).Theseresultssuggestthatacountry'seconomic constraints in fostering fiscal discipline, the rela- performance partly reflects the degree to which sub- tionship between subnational borrowing and a national governments control their revenues. country's overall fiscal sustainability, and the result- In a study of the United States, Xie et al. (1999) ing need for managing financial risk. showed that further fiscal decentralization with the aim of boosting efficiency and economic growth would in fact harm growth. Akai and Sakata (2002) Links between Fiscal refuted that finding, incorporating a more elabo- Decentralization and rate definition of fiscal decentralization. Their Economic Growth study examined the growth impact of giving sub- In theory, devolving fiscal responsibilities can national governments the authority to raise taxes improve public sector efficiency, boost competition and spend public funds.5 Since local governments among subnational governments in delivering pub- do not necessarily spend locally collected revenues lic services, and stimulate economic growth (Bird locally, the authors found that they may not in fact and Wallich 1993). These potential benefits reflect have tax autonomy (Zhang and Zou 2001). the belief that subnational governments can best Martinez-Vazquez and McNab (2001) con- make growth-promoting public expenditures in firmed the inconclusive relationship between fiscal sectors such as education, health, and infrastruc- decentralization and economic growth and the ture, because they have better information on local importance of individual country circumstances-- circumstances and interregional differences (Oates including noneconomic ones--in determining 1993).1 causality. Lin and Liu's analysis of China (2000), East Asia's spectacular precrisis growth and using provincial-level data from 1970 to 1993, sug- postcrisis rebound have depended significantly on gests that fiscal decentralization has helped spur prudent macroeconomic management. However, that country's impressive growth over the past 20 or empirical research has been inconclusive regarding 30 years.6 However, Zhang and Zou (2001) found a any causal relationship between decentralization negative relationship between fiscal decentraliza- and growth in developing countries. tion and economic growth in China.7 In a study of 46 countries from 1970 to 1989, Davoodi and Zou (1998) found a negative relation- Subnational Fiscal Discipline ship between fiscal decentralization and economic and Budget Constraints growth in developing countries, and no relation- ship in developed countries.2 The authors explain Determining the degree of fiscal discipline among that "conventional wisdom points towards positive subnational governments in East Asia requires growth effects of capital and infrastructure spend- identifying the budget constraints they face. Doing ing and negative growth effects of welfare and cur- so in countries like Cambodia--where subnational rent spending." This implies that in developing governments have little spending and revenue countries--where subnational entities spend a authority and rely almost entirely on transfers from larger proportion of their budgets on wages and the center--is relatively easy. At the other extreme, salaries and social welfare--decentralized systems about half or more of the resources available exhibit slower long-term growth per capita.3 The to China's subnational governments come from Achieving Fiscal Sustainability 55 off-budget sources, over which they have virtually Information on these quasifiscal transactions total control.8 In Vietnam, off-budget accounts remains weak, and work on improving this infor- such as the Social Security Fund, the Health Insur- mation base is just beginning, even in China. ance Fund, and the National Development Fund Finally, subnational governments in many coun- constrain the budget autonomy of subnational tries have faced difficulties in planning their activi- governments. ties and managing their finances owing to delays in In some East Asian countries, state-owned allocation decisions and a lack of predictability enterprises at the subnational level receive off- regarding intergovernmental transfers. In Thai- budget resources in the form of deferred taxes or land, other than knowing with certainty that their arrears accruals on debt service and other contrac- nongrant revenues will equal those of the previous tual payments. This is a particular concern in China fiscal year, local governments find it difficult to and Vietnam. In Thailand, this problem is miti- predict their shared tax revenue allocations. This is gated by the fact that subnational governments do mainly due to delays in establishing the criteria for not typically own enterprises, and by the inclusion distributing the allocations from the center to local of all state-owned enterprises in the "consolidated governments.10 public sector's" budgetary accounts. Extrabudgetary revenues and expenditures Subnational Borrowing and among subnational authorities make their true Fiscal Sustainability budget constraints difficult to ascertain. This is especially problematic when data are not reported Fiscal sustainability means that a subnational gov- in a timely manner and are often incomplete, as in ernment covers its expenditures out of its own rev- most East Asian countries. Thus, when govern- enues, reducing its dependence on borrowing and ments rely substantially on extrabudgetary funds, transfers from the center (Bird 2003). To determine national budgets give only a partial picture of fiscal whether a subnational government's plans are fis- realities. cally sustainable, analysts need accurate informa- Enforcing hard budget constraints among subna- tion on revenues and expenditures at the central, tional governments requires clear expenditure regional, and local levels. Analysts also need to assignments, formula-based transfer systems, local understand the interplay between intergovernmen- revenues, prudent subnational borrowing rules, and tal grants and government borrowing--that is, how good financial reporting (see box 1.3 in chapter 1). hard the subnational budget constraint truly is. Noneof theEastAsiancountriesexaminedheremeet Careful analysis of country-specific intergovern- these prerequisites. Such a situation can lead to per- mental relations, and the resulting incentive frame- verse incentives to overspend, accumulate arrears, work, should accompany any analysis of fiscal and overborrow. Key among such incentives is the sustainability. prevailing practice of higher-level financial bailouts Subnational governments in most of the East for subnational governments that are already in or Asian countries examined here rely heavily on even heading toward default. This is of particular intergovernmental transfers. The large bailouts concern if the subnational government is large, as is during the 1997 financial crisis and since, such as often the case with soft budget constraints.9 the recent recapitalization of Chinese state-owned A country's system of intergovernmental fiscal banks in 2004, have undermined fiscal sustainabil- management (or lack thereof) may also motivate ity in East Asia by softening the budget constraints subnational authorities to keep their transactions imposed on subnational governments. Discre- off-budget--especially richer provinces that do not tionary transfers to deficit subnational govern- want to cede part of their revenue base to the center ments in China, Indonesia, and Vietnam, and for tax sharing. Effective fiscal decentralization unclear assignment of responsibilities throughout requires an institutional structure that minimizes the region, have compounded this result. such adverse expectations. In China, "fee-to-tax" Many subnational governments in East Asia also reforms aim to bring some off-budget subnational have access to onlending from the central govern- revenues within the budgetary umbrella, but much ment and donors. Such lending is typically in the remains to be done in most East Asian countries. form of sovereign guaranteed external loans that 56 East Asia Decentralizes the central government contracts but channels to borrowing, subnational governments have incen- subnational entities at different interest rates. tives to run unsustainable deficits. Onlending interest rates and terms vary across Fiscal, political, and financial institutions that countries. In China, the final borrower bears the strengthen competition at the local level, especially entire foreign-exchange risk associated with these for capital, can promote hard budget constraints.13 loans, while in other countries the center assumes If institutions directly or indirectly suggest that the some or all of this risk. Most subnational borrow- central government will step in to cover subna- ing in Indonesia has occurred through central gov- tional liabilities in the case of default, they may ernment onlending mechanisms on terms that are encourage subnational governments to "overbor- highly favorable to the center. However, the repay- row, overspend, or undertax" (Rodden 2000a). ment record of Indonesian subnational borrowers The Latin American experience suggests that has been poor.11 Thailand allows subnational gov- subnational governments that are subject to hard ernments to borrow from domestic banks and budget constraints are more likely to tax and spend bond markets, subject to legal ceilings in any given prudently (Bird 2001). Fiscal sustainability is also a budget year. In Indonesia, the Ministry of Finance forward-looking concept, in that it requires accu- sets the aggregate limits to regional borrowing for a rate assumptions about revenues and expenditures particular fiscal year in August of the previous year, and key economic variables. Making meaningful and no direct borrowing from foreign sources is assumptions that are palatable to policy makers and permitted. Instead, all such borrowing occurs their constituents requires a good understanding of through onlending arrangements with the min- institutional and country-specific details. istry. Subnational governments in Vietnam may not In Indonesia, for instance, only after it passed borrow at all (see table 3.1). Laws 22 and 25 on intergovernmental fiscal rela- International experience since the early 1980s, tions in 1999 and issued implementing regulations especially in Latin America, suggests that without (PP107) in 2000 could regions borrow without appropriate accountability and transparency mech- strict approval from the center.14 Subnational anisms, decentralization can encourage dangerous domestic borrowing is now subject to a rule-based opportunistic behavior by state and local authori- approach and central government approval. For- ties. If left unchecked, such opportunism could eign borrowing is not allowed, except through undermine macroeconomic stability. The most onlending from the central government or donors. vivid manifestation of this phenomenon is the soft- In view of Indonesia's public debt burden, how- ening of subnational budget constraints (Rodden ever, and as the country establishes local financial 2000a; World Bank 2002). Avoiding this risk management structures, a ministerial decree tem- depends on the ability of the central government to porarily banned any subnational borrowing until prevent subnational authorities from passing their the intergovernmental fiscal relations Law 25/99 liabilities to higher-level governments.12 This, in was revised. This ban has been in effect since 2000 turn, requires institutional mechanisms to disci- but was to be lifted at the end of fiscal year 2004, pline borrowing by state and local governments. after the implementation regulations of the revised Examining experiences in Argentina, Brazil, and Law 25 are in place. The goal is to maintain fiscal India, Rodden et al. (2003) show that unsustainable discipline by strengthening both market and rule- subnational deficits emerge when provinces have based mechanisms--highlighting the fact that powerful representatives, when they depend heavily countries may sometimes require strong, centrally on intergovernmental transfers, and when they imposed fiscal constraints, especially in the early have autonomous access to sources of deficit stages of decentralization. Subnational govern- financing. The latter can include bonds, loans from ments may rely on short-term borrowing (with domestic banks (which may themselves be state- maturities of less than 12 months) to manage their owned, as in China), nonpayment of employee cash flow. In theory, as in China, the Indonesian wages, and contingent liabilities. If these exist central government can intercept general grant where the central government cannot commit to funds if a region fails to meet its debt service a no-bailout policy, or cannot limit subnational obligations. TABLE 3.1 Subnational Borrowing Practices Summary of practices Cambodia China Indonesia Philippines Thailand Vietnam Borrowing power LGs cannot borrow. LGs may not borrow LGs can borrow or LGs can borrow or LGs cannot borrow, Provinces and cities and practices Stipulated according against general issue bonds in issue bonds; some domestically or with provincial (excluding bor- to central govern- revenues or issue domestic market prudential from abroad, status can borrow rowing from cen- ment regulations. bonds. subject to rules restrictions. without prior within prudential tral government) LGs can and do and central gov- approval limits and as borrow through ernment approval. from CG. approved by CG. asset-holding Temporary freeze on or project all borrowing up companies. to 2005. Borrowing from LGs may not Borrowing by LG LGs borrow mainly LGs borrow exclu- LGs not allowed to the government/ borrow. (mainly special from CG through sively from gov- borrow without donor onlending service units) donor onlending ernment financial prior approval from China and from budget. institutions; from CG. Development 50% of latter loans largely donor Bank; commercial in arrears; 63% of onlending. bank loans to loans to LGs in GFIs monopolize "off-budget" arrears. depository bank funds. New onlending business, so de terms being facto IRA intercept negotiated. has led to good Under new PP107, LG repayment CG can intercept history. DAU. Terms of onlending loans by GFI to be in line with those of local commer- cial banks, but in practice GFI often sets a lower rate. (Continued) 57 58 TABLE 3.1 (continued) Summary of practices Cambodia China Indonesia Philippines Thailand Vietnam Banking sector Not much banking LG-owned LGs borrow short Private commercial LGs are only now LG-owned compa- sector­related companies borrow term from banks bank lending to exploring oppor- nies borrow from activity; basically primarily from to cover cash flow. LGs virtually tunities to borrow banks to finance a cash economy. commercial banks, Banking sector still nonexistent. from local com- infrastructure Most banks mostly 3­5 years; under restructur- Land Bank of the mercial banks and projects. foreign-owned. sometimes 10. ing; 22 of 27 BPDs Philippines, the public revolving Commercial banks National Bank of Interest rates recapitalized. largest provider funds. purchased HCMC Cambodia regulated by PBC. of credit to LGs, LGs mainly borrow private placement opened dollar- Banking sector is now uses its own from local bond. denominated burdened by high resources (as well development Gradual liberaliza- accounts for rate (40%) of as donor funds): funds monitored tion of interest commercial banks nonperforming 5 years. by Ministry of rate since 1996; in 1998; deposits loans; concern for Privatized Philip- Interior. State Bank of with the bank are the ability of LG pines National Vietnam (SBV) remunerated at companies to Bank still lends to removed lending- 7/8 of SIBOR. depend on LGs (eligible to be rate ceiling. rollover of bank an LG depository): SBV launched loans. 4­7 years. banking-sector New GFI players in restructuring LG credit. program in 2001, phasing in SOCB recapitalization. Capital markets LG-owned compa- Defaults of munici- 6 LG banks: West 13 LG bond issues LGs cannot go HCMC issued GO nies cannot issue pal bonds in early Java, East Java, since 1991: 4 directly to capital bonds in 1995 bonds. 1990s led to ban Central Java, issues guaranteed markets. (D 30 billion, Credit to the private on LG bond North Sulawesi, by Home Guaran- Debt financing by 3-year, private sector is 7 percent issuance. West Sumatra, tee Corp. (govern- LGs has been placement) for of GDP, loans are LG companies issue and PT Bank DKI ment agency), 8 limited (some toll-road project; mostly short term. bonds (Shanghai: issued 12 bonds issues guaranteed revenue bonds is preparing Mainly providing 8 years); strict (1991­2000); by LGUGC; mostly issued). another. (Decree import/export procedures for the 3­7 years. 7-year. 93/2001 allowed financing and issuing corporate LGUGC established HCMC bond working capital to bonds (single A or in 1998, owned issuance.) trade and service better for issues 51% by Bankers Ministry of Finance sectors. greater than Y Association and approval required 100 million). 49% by Develop- for bond issues; ment Bank of the market-rate Philippines. pricing unlikely. HCMC Securities Exchange estab- lished in 2000; bond is 10% of transaction vol- ume; the Hanoi exchange opened in early 2005. Monitoring of local Financial accounting Government finan- Regional financial Ministry of Finance Ministry of Informa- Central budget government centralized in cial information information pilots new system tion monitors sub- guidelines. credit National Treasury system being system being for LG fiscal/ national borrow- City budget and provincial piloted. prepared; Ministry financing ing from local information not branches. LG credit not of Finance starts reporting. development disclosed. Main financial state- assessed. mapping LG fiscal GFI appraises central funds and public SOCB evaluates ment for LG Credit-rating capacity? government revolving funds. large-scale expenditure man- agencies exist. transfer and LG projects. agement pro- tax base. duced by Ministry LGUGC has internal of Finance Budget credit-rating Dept. system. Source: World Bank. Note: BPD state-owned commercial banks; DAU dana alokasi unam, the main intergovernmental fiscal transfer mechanism in Indonesia; HCMC Ho Chi Minh City; GO government order; IRA Internal Revenue Allotment; PBC People's Bank of China; LG local government; CG central government; GFI government financial institution; LGUGC local government unit guarantee corp.; SOCB state-owned commercial bank. 59 60 East Asia Decentralizes In Thailand, local governments may borrow do not intend such an outcome, or if local gov- domestically and internationally, with prior ernments do not use the allocated resources effi- authorization from the cabinet. These governments ciently, the resulting trends will undermine fiscal may issue debt securities and borrow from official, sustainability. external bilateral creditors for development proj- For such rules to be credible and sustainable, ects. In practice, local debt financing is somewhat they must be part of a well-articulated fiscal frame- limited, including from domestic capital markets. work that improves the government's position over The primary source of borrowing has been local the long term. Such a framework includes clear development funds managed by the Ministry of intergovernmental fiscal relations, appropriate sub- Interior. Subnational governments have more national tax structures, and public pensions.20 recently borrowed from commercial banks and The central government needs to clearly define public revolving funds. accountability and establish financial management In Vietnam, subnational governments may not practices to enforce these rules.21 In fact, the tem- run fiscal deficits. Provinces may borrow, but only porary ban on subnational borrowing in Indonesia domestically, by issuing project investment bonds, is a prudent interim measure until the country or by borrowing from the Development Assistance establishes a more solid framework for regional Fund.15 Provinces may use these funds only for borrowing and a regional financial information projects with prior approval from the Provincial system. A more solid framework for subnational People's Council under the five-year provincial borrowing must include procedures for handling Public Investment Plan. The province must also regional default. Otherwise, without a credible legal allocate funds for debt service in future budgets or regulatory threat, the center will end up paying until the debt matures.16 Local state-owned enter- the bill.22 prises that provide essential services, such as waste disposal, water, electricity, and transportation, may Managing Fiscal Risks borrow from both external and domestic sources. All external borrowing is subject to approval by the Rules and administrative controls can help reduce central government, which provides a sovereign the risks of subnational borrowing. Key measures guarantee. include strengthening the intergovernmental fiscal To promote responsible subnational borrowing, system and, when the situation warrants, requiring some East Asian governments have disseminated ex ante authorization and ex post monitoring. For well-defined, transparent fiscal rules. The rationale instance, the central government could set annual is that all borrowing decisions should take into limits on the debt of individual local governments, account the fiscal implications for future genera- review individual loans, including their terms and tions.17 The issue of implementation remains to be conditions, and centralize all borrowing, with addressed, however. onlending to local governments.23 Other measures In Indonesia, regulations implementing the to encourage fiscally sustainable borrowing include decentralization framework limit the debt-to- prohibiting subnational authorities from issuing revenue ratio to 75 percent of the previous year's guarantees (see table 3.2 for the kinds of explicit general revenue, and the debt service-to-revenue and implicit guarantee mechanisms that may exist ratio to 40 percent in any given budget year. Regula- at subnational levels), and imposing ceilings on the tions also govern onlending to the regions.18 In the net worth or loan portfolio of borrowers. Indonesia Philippines, the central government limits transfers and Thailand have established ceilings on debt or to local governments to 40 percent of their internal debt service as a share of local revenues. Other revenues from three years before.19 Since the ratio countries, such as Vietnam, require local govern- of revenue to gross domestic product has been ments to balance their budgets and restrict their falling, this rule implies that some local govern- borrowing to specific purposes, such as capital ments may receive higher transfers than intended, investment. or than the central government can afford. Declin- Countries can also rely on the market to regulate ing trade taxes also build upward bias into the rev- subnational borrowing. A market-based system enue share of local governments. If policy makers requires minimum legal and regulatory structures, Achieving Fiscal Sustainability 61 TABLE 3.2 Subnational Fiscal Risks Direct Contingent Liabilities (obligation in any event) (obligation if a particular event occurs) Explicit Government Local government debt. Local government provides liability is Arrears (if legally binding). guarantees for debt and other recognized Nondiscretionary budgetary obligations of financial and by law or spending. nonfinancial enterprises and contract. other entities. Local government insurance schemes (such as crop insurance). Implicit A "moral" Capital and recurrent costs of local Claims arising from local obligation on public investment projects. government letters of comfort. the part of the The cost of future benefits under Claims by failing local financial government local social security schemes. institutions and other entities. that mainly Claims related to enterprise reflects public restructuring and privatization. expectations Claims by beneficiaries of failed and pressure by social security or other funds, interest groups. beyond any guaranteed limits. Claims related to local crisis management, such as public health, environment, and disaster relief. Source: Brixi and Mody 2002. such as supervision and disclosure practices; guide- Subnational governments that relax their budget lines for issuing, settling, and repaying debt; bank- constraints contribute to public sector deficits and ruptcy procedures (including creditor remedies); threaten national solvency. Hidden budget channels protection against disruption of essential public include off-budget borrowing; arrears on civil serv- services; and measures to prevent moral hazard. ice wages and payments to suppliers and other levels East Asian countries are only now establishing of government; indirect liabilities through public these institutional structures. enterprises or publicly owned banks, which are often Another market instrument for reducing the insolvent; and other contingent liabilities such as credit risk of subnational borrowing is regular unfunded pension and provident funds. Soft budget monitoring of creditworthiness. Private sector enti- constraints and the expectation of central bailouts ties can help investors by rating the likelihood that contribute to moral hazard. Data on these "hidden subnational governments will default. Such credit deficits" in East Asia are just becoming available ratings should reflect both the capacity and the (Kharas and Mishra 2001). Thus central govern- willingness of debt issuers to make timely payments ments lack the information they need to monitor the on both principal and interest. Key elements of fiscal risks of subnational governments. In most East creditworthiness include the subnational govern- Asian countries, the Ministry of Finance typically ment's economic base (net worth), revenue auton- receives regular reports on budgetary revenues and omy and stability, revenue-expenditure balance, expenditures on a cash basis from subnational gov- intergovernmental fiscal relations, the subnational ernments. The ministry does not, however, have debt burden, and contingent liabilities. Financial access to timely information on many extrabudgetary management practices in the region matter, as do and off-budget capital expenditures and borrowing, guarantees, insurance, and other mechanisms to or on local guarantees, financial institutions, pen- enhance the credibility of subnational borrowing. sion funds, employment insurance funds, and other However, credit ratings of subnational entities are transactions that could generate liabilities (Ma and not yet available in most East Asian countries. Brixi 2002). Anecdotal evidence on provincial-level 62 East Asia Decentralizes off-budget and contingent liabilities abounds in vehicles that carry explicit or implicit guarantees some countries. from the central government raise the latter's risk. In China--where direct and indirect support of Yet estimating the scope of the contingent liabilities expanded investment and credit to subnational that subnational spending and borrowing impose entities has driven much recent growth--reliance on the central government is difficult, especially if on banking and off-budget funds has been impor- no one government agency is cataloguing these tant.24 Poor monitoring of such investment implies transactions. East Asian countries are just begin- that the health of the financial sector is at risk. Sig- ning to undertake this task, at least for large contin- nificant investment in industries such as cement, gent liabilities. China and Thailand have been steel, and aluminum seems even riskier, given sub- building the capacity of the central government to stantial excess capacity. manage overall public debt. Table 3.2 outlines typical sources of contingent Early warnings, such as those used in Brazil, liabilities, which can be explicit or implicit. For Colombia, and the United States, can provide a instance, the explicit fiscal burden from rising good starting point for monitoring the fiscal risks safety net expenditures may generate liabilities for of subnational borrowing, but such indicators may the central or provincial government. Also, what not reliably reflect future financing pressures.27 Ma may be "contingent" for the central government (2001) has proposed a composite indicator that may be a direct liability for a provincial or subna- reflects both fiscal pressures and the current fiscal tional government. Cataloguing such liabilities is position of subnational governments. Information an essential first step in establishing a system to on their assets and liabilities, exposure to market assess the obligations and fiscal risks of local and rollover risks, capacity for managing these governments.25 China, Indonesia, Thailand, the finances, and the compatibility between revenue Philippines, and Vietnam are now embarking on and spending responsibilities will enable informed this important endeavor. judgment concerning fiscal risks that may need As provincial and municipal governments gain immediate attention. Without sound fiscal report- greater authority to tap domestic and international ing and auditing at the subnational level, even the financial markets, the financial risk at subnational best-designed early-warning system will not be levels will also grow--and will need managing. effective. Countries need clear rules for dealing Weaknesses in regulation and oversight have led to with debt-distressed subnational governments. the proliferation of off-budget financing, govern- Vietnam is examining the fiscal risks of borrowing ment guarantees, and other contingent liabilities. by state-owned enterprises, but most East Asian Decentralization has given subnational governments countries are just beginning to tackle this issue. a greater role in managing and delivering public services, and more budgetary responsibility for civil Conclusions service pensions and provident funds. All these fac- tors exert a significant impact on the quality of serv- A piecemeal approach to intergovernmental fiscal ices and expenditure mix of local governments. reform is unlikely to succeed because it will not Countries need to estimate the costs of deliver- take into account the interdependence between ing devolved responsibilities so they understand the transfers and revenue assignments, or between potential fiscal risks. For example, if Thailand allo- expenditure assignments and own-source revenues cates 20 percent of revenues to local authorities but and transfers. To enforce hard budget constraints, does not devolve commensurate responsibility for institutions must clearly and credibly convey the expenditures, then the central government bears a message that local governments will bear the costs significant risk of assuming the resulting deficit.26 and benefits of their fiscal decisions. One such These risks will grow if service delivery declines institution is the capital market, which in theory owing to capacity constraints among subnational rations access to capital among subnational bor- governments. In China and Indonesia, where sub- rowers based on the soundness of their fiscal deci- national governments have more expenditure sions. Institutions that ensure that public officials responsibilities than revenue-raising authority, must answer to the needs and aspirations of the subnational borrowing and off-budget financing local populace are essential as well. However, such Achieving Fiscal Sustainability 63 institutions can work only if local governments appropriate early-warning systems for fiscal risks. have considerable fiscal autonomy, and when the East Asian countries have begun work in these crit- central government makes a credible commitment ical areas, but success will require consistent effort that it will not provide bailouts (World Bank over the short and medium term. 2000b). None of the countries in East Asia now meet these conditions for enforcing hard budget Endnotes constraints. East Asian countries need a comprehensive 1. This theory assumes that labor and capital mobility will ensure competition among subnational governments for approach to intergovernmental fiscal reform, tack- effective public sector service delivery, as well as a match ling subnational expenditures and revenues at the between the preferences of local citizens and governments. same time. Subnational governments must have 2. Davoodi and Zou (1998) define fiscal decentralization in terms of spending by subnational governments as a fraction enough revenues to implement their spending of total government spending. Fiscal decentralization rises if responsibilities (see chapters 5 and 6). Toward spending by state and local governments expands relative to this end, incentives that encourage cooperation spending by the central government. The authors use the between officials at different levels are essential. average growth of real per-capita output over 5-year and 10-year periods as a proxy for long-run growth. China and Indonesia need such incentives, for 3. Davoodi and Zou (1998) did not use disaggregated subna- example. tional data to determine which province or region spent Central governments need to carefully monitor more on capital and infrastructure relative to others. The authors also noted that countries may not realize the effi- subnational borrowing and the resulting fiscal ciency gains of fiscal decentralization if central authorities risks in order to maintain fiscal discipline and pru- constrain subnational revenue collection and spending, dent macroeconomic management.Before countries and if local citizens do not elect local officials. Labor and capital mobility may not be as easy as theory assumes. give subnational governments free rein to borrow-- 4. A significant part of spending in these sectors often occurs domestically or abroad--they need to make infor- under "current expenditure." mation critical to analyzing subnational creditwor- 5. See chapter 2 in Litvack et al. (1998) for a discussion of the thiness available to stakeholders.Appropriate checks distinction between decentralization, deconcentration, and delegation. and balances must ensure that these data are reliable 6. Lin and Liu (2000) found that rural reform, development and consistent across provinces. Domestic and of the nonstate sector, and capital accumulation have also foreign capital markets play an important role in been driving forces in China. 7. In the same study, the authors found a positive relationship diversifying local government funding sources and between fiscal decentralization and provincial economic tracking subnational creditworthiness. East Asian growth in India. Questions about the appropriateness of countries--especially those with weak financial the "decentralization variable" again arise. See Ebel and Yilmaz (2003). systems--will need to implement a complex set of 8. See World Bank (2002). Subnational governments in China institutional reforms to support these efforts. also often engage in commercial activities to supplement Finally, to achieve fiscal sustainability, countries their revenue, using land resources and enterprise assets as will also need to reform governance, public enter- their investment capital. Under a model pioneered in Shanghai, many municipalities are creating "corporations" prises, sectors such as power, and the intergovern- to manage public resources and debt obligations. These mental fiscal structure to ensure hard budget corporations may "enhance their debt servicing capacity" constraints for subnational governments. Central by engaging in profit-making activities. The General Corporation of Shanghai Municipal Property Devel- governments also must monitor contingent liabili- opment is one of the first such vehicles, created to ties to ensure prudent fiscal management. Over the help finance the city's enormous need for facilities and longer term, a database and analytical indicators infrastructure. 9. Wildasin (1997) showed that a local government's ability to can form the basis for a credit-rating system for extract a bailout from the central government depends on local governments. Such a system is critical to the former's size. Larger subnational governments thus opening up subnational borrowing and developing tend to operate under softer budget constraints, creating a municipal bond market.28 Also important in incentives for overspending and overborrowing if not ade- quately monitored by the center. minimizing future surprises from the decisions of 10. For example, although Thailand's central government set subnational governments is a clear division of grant allocations well in advance of fiscal year 2002, it did responsibilities across levels of government, formal not disburse these grants until the very end of the fiscal year because of delays in establishing the allocation rule. channels for reporting contingent liabilities 11. Empirical evidence in Lewis (2003) suggests that local and analyzing their potential fiscal impact, and governments have borrowed "well within their fiscal 64 East Asia Decentralizes capabilities to repay," but that they have largely been 25. Such a system can potentially be used to generate a ranking unwilling to repay these debts. The key issue is therefore of local governments based on their fiscal health, and to one of credit risk rather than fiscal sustainability of subna- enable the central government to decide when to intervene tional borrowing in Indonesia. and provide emergency assistance (Ma and Brixi 2002). 12. For example, state-level defaults on debt payments helped 26. These risks will be compounded as the local share of rev- trigger Brazil's financial crisis in 2000. See Rodden 2000a. enues grows to 35 percent by 2006. 13. Rodden et al. (2003) draw on the experiences of seven 27. See World Bank (2002). Brazil had imposed limits on sub- developing and transition economies and four OECD national borrowing, Colombia had a "traffic light system," countries in identifying institutional arrangements associ- and Ohio in the United States has a "fiscal watch program." ated with soft and hard subnational budget constraints. 28. See Ma and Brixi (2002) for relevant experiences in This exercise provides lessons for East Asian countries. Australia, Brazil, Colombia, New Zealand, and the 14. States in Indonesia can now borrow for projects that gener- United States, and a list of risk indicators for monitoring ate a "direct or indirect" financial return, per Implementing the contingent liabilities of subnational governments. Regulation 107 of 2000. However, the central government can restrict borrowing if "general conditions" warrant such Bibliography a restriction, as in the temporary ban by ministerial decree in 2001. Akai, Nobuo, and Masayo Sakata. 2002. "Fiscal Decentralization 15. The Vietnamese Budget Law stipulates that, for a given Contributes to Economic Growth: Evidence from State- year, a province can borrow up if its outstanding debt will Level Cross-Section Data for the United States." Journal of not exceed 30 percent of its capital budget. Until 2003, local Urban Economics 52: 93­108. governments had issued only two project bonds, both by Bird, Richard. 2003. "Fiscal Flows, Fiscal Balance, and Fiscal Ho Chi Minh City (HCMC). In 2003 the central govern- Sustainability." World Bank Institute, Washington, DC. ment approved new HCMC urban infrastructure bonds. In Bird, Richard, and Christine Wallich. 1993. "Fiscal Decentraliza- September and October 2003, HCMC issued two-year and tion and Intergovernmental Fiscal Relations in Transitional five-year bonds at annual interest rates of 8.52 percent Economies: Towards a Systematic Framework of Analysis." and 9 percent, respectively. In December 2003, HCMC Working Paper. Country Economics Department, World issued another five-year urban bond at 8.5 percent interest. Bank, Washington, DC. As of October 2003, provincial borrowing amounted to Brixi, Hana Polackova, and Ashoka Mody. 2002. "Dealing with 0.43 percent of gross domestic product (D 2.5 trillion). Government Fiscal Risk: An Overview." In Government at 16. The Ministry of Finance closely monitors this commitment Risk, ed. Hana Polackova Brixi and Allen Schick. Washington, each year, requiring all subnational authorities to report DC: World Bank. their debt, interest and principal payments, and borrowing Burki, Shahid Javed, Guillermo E. Perry, and William R. plans by type of instrument. In practice, subnational gov- Dillinger. 1999. "Beyond the Center: Decentralizing the ernments have delayed providing this information. State." Viewpoints. Washington, DC: World Bank, World 17. Hence, these rules are often derived using the fiscal sustain- Bank Latin American and Caribbean Studies. ability approach based on the government's "lifetime Davoodi, Hamid, and Heng-fu Zou. 1998. "Fiscal Decentraliza- budget constraint." tion and Economic Growth: A Cross-Country Study." Jour- 18. Decree KMK35 of 2003 in Indonesia stipulates that nal of Urban Economics 43: 244­57. onlending is allowed only for regional government projects Ebel, Robert D., and Serdar Yilmaz. 2003. "On the Measurement that generate revenue. The decree does not specify how to and Impact of Fiscal Decentralization." In Public Finance in handle a default by a subnational entity, except in the case Developing and Transitional Countries: Essays in Honor of of loans from the center, where deductions from the gen- Richard Bird, ed. Jorge Martinez-Vazquez and James Alm. eral grant would cover the debt service. KMK35 is currently Cheltenham, UK: Edward Elgar Press. under review. Kharas, Homi, and Deepak Mishra. 2001. "Fiscal Policy, Hidden 19. This means that the intergovernmental transfer equation Deficits and Currency Crises." In World Bank Economists' excludes trade taxes. Forum, ed. S. Devarajan, F. H. Rogers, and L. Squire. 20. See Kopits (2001). Adopting fiscal reporting in accordance Washington, DC: World Bank. with the IMF's Code of Fiscal Transparency would signal Kopits, George. 2001. "Fiscal Rules: Useful Policy Framework credibility of rules. or Unnecessary Ornament?" IMF Working Paper 01/145. 21. These include best-practice accounting standards and effec- Washington, DC: International Monetary Fund. tive monitoring and reporting of government liabilities. Lewis, Blane D. 2003."Local Government Borrowing and Repay- 22. Where the central government tightly controls local spend- ment in Indonesia: Does Fiscal Capacity Matter?" World ing decisions, as in Cambodia, China, Vietnam, and to Development 31 (6): 1047­63. some extent Thailand, local governments are more likely to Lin, Justin Yifu, and Zhiqiang Liu. 2000."Fiscal Decentralization credibly shift blame for a fiscal crisis to the center (World and Economic Growth in China." Economic Development Bank 2000b). and Cultural Change 49 (1): 1­21. 23. See Burki et al. (1999). Litvack, Jennie, Junaid Ahmad, and Richard Bird. 1998. 24. The fixed-assets investment is driven by local governments "Rethinking Decentralization in Developing Countries." rather than the central government. For the first half of Sector Studies Series. World Bank, Washington, DC. 2003, central government projects amounted to only McKenzie, Kenneth J. 2003. "A Note on the Sustainability of Ą 184.8 billion (12.3 percent), a decline of 7.7 percent over Fiscal Policy in Subnational Governments." World Bank the previous year. However, local government projects Institute, Washington, DC. amounted to Ą 1,322.4 billion (87.7 percent), an increase of Ma, Jun. 2001."Monitoring the Fiscal Risks of Subnational Gov- 41.5 percent over the same period in the previous year. ernments." In Government at Risk, ed. Hana Polackova Brixi Most of this investment is in infrastructure projects. and Allen Schick. Washington, DC: World Bank. Achieving Fiscal Sustainability 65 Ma, Jun, and Hana Polackova Brixi. 2002."Monitoring the Fiscal Relations." Policy Research Working Paper No. 1843. World Risks of Subnational Governments." PREM Notes No. 6. Bank, Washington, DC. Washington, DC: World Bank. World Bank. 2000a. "Thailand: Public Finance in Transition." Martinez-Vazquez, Jorge, and Robert M. McNab. 2001. "Fiscal Report No. 20656-TH. World Bank, Washington, DC. Decentralization and Economic Growth." Working Paper ------. 2000b. "Decentralization and the Challenge of Hard 01-1. Georgia State University, Andrew Young School of Budget Constraints." PREM Notes No. 41. Washington, DC: Policy Studies. World Bank. Oates, Wallace E. 1993. "Fiscal Decentralization and Economic ------. 2000c. "Vietnam: Managing Public Resources Better." Development." National Tax Journal 42 (2): 237­43. Public Expenditure Review Report No. 24469, annex B. Rodden, Jonathan. 2000a."Decentralization and the Challenge of World Bank, Washington, DC. Hard Budget Constraints." PREM Notes No. 41. Washington, ------. 2002. "China: National Development and Subnational DC: World Bank. Finance: A Review of Provincial Expenditures." Report ------. 2000b. "The Dilemma of Fiscal Federalism: Hard and No. 22951-CHA. World Bank, Washington, DC. Soft Budget Constraints around the World." Massachusetts ------. 2003a. "Decentralizing Indonesia." Report No. 26191- Institute of Technology, Cambridge, MA. IND. World Bank, Washington, DC. Rodden, Jonathan, Gunner Eskeland, and Jennie Litvack, eds. ------. 2003b. "Philippines: Improving Government Perfor- 2003. Decentralization and Hard Budget Constraints. mance: Discipline, Efficiency and Equity in Managing Public Washington, DC: World Bank. Resources." Public Expenditure, Procurement, and Financial Shah, Anwar. 2004. "Fiscal Decentralization in Developing Management Review Report No. 24256-PH. World Bank, and Transition Economies: Progress, Problems, and the Washington, DC. Promise." Research Working Paper No. 2832. World Bank, ------. 2004. East Asia Update. Washington, DC: World Bank. Washington, DC. ------. 2005. Global Economic Prospects. Washington, DC: Ter-Minassian, Teresa, and John Craig. 1997. "Control of World Bank. Subnational Borrowing." In Fiscal Federalism in Theory Xie, Danyang, Heng-fu Zou, and Hamid Davoodi. 1999. "Fiscal and Practice, ed. Teresa Ter-Minassian. Washington, DC: Decentralization and Economic Growth in the United International Monetary Fund. States." Journal of Urban Economics 45 (2): 228­39. Weist, Dana. 2001. "Thailand's Decentralization: Progress and Zhang, Tao, and Heng-fu Zou. 1999. "Fiscal Decentralization, Prospects." Paper prepared for the King Prajadipk Institute Public Spending, and Economic Growth in China." Journal Congress III on Decentralization and Local Government in of Public Economics 67: 221­40. Bangkok, Thailand. ------. 2001."The Growth Impact of Intersectoral and Intergov- Wildasin, David E. 1997. "Externalities and Bailouts: Hard ernmental Allocation of Public Expenditure: With Applica- and Soft Budget Constraints in Intergovernmental Fiscal tions to China and India." China Economic Review 12: 58­81. 4 Fiscal Disparities in East Asia: How Large and Do They Matter? Bert Hofman and Susana Cordeiro Guerra The decentralization of revenue sources and expen- the migrants could have been employed more pro- diture responsibilities to subnational levels of gov- ductively elsewhere.1 Finally, large disparities in ernment can increase the efficiency of spending, public service delivery may cause social unrest in and increase participation in decision making by regions that are left behind, and could undermine local constituents. It is also true, however, that the sense of unity in a country. Such diversity, cou- decentralization can result in inequities in service pled with growing disparities across regions, may delivery among citizens of the same country constitute the basis for regional insurrections such depending on where they live. While decentraliza- as in the Muslim regions of Mindanao in the tion need not cause these inequities, devolution of Philippines, and in West Papua in Indonesia revenue sources, combined with disparities in (Hill 2000). endowments of regions, is likely to lead to dispari- Most governments take an interest in the level ties in fiscal resources at the subnational level. and distribution of public services provided to If left alone, these disparities could lead to either their citizens, even in a decentralized system. For lower levels of services in fiscally poor regions or some countries, this is reason enough not to decen- higher tax rates for similar levels of government tralize the financing or provision of services services in those regions. In turn, disparities in deemed critical to national goals. Indeed, income service delivery could thwart poverty alleviation redistribution is seen primarily as a central func- efforts, as public services that are usually provided tion because central governments are regarded as at the subnational level, such as primary health and better able to manage one of the key instruments of education, are critical in empowering the poor. this goal: a progressive income tax.2 Yet, at the same Disparities in the level and quality of these services time, services that are likely to affect income distri- could therefore eventually perpetuate inequalities bution, poverty alleviation, or similar national in income levels, or at least delay conversion of goals are decentralized. Moreover, many govern- incomes across regions. Large fiscal disparities ments have entered into commitments on the out- could also induce migration to regions that have comes of services that are routinely at least partially higher revenues and better services, even though decentralized to subnational levels of government.3 67 68 East Asia Decentralizes So they should be concerned about the level and also tend to be powerful regions, and taking distribution of fiscal resources among levels of gov- resources away from them to give to poorer regions ernments, as those resources enable subnational may simply be politically unfeasible. governments to deliver services. Most East Asian governments care about equi- Even if governments care about the distribution table services to their people, and thus take an of services, they may not take policy action to cor- interest in the distribution of fiscal resources rect the distribution of fiscal resources among sub- among subnational governments, which deliver national governments--known as horizontal fiscal many of these services. Countries such as Indonesia imbalances. This could occur for two reasons. First, have included subnational fiscal equity as an policy makers may count on market adjustments. explicit goal in their Constitution.5 Other coun- Thus, for some countries, migration is considered a tries, such as China and the Philippines, include countervailing, equalizing force, driving people to strong commitments on equal access to services in the constituency that delivers the most beneficial their Constitution, whereas the delivery of many of level of public services at tax rates deemed appro- these services is devolved to subnational govern- priate by the constituents. Second, legitimate policy ments.6 Similarly, international commitments to tradeoffs need to be weighed: an aggressive transfer public service outcomes, such as the Millennium policy may be seen as dampening needed incentives Development Goals, often relate to services deliv- for increasing own-revenue mobilization by subna- ered at the subnational level. These commitments tional governments. Such a short-term tradeoff imply that the state should care about whether sub- deemphasizes equalization to provide for an own- national governments are capable of delivering revenue foundation for a future of reduced fiscal such services, and thus a concern about subna- disparities and transfer dependency.4 tional fiscal capacity and its distribution over sub- Similarly, an equalization-only policy (perhaps national entities. In fact, most East Asian countries just in the early stages of decentralization) may show that they care to some extent by having some compete with broader considerations of efficiency form of fiscal equalization mechanism in place. and growth on the nation's agenda. In China, for This chapter reviews the causes and conse- instance, the coastal development strategy of the quences of fiscal inequities among subnational gov- 1980s and 1990s deliberately left more resources in ernments in East Asia. It shows that endowments regions with stronger growth prospects. Using data and fiscal capacities among subnational entities are from the 1985 to 1998 period of fiscal decentraliza- large, and that these inequities translate into large tion in China, Qiao et al. found that inequality in fiscal inequities before equalization. The transfer the distribution of fiscal resources across provinces systems now in place in East Asian countries are was positively related to higher economic growth, only mildly equalizing, so that even after equaliza- and that that higher growth, in turn, led to greater tion the remaining fiscal disparities are still large. inequality (Qiao et al. 2003). Yan found a similar The chapter discusses the possible consequences of tradeoff between growth and equity in China, and these disparities for service delivery inputs, out- further showed that the fiscal reforms in 1985 and puts, and outcomes. It concludes with some guide- 1994 did not contribute to increased equalization lines on how to increase transparency on inequali- (Yan 2003). ties at the subnational level, and how to make In other countries, governments compensate for transfer systems more equalizing without under- horizontal fiscal imbalances not through redistri- mining the incentives for local governments to raise bution, but by centrally providing certain services their own revenues and spend their resources in poor regions while leaving rich regions to fend efficiently. for themselves. This approach may indeed also be beneficial from an efficiency point of view, as some Fiscal Disparities in East Asia evidence suggests that the center is better at some services critical for poverty alleviation, including East Asian countries show a wide variety of natural the targeting of a social safety net (Ravaillion endowments, economic opportunities, levels of 1998). Finally, redistribution of fiscal resources may development, and poverty (see figure 4.1). Indeed, remain limited for political reasons: rich regions the richest province in Indonesia has 17 times the Fiscal Disparities in East Asia: How Large and Do They Matter? 69 FIGURE 4.1 Percentage of Population Living below PPP$2 a Day in 2002, East Asia and Pacific gross domestic product (GDP) per capita as the Indeed, East Asian countries reveal large dispar- poorest province; in China that figure is 11, and in ities in subnational revenue-generating capacity Vietnam it is 9.5.7 These disparities in per capita (table 4.1). In fact, disparities in own-revenues are income have been relatively stable (Hill 2000; larger than disparities in income per capita. In Garcia-Garcia and Soehstianingsih 1998). In such China, for instance, own-revenues per capita of the an environment, devolving expenditure responsi- richest province are 15 times those of the poorest bilities and revenue sources could lead to inequities region. These disparities are even higher in Indone- because of regional disparities in revenue potential sia, the Philippines, and Vietnam.8 In the latter and the cost of delivering services. three countries, natural resource revenues collected 70 East Asia Decentralizes TABLE 4.1 Disparities in Provincial Revenues before Grants, per Capita (US$, latest available year) China Indonesia Philippines Vietnam Maximum 283.2 59.5 7.6 343.4 Minimum 18.1 3.8 0.2 6.5 Average 55.7 12.1 1.5 36.9 Max./min.a 15.7 15.7 35.4 53.0 Standard deviation 54.8 11.6 1.4 61.1 Coefficient of variationb 0.98 0.96 0.97 0.79 Sources: SABER Database; Indonesia Regional Fiscal Information System. Note: 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 figures for Indonesia include own-revenues and shared revenues. Since the Internal Revenue Allotment in the Philippines is an equalizing transfer, it is excluded from revenues. A breakdown of revenue for Thailand was not available at a comparable subnational level. a. The figures represent the ratio in revenues between the province with the highest revenues and the province with the lowest revenues. b. The coefficient of variation is the standard deviation divided by the average. TABLE 4.2 Disparities in China's Provincial Fiscal Outcomes, per Capita, 1979­2002 (yuan) Revenues Expenditures 1979 2002 1979 2002 Maximum 1,525 4,363 274 5,307 Minimum 12 274 34 655 Average 143 806 98 1,621 Max./min. 127 16 8 8 Standard deviation 293.5 939.2 67.4 1,218.4 Coefficient of variation 2.0 1.2 0.7 0.8 Source: Qiao et al. 2003. Note: The negative minimum value for China reflects the definition of revenues, which counts subsidies that cover losses from state enterprises as negative revenue. or shared by the center with the regions exacerbates slightly, with the interprovincial coefficient of apparent inequalities. Below the provincial level, variation rising from 0.7 to 0.8; revenue disparities inequalities grow even larger. In Indonesia, for have fallen from 2.0 to 1.2. Yet the latter continue to instance, the richest local government had 50 times be large, with Shanghai having almost 16 times the the own-revenues of the poorest local government per capita revenue of Tibet. Tibet's per capita expen- in 2001, and the richest county in Gansu province diture was similarly 8 times that of Hainan in 1979, had 82 times the per capita revenues of the poorest while Shanghai's is 8 times that of Henna today. province (World Bank 2003b and World Bank In Indonesia, disparity in total revenues after 2002a).9 grants was nearly as large in 2002 as in 1994 These interregional fiscal disparities are not (see table 4.3). Disparities in own-revenues across a recent phenomenon. In China, data show that provinces were already rising before decentraliza- large fiscal disparities have persisted over time tion began in 1999. Disparities in own-revenues plus (see table 4.2).10 Expenditure disparities have risen shared revenues have risen since decentralization, Fiscal Disparities in East Asia: How Large and Do They Matter? 71 TABLE 4.3 Variation in Revenues across Provinces in Indonesia, 1994­2002a 1994 1999 2000 2002 Own-revenues 0.87 0.88 0.98 0.90 Shared taxes 0.73 0.70 1.28 0.86 Shared nontaxes 0.74 0.74 1.24 1.18 SDOb 0.79 0.51 0.64 0.61 INPRESc 0.69 0.51 0.68 0.64 Total revenue disparity 0.68 0.51 0.68 0.66 Source: SABER Database. a. The figures represent coefficients of variation (see table 4.1). b. SDO (subsidi daerah otonom), or Subsidy for Autonomous Region. c. After 2001, INPRES (instruksi presiden), Presidential Instructions (Grant for Regional Development), was made equal to the DAU (dana alokasi umum), or General Allocation Grant. especially because of shared revenues from natural The central government distributes the DAU resources. according to a formula that takes both revenue capacity and expenditure needs into account. Rev- enue capacity is defined as potential own-source Equalization Mechanisms revenues plus shared tax revenues, plus 75 percent China, Indonesia, the Philippines, Thailand, and of shared natural resource revenues.12 The central Vietnam rely on grant systems to address fiscal dis- government defines expenditure needs based on parities (see table 4.4). These systems rely on a for- population, poverty rate, land area, and construc- mula to determine distribution, and three of the tion costs as an indicator of "geographical circum- five use a formula to determine the resource pool. stances." Distribution of the DAU is based partly on Three of the systems take into account both the past spending patterns--largely to accommodate revenue capacity and the expenditure needs of local the transitory impact of the 2001 decentralization. governments, whereas the Filipino and Thai sys- A lump sum per region also plays a role in the tems consider only expenditure needs. allocation. The new earmarked grant system (DAK) The distribution pools vary greatly from coun- is still small compared with general grants-- try to country. In Indonesia and the Philippines, amounting to about 3 percent of total grants. How- equalization grants account for the largest share of ever, it also includes an element of equalization. grants from the center to local governments, while Regions with low fiscal capacity pay only 10 percent earmarked grants dominate in Thailand and China. in matching funds, whereas those with high fiscal In some countries earmarked grants include equal- capacity pay up to 50 percent. izing elements, but in others they exert a counter- In China, the central government dedicates an equalizing effect. Although numbers are hard to ad hoc amount to transfers to the 16 poorest come by, the equalization system can also include provinces. Although the 1994 Tax Sharing System central spending that exerts a regional impact. introduced an equalization scheme based on vari- In Indonesia, the equalization grant--dana ables such as provincial GDP, student-teacher alokasi umum (DAU)--is the mainstay of the inter- ratios, number of civil servants, and population governmental fiscal system. DAU funding consists density, the scheme is still in "transitional" status.13 of 25 percent of central revenues after tax sharing The scheme also remains small: each beneficiary with the regions.11 Of this amount, 10 percent goes province receives only a fraction of its fiscal needs to the provincial level, which plays a relatively minor as determined by the formula, and in 2001 the role in public services, while 90 percent goes to local scheme accounted for only 3 percent of total cen- governments. The DAU finances some 70 percent of tral transfers (World Bank 2004e). Although the local government spending and some 50 percent reform marked the first time that the government of provincial spending. explicitly budgeted an equalization grant, it may 72 East Asia Decentralizes TABLE 4.4 Equalization in Intergovernmental Transfer Systems Indonesia China Philippines Thailand Vietnam Equalization All regions 16 regions All regions Allocated based Allocated to grant receive an receive a small receive a fixed on discretion to jurisdictions principles equalizing equalizing share of central Provincial where approved general grant. grant, government tax Administrative expenditure amounting to revenues known Organization budgets exceed only 3% of all as the Internal (PAO), the sum of transfers in Revenue municipalities, own-revenues 2001. Allotment (IRA). and Tambon and the 100% Administrative retention of Organization all shared (TAO), and then revenues. further allocated to individual localities based on formula. Formula- Yes; 25% of No; level of Yes; central No; total Partially; based source? actual central funding government amount of formula based government decided by sets IRA of local different types on calculated revenues after annual budget governments at of grants varies budget transfer revenue based on ad 40% of average annually. The between the sharing. hoc principles. internal tax system is center and collections three moving to a provinces. years before the formula-based Expenditure current year. system. needs are negotiated. Main features Based on Expenditure IRA divided 5% reserved for Subnational of formula expenditure needs and among unfunded (SNG) needs and revenue provinces devolving expenditure revenue capacity based (23%), functions and needs minus capacity, on formula cities (23%), 95% for local total revenue but 50% derived from municipalities authorities, from taxes determined by regression (34%), and including are shared transitional analysis on barangays PAOs (7%), 100 percent elements. "standard (20%). IRA municipalities with SNGs. Expenditure budget." Relies allocated based (52%), and Total revenues needs reflect on variables on population TAOs (41%). shared between population, such as (50%), land Allocation central and poverty rate, provincial GDP, area (25%), and across local SNG budgets. land area, and student-teacher equal sharing authorities is construction ratios, number (25%). based on equal cost index. of civil servants, share (25%), Revenue and population population capacity density. (30%), area estimated as (5%), invert to standardized local revenues, own-revenues excluded grants (based on (20%), and average efforts), invert to plus shared tax specific grants revenues, plus received (20%). 75% of natural resource revenues. Fiscal Disparities in East Asia: How Large and Do They Matter? 73 Indonesia China Philippines Thailand Vietnam Equalizing Weak, owing to Weak, owing to Medium; IRA Weak; lack of Medium; properties transitional limited equalizing transparency improved (weak, elements and resources. effect is not in allocation from ad hoc medium, imperfections in enough to leads to negotiated strong) formula. counteract self-interested transfers. disparities in tax politics. Delays Introduced base because it in allocation clear objectives weakly decisions have and stability by compares undermined fixing formulas, expenditure local planning decreasing needs and and financial the role of revenue management. bargaining. capacity. Equalization Yes; special No; special No; matching Yes; other No; absence through allocations grants deequal- grants from general grants of conditional specific depend on izing, such as central agencies include grants, but grants fiscal capacity tax rebates for (augmentation tax effort targeted (ad hoc or but are small. state enterprise funds) are promotion, national conditional)? support. Lacks proportionally local good programs monitoring small and governance function mechanisms. usually subject promotion, similarly (see to political devolution of below). interests. compensatory functions, train ticket compensation, local development, and education. Specific grants include educational (capital projects) and development projects allocated on a project basis. Other Distribution of Transitory Targeted equalizing revenues from grants "national elements of natural determined on programs" fiscal system resources ad hoc basis channeled to regions according through bordering to annual provincial producing budgetary budgets in regions. demands. support of poorest communes, reforestation, and national health program. Sources: World Bank 2000, 2002a, 2003b; Manasan 2002; World Bank background papers on Vietnam and Thailand, 2003. 74 East Asia Decentralizes have only symbolic significance until the govern- provinces and cities receive 23 percent, municipali- ment allocates more funds (World Bank 2002a). ties receive 34 percent, and barangays receive Earmarked grants--which account for more than 20 percent.The share within each tier of government 95 percent of all grants--include no equalizing ele- reflects three basic criteria: population (50 per- ment. Two-thirds are "tax return grants," which cent), land area (25 percent), and equal sharing amount to revenue sharing on a derivation basis. across provinces (25 percent). Before the LGC, the In Vietnam, the tax sharing-cum-contracting relative weights of these criteria differed: popula- system equalizes. The revised State Budget Law tion (70 percent), land area (20 percent), and equal (2002) gives the equalization transfer to jurisdic- sharing (10 percent). The greater emphasis on tions whose approved expenditure needs exceed the equal sharing (from 10 to 20 percent), coupled with sum of own-revenues and 100 percent retention of the fall in population weight (from 70 to 50 per- all shared revenues (World Bank 2004e). According cent), shows a shift in equity concerns. to the equalization formula, the local tax adminis- tration branch determines revenue potential based Do Grant Systems in on collections from previous years, factoring in any East Asia Equalize? changes in tax policy and expected economic growth (Martínez-Vázquez 2003). Expenditure needs are Determining whether these grant systems equalize based on per capita expenditure norms rather than requires defining "equalization." Common use physical standards as prior to 1996, which were defines fiscal equality as "the capability of subna- unaffordable. Regions may keep revenues in excess tional governments to deliver similar levels of of agreed shares. Beyond the equalization transfer, servicesatsimilarlevelsof taxes"(Searle2002).While Vietnam has no other conditional grants except a such a definition points to how an equalization grant series of "national programs" aimed at the poorest might best be designed, determining whether it communes and to fulfill reforestation and health equalizes requires information on differences in the goals. The government channels these programs to cost of services--which can be large--among sub- recipients through provincial budgets. national levels of government, and these data are In Thailand, general grants include a fiscal equal- not yet available.16 This chapter therefore uses a ization grant.14 Fiscal equalization grants rely on a simpler approach, asking whether disparities among strict formula based on equal share (25 percent), subnational revenues drop as a result of intergovern- population (30 percent), area (5 percent), invert to mental grants. The answer is that indeed, for all the local revenues excluding grants (20 percent), and countries reviewed, the distribution of revenues per invert to specific grants received (20 percent). A capita becomes more equal after transfers, although second type of grant, known as specific grants, their equalizing effect varies significantly, being covers educational capital investment and develop- strongest in Vietnam and weakest in the Philippines ment projects based on project criteria. Although (see figure 4.2 and tables 4.1 and 4.5). grants were the largest source of intergovernmental Comparing revenues and expenditures per transfers in 2003--accounting for 38 percent of capita against income per capita provides further local revenues--the central government does not evidence of the equalization properties of transfer determine allocations until well after the fiscal year systems. If the income elasticity of expenditures is begins. This leads to a lack of transparency and an lower than the income elasticity of revenues, it can extremely politicized system, and creates planning be argued that the transfer system is equalizing.17 and budgeting difficulties for local governments. This is indeed the case for the East Asian countries Grants do not alleviate the fiscal imbalance between for which enough data are available, suggesting that the central and subnational governments. grant systems equalize (see table 4.6). Under the Local Government Code (LGC) in the Even after transfers, though, per capita revenue Philippines, local governments receive a fixed share disparities remain large. The richest province in the of central tax revenues, known as the Internal Rev- Philippines has 28 times more revenues per capita enue Allotment (IRA). The LGC sets the aggregate than the poorest one, while the numbers for China, IRA share at 40 percent of subnational revenues Indonesia, and Vietnam are 8, 10, and 22, respec- three years before the current year.15 Of this share, tively. For comparison, in the United States, the Fiscal Disparities in East Asia: How Large and Do They Matter? 75 TABLE 4.5 Disparities in Province-Level Revenues after Grants, per Capita (US$, latest available year) China Indonesia Philippines Vietnam Maximum 444.4 431.4 117.5 393.1 Minimum 42.8 39.8 4.2 25.1 Average 100.7 106.3 14.8 65.9 Max./min. 10.4 10.8 28.1 15.7 Standard deviation 83.1 78.9 13.3 65.2 Coefficient of variation 0.82 0.74 0.92 0.42 Sources: SABER Database; Indonesia Statistical Yearbook, China Statistical Yearbook; Philippine Statistical Yearbook; Vietnam Statistical Yearbook; Ministry of Finance (Vietnam); Indonesia Regional Fiscal Information System; Public Expenditure Reviews; Regional Expenditure Reviews; authors' calculations. Note: The table presents consolidated province-level data. The revenue numbers include own- and shared revenues for Indonesia. A breakdown of revenue for Thailand was not available at a comparable subnational level. FIGURE 4.2 Disparity in Provincial per TABLE 4.6 Evidence of Equalization Capita Revenues before and after Transfers Income Income elasticity of elasticity of 1.2 revenue expenditures 1.0 China 1.03 0.61 Indonesia 0.81 0.57 0.8 Vietnam 1.26 0.72 variation of 0.6 Source: Authors' calculations. Note: The figures are the estimated parameters 0.4 for the log of income per capita in a regression that regresses the log of revenues (expenditures) coefficient 0.2 per capita against the log of income per capita and a constant. All the reported elasticities are 0 significant at the 1 percent level. China Indonesia Philippines Vietnam Sources: SABER Database; China Statistical Yearbook; Finance Yearbook of China; Indonesia Regional Fiscal Information System; Philippine Statistical Yearbook; Vietnam Statistical Yearbook; Vietnam Ministry of Finance; Public Expenditure Reviews; Regional subnational entity, the larger the measured inequal- Public Expenditure Reviews; authors' calculations. ity. In Indonesia, the richest local government had 30 times the revenue per capita of the poorest local government in 2002, while the comparable number poorest state has about 65 percent of the revenues for 2001 was more than 50. Within China's Gansu of the average state. In Germany, the Finanzausgle- province, the per capita revenues of the richest ich subsidizes any state falling below 95 percent of county were 37 times those of the poorest in 1999 the average level (and taxes any state receiving more (Hofman et al. 2003; World Bank 2002a). than 110 percent). In Brazil, the richest state has 2.3 Intraprovincial disparities thus appear to be larger times the revenues per capita of the poorest state. than interprovincial ones. In Russia, disparities are larger: the richest of 89 A number of studies confirm the modest impact regions has revenues per capita some 40 times on equalization of intergovernmental fiscal sys- higher than the poorest (World Bank 2002b; tems. A recent study of the Philippines suggests that Martínez-Vázquez and Boex 1998). However, Rus- the IRA's equalizing effect has not countered dispar- sia's regions are smaller than the average East Asian ities in tax base across local governments (Manasan province, and indications are that the smaller the 2002). The same study found that the IRA had a 76 East Asia Decentralizes counterequalizing effect in provinces from 1995 to subnational level are as scarce as fiscal data, but 2000, and in municipalities from 1999 to 2000. In Human Development Indicators and informa- Indonesia, Lewis (2003) found the DAU distribu- tion on persons per hospital bed, literacy, and life tion to be equalizing, but less than warranted by expectancy are available for most countries under revenue capacity and expenditure needs alone, review.18 The first indicator--persons per hospital whereas Hofman et al. (2003) elaborated on the bed--measures service inputs, whereas the other DAU's weak equalization performance. indicators measure outcomes. These measures show Vietnam's fiscal system seems to be more equal- large variation across and within countries,although izing: the country redistributes a substantial pro- less than the fiscal indicators reveal.19 In terms of portion of revenues collected in wealthier provinces persons per hospital bed, Indonesia's lowest-ranking to poorer ones (World Bank 2000). Rao (2001) uses province is 7.6 times worse than the best-ranking cross-section estimates to show that elasticity is province, while the same numbers for Vietnam, much lower for revenues than for expenditures. China, the Philippines, and Thailand--3.3, 3.4, 4.4, This implies a significant degree of equalization, in and 6.8, respectively--are better but still discourag- line with the steep drop in provincial inequality ing.20 China's highest-ranking province is almost 3 presented here. For China, Yan (2003) shows that times as literate as the lowest-ranking one, while fiscal reforms in 1985 and 1994 have not alleviated Indonesia (1.34) and Vietnam (1.9) follow close the unequal impact of market reform, and indeed behind. Human development and life expectancy appear to have exacerbated regional disparities. indicators confirm this trend, with Vietnam's best Indeed, evidence reveals that earmarked grants province ranking nearly twice as high as the lowest are mostly flowing to richer provinces (Ahmad province on both indicators. The figures for China et al. 2000; World Bank 1993). Disparities in per and Indonesia are only slightly better (see tables 4.7 capita revenues after grants translate into dispari- to 4.10). ties in per capita expenditures, although not on a To assess whether there is a systemic relation one-to-one basis. These disparities may reflect dif- between fiscal disparities and disparities in service ferences in access to borrowing or in the use of delivery indicators, we used a simple regression reserves. approach.Weusedtwospecifications(seetable4.11). Specification 1 is a regression of social outcome indicators--HDI, persons per hospital beds, life Do Fiscal Disparities Matter? expectancy, and literacy rates--as a function of a Policy makers may focus on fiscal disparities for a constant and expenditures per capita. In specifica- number of reasons. However, they matter if they tion 2, we also include per capita GDP as an translate into large disparities in service delivery explanatory variable, to control for potential corre- and outcomes. Indicators of service delivery at the lations between provincial per capita income and TABLE 4.7 Disparities in Province-Level HDI Indices, 1997­2002a China Indonesia Philippines Vietnam Maximum 0.85 0.73 0.65 0.84 Minimum 0.52 0.54 0.39 0.49 Average 0.70 0.64 0.56 0.74 Max./min. 1.64 1.34 1.65 1.72 Standard deviation 0.07 0.04 0.11 0.06 Coefficient of variation 0.10 0.06 0.19 0.08 Sources: United Nations Development Programme (UNDP) Human Development Indicators (HDI); Government of Indonesia 2002; National Statistical Coordination Board 2003; Republic of China, multiple years; Republic of Vietnam, multiple years. a. Data are not available across all provinces for Thailand. Data for the Philippines are at the regional level. Fiscal Disparities in East Asia: How Large and Do They Matter? 77 TABLE 4.8 Disparities in Province-Level Persons per Hospital Bed, 1997­2002a China Indonesia Philippines Thailand Vietnam Maximum 650.34 3,969.41 3,502.49 1,114.00 867.93 Minimum 189.73 524.30 796.71 165.00 261.20 Average 405.34 1,862.43 2,125.25 579.65 524.77 Max./min. 3.43 7.57 4.40 6.75 3.32 Standard deviation 105.35 903.95 1,536.92 494.75 152.94 Coefficient of variation 0.26 0.49 0.72 0.85 0.25 Sources: UNDP Human Development Indicators; Government of Indonesia 2002; National Statistical Coordination Board 2003; Republic of China, multiple years; Republic of Vietnam, multiple years. a. Data are for the most recent year available. Data for the Philippines are from the regional level. TABLE 4.9 Disparities in Province-Level Life Expectancy, 1996­2002a China Indonesia Philippines Thailand Vietnam Maximum 78.14 71.10 73.70 79.80 95.00 Minimum 64.37 57.80 60.32 61.60 50.00 Average 71.24 66.21 70.79 72.34 81.66 Max./min. 1.21 1.23 1.22 1.30 1.90 Standard deviation 3.19 3.23 3.17 35.76 7.64 Coefficient of variation 0.04 0.05 0.05 0.49 0.09 Sources: UNDP Human Development Indicators; Government of Indonesia 2002; National Statistical Coor- dination Board 2003; Republic of China, multiple years; Republic of Vietnam, multiple years. a. Data are for the most recent year available. Data for the Philippines are at the regional level and for males only. Data for Vietnam are for males only. TABLE 4.10 Disparities in Province-Level Literacy Rates, 1994­2002a China Indonesia Philippines Vietnam Maximum 93.55 97.80 98.80 96.90 Minimum 33.82 72.80 73.50 51.30 Average 83.19 89.34 92.27 88.16 Max./min. 2.76 1.34 1.34 1.89 Standard deviation 11.04 6.57 5.92 8.32 Coefficient of variation 0.13 0.07 0.06 0.09 Sources: UNDP Human Development Indicators; Government of Indonesia 2002; National Statistical Coordination Board 2003; Republic of China, multiple years; Republic of Vietnam, multiple years. a. Data are for the most recent year available. Data are not available across all provinces for Thailand. Data for the Philippines refer to the regional level. expenditures, and their implications for service the correlation, although positive, is not significant. outcomes.21 The reason may be that education financing still falls Specification 1 suggests a modestly significant partly under the auspices of the central government, correlation between social indicators and subna- given recent efforts to recentralize education expen- tional expenditures.In China andVietnam,the indi- ditures. In Vietnam, the only exception to the corre- cators usually correlate significantly with expendi- lation between social indicators and subnational tures. In China, the only exception is literacy rates: expenditures is persons per hospital bed. This 78 East Asia Decentralizes TABLE 4.11 Expenditure Disparities and Service Outputs and Outcomes at the Subnational Level Dependent variable and specification China Indonesia Vietnam HDI 1 0.10* 0.01 0.10* HDI 2 0.01 0.002 0.001 Persons per hospital bed 1 0.48* 0.29* 0.04 Persons per hospital bed 2 0.39* 0.15 0.15 Life expectancy 1 0.04* 0.02* 0.04* Life expectancy 2 0.02* 0.01 0.02 Literacy 1 0.03 0.02 0.06** Literacy 2 0.03 0.01 0.02 Source: Authors' calculations. 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 1 (y * expenditure per capita), and a constant, expenditures per capita, and provincial GDP per capita in specification 2 (y * expenditure per capita * GPP per capita. The 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). * indicates significance at the 5 percent level ** indicates significance at the 10 percent level. reflects the relatively modest government role in Another possible explanation for the weak corre- financing health care.22 Given the predominance of lation between fiscal disparities and service indica- private funding for health services, subnational rev- tors is that the latter change slowly over time. Only enues would not make a tangible difference in out- China has data that allow for testing the hypothesis comes. Infant mortality might therefore be a better that indicators deteriorate in regions with persist- outcome measure in this case because it relates ently low spending, compared with regions that directly to health services, unlike life expectancy, spend more. In this third set of regressions, we use a which may imply other variables such as food longer time series for provincial per capita expendi- source. tures (1985­2000) as the explanatory variable for In Indonesia, unlike in China and Vietnam, the service delivery indicators (see table 4.12). We also results are significant only for persons per hospital add a fifth social indicator--combined student bed and life expectancy. Such relatively modest enrollment for primary, secondary, and tertiary results may reflect the fact that the decentralization schools--to further test our hypothesis.25 In this process is just beginning in these countries. case, excluding life expectancy, both specifications In specification 2, per capita subnational expen- 1 and 2 suggest a significant correlation between the ditures do not correlate with service outputs and four outcome indicators and subnational expendi- outcomes. A possible explanation for these ambigu- tures across time, suggesting that persistent fiscal ous results may be multicollinearity,23 which reduces inequities do matter.26 the efficiency of the estimation, although the esti- In sum, although the variation in outcome indi- mator remains unbiased.24 cators is much less than the variation in revenues, In addition, the outcome indicators used in this there is some evidence that outcomes are affected chapter are naturally limited in the power of their by different levels of subnational revenues across estimation. Most provinces score highly in literacy provinces. In the short run, effects seem dominated andlifeexpectancyrates,forexample.Unlikeincome, by differences in income per capita, but evidence most of these outcome indicators are bounded, so on China suggests that persistent fiscal inequalities convergence over time should be the norm, which result in inequalities in social indicators beyond presents difficulties with the estimation. those that can be explained by income per capita. Fiscal Disparities in East Asia: How Large and Do They Matter? 79 TABLE 4.12 Fiscal Disparities and Public Bank 2004a). Services can also fail poor people due Services Outputs and Outcomes in China, to a lack of demand. In this case, the geographical 1985­2000 distribution of resources matters because there may be insufficient demand for services in certain Variable and Specification regions and locations. Ultimately, whether spend- HDI 1 0.11* ing will materialize into better services and out- HDI 2 0.003 comes will depend if it is coupled with accountabil- Persons per hospital bed 1 0.50* ity systems (strong performance management and Persons per hospital beds 2 0.40* producer accountability) and appropriate policies Life expectancy 1 0.06* (Pritchett 2004). Life expectancy 2 0.02 Literacy 1 0.06* Literacy 2 0.07* Conclusions Combined enrollment 1 0.13* Combined enrollment 2 0.08* This chapter shows that fiscal disparities among subnational governments in East Asia are consider- Source: Authors' calculations. able. Equalization mechanisms diminish subna- Note: The table presents the results of regression tional fiscal disparities, yet they remain large. We analyses with the output or outcome indicator find considerable disparities in service outputs and as a dependent variable. Dependent variables are a constant and expenditures per capita over outcomes as well, but data and methodological 15 years in specification 1 (y * expen- issues prevent us from establishing a strong link diture per capita over 15 years), and a constant, with large fiscal disparities. For the one country expenditures per capita over 15 years, and provincial GDP per capita in specification 2 with enough data (China), we find that persistent (y * expenditure per capita * GPP fiscal disparities do seem to affect health and edu- per capita). Results exclude Chongqing, Tibet, cation outcomes. and Qinghai. * indicates significance at the 5 percent level; The reasons why inequalities may persist are ** indicates significance at the 10 percent level. numerous. First, expenditure needs may vary sig- nificantly. Such variation could reflect differences in costs or the fact that some regions provide more services than others. Persistent inequalities may Although public spending can be critical in also reflect the central government's emphasis on improving health and education outcomes, there revenue mobilization. Too much equalization--if are many reasons why increased public spending not well designed--could reduce the incentives does not necessarily translate into better services for subnational governments to mobilize own- and better outcomes. How (effectiveness) and revenues,underminingoveralltaxrevenues.Inequal- where (sector allocation, levels of government, geo- ities among regions could also induce people to graphical distribution) resources are used deter- migrate to regions with better economic prospects. mines whether they lead to positive outcomes. For Poor regions may be less capable of handling funds example, Thailand and Peru both increased public than rich ones, or less concerned with poverty alle- spending on primary schooling, yet completion viation than the center. Rich regions are also pow- rates fell in Thailand while they rose in Peru. Con- erful, and they do not like to lose out to poorer versely, health spending in Mexico and Jordan regions. Reasons for persisting inequalities in the diverged, but child mortality fell in both countries East Asian context include the coastal development (World Bank 2004a). Indeed, effectiveness in the strategy in China, which allowed some provinces to use of funds varies widely across countries. Studies get rich first, and Indonesia's policy of channeling show that the nonpoor typically receive a dispro- revenues from natural resources to regions with portionate share of public spending (Filmer 2003). separatist tendencies. Yet, even if spending is well targeted, it does Based on these preliminary findings and the not always reach frontline service providers (teach- broader implications of interregional inequity, ers, doctors), and when it does, providers must we offer recommendations to address such inequity be both present and effective in their jobs (World in four areas: information management, transfer 80 East Asia Decentralizes systems, the role of the central government, and Country-specific recommendations include: further research and analysis. The policy debate on fiscal inequalities--and · China's transfer system could aim to reflect local intergovernmental fiscal relations more generally-- revenue capacity and expenditure needs in a requires more data. Without better information on rules-based manner, rather than through ad hoc the size of inequalities and how they have evolved, distribution. Besides a more transparent system, informed debate over which are acceptable is China should also aim for simpler equalization impossible. Each East Asian country suffers from a mechanisms--perhaps by combining various dearth of data. Those analyzed here are far from equalization grants into one overarching grant. perfect, and we could not include other countries To further enhance both transparency and sim- because they lack fiscal data at the subnational plicity, the country could systematically identify level, especially below the first subnational tier. total equalization funds, based on tradeoffs Data on differences in the cost of delivering services between equalization, growth, and incentives for are also critical in assessing interregional dis- local revenue mobilization and expenditures. parities. Governments must set up monitoring · Although Indonesia's transfer system introduces systems--an undertaking that requires significant the notion of expenditure needs and revenue resources. Indonesia maintains a database that-- capacity through its equalization grant, it should supported by the legal requirement that regions define a more equalizing DAU by phasing out report--includes fiscal information on most of 410 the transitional elements and the "hold harm- local governments. In China, subnational govern- less" portion of the allocation. In terms of trans- ments have fiscal data, but the central government parency and simplicity, Indonesia should focus has little of this information. Some countries also on more consistent treatment of natural resource need to adjust their accounting and budget classifi- revenues in revenue sharing and the equaliza- cation systems. China, for instance, classifies signif- tion formula. The country should introduce a icant fiscal resources as "extrabudgetary" and infor- selective system of specific grants combined with mation on those funds is even scarcer, even though an onlending window to promote local financ- they fundamentally alter conclusions on fiscal ing of national priorities. disparities. · In Vietnam, negotiation and discretion remain, More data analysis also needs to inform the primarily for "surplus" provinces and local gov- policy debate. Countries should aim to regularly ernments. The central government may wish to review the results of their intergovernmental fiscal consider introducing conditional grants or systems, including fiscal and service disparities. For other transfer instruments--with or without example, following the highly successful example of matching provisions--to create incentives and South Africa, Indonesia published the first of what provide financing for expenditures in priority will be a regular intergovernmental fiscal review in areas such as health and education. Local gov- March 2004. Such reports will allow policy makers ernments could also use such funding for social to evaluate the intergovernmental fiscal system. assistance programs, along with grants for capi- Regardless of whether more or less fiscal equaliza- tal infrastructure.27 tion is desirable, countries have significant scope to improve the design of their intergovernmental sys- Our data show that although huge fiscal dispari- tems. Central governments must determine their ties exist across subnational governments, inequali- equalization goals and priorities regarding income ties in health and education outcomes are relatively levels, fiscal capacity, expenditure needs, and per modest. Why? What are the countervailing forces capita revenues (Bahl 2000). The goals of equaliza- within each country or groups of subnational tion grants are often unclear: some grants embody provinces and municipalities? Do people tend to features of earmarked grants, for example, usually migrate from one region to another? Is there a push based on temporary considerations.A more compre- for asymmetrical decentralization? What are gov- hensive goal would require each local government to ernment preferences, and what role do they play in deliver a minimum level of goods and services. policy making? Debate regarding the center's role Fiscal Disparities in East Asia: How Large and Do They Matter? 81 and the impact of its programs is also critical. cial crisis without any consultation. Despite a Supreme These and many other questions present challeng- Court ruling in favor of local governments, disagreement remains on whether the formula should be subject to appro- ing lines for further empirical analysis. priation or decreased as part of the congressional appropri- ation process. See Manasan 2002. Endnotes 16. The United States has seen some success in using these data for grant design. See Rafuse 1992. 1. Migration can be considered excessive in an economic 17. This ignores the effect of borrowing. However, since richer sense, if the marginal productivity of a worker would be regions are likely to have more access to borrowing, the higher in his place of origin, or if the congestion costs in the conclusions on equalization are likely to hold. region of destination are larger than the private benefits 18. This chapter adopts the UNDP procedure for computing obtained from the better services; see Ahmad and Craig in HDI. Provincial HDI is based on three indicators: longevity, Ter-Minassian, ed., 1997. as measured by life expectancy at birth; education attain- 2. For an extensive discussion of this point, see Tanzi and ment, as measured by a combination of adult literacy (two- Shuknecht 1995 and Prud'homme 1995. thirds weight) and combined primary, secondary, and terti- 3. A review, assessing constitutional rights to education and ary enrollment ratios (one-third weight); and standard of health care in 187 countries, concludes that of the 165 living, as measured by real GDP per capita in terms of pur- countries with written constitutions available, 116 made chasing power parity (in dollars) (UNDP 1999). We broadly reference to a right to education and 73 to a right to health define persons per hospital bed in terms of public hospitals. care (Gauri 2003). (In Indonesia, the indicators include both special and 4. In the Philippines, for example, IRA allocations have some- general hospitals.) We define literacy rate as the percentage what detracted from LGUs' incentive to improve the collec- of people aged 15 and above who cannot read and write a tion of own-source revenues (World Bank 2003a). short, simple statement about their everyday life (World 5. Art. XVIII (a), sub. 2, of the Indonesian Constitution states, Bank 2003c). We define life expectancy as the number of "The relationship in finances, public services, utilization of years a newborn infant would live if prevailing patterns of natural resources and other resources . . . shall be regulated mortality remained the same throughout its life (World and executed fairly and equitably based on the law." Bank 2003c). 6. Art, II, sections 9 and 10, of the Philippine Constitution 19. Variation is measured through the coefficient of variation, states, "The State shall promote a just and dynamic social which is the standard deviation divided by the average. order that will ensure the prosperity and independence of 20. Although maximum-to-minimum ratios within countries the nation and free the people from poverty through poli- capture the same trends as more sophisticated measures cies that provide adequate social services, promote full (such as the gini coefficient), those ratios are more respon- employment, a rising standard of living, and an improved sive than other measures to the tail-end of the distribution quality of life for all. The State shall promote social justice across countries. in all phases of national development." 21. Specification 2 aims to address the problem of reverse 7. The figure for Vietnam excludes the oil-producing region causality between expenditures (revenues) and outcomes. of Ba-Ria Vung Tao. Otherwise, reverse causality would suggest the possibility 8. Fiscal disparities can differ depending on the indicator of that our regression estimates may be upwardly biased. The inequality. A future version of this study will also use the third set of regressions, which examines a longer time series population-weighted Theil indicator to calculate fiscal for China, will address the possible problem of multi- disparities. This will better measure the impact of fiscal dis- collinearity (when two or more independent variables are parities from the individual's perspective. approximately linearly related) in specification 2. 9. The level of fiscal disparities is sensitive to cost differentials 22. "In 1998, the public expenditure share in aggregate health across provinces. However, data on provincial cost of living spending was only about 20 percent, with households are not readily available for these countries. accounting for the remaining 80 percent" (World Bank 10. Data limitations, along with changes in these countries' 2000). decentralization processes, make a similar analysis very dif- 23. The problem of multicollinearity is when two or more ficult to perform for other countries in our sample. independent variables are approximately linearly related in 11. In principle, the law prescribes this as 25 percent of actual the sample data. revenues, but the DAU has so far been based on budgeted 24. Indeed, expenditures are highly correlated with GDP for revenues. China (.67), Indonesia (.42), and Vietnam (.67) (with sig- 12. For detail, see Hofman et al. 2003. nificance at the 1 percent level). It would be interesting to 13. As a result of the 1994 Tax Sharing System, revenue-sharing see if the same results hold at even lower levels of govern- arrangements changed from a negotiated, mildly equalizing ments. The only exception to the rule is China. system to one based on strict tax assignments (Wong 2002). 25. As shown in the previous set of regressions, life expectancy For more detail, see Ahmad et al. 2000. yields ambiguous results, as it is affected by a broad range 14. Besides fiscal equalization, general grants promote tax of factors for which we do not control. Infant or child mor- effort and local good governance, devolution of compul- tality would therefore be a better indicator. Lacking those sory functions, train ticket compensation, local develop- data, we have used combined gross enrollment as our fifth ment, and education. Exceptions to the formula-based allo- outcome indicator in this final set of regressions. cation criteria are local good governance, train ticket 26. The only exception to the results is HDI in specification 2, compensation, and local development. yet this is understandable, as GDP is one of its components. 15. Although the aggregate IRA share is fixed by law, the central 27. Martínez-Vázquez argues for the need to structure trans- government withheld 5 percent after the 1998 Asian finan- fers at the subprovincial level (2004b). 82 East Asia Decentralizes Bibliography ------. 2002. "Spatial Disparities in Developing East Asia: A Survey." Asian-Pacific Economic Literature 16 (1): 10­35. Ahmad, Ehtisham, and John Craig. 1997. "Intergovernmental Hofman, Bert, Kadjatmiko, and Kai Kaiser. 2003. "Fiscal Transfers." In Fiscal Federalism in Theory and Practice, Equalization in Indonesia's 2001 `Big Bang' Decentraliza- ed. Teresa Ter-Minassian. Washington, DC: International tion." Paper prepared for conference at Georgia State Uni- Monetary Fund. versity, Can Decentralization Help Rebuild Indonesia?, Ahmad, Ehtisham, Li Keping, and Thomas Richardson. 2000. Atlanta. "Recentralization in China?" Conference on Fiscal Decen- Jimenez, Emmanuel, and Donald Cox. 1995. "Private Transfers tralization, November. and the Effectiveness of Public Income Redistribution in the Bahl, Roy. 2000. "Intergovernmental Transfers in Developing Philippines." In Public Spending and the Poor: Theory and and Transition Countries: Principles and Practice."Washing- Evidence, eds. Dominique Van de Walle and Kimberly Nead. ton, DC: World Bank, Urban and Local Government, Washington, DC: World Bank. Municipal Finance. Lewis, Blane. 2003. "Local Government Borrowing and Repay- Bahl, Roy, and Jorge Martínez-Vázquez. 2003."Fiscal Federalism ment in Indonesia: Does Fiscal Capacity Matter?" World and Economic Reform in China." Atlanta, GA: Georgia State Development 31 (6): 1047­63. University, International Studies Program. Manasan, Rosario. 2002. "Fiscal Decentralization: The Case Balisacan, Arsenio M., Ernesto M. Pernia, and Abuzar Asra. of the Philippines." Background Paper for World Bank­ 2003. "Revisiting Growth and Poverty Reduction in Indone- sponsored course on Decentralization and Intergovernmen- sia: What Do Subnational Data Show?" Bulletin of Indonesian tal Fiscal Reform in East Asia, Bangkok, June 6­7. Economic Studies 39 (3): 329­51. Martínez-Vázquez, Jorge. 2003. "Improving the Design of Fiscal Birdsall, Nancy, and Estelle James. 1990."Efficiency and Equity in Decentralization in Vietnam." Presentation at Government Social Spending: How and Why Governments Misbehave." of Vietnam/World Bank workshop, Ha Tay Province, Washington, DC: World Bank; Policy, Research, and External February. Affairs; Population and Human Resources Operations. ------. 2004a. "The Design of Fiscal Decentralization: Princi- Bourguignon, François. 2003. "Transition of China's Northeast: ples and Best International Practice." Presentation to Public The Need for Combining Regional and National Policies." Expenditure Review and Integrated Fiduciary Assessment Paper for seminar on Development Strategy for Northeast Mission, Hanoi, April. China, Shenyang, Liaoning Province, China, December 3­4. ------. 2004b. "Improving the Design of Fiscal Decentraliza- Brodjonegoro, Bambang. 2000. "Regional Autonomy and Fiscal tion in Vietnam." Prepared for World Bank Vietnam Public Decentralization in Democratic Indonesia." Hitotsubashi Expenditure Review. Washington, DC: World Bank. Journal of Economics 41: 111­22. Martínez-Vázquez, Jorge, and Jameson Boex. 1998. "Fiscal Buchanan, James M. 1950."Federalism and Fiscal Equity." Amer- Decentralization in the Russian Federation: Main Trends ican Economic Review 40 (4): 583­99. and Issues." Washington, DC: World Bank, Economic Devel- Ebel, Bob. 2003. "Intergovernmental Transfers." Presentation at opment Institute. workshop on Decentralization and Intergovernmental Fiscal Martínez-Vázquez, Jorge, and Robert M. McNab. 2003. "Fiscal Reform. Washington, DC: World Bank, March 24­26. Decentralization and Economic Growth." World Develop- ------. 2003. "Subnational Revenues and Intergovernmental ment 31 (9): 1597­1616. Relations." Presentation at workshop on Decentralization McLure, Charles E., Jr. 1999. "The Tax Assignment Problem: and Intergovernmental Fiscal Reform. Washington, DC: Conceptual and Administrative Considerations in Achieving World Bank, March 24­26. Subnational Fiscal Autonomy."Washington, DC: World Bank. Filmer, Deon. 2003. "The Incidence of Public Expenditures on Milanovic, Branko. 2001. "Conflict between Horizontal Equity Health and Education." Background note for World Develop- and Maximum Poverty Reduction: How Best to Allocate ment Report 2004: Making Services Work for Poor People. Funds to Regions? An Empirical Analysis." Washington, DC: Washington, DC: World Bank; and New York: Oxford Uni- World Bank. versity Press. National Statistical Coordination Board. 2003. Philippine Statis- Garcia-Garcia, Jorge, and Lana Soehstianingsih. 1998. "Why Do tical Yearbook. Manila: National Economic and Development Differences in Provincial Income Persist in Indonesia?" Authority. Bulletin of Indonesian Economic Studies 34 (1). Pritchett, Lant. 2004. "Towards a New Consensus for Addressing Gauri, Varun. 2003. "Social Rights and Economics: Claims to the Global Challenge of the Lack of Education." Copenhagen Health Care and Education in Developing Countries." Consensus Challenge Paper. Washington, DC: World Bank, Development Research Group. Pritchett, Lant, and Jonah Gelbach. 2000."Indicator Targeting in Gill, Nicholas, and Andres Rodriguez-Pose. 2003. "Is There a a Political Economy: Leakier Can Be Better." Washington, Global Link between Regional Disparities and Devolution?" DC: World Bank. Research Papers in Environmental and Spatial Analysis 79. Pritchett, Lant, Deon Filmer, and Jeffrey Hammer. 1999."Health London: London School of Economics and Political Science, Policy in Poor Countries: Weak Links in the Chain." Wash- Department of Geography and Environment. ington, DC: World Bank. Government of Indonesia. 2002. Statistik Indonesia (Statistical Prud'homme, Rémy. 1995. "The Dangers of Decentralization." Yearbook of Indonesia). Jakarta: Badan Pusat Statistik. World Bank Research Observer 10 (2): 201­21. Hegedus, Jozsef. 2001. "Creating Change Networks for Local Qiao, Baoyun, Jorge Martinez-Vazquez, and Yongsheng Xu. Governance." Presentation for World Bank Institute/United 2003. "Growth and Equity Tradeoff in Decentralization Pol- Nations Development Programme, Initiative for Central icy: China's Experience." Atlanta, GA: Georgia State Univer- Asia, Bratislava, July 30­August 3. sity, International Studies Program. Hill, Hal. 2000."Intra-Country Regional Disparities." Paper pre- Rafuse, Robert. 1992. "Representative Expenditure." Washing- sented at Second Asian Development Forum, Singapore, ton, DC: U.S. Advisory Commission on Intergovernmental June 6­8. Relations. Fiscal Disparities in East Asia: How Large and Do They Matter? 83 Rao, M. Govinda. n.d. "Fiscal Assignment and Service Delivery." Van de Walle and Kimberly Nead. Washington, DC: World Presentation for Institute for Social and Economic Change, Bank. Bangalore, India. Wong, Christine. 2002. "Issues of Equalization in China." ------. 2001. "Challenges in Fiscal Decentralization: An Asian Presentation for workshop on Intergovernmental Fiscal Perspective."Paper presented at conference on Public Finance Relations in East Asia. Washington, DC: World Bank. in Developing and Transition Countries, Atlanta, GA. World Bank. 1993. "China: Budgetary Policy and Intergovern- Rao, M. Govinda, Richard M. Bird, and Jennie Litvack. 1998. mental Fiscal Relations." Report No. 11094-CHA. Washing- "Fiscal Decentralization and Poverty Alleviation in a Transi- ton, DC: World Bank. tional Economy: The Case of Vietnam." Asian Economic ------. 2000. "Managing Public Resources Better." Vietnam Journal 12 (4): 353­78. Public Expenditure Review. Washington, DC: World Bank. Ravallion, Martin. 1993. "Poverty Alleviation through Regional ------. 2002a. "China: National Development and Subnational Targeting: A Case Study for Indonesia." In The Economics of Finance: A Review of Provincial Expenditures." Washington, Rural Organization: Theory, Practice and Policy, eds. Karla DC: World Bank. Hoff, Avishay Braverman, and Joseph Stiglitz. Washington, ------. 2002b. "Brazil: Issues in Brazilian Federalism." Wash- DC: World Bank; and New York: Oxford University Press. ington, DC: World Bank, Latin America and the Caribbean ------. 1998. "Reaching Poor Areas in a Federal System." Policy Region, Poverty Reduction Economic Management Sector Research Working Paper. Washington, DC: World Bank, Unit, Brazil Country Management Unit. Development Research Group. ------. 2003a. "Philippines: Improving Government ------. 1999a. "Is More Targeting Consistent with Less Spend- Performance--Discipline, Efficiency, and Equity in Manag- ing?" Washington, DC: World Bank. ing Public Resources." Washington, DC: World Bank, ------. 1999b. "Are Poorer States Worse at Targeting Their Poverty Reduction Economic Management, East Asia and Poor?" Economic Letters 65: 373­77. the Pacific Region. Ravallion, Martin, and Gaurav Datt. 1990."Regional Disparities, ------. 2003b. "Decentralizing Indonesia." Regional Public Targeting, and Poverty in India." Washington, DC: World Expenditure Review. Washington, DC: World Bank, Poverty Bank; Policy, Research, and External Affairs; Population and Reduction Economic Management, East Asia and the Pacific Human Resources Operations. Region. Republic of China. Multiple years. Statistical Yearbook of the ------. 2003c. "World Development Indicators." Washington, Republic of China. Taiwan: Directorate-General of Budget, DC: World Bank. Accounting, and Statistics. ------. 2004a. World Development Report 2004: Making Services Republic of Vietnam. Multiple years. Statistical Yearbook. Hanoi: Work for Poor People. Washington, DC: World Bank; and Statistical Publishing House. New York: Oxford University Press. Searle, Bob. 2002. "Federal Fiscal Relations in Australia." Torino, ------. 2004b. "Chinese 11th Five-Year Plan." Concept note. Italy: International Centre for Economic Research. Washington, DC: World Bank, Development Economic Smoke, Paul. 2001. "Fiscal Decentralization in Developing Research Group. Countries: A Review of Current Concepts and Practices." ------. 2004c."2004 Public Expenditure Review and Integrated Democracy, Governance and Human Rights Programme Fiduciary Assessment." Aide mémoire, Multi-Donor Paper No. 2. Geneva: United Nations Research Institute for Mission, Hanoi, April 5­16. Social Development. ------. 2004d."Inequality in Latin America and the Caribbean: Tanzi,Vito, and L. Shuknecht. 1995."The Growth of Government Breaking with History?"Washington, DC: World Bank, Latin and the Reform of the State in Industrial Countries."Working American and Caribbean Studies. Paper. Washington, DC: International Monetary Fund. ------. 2004e."Managing Decentralization in East Asia."Wash- Ter-Minassian, Teresa. 1997. Fiscal Federalism in Theory and ington, DC: World Bank, Poverty Reduction Economic Practice, International Monetary Fund. Washington, DC: Management, East Asia and Pacific Region. International Monetary Fund. Yan. 2003. "Regional Equalization in China's Public Finance." UNDP (United Nations Development Programme). 1999. China Unpublished paper. Human Development Report. New York: United Nations. Zhang, Zhihua, and Jorge Martínez-Vázquez. 2003."The System Van de Walle, Dominique. 1995. "Incidence and Targeting: An of Equalization Transfers in China." International Studies Overview of Implications for Research and Policy." In Public Program Working Paper No. 03-13. Atlanta, GA: Georgia Spending and the Poor: Theory and Evidence, eds. Dominique State University, Andrew Young School of Policy Studies. 5 Public Expenditure on the Frontline: Toward Effective Management by Subnational Governments Edward Mountfield and Christine P. W. Wong East Asia has been home to some of the world's Of the five most populous low- and middle- most centralized public expenditure management income countries in East Asia--hereafter referred systems. From a macroeconomic perspective, this to as the EA51--four are designing programs to approach may have had its merits--for example, transfer expenditure responsibilities to lower-level allowing for a rapid fiscal response during the 1998 governments. Indonesia is implementing a "Big crisis. However, such top-down approaches have Bang" decentralization program. Thailand has typically not been conducive to locally responsive committed to implementing a gradual but poten- service delivery (World Bank 2000d, pp. 31­33). tially major decentralization program. Vietnam has Countries have often achieved fiscal discipline at recently legislated to shift greater expenditure the expense of effectiveness and efficiency. To authority and responsibility to provincial People's accomplish broader national objectives--not only Councils. And the Philippines continues to explore macroeconomic stabilization but also poverty policy options for strengthening its comprehensive reduction and effective and efficient local service program of decentralization started a decade ago. delivery--public expenditure management is mov- The fifth country, China, is alone in moving in the ing closer to the frontline. opposite direction: since the late 1990s it has insti- Intergovernmental reform of expenditure man- tuted a program to reform public expenditure agement presents some major opportunities. Local management that will--at least in the short run-- management of spending can support service increase central scrutiny of local spending. How- delivery that is more responsive to the needs and ever, since China has gone furthest among the wishes of local people and more efficient given EA5 in assigning spending responsibilities to local conditions. Equally, there are significant subnational governments, the recent changes threats: duplication, poor coordination, growing should be seen as part of a continuing effort to inequity, and even the collapse of essential services. find the right balance between central control 85 86 East Asia Decentralizes and local flexibility, rather than a rejection of followed by Vietnam, with 48 percent. At the lower decentralization. end is Thailand, where subnational spending is just Given the EA5's relatively recent moves to 10 percent of the total. The East Asian archipelago decentralize expenditure management, deriving countries are somewhere in the middle: the conclusions as to whether these countries have Philippines is at 25 percent and Indonesia reached "failed" or "succeeded" in terms of service delivery 31 percent after Big Bang decentralization. Table 5.1 would be premature. Instead, this chapter reviews compares the subnational expenditure shares of the experiences so far with decentralizing expenditures EA5 with those of other countries. in the EA5. The chapter presents some empirics on In a study of some 100 countries,Roy Bahl and his expenditure decentralization and provides an ini- colleagues at Georgia State University found the tial assessment of the nascent expenditure manage- share of subnational expenditures to be positively ment systems at subnational levels in these coun- correlated with per capita gross domestic product tries. Finally, the chapter analyzes the extent to (GDP) and land area, and negatively correlated with which the incentive framework and institutional ethnic diversity.2 Transition economies as a group environment in each country are likely to foster also tend to be more decentralized than expected successful subnational expenditure management. from the other variables. Led by the two transition economies of China and Vietnam, the EA5 have decentralized more than the average 14 percent of Subnational Expenditure Management total expenditures that Bahl et al. found for develop- in the EA5: Some Empirics ing countries, but have decentralized less than the The extent to which EA5 countries have decentral- average of 35 percent in developed countries. Even ized public spending varies. By the conventional consideringthesefactors,however,theauthorsfound measure of subnational shares of total spending, that China and Vietnam were more decentralized China is by far the most decentralized of the five, than expected in the 1990s, Indonesia and Thailand with subnational spending at 69 percent. It is less,andthePhilippinesexactlyattheexpectedlevel.3 TABLE 5.1 Expenditure Decentralization in East Asia in Comparative Perspective Subnational expenditure (% of national expenditure) China 2002 69 Indonesia 2002 32 Philippines 2001 26 Thailand 2001 10 Vietnam 2002 48 Developing countries 1990s 14 Transition countries 1990s 26 OECD countries 1990s 32 Other large countries, 1990s: Germany 40 India 46 Japan 61 Pakistan 29 Russia 38 United States 46 Sources: Bahl 2002; World Bank 2003a; staff estimates. Note: OECD Organisation for Economic Co-operation and Development. Public Expenditure on the Frontline: Toward Effective Management by Subnational Governments 87 However, while subnational spending share is a governments are responsible for not only providing convenient measure for comparing decentraliza- but also financing their service responsibilities, with tion across countries, it is often misleading when little support from either earmarked or equalization the locus of spending does not coincide with the transfers. level of decision making, as has often been the case Differences in how countries report the division in East Asia (Bahl 2002). Vietnam is a perfect illus- of responsibilities may also muddy comparisons. tration of this shortcoming. Despite its high subna- Such differences may simply reflect different trans- tional spending share, Vietnam was--until it fer instruments. A central government that ear- implemented the new State Budget Law in January marks transfers for education would more likely 2004--formally one of the least decentralized report education as a joint responsibility than a countries in the world. Local governments were central government that provides block grants to essentially carrying out deconcentrated functions local governments. at the behest of the central government, which Given the heterogeneity of the EA5, another determined service levels and standards. dimension is worth mentioning: multiple levels of Another conventional yardstick for measuring subnational government. China's intermediate level decentralization is the pattern of expenditure of government is larger than that of most countries: assignments. Table 5.2 shows expenditure assign- each such government has an average of 45 million ments for the EA5 and other countries. China people. Prefectural units--mostly municipalities-- again appears to be the most decentralized of the have an average population of 3.7 million, making five. Compared with the other countries, China them as large as or larger than provinces and states has transferred a much larger portion of expendi- in many countries. Considering the size of countries ture responsibilities to local governments, includ- themselves is also important in drawing inferences ing several big-ticket items--not only education about state versus local service provision. Even and health, which are common local functions in accounting for size, China is extreme in having many countries, but also unemployment benefits transferred so many responsibilities to the lowest and pensions. The latter make China highly levels: municipalities and counties--not provinces unusual. Pensions and unemployment benefits or the central government--are responsible for often exceed the financing capacity of subnational social welfare provision. governments, and their cyclical nature makes them often inappropriate for such governments. In Public Expenditure Management almost all other countries, these responsibilities Processes in the EA5 are either central or shared with state govern- ments. Indonesia and Thailand are also quite East Asian countries have substantially expanded decentralized compared with other countries, with the extent to which subnational levels manage pub- education and health assigned solely to local gov- lic expenditure. The most critical expenditure man- ernments. In the Philippines, health is a joint agement processes, which are nascent and still responsibility of state and local governments, evolving, include the following: while education remains a central government Policy analysis and planning. The policy respon- responsibility. sibilities of government departments and the legis- Table 5.2 shows expenditure assignments, but lature must be clear, appropriate, and transparent. again we must be cautious in interpreting such An effective policy analysis and planning process summary tables. The table identifies only the level should identify clear development goals and priori- of government responsible for providing services, ties, provide a realistic fiscal framework for public not necessarily the one that finances the services. expenditure rather than encouraging a "shopping Indonesia, for example, assigns costly functions like list" approach, and inform planning with sound, education and health to subnational levels, but evidence-based policy analysis. Such a process finances them through specific transfers from the should also involve sector departments within the central government. Subprovincial governments in subnational government rather than just planning China, in contrast, receive limited transfers from and finance departments; provide a mechanism for which to fund their expenditure mandates. Local the central government to coordinate policy with 88 TABLE 5.2 Expenditure Assignments in East Asian and Comparison Countries Environment Foreign and natural Unemployment Industry and Social Country Defense affairs resources insurance agriculture Education Health welfare Police Highways EA5 countries: China F F F, S, L L F, S, L L L L L F, S, L Indonesia F F .. .. L L L L F F, S, L Philippines (1) F F .. .. .. F F F .. .. Philippines (2) F F .. .. .. F, S, L F, S, L S, L S, L .. Thailand (1) F F .. .. .. L L .. L L Thailand (2) F F .. .. .. L L .. L L Vietnam F F F, S, L .. F, S, L S, L S, L F, S, L F F, S, L Comparison countries: India (1) F F F, S F, S F, S F, S S F, S S F India (2) F F F, S F, S F, S F, S, L S, L F, S S F Japan (1) F F .. .. L F, L F, L F, L L .. Japan (2) F F .. .. .. L F, L F, L F, L L Malaysia (1) F F L .. F, S F F, S F, S F F Malaysia (2) F F L .. F, S F F, S F, S F F Countries that assign constitutional powers: Canada F F F, S F, S C S S(F) F, S F, S S United States F, S F F, S F, S S S, F S(F) F, S F, S Switzerland F F C C F, S C, F, S S, C F, C S F, S Australia F, S F F, S C S, C F, S F, S C S, F F, S Germany F F, S C C C C, S C, F, S C C, S C Austria F F F, S F F F, S C, F, S C F, S F, S Sources: World Bank 2003a, annex 4; other World Bank information. Note: 1 responsibility, 2 provision, F federal/national, S state/province, L local, C concurrent, (..) not identified. Public Expenditure on the Frontline: Toward Effective Management by Subnational Governments 89 subnational governments; and institute processes holders. And governments should monitor budget that allow for extensive consultation with service outputs and outcomes to the extent possible. users, frontline providers, donors, and civil society. Auditing and evaluation. A clear distinction Finally, both the planning framework and the should be made between internal and external underlying analysis should be made public and audit. End-year accounts of subnational expendi- widely disseminated, and the planning framework tures should be audited under a reasonably rapid updated regularly--ideally annually, as part of the timetable, ideally by an independent auditor general budget cycle. or under his or her supervision. Audit reports Budget formulation. An effective budget cycle should be scrutinized by the legislature and made should have a logical sequence and timing and public in a timely fashion. When the auditor general build on the planning framework. The budget issues an adverse report, policy makers need to take should also reconcile development goals and pol- appropriate follow-up action. Audits should include icy priorities with a realistic assessment of fiscal budget outputs and outcomes as well as financial constraints. The budget should cover all subna- inputs to the extent possible. They should also tional expenditures, including capital and current systematically evaluate a sample of projects and pro- expenditures as well as subsidies, transfers, inter- grams, with the results drawn upon in planning and est, and net lending. The budget should explicitly budget formulation for subsequent years. incorporate the use of extrabudgetary funds, including those financed by donors; minimize ear- Assessing East Asian Countries marking of resources; and clarify the future budget implications of investment decisions. Once again To what extent do the EA5 countries have these all stakeholders should be involved, with budgets core expenditure management processes in place? widely disseminated and enough time allowed To answer that question, we reviewed published for consultation and debate before legislative and unpublished reports on expenditure decentral- approval. ization in these countries.4 We found that such Budget execution. Budget execution should processes generally remain weak at the subnational adhere as closely as possible to the estimates level (see table 5.3). Planning processes are discon- approved by the legislature (although under clearly nected from fiscal planning and budgeting prescribed circumstances the finance agency may processes, and poorly coordinated with planning have the authority to change departmental alloca- at other levels of government. Budget formulation tions during the fiscal year). Annual cash require- suffers from major delays and is fragmented, with ments for each spending unit should be forecast parallel budgets for investment projects and recur- so cash management is efficient and consistent rent expenditures managed by separate institu- with releases and authorizations to spend. Arrears tions. Budget execution is slow, with long delays in should not be significant as a proportion of total effecting appropriations, significant divergence spending, and planning and control systems should between approved appropriations and outturns, protect against overspending, provide effective and corruption in procurement. Monitoring and crosschecks between human resource management accounting systems often produce data that are nei- systems and payroll, and establish competitive and ther timely nor accurate. Auditing is typically weak, transparent procurement. and evaluation almost nonexistent. Monitoring and accounting. National law needs Part of the explanation is that countries such as to establish financial reporting and accounting Indonesia, Thailand, and Vietnam are just begin- rules for subnational governments that reflect rec- ning to decentralize expenditures. However, time ognized accounting standards. Subnational govern- does not automatically solve these problems: China ments need to collect comparable data, to facilitate and the Philippines are still struggling to build the policy analysis as well as management of national institutions they need despite starting one and two fiscal targets. In-year and end-year statements decades ago. should be produced shortly after the end of the fis- We provide further assessment, with country- cal year and made routinely available to all stake- by-country examples, on the next page. 90 East Asia Decentralizes TABLE 5.3 Strength of Core Expenditure Management Processes at Subnational Levels China Indonesia Philippines Thailand Vietnam (1980) (2001) (1992) (1999) (1994­1996) Policy analysis and planning Budget formulation Budget execution Monitoring and accounting Auditing and evaluation Source: Authors' assessment. Note: strong, medium, weak. Start date of decentralization in parentheses. Policy Analysis and Planning. Vietnam has an recurrent budgeting at each level of government. established tradition of planning at subnational as For example, a provincial Finance Bureau deter- well as national levels. Provincial Planning and mines the annual capital provision, but a provincial Investment departments coordinate inputs to a Planning Commission allocates these resources Five-Year Plan and a five-year Public Investment between projects. These agencies rarely use cost- Program (PIP). The national Five-Year Plan and benefit analysis or other selection techniques. PIP then consolidate these plans. However, prob- In Indonesia, the center has traditionally domi- lems persist. First, under Vietnam's system of "dou- nated planning processes. These processes have ble subordination," reporting arrangements remain started to evolve since the Big Bang decentralization confused: the provincial departments of Planning of 2001, and the government has introduced a and Investment report not only to the local People's number of subnational planning instruments. Par- Committee and People's Council, but also to the ticipation of civil society at the village level has central Ministry of Planning and Investment, and greatly increased, reflecting new regulations that through it to the National Assembly. Second, the prescribe this participation. Planning processes are local planning process occurs with little reference thus both top-down and bottom-up. However, to fiscal constraints. Provinces submit investment major weaknesses remain. It is unclear how and plans that would typically add up to much more when the top-down and bottom-up planning than the resources available. Third, the planning processes are integrated, for example. There is a process encompasses only the investment budget large amount of duplication between these plan- (capital projects plus the often donor-funded ning processes, as well as some inconsistency. The development projects), paying little attention to the planning process is largely devoid of considerations cost implications of development expenditures or of affordability: only the annual regional plan takes the future budgetary implications of new spend- fiscal considerations into account. This results in ing commitments. Vietnam has started piloting much disappointment at all levels of government: medium-term expenditure planning, both at district governments must make difficult choices national and provincial levels, bringing planning of and sharp cuts in the wish lists of villages, bureaus, capital and recurrent expenditures under a single and the district as a whole. Also similar to Vietnam framework. However, establishing this system fully is the focusing of planning on the development will take some years. budget and capital projects rather than on the In China, provincial Planning Commissions budget as a whole. Altogether, this situation is address the priority areas declared by the central inconsistent with Indonesia's plans to require government in its Economic Development Plans. "performance budgets" from regional governments As in Vietnam, capital budgeting is separate from under Government Regulation 105 2000. However, Public Expenditure on the Frontline: Toward Effective Management by Subnational Governments 91 only a few regions, such as Semarang, have prepared of three years will also promote local budget to move in that direction. The lack of clarity regard- planning. ing the postdecentralization role of the national Although seemingly coherent, this subnational development planning agency, BAPPENAS, further budget process has continuing weaknesses. Perhaps confuses these processes. the most significant problem is the separate In the Philippines, the Local Government Code Department of Finance and Department of Plan- prescribes a participatory approach to planning, ning and Investment at the provincial level, and the but compliance is poor. The code mandates that relatively weak coordination between these finance each local government have a comprehensive, mul- and planning functions. This produces a discon- tisectoral development plan formulated by its Local nect between planning and budgeting processes, Development Council (LDC) and approved by its and between recurrent and capital spending, with legislature. However, fewer than half of local gov- plans focused on capital investments and little for- ernments have an LDC in place. According to the ward planning for recurrent spending. The rural code, one-quarter of LDC members should come transport sector, for example, has recently attracted from nongovernmental organizations (NGOs) and major capital investment: only 269 communes now people's organizations. However, one survey found have no road access to district centers. However, a that fewer than one-third of local development lack of maintenance provisions means that many plans have benefited from meaningful input from roads fall into disrepair soon after construction is NGOs and people's organizations. The planning complete. Requiring communities to contribute to that does exist focuses heavily on capital projects. road maintenance places a disproportionate bur- Some cities and provinces report having medium- den on the poor: World Bank staff estimate that and long-term development plans. However, most such annual contributions equal 9.8 percent of the smaller local governments have only an Annual annual poverty line in the relatively poor Northern Investment Plan. Project prioritization is usually an Uplands, compared with 4.6 percent in the country ad hoc process, conducted with little reference to as a whole. costs and benefits. Local officials report that their In China, the 1994 Budget Law governs the investment plans are formulated independently of budget formulation process at central and local lev- regional and national investment plans, and vice els, but major weaknesses in subnational budgeting versa. persist. The period allowed for preparing the budget is too short, often beginning only two Budget Formulation. In Vietnam, provincial months before the start of the budget year in Departments of Finance manage budget formula- January. This has a cascading effect through the tion at subnational levels, supervised by the provin- five-tiered system, since each tier has to wait for cial People's Council. Province-level spending units information from the higher level before it can for- supervised directly by the provincial government mulate its own expenditure and transfer plans. submit their spending bids to the provincial Because capital subsidies usually require matching Department of Finance. At the district level, funds, subnational governments have to hold large Bureaus of Finance filter the spending bids of reserves. Budget formulation occurs on an annual district-level spending units and communes. The basis, and the budget is not explicitly linked to a provincial People's Councils previously submitted medium-term strategy or a multiyear plan. Instead, their budgets to the central Ministry of Finance for budgeting is essentially incremental, characterized review and approval by the National Assembly. by passive distribution of resources among exist- Under the new State Budget Law, however, the ing structures and services. Extrabudgetary funds provincial People's Councils have enhanced finance portions of expenditures in most subna- authority to approve provincial spending plans, tional departments, and many local governments with the National Assembly responsible principally finance investments off-budget. The main budget for approving revenue shares and transfers from includes considerable earmarking. the center, plus the consolidated budget. Transfers Budget formulation in Indonesia's regions from the center to the provinces for stable periods remains principally an administrative exercise, 92 East Asia Decentralizes largely guided from the center, with district govern- However, only a small proportion of Thailand's ments and regional councils getting involved late in public spending is managed at the local level. the process. All spending units submit their budget proposals to the local government's budget com- Budget Execution. In Vietnam, provincial People's mittee, which in East Lombok, for example, Councils approve subnational budgets. However, includes representatives of the regional secretary, provincial and district branches of the State Trea- the finance bureau, the revenue office, the planning sury Department--part of the central Ministry of bureau, and the legal bureau. The budget commit- Finance--supervise financial execution of the tee reviews revenue estimates and spending plans approved budget. Treasury offices are responsible before finalizing the budget proposal around Octo- for issuing line-item allotments, approving checks ber for submission to the Regional Council. At least drafted by spending units against those allotments, in some provinces, the Regional Council approves maintaining records of unspent balances, and com- the budget before the start of the budget year in piling periodic financial reports. These mechanisms January. However, regional budget preparation have proven broadly effective in preventing over- during the first few years of decentralization has spending, helping Vietnam earn its reputation for been compressed as a result of late approval of the fiscal prudence. Although procedures exist for real- national budget, which determines transfers from locating the budget during the fiscal year, divergence the center. between budget plans and outputs is smaller than These processes are further complicated by con- in most countries at similar levels of development. tinuing confusion regarding relationships between However, multiple layers of control also contribute national and subnational governments, and to delays in projects and programs. Furthermore, between agencies at each level. According to regula- a portion of subnational expenditures--although tion 105 2000, the Ministry of Home Affairs is still often eventually accounted for by Treasury-- in charge of issuing budgeting guidelines, but the remains outside the budget execution process. Ministry of Finance sets standards for financial Indonesia has established procedures for appor- management throughout the public sector. tioning and executing the approved budget. How- In the Philippines, the integrity of local budget- ever, delays in issuing warrants pose a significant ing is distorted, as in Indonesia, by poor revenue problem. The main reasons for these delays are lack estimates during the budget formulation process. of information about revenues from the center In particular, the Internal Revenue Allotment (IRA) and lack of clarity of central regulations regarding supplied by the central Department of Budget and decentralized funds. In the first year of decentral- Management has been highly unreliable: in 1998 ization, many regional governments were forced to the IRA estimate was 9 percent short of actual engage in a major budget adjustment. The main releases. The shortfall dropped to 2 percent and cause was a centrally mandated increase in civil 6 percent in 1999 and 2000, respectively, but shot service wages announced in July 2001 (although up to 15 percent in 2001. Budgeting is largely con- contingency transfers offset the impacts of this ducted in an incremental fashion. Community par- increase). ticipation is limited: budget hearings are officially In China, likewise, budget execution at local lev- open to all interested parties, but local governments els has been characterized by slow disbursement. often do not circulate the budget document or Late approval of the budget and the long delay in expenditure statements. effecting appropriations mean that spending units In Thailand, annual budgeting has been rela- cannot plan their spending efficiently. A significant tively well-established in local administrations portion of expenditures thus occurs during the last since it was first instituted. Local authorities months of the year. prepare their budgets in conformance with central Procurement processes remain weak and poorly mandates. Executive bodies usually ready their administered at national and subnational levels in budget documents between May and July and sub- most EA5 countries. In Vietnam, the Ministry of mit them to the local council in August, with final Planning and Investment formally oversees pro- endorsement by provincial governors or district curement at all levels of government, but its scrutiny heads by the start of the budget year in October. is weak and indirect, and a clarifying Procurement Public Expenditure on the Frontline: Toward Effective Management by Subnational Governments 93 Ordinance has yet to be approved. In Indonesia, Law tems. Each country is implementing new financial 25 of 1999 (now Law 32 of 2004) allows regions to management information systems that should establish their own procurement regulations, with strengthen overall government reporting. upcoming tenders published in local newspapers. In China and Indonesia, each budget level main- However, Presidential Decree 18 of 2000 allows for tains its own accounting system.Both countries have preferential treatment of local bidders, and many established some reporting requirements, but insti- regional officials see procurement as a means to tutional confusion over the powers of central agen- support local contractors rather than a way to cies to specify and enforce these requirements has obtain the best price. In the Philippines, the Local undermined them. In Indonesia, for example, the Government Code decentralized procurement to decentralization law splits subnational responsibility local governments at a time when the legal and for financial management between two ministries-- institutional framework was unclear and nontrans- Finance and Home Affairs. In both countries, the parent. Although the country has since strength- quality of reporting is poor, financial reports are dif- ened the national procurement framework, local ficult to compare, and consolidating these reports policies and practices still raise concerns, including is difficult. This situation reflects poor financial delays, excessive local preference, and contract price management information systems. For example, negotiation, which often becomes an entry point for in Indonesia, subnational units submit quarterly corruption. In Thailand, local politicians often budget reports and financial accounts to the Min- intrude in the procurement process. istry of Finance mostly on paper. These two coun- tries are investing in strengthening their financial Monitoring and Accounting. Experience from management information systems. However, success around the world has underscored the importance will depend on their ability to resolve issues regard- of having the central government set and enforce ing the authority of central government to impose standards for financial reporting. National law accounting systems on subnational governments. needs to provide reporting and accounting rules for both the central and subnational levels. Such provi- Auditing and Evaluation. In Indonesia, local audit sions help ensure that subnational governments arrangements remain confused. Law 25 of 1999 generate financial reports that are comparable with determined that"prevailing regulations"should gov- those of other localities. They also help ensure that ern audits of local budgets. These regulations the central government can generate data for the assigned the central government's internal auditor, entire government, for both analyzing policy and the Supreme Audit Authority, the Inspectorate Gen- managing fiscal targets. Local capacity constraints eral of the province, and the Inspectorate General may also require the central government to lead in of Home Affairs as external auditors. Presidential this technically complex area. However, under a Decree 74 of 2001 has since assigned three internal systematic program, local governments can often auditors the right to audit local budgets: the local quickly develop their capacity to fulfill this task. government's auditor,the province's auditor,and the China, Indonesia, the Philippines, and Vietnam inspectors general of line ministries on technical all have programs to modernize their financial aspects. According to Law 5 of 1973, the Supreme management information systems at national and Audit Agency--the only external auditor in the subnational levels. In the Philippines and Vietnam, country--has the authority to audit all levels of gov- the central government has the legal authority to ernment. However, a draft law submitted to Parlia- set and enforce financial reporting standards. Both mentinSeptember2000putsthisauthorityindoubt. countries also have centrally specified charts of In Vietnam, the State Audit of Vietnam (the accounts, accounting systems, and reporting proce- external auditor) has existed for less than a decade. dures. However, while these systems can produce The Ministry of Finance recently issued 21 new general government reports, they are not timely or auditing standards that reflect international stan- accurate. For example, end-year accounts are not dards and strengthened reporting to the National available in Vietnam until 18 months after the end Assembly. Formally, the State Audit of Vietnam of the fiscal year. In the Philippines, local financial includes provinces and has five regional offices. In management is still based mainly on manual sys- practice, however, the role of the state auditor at the 94 East Asia Decentralizes subnational level has not been established, and governments have an important role to play. Unless most auditing is performed by the State Inspection they establish appropriate institutions and incentive Office (the internal auditor) and the network of frameworks, successful subnational expenditure provincial inspection offices. management processes are unlikely to develop. Key In the Philippines, the Commission on Audit factors include the following: has the power to audit local governments. However, Clear assignment of responsibilities. Clarity is the internal audit function is nonexistent in most essential regarding which expenditures each level of local governments. Until recently, the commission's government is responsible for. According to regional directors were overstretched, personally Wallace Oates's "decentralization theorem," "each supervising the audit of all government agencies in public service should be provided by the jurisdic- the region, although the government took steps in tion having control over the minimum geographic 2002 to address that problem. area that would internalize the benefits and costs of Auditing is a weak link in China's subnational such provision."5 The European Union has adopted expenditure management. Local audit bureaus do the "subsidiarity" principle, which asserts that ex post audits of local budgets. Audit bureaus may lower levels of government should have responsi- also select particular departments for a more in- bility for spending and delivering services unless depth audit and are also charged with auditing there is a convincing case for assigning that respon- extrabudgetary funds. Each local government at the sibility to higher levels. province level and below has its own audit bureau Matching of resources to responsibilities. The with similar responsibilities. However, although resources available to each subnational government local audit bureaus also report to the People's Con- shouldreflectthecostsof theservicesitmustprovide. gress, they are under the direct authority of the Such a match can occur through some combination executive branch, compromising their indepen- of tax and revenue assignment, tax- and revenue- dence. They are also underresourced and in need of sharing agreements, unconditional grants, condi- staff training. tional block grants (transfers subject to conditions In Thailand, only a small number of large-scale or service standards), targeted grants for specific local authorities, such as the Bangkok Metropolitan purposes or projects, and appropriate borrowing Administration, Pattaya City and Nakorn munici- authority. These resource flows should be as stable palities, have internal auditors. Two external audit- andpredictableaspossibletofacilitatelocalplanning. ing units--the Office of Auditor General and Matching of authority to responsibilities. Subna- the provincial auditing teams--have considerable tional governments will build stronger expenditure credibility but limited capacity to provide auditing management processes--and deliver services more services to all local authorities. effectively--if they have the necessary degree of In most EA5 countries, subnational budget eval- authority over those resources. Central govern- uation is still in the earliest stages of development. ments sometimes pass functions and resources to In China, local budget evaluation barely exists. In subnational governments but then deny them the Vietnam, the government has conducted three control they need to deliver responsive and high- provincial expenditure reviews and is planning quality services. Such overconstrained approaches several more. In Indonesia, the head of a region is rest on the often false assumption that central offi- required to present an annual accountability report cials have better information and sharper incentives to the Regional Council. However, because the than subnational officials. Such approaches distort budget documentation does not specify measurable expenditure management processes and impede the targets, evaluating outputs and outcomes is difficult. ability of local governments to innovate. Above all, micromanagement and second-guessing blur the Subnational Expenditure lines of accountability between tiers of government. Management: Getting Institutions Local capacity. Subnational governments need and Incentives Right the capacity to develop strong processes for manag- The institutional environment for subnational pub- ing expenditures for cost-effective service delivery. lic expenditure management will play an important In deciding how far to decentralize responsibility part in determining its evolution in the EA5. Central for public spending, central governments need to Public Expenditure on the Frontline: Toward Effective Management by Subnational Governments 95 TABLE 5.4 Do EA5 Countries Meet Conditions for Effective Decentralization of Expenditure Management? China Indonesia Philippines Thailand Vietnam (1980) (2001) (1992) (1999) (1994­96) Clear assignment of responsibilities Matching of resources to responsibilities Matching of authority to responsibilities Local capacity Local accountability National accountability Source: Authors' assessment. Note: high, medium, low. Start date of decentralization in parentheses. assess local capacity. Nevertheless, substantive published and unpublished reports on expenditure capacity is likely to develop only through "learning decentralization. At present, as table 5.4 shows, by doing," in which subnational governments are none of the EA5 countries score highly on any of given some budget to manage. Central government these conditions. Assignment of responsibility is has an important role to play in creating such an often fuzzy. Resources and authority over those environment, as well as in mentoring and training resources are often poorly matched with responsi- local officials. bilities. Local capacity and accountability--to both Local accountability. If subnational expenditure local people and central government--are often management is to translate into cost-effective serv- limited. ices, local governments need to be accountable to The following sections elaborate on this citizens and their institutions. Such accountability assessment. includes external auditors and representative local assemblies, public interest bodies and civil society, Clear Assignment of Responsibilities. A lack of and individuals with concerns and grievances. Ana- clarity in assigning responsibilities is a common lysts often argue that boosting the share of local problem in the EA5. In Indonesia, these problems expenditures financed by own-revenues enhances are due mainly to hasty implementation of sweeping local accountability (although this argument has changes and the inexperience of reformers. This deficiencies in countries where only a small pro- experience illustrates the difficulty of applying theo- portion of citizens pay direct taxes). retical concepts of decentralization. Based on the National accountability. Because subnational subsidiarity principle, the decentralization law (Law administrations draw on transfers and deliver serv- 20) of 1999 gave all "authorities" to local govern- ices with national impacts, central governments ments except those specifically assigned to the center must find a way to hold them accountable but stop and regions. However, since "authorities" are short of micromanagement.Toward that end,central broader than functions,the result is much confusion and local governments could create multiyear over who is responsible for what, from legislation to "contracts" covering both expenditure and revenue planning to implementation. Further confounding assignments that include performance criteria and the assignment of responsibility is the fact that cen- minimum service standards. tral line ministries still account for a significant amount of local spending. Adding to the confusion and disputes, Law 22 of 1999 called for adjusting Assessing East Asian Countries sectoral laws and regulations to conform to the new To what extent have EA5 countries established these decentralization framework--in effect setting aside conditions for success? Once again we reviewed the functional responsibilities and operational 96 East Asia Decentralizes TABLE 5.5 Distribution of Expenditures and Revenues in China, by Administrative Level (percent) Expenditures Revenues 1999 China Hebei Gansu Hunan Jiangsu China Hebei Gansu Hunan Jiangsu Province 28.2 26.4 32.8 31.6 25.8 21.2 20.6 16.5 13.6 16.2 Prefecture 30.2 23.7 19.5 22.4 34.7 35.4 23.4 24.9 27.8 43.8 Countya 41.5 39.2 34.3 37.8 39.5 43.4 35.8 39.9 35.2 40.0 Township 10.7 13.4 8.9 0.0 20.2 18.7 23.4 0.0 Change since 1994/95 (in percentage points) Province 1.8 6.8 0.5 5.6 5.8 4.1 0.9 1.4 0.4 11.3 Prefecture 1.1 1.6 2.6 1.1 0.1 5.6 1.9 3.0 3.3 1.8 Countya 0.6 5.9 0.2 2.4 5.9 1.5 2.5 0.6 0.8 9.5 Township 11.0 1.9 3.8 0.0 1.5 3.8 3.6 0.0 Source: World Bank 2002a, table 8. a. County figures for China include townships. details that sectoral and ministerial decrees usually able. Second, this murkiness has worsened the fiscal contain. Legal battles have ensued, as some central status of the lowest tiers and adversely affected ministries sought to exempt their agencies from service delivery. This has been evident in rural pro- decentralization laws, at least until local govern- vision of basic education: provinces, prefectures, ments were ready to take on the functions.6 and counties have done little or nothing to help In China and Vietnam, the lack of clarity stems rural townships that lack the resources to imple- from a second dimension of the assignment prob- ment the national policy of providing nine years of lem: how to distribute functions among levels of free education. government in a multitiered administrative setting. Table 5.5 shows what happens in the absence of In both countries, the administrative systems oper- formal rules. For China as a whole, and for three of ate as nested hierarchies, in which the central gov- the four provinces in the sample, counties and ernment sets rules only for provinces, which set townships together lost expenditure shares from rules for districts and communes.7 1994 to 1999, indicating that they had fewer China's 1994 Budget Law spells out in broad resources to finance their responsibilities, which principles the division of functions between central did not change. These trends confirm the com- and subnational governments, but is silent on the plaint in many localities that each administrative division of labor between tiers of subnational gov- level tries to capture more revenues by redefining ernment. This leaves essentially all decentralized how to share "local" taxes with the level below. functions as concurrent assignments for the vari- Since the county and township levels together pro- ous subnational tiers. In practice, China typically vide the bulk of basic services, including 70 percent assigns the responsibilities based on economies of of total public expenditures on education and scale and the subsidiarity principle. For example, 55 percent of expenditures on health, this trend provinces operate universities and large hospitals, could significantly undermine services, especially while lower-tier governments run primary and in rural areas. Provinces, in contrast, have gained middle schools and small hospitals and health significant expenditure shares even though they are clinics. not directly involved in providing services or redis- This approach has led to two undesirable out- tributing income. comes in China over the past decade. First, expen- A third result of the lack of clear assignments diture assignments are murky, with a good deal of occurs when central ministries and departments uncertainty about which level of subnational gov- resist giving up their authority under decentraliza- ernment is responsible for what. With all subna- tion. In Vietnam, the roles of sectoral ministries tional levels jointly responsible, no one is account- and provinces in prioritizing expenditures are Public Expenditure on the Frontline: Toward Effective Management by Subnational Governments 97 unclear. Ministries establish unrealistic service huge deficits, they devolved more than enough rev- norms that provinces largely ignore. This phenom- enues in 2001 to match expenditure responsibili- enon is common in China and the Philippines as ties. In the Philippines, decentralization has also well. In fact, national standard setting on civil serv- transferred more resources than responsibilities to ice wages and salaries, and even nationally man- the barangays (fourth tier), specifying that they can dated wage increases, are common in all EA5 coun- levy a tax of up to 1 percent of gross receipts on tries to varying extents. This is extremely disruptive businesses and collect fees, but without specifying to local budgeting, given the large share of local that they perform any significant duties(Azfar et al. expenditures absorbed by personnel costs (see 2000). chapter 7). China reveals the opposite mismatch: subna- The realities of service delivery are often such tional governments account for a much larger share that simplistic assignment of complex services to a of expenditures (70 percent) than revenues (45 per- single level of government is not practical. Most cent). Central transfers--which finance about countries have overlapping assignments. However, 45 percent of subnational expenditures on clarifying responsibility for regulation, financing, average--fill the vertical fiscal gap. However, the implementation, provision, and maintenance of transfer system is not equalizing, often giving more assets is still important, as is ensuring that these to rich provinces than to poor ones (see chapter 4). assignments are clear to all, including the public. Thus, resource gaps emerge in poor regions with Most of the EA5 countries have considerable scope inadequate self-financing, and at lower tiers of for improvement in this regard. the administrative hierarchy such as counties and townships. Since fiscal capacities vary greatly across Matching of Resources to Responsibilities. The localities--reflecting the uneven distribution of literature on fiscal federalism implicitly assumes economic activities--the result has been large and that local governments are largely financing their growing service disparities between urban and own services--hence the choice between high-tax, rural areas, and between regions. high-service packages and low-tax, low-service As interregional disparities in spending have packages across localities. The reality is that there is risen sharply since the mid-1990s, public services no a priori reason why, under a well-designed inter- have gradually deteriorated in poor regions (World governmental system, the expenditure needs of Bank 2002a). In Yunnan Province on China's subnational governments will match their ability to southwestern border, 106 of 127 counties were generate own-revenues to meet those needs. Thus, reportedly unable to meet budgeted expenditures intergovernmental transfers play a crucial role, in 1995. In neighboring Guizhou, the poorest both vertically (in determining whether local gov- province in China, many counties could not meet ernments have the resources to perform their payroll (Dai Xiaoming 1997). In the Liangshan Yi assigned responsibilities) and horizontally (in minority Prefecture in Sichuan Province, this led to keeping interregional disparities to acceptable the elimination of free medical care and epidemic levels). prevention programs, among other services. Many All the EA5 countries face problems coordinat- clinics and health stations closed, and epidemic dis- ing the decentralization of revenues and expendi- eases thought to have been wiped out reappeared. tures, albeit for different reasons. In Thailand, the In 1996, only 40 percent of Yi children attended transfer of revenues has outpaced the transfer of school, a figure that dropped to 10 percent in responsibilities. In principle, the laws provide local poorer villages (Heberer 2001). With ambitious governments with minimum transfers, but the cen- mandates for many public services, local govern- tral bureaucracy has been slow to decentralize com- ments must raise their own funds off-budget, rely- mensurate functions. As a result, in mid-2003, local ing heavily on user charges to finance primary edu- revenues were about 22 percent of national rev- cation, and on the sale of medicines, supplies, and enues, while the country has decentralized only other assets to finance public health. about 4 percent of expenditures. Fiscal pressures have also reduced the resources In Indonesia, even though central authorities available to local governments in the Philippines. scrambled to decentralize expenditures to avoid With revenues falling from 18.7 percent of gross 98 East Asia Decentralizes BOX 5.1 The Backdoor Route to Decentralization in China While straining under diminishing fiscal resources that annual fee payments averaged almost but facing growing demands for services, local Ą 11,000 in the nine cities surveyed, while taxes governments in China have turned to a plethora were less than Ą 1,500. The ratio of fees to taxes of fees and levies on local constituents to bridge ranged from a high of 13 to a low of 4, with the the budget gap. These include education sur- highest in smaller cities. These fees were almost charges, traffic safety fees, family planning fees, entirely outside the purview of the budget, as dif- and contributions to National Day celebrations ferent departments collected and used the fees and preparation for the Olympic Games. The until public expenditure management reforms central government sanctioned these efforts, and implemented since 2002. Once authorized, the encouraged and even urged local governments reporting requirements for fees were lax, and to "find local solutions." finance departments were often unable to track Today local governments depend on extra- total collections and how they were used. Expen- budgetary resources to finance nearly half their diture management reforms are attempting to expenditures, on average.a In 1998, for exam- bring these extrabudgetary revenues under ple, the budget financed an average of only tighter supervision, but resistance is reportedly 53 percent of China's expenditures in education; widespread. tuition, fees, social contributions, and profits These extrabudgetary funds provide a back- from school-run enterprises provided the rest. In door route to decentralization in China, as they the health sector in 1999, budgetary appropria- give local governments real autonomy on both tions provided only 11 percent of total operating the revenue and expenditure sides, which they revenues in public hospitals and clinics, while lack under the formal budget. However, these out-of-pocket payments by patients accounted funds are not transparent, and the user charges for 59 percent. from which they are financed are often highly A 1999 survey in China's Shanxi Province regressive. illustrated the dominance of fees over taxes in some sectors and localities. The survey found a. Wong 1998; Fan 1998; and World Bank 2002a. national product in 1997 to 14.6 percent in 2001 often have functional responsibility without the and continuing to fall through 2003, the resulting authority to manage related public expenditures. budget deficits and rising interest costs have This undercuts the promised benefits of both devo- squeezed social spending and the ability of local lution and delegation strategies (see box 5.2). governments to fulfill their responsibilities. The One dimension of this problem is a lack of local decline is exacerbated by budgetary inflexibility at authority regarding sectoral spending allocations. the sectoral level. For example, in education, the Directives by the central government often tie a sig- ratio of personnel costs to the total recurring nificant proportion of revenues and resources budget rose to 92.4 percent in 2001, severely limit- transferred to local governments to specific sectors, ing room to maneuver (World Bank 2003e). functions, or services. None of the EA5 except China has assigned local In Vietnam, for example, provincial finance governments significant revenue bases, and China departments must allocate their budgets in did so through the "backdoor"--that is, largely off- accordance with norms specified by central line budget (see box 5.1). ministries. However, budgets allocated to provinces have not enabled them to meet all the obligations Matching of Authority to Responsibilities. imposed by the center. Provincial finance depart- Another significant shortcoming in the institu- ments, in consultation with the People's Commit- tional framework and incentive environment for tee and People's Council, have therefore used their subnational expenditure management has been fre- discretion in allocating funds between sectors and quent mismatch between authority and responsi- spending units. A new State Budget Law, which bility. Local governments and service providers took effect in January 2004, consolidated this Public Expenditure on the Frontline: Toward Effective Management by Subnational Governments 99 BOX 5.2 China: Devolution by Default Rather Than by Design During China's transition to a market economy, from the budget to the banking system reduced decentralization has occurred more by neglect central spending, while wage increases pushed than by design. Subnational expenditures have up the cost of labor-intensive services, which risen from about 45 percent of the total on the were mostly local. Safety net expenditures--also eve of transition to around 70 percent today. local responsibilities--rose rapidly with retrench- However, this increase reflected changes in the ment in the state-owned enterprise sector, composition of expenditures and relative prices which entailed unemployment stipends, early rather than a major change in expenditure retirements, and pension payouts. As a result, assignments. local expenditures grew relatively. As in other Soviet-type economies, in China Behind these numbers, real decentralization the central government was responsible for had in fact occurred. While local governments national defense, economic development are performing largely the same functions as (capital spending, research and development, before, under the planned economy they had industrial policy, and universities and research acted as agents in fulfilling deconcentrated func- institutes), and national institutions such as the tions of the central government, which bore the judicial system. The central government had del- costs through revenue-sharing arrangements. egated to local governments responsibility for During the transition, however, incremental day-to-day public administration and social serv- changes in revenue-sharing arrangements ices, such as education (except universities), weakened and then severed the link between public safety, health care, social security, hous- retained revenues and expenditure needs, and ing, and other local and urban services. A large local governments became responsible for portion of central spending focused on making financing their assigned functions from own- capital investments and financing state-owned revenues. This devolution had occurred without enterprises. When the transition began, the fanfare in the mid-1980s and was later codified transfer of financing for state-owned enterprises in the Budget Law of 1994. de facto discretion. Recognizing this reality in law ments. These constraints have contributed to will help clarify accountability and strengthen local China's backdoor decentralization, whereby local transparency. But some constraints on provinces governments have moved more and more funds remain. They must spend 15 percent of their bud- outside the purview of the budget and fiscal gets on the national priority sector of education by authorities. 2000, 18 percent by 2005, and 20 percent by 2010. In Indonesia, the central government continues They must also spend 2 percent of their budgets on to enforce regulations that determine spending on science and technology. certain items, including the Education Law, which In China, a multiplicity of laws stipulating calls for providing nine years of education to all, spending increases in certain sectors and regions with free primary schooling. Government regula- constrain local governments. For example, the cen- tions 109 and 110 of 2000 also regulate the ceiling tral government required education expenditures for spending by the head of a region and the to rise to 4 percent of GDP by 2000, that agricul- provincial representative assembly, but these regu- tural spending must grow faster than revenues, and lations are widely ignored. that spending on propaganda and culture should Another common problem is a lack of authority be no lower than overall revenue growth. The cen- transferred from the central or local finance func- ter also required that spending on science and envi- tion to local service providers, which have little dis- ronmental protection each rise to 1.5 percent of cretion in using resources. In Vietnam, provincial GDP by 2000, and that health care spending keep finance departments required spending units to pace with revenue growth, with per capita spending agree to every detail in their budgets and obtain rising from Ą 2.6 to Ą 4 by 2000. formal approval before reallocating spending. Reconciling all these constraints and unfunded However, recent decrees have gradually delegated mandates is almost impossible for local govern- greater budgetary flexibility, within a fixed block 100 East Asia Decentralizes BOX 5.3 The Block Grant Experiment in Ho Chi Minh City In 1999, the Vietnamese government authorized Reducing overstaffing. Most agencies cut their a pilot to introduce block grant budgeting in 10 staff by around 15 percent compared with their administrative units (districts and departments) in quota, and some made bigger cuts. However, Ho Chi Minh City. An early World Bank­sponsored most staff were transferred rather than retired or assessment found that the pilots had made signif- retrenched. icant progress in meeting their objectives: Raising incomes of employees by reapplying Restructuring departments and streamlining savings. Departments used gross savings to raise administrative procedures. In most pilot agencies, incomes, with 70 percent going to salaries, restructuring and streamlining had already 20 percent to a bonus fund, and 10 percent to begun. The pilot gave a further boost to this, a staff welfare fund. although centrally prescribed functional obliga- Increasing transparency. Departments launched tions limited the scope for restructuring. various initiatives to monitor service delivery stan- Reducing administrative costs. Reported gross dards and customer satisfaction. No evidence financial savings in the first year of the pilot ranged suggests that service quality declined in the pilot from 13 percent to 29 percent, based on adminis- agencies. trative economies as well as staff reductions. Because departments used the savings to supple- Source: Bartholomew et al. 2005. ment salaries, as planned, net savings were zero. grant, to spending and service delivery units.8 Units subnational governments carefully. Checks and bal- will have the freedom to shift resources between ances need to be in place if the move away from recurrent budget lines and--crucially--to use line-item budgeting is not simply to replace one savings from staffing cuts or lower operational form of inefficiency with another. As they lift line- costs to finance salary supplements. Selected item budget constraints, central governments need spending units in Ho Chi Minh City have piloted to create alternative mechanisms such as stronger this approach over two years, and a recent World capacity and greater ex post accountability for the Bank­sponsored assessment concluded that the use of resources and service performance. Strong pilots may have resulted in significant staffing financial management systems are essential for pre- reductions and savings (Bartholomew et al. 2005). serving fiscal discipline. We turn next to issues of (See box 5.3 for further details on this experiment.) capacity, accountability, and reporting. Nominally decentralized expenditure manage- ment regimes involving high levels of central con- Local Capacity. Local capacity for managing fiscal trol are likely to be both inefficient and ineffective. resources varies greatly, not only across administra- Such regimes will tend to make the allocation of tive tiers but also within each tier. As the Ho Chi resources overly rigid and encourage compliance Minh City block grant experiment suggests, the with rules and red tape rather than a focus on administrative capacities of large cities may some- responsiveness and service outputs and outcomes. times be equal to those of the central government. Such regimes also remove incentives among service That story also clearly reveals that mobilizing these providers to find savings, because officials cannot capacities to improve resource allocation and serv- transfer savings within expenditure items or ice delivery depends critically on establishing subitems to other categories but must forfeit them appropriate incentives and constraints. to the treasury. Above all, micromanagement and For the most part in the EA5, however, local second-guessing blur the lines of accountability for governments typically have weaker capacity in results between the finance function and the spend- expenditure management skills, such as plan- ing unit. ning, budgeting, execution, audit, and procure- Nevertheless, central governments need to ment. In China, local public expenditure manage- manage the passing of budgetary authority to ment is extremely poor. Budgeting is passive and Public Expenditure on the Frontline: Toward Effective Management by Subnational Governments 101 input oriented rather than results oriented. Budget pressure from upper levels to comply with UCE, formulation is almost totally inertia-driven, as it townships in poor regions have not met targets adds incrementally to past year's allocations with- (Wong et al. 1995).10 At the same time, the strain of out focusing on goals, outcomes, or performance. trying to do so has led townships to impose escalat- Accounting standards are lax, and waste of ing fees and involuntary contributions on rural res- resources is common (World Bank 2002a). That idents, spawning a rising tide of collective protests these problems persist after more than two decades through the 1990s.11 Thus, decentralizing basic edu- of economic decentralization suggests that their cation to townships has hindered the priority root cause lies elsewhere. The inefficiencies of local national program of universal compulsory educa- expenditures can be explained by the bifurcated tion while imposing high tax burdens on local Chinese fiscal system: local officials are often not populations. interested in improving financial management of budgets over which they lack real control; they focus Local Accountability. Institutions for managing instead on developing extrabudgetary resources expenditures focused on effective and responsive over which they exercise near-total control. service delivery are likely to evolve only if institu- Another dimension of the problem is that coun- tions also hold local governments accountable for tries have sometimes made "wrong assignments" by the way they use resources. As previous sections devolving responsibility for providing services to make clear, however, local governments cannot be levels of government that do not have the resources held accountable for nonperformance if they have and administrative capacity to respond. Again, not been given clear assignments, if they do not China provides a good example. Throughout the have the resources and authority to respond, or if history of the People's Republic, rural basic educa- their assignments and capacities are poorly tion has been the responsibility of township gov- matched. ernments, which average 27,000 in population, and One of the key assumptions of decentralization their antecedent, agricultural people's communes.9 is that local governments are more responsive Under the collective economy, public funds of the because they are "closer to the people." The litera- collectives financed rural education, with teachers ture on fiscal decentralization tends to emphasize mostly paid in "work points," or shares in the col- bottom-up mechanisms of "voice" (elections) and lective's net income. Under that system, the level "exit" (mobility), both exercised by the electorate. and quality of education varied among collectives, In the nascent democracies of Indonesia and Thai- depending on their resources and allocative land, these mechanisms may take time to develop. choices. Through the 1980s the government Even in the Philippines, where grassroots democ- imposed rising standards on rural schools and racy began a decade earlier, a 1999 study found that teachers, and by the late 1980s directed that all rural a lack of information limited popular influence on teachers be upgraded to the status of public local decision making--citizens generally knew less employees. This greatly raised the cost of providing about local government than about national gov- rural education, as the average salary for teachers in ernment. While citizens rely on the media for infor- state-run schools was several times that of teachers mation on the national government, they rely in rural "community" schools. largely on local officials and personal contacts for Expenditure pressures on townships rose further information on local government, as the media when the government introduced a Compulsory tend to focus only on "big news." In turn, local offi- Education Law calling for nine years of Universal cials have little awareness of the preferences of local Compulsory Education (UCE) by 2000--a level citizens, although municipal officials were more that far surpassed provision in most rural areas. aware than provincial officials. In the Philippines, Townships--primarily agricultural units--have no decentralization also does not seem to have significant tax base, except for the lucky ones in improved civic discipline despite the democratic coastal provinces, where township and village enter- setup. Measured corruption remains high and has prises thrive. For them, education is by far the grown under decentralization (Azfar et al. 2000). largest budgetary outlay, which even in the early Analysts also often assert that local taxing author- 1990s absorbed 40­60 percent of the total. Despite ity enhances local accountability: when local citizens 102 East Asia Decentralizes are paying directly for public services, they devote has been far less successful in meeting other targets, more attention to local politics and officeholders. however, such as exerting aggregate fiscal discipline, However, this assertion lacks empirical justification ensuring the delivery of vital services to all citizens, anywhere in the world, is conceptually suspect to the and protecting vulnerable social groups. extent that it elides citizens and taxpayers, and is Although Vietnam remains a one-party state, particularly questionable in the EA5, where only a the National Assembly is growing in importance, small proportion of citizens pay direct taxes. reviewing government plans, budgets, and imple- The exit option is not an effective mechanism mentation and exercising its constitutional author- for holding local governments accountable in ity as the highest organ of the state. The bureau- countries where government is the sole provider of cracy sees the National Assembly as a significant many services. This is true in China and Vietnam, source of authority and a potential arena for inde- where private providers are just emerging in vital pendent criticism of government performance. services such as education and health care, and However, the People's Councils at provincial levels where government monopolies and administrative and below have yet to emerge in all but the major restraints hinder the development of many serv- cities as significant checks on the executive. The ices. In these countries, people's option of moving memberships of the People's Councils (the legisla- to another jurisdiction ("voting with the feet") is ture) and the People's Committees (the executive) constrained. Political culture also hinders the often overlap, and their capacity is often limited. responsiveness of local governments and service Given the fragile nature of local democracy in providers. Officials who are used to responding to EA5 countries, the central government plays an rules and regulations will need time to become important role in managing decentralization. It more client oriented. must have strong capacity to monitor and evaluate decentralization, and it must set up mechanisms to National Accountability. The transition eco- hold local governments accountable in fulfilling nomies of China and Vietnam, and to some extent their responsibilities. But central governments the other countries in the EA5, also have top-down must balance these imperatives against the need to mechanisms to hold local governments account- give local governments autonomy and incentives, able. A study of transition economies in Eastern and the need to guard against micromanagement Europe and the former Soviet Union found that and the reassertion of authority. intermediate levels of government tended to The availability and timeliness of information respond to higher levels of government rather than on subnational governments is weak for most to local people (Wetzel 2001). countries in the world, and the EA5 is no exception. In China and Vietnam, through the nomen- Improving local transparency and information klatura system of appointing the top officials, the flows is vital to building subnational accountability Communist Party continues to wield control over to both citizens and higher-level governments. senior civil servants.In China,the government relies on elaborate systems of evaluation and performance Conclusion bonuses for top officials at all levels of government to induce compliance on priority goals such as Building sound public expenditure management growth, family planning, and prevention of social processes at the subnational level in the EA5 is a unrest. These mechanisms have been spectacularly large and daunting program. The success of that successful in ensuring that local officials promote program will depend on establishing appropriate growth and invest in basic infrastructure. Judging institutions and incentives. In other parts of the from the remarkable record of high and sustained world, expenditure management processes and the growth rates over the past two decades, China is supporting institutions and incentives evolved over arguably a successful case where the central govern- decades and even centuries. The East Asian experi- ment gets more of what it wants by giving incentives ence is still unfolding. Nevertheless, a number of to local governments to fulfill the central agenda. preliminary conclusions emerge from this review of This is the argument of the so-called "market- the region's early experiences in decentralizing preserving fiscal federalism" school. That approach public expenditures. Public Expenditure on the Frontline: Toward Effective Management by Subnational Governments 103 First, expenditure management processes at the clearly and match resources and authority to subnational level are generally weak. Planning responsibility. These efforts should go hand in processes are disconnected from fiscal planning hand with strengthening capacity and accountabil- and budgeting processes, and poorly coordinated ity institutions at both local and central levels. with planning at other levels of government. Bud- Third, decentralizing expenditure appears to get formulation suffers from major delays and is have yielded some benefits. In particular, given flex- fragmented, with parallel budgets for investment ibility and the right incentives, local governments projects and recurrent expenditures managed by have shown significant capacity to innovate, both in separate institutions. Budget execution is slow, with responding to local demands and conditions and in long delays in effecting appropriations, significant achieving cost savings. Innovations in service deliv- divergences between approved appropriations and ery associated with flexibility given to frontline serv- outputs, and corruption in procurement. Monitor- ice providers by subnational governments in ing and accounting systems often produce data that Vietnam and China are cases in point. Such innova- are neither timely nor accurate. Auditing is typi- tion has occurred despite imperfect conditions for cally weak and evaluation almost nonexistent. decentralization and its incomplete nature. How- Part of the explanation is that countries such as ever, major distortions have emerged as well. In par- Indonesia, Thailand, and Vietnam are just begin- ticular, inequalities in service provision have often ning to decentralize expenditures. However, time grown. EA5 governments need to pay greater atten- does not automatically solve these problems: China tion to the impacts of decentralized management of and the Philippines are still struggling to build the public expenditures and service delivery on poverty needed institutions despite starting one and two reduction and equity. The process for learning decades ago. There is no regional panacea: every through local innovations must be formalized. Both country will need to find its own way forward. positive and negative experiences need to be ana- Nevertheless, all the EA5 countries clearly need to lyzed, compared, and disseminated, and both local prioritize strengthening core expenditure manage- and central authorities must respond flexibly to ment processes at subnational levels as part of facts on the ground. Central governments--as well wider decentralization programs. In so doing, as regional groups such as the Association of South countries need to devote attention to budget for- East Asian Nations (ASEAN) and international mulation and execution, and focus on "getting the organizations such as the World Bank--have an basics right" rather than attempting to jump to any important role to play in capturing and disseminat- perceived world "best practice."12 Most EA5 coun- ing such lessons. tries should pay early attention to coordinating Fourth, perhaps the defining challenge in EA5 subnational planning and budgeting, integrating countries is to balance the need for bottom-up local extrabudgetary resources into the budget pro- discretion with the need for top-down direction in cess, and strengthening information on financial subnational expenditure management. In country management. after country,central ministries continue to overplay Second, none of the EA5 countries score highly their role by creating unfunded mandates and on the institutions and incentives needed to man- impinging on the necessary discretion of local gov- age subnational expenditures effectively. Assign- ernments. At the same time, central governments ment of responsibilities is often fuzzy. Resources play an important role in guaranteeing minimum and the authority to manage those resources are service standards, transparency, and accountability; often poorly matched with responsibilities. Local in transferring resources between regions; in ensur- capacity and accountability to both local people ingcoordinationbetweencentralandlocalinitiatives and central government are often limited. Although as well as between local initiatives; and in creating EA5 countries can take immediate steps to incentives for developing local expenditure manage- strengthen expenditure management processes, ment capabilities. This is particularly the case given governments cannot expect healthy local institu- the weaknesses of local democracy in EA5 countries. tions to develop organically until they have Efforts to decentralize public expenditures need to addressed these structural issues. Over the longer focus on redefining the role of central ministries as term, central governments need to assign functions well as developing new roles for local agencies. 104 East Asia Decentralizes Endnotes Vietnam's Experiment with Block Grant Budgeting." Inter- national Journal of Public Administration. 28 (3/4). 1. Our study does not include Cambodia. Bird, Richard. 2003."Local and Regional Revenues: Realities and 2. The study sample included countries for which subnational Prospects." Mimeo. University of Toronto, Toronto. budgetary data were available from the International Bird, Richard, Jennie Litvack, and M. Govinda Rao. 1995."Inter- Monetary Fund, the World Bank, the Organisation for governmental Fiscal Relations and Poverty Alleviation in Economic Co-operation and Development, and other Vietnam." Policy Research Working Paper No.1430. World sources. Bank, Washington, DC. 3. Bahl's regression results (2002) showed a "decentralization Burki, Shahid Javed, Guilhermo E. Perry, and William R. effort" of 2.6 for Vietnam, 1.4 for China, 1.0 for the Philip- Dillinger. 1999. "Beyond the Center: Decentralizing the pines, 0.9 for Indonesia, and 0.6 for Thailand. The author State." Report No. 19616. Latin American and Caribbean measured decentralization as the ratio of actual subna- Studies, World Bank, Washington, DC. tional expenditures to "predicted" expenditure share. Dai Xiaoming. 1997. "Concerning Fiscal Autonomy in Minority 4. In the cases of China and Vietnam, we drew on our own Nationalities Autonomous Regions and the Legal Issues." interactions with central and subnational governments. We Nationalities Research (minzu yanjiu) 6: 8­17. also benefited from conversations with colleagues in the Faguet, Jean-Paul. 1997. "Decentralization and Local Gov- World Bank's East Asia and Pacific Region. ernment Performance." Centre for Economic Performance 5. The classic formulation of this can be found in Oates 1972. and Development Studies, London School of Economics, 6. Examples include the ministries of Land Management and London. Investment Approval (World Bank 2003a). Fan, Gang. 1998. "Market-Oriented Economic Reform and the 7. Russian-educated Vietnamese often refer to this nested Growth of Off-Budget Local Public Finance." In Taxation in hierarchy as the "Matrouchka" system. Modern China, ed. Donald Brean. New York: Routledge. 8. Government of Vietnam, Decision 192/2001/QS-TTg: 17, Heberer, Thomas. 2001. "Nationalities Conflict and Ethnicity in Decision of the Prime Minister on the Expansion of the the People's Republic of China, with Special Reference to the Pilot Block Grant Scheme, 2001; Government of Vietnam, Yi." In Perspectives on the Yi of Southwest China, ed. Stevan Decree 10/2002/NS-CP, Decree Regarding Financial Regu- Harrell. Berkeley: University of California Press. lation of Service Delivery Agencies Whose Operation Kitchen, Harry, and Enid Slack. 2001."Providing Public Services Generate Revenue, 2002. in Remote Areas." Paper prepared for Fiscal Federalism in 9. Townships were traditionally units of rural government. Russia. World Bank Institute, Washington, DC. They were replaced by people's communes during the Martinez-Vazquez, Jorge, and Charles E. McLure. 1998. "Inter- period of collective agriculture. When the communes were governmental Fiscal Relations in Vietnam." Andrew Young disbanded in 1983, townships were restored as units of School of Policy Studies, International Studies Program, government. Georgia State University, Atlanta, GA. 10. At year-end 2000, the deadline for reaching the targets, 500 Mountfield, Edward. 2001. "Public Financial Accountability of China's 2,100 counties had not yet achieved UCE, and in East Asia." East Asia Update. Washington, DC: World another 600 counties needed help in shoring up their UCE Bank. status. In Gansu province in the northwest, 35 of 86 coun- Oates, Wallace E. 1972. Fiscal Federalism. New York: Harcourt ties had not achieved UCE. Among them, 6 counties were Brace Jovanovich. not yet providing even six years of primary schooling Prud'homme, Rémy. 1995. "The Dangers of Decentralization" (World Bank 2002a). World Bank Research Observer 10 (2): 201­20. 11. Responding to these protests, in 2001 the government Schick, A. 1998. "A Contemporary Approach to Public Expendi- introduced a major reform program to eliminate all rural ture Management." World Bank, Washington, DC. fees, and is incrementally addressing excessive decentraliza- Weist, Dana. 2001. "Thailand's Decentralization: Progress and tion to rural townships of key functions such as basic Prospects." Paper prepared for the King Prajadhipok's Insti- education and health. tute (KPI) Third Annual Congress on Decentralization and 12. On the case for "getting the basics right" in public expendi- Local Government in Thailand. ture management, see Schick 1998. Wetzel, Deborah. 2001. "Decentralization in the Transition Economies: Challenges and the Road Ahead." Report Bibliography No. 25132. Poverty Reduction and Economic Management Unit, Europe and Central Asia, World Bank, Washington, DC. Azfar, Omar, Tugrul Gurgur, Satu Kahkonene, Anthony Wong, Christine P. W. 1998. "Fiscal Dualism." In Taxation in Lanyi, and Patrick Meagher. 2000. "Decentralization and Modern China, ed. Stevan Harrell. New York: Routledge. Governance: An Empirical Investigation of Public Service ------. 2000. "Central-Local Relations Revisited: The 1994 Delivery in the Philippines." Mimeo. College Park, MD: Tax-Sharing Reform and Public Expenditure Management Center for Institutional Reform and the Informal Sector, in China." China Perspectives 31: 52­65. University of Maryland, College Park, MD. Wong, Christine P. W., C. Heady, and W. T. Woo. 1995. Fiscal Bahl, Roy. 2000. "Intergovernmental Transfers in Developing Management and Economic Reform in the People's Republic of and Transition Countries: Principles and Practice." Report China. New York: Oxford University Press. No. 21097. Urban and Local Government, World Bank, World Bank. 2000a. "Managing Public Expenditures for Washington, DC. Better Results." Report No. 20342-CHA. World Bank, ------. 2002. Presentation at the World Bank and State Council Washington, DC. Development Research Center workshop on rural public ------. 2000b. "Thailand: Public Finance in Transition." World finance, Beijing. Bank, Washington, DC. Bartholomew, Ann, Stephen Lister, Ed Mountfield, and Nguyen ------. 2000c. "Vietnam: Managing Public Resources Better." Van Minh. 2005. "Crossing the River by Feeling the Stones: Report No. 24469. World Bank, Washington, DC. Public Expenditure on the Frontline: Toward Effective Management by Subnational Governments 105 ------. 2000d. World Development Report: Entering the 21st World Bank. 2003c. "Philippines: Decentralization and Service Century. Washington, DC: World Bank. Delivery--From Promise to Performance. Cross-Cutting ------. 2001. "Vietnam: Country Financial Accountability Themes in Local Government Functioning." Report Assessment." World Bank, Washington, DC. No. 22181. World Bank, Washington, DC. ------. 2002a. "China: National Development and Subnational ------. 2003d. "Philippines: Decentralization and Service Finance: A Review of Provincial Expenditures." China Sub- Delivery--Strengthening Local Government Financing and national Expenditure Review Report No. 22951. World Resource Management in the Short Term. Cross-Cutting Bank, Washington, DC. Themes in Local Government Functioning." Report ------. 2002b. "Implementing Decentralization in Thailand: No. 26104-PH. World Bank, Washington, DC. The Way Forward." Capacity-Building Project. World Bank, ------. 2003e. "Philippines: Improving Government Washington, DC. Performance--Discipline, Efficiency, and Equity in Manag- ------. 2002c. "Vietnam: Delivering on Its Promise." Develop- ing Public Resources." World Bank, Washington, DC. ment Report No. 25050. World Bank, Washington, DC. ------. 2003f. Regional public expenditure reviews. World ------. 2003a. "Decentralizing Indonesia." Report No. 26191. Bank, Washington, DC. World Bank, Washington, DC. ------. 2004. World Development Report: Making Services Work ------. 2003b. "Fiscal Decentralization in Thailand." Decen- for Poor People. Washington, DC: World Bank; and New tralization flagship background note. World Bank, York: Oxford University Press. Washington, DC. 6 Subnational Own-Source Revenue: Getting Policy and Administr ation Right Robert R. Taliercio Once countries decide to decentralize--whether This chapter follows this logic. It begins by ana- gradually, as in Thailand and Vietnam, or with an lyzing the links between revenue assignment and initial dramatic change, as in the Philippines and autonomy by country and type of revenue, and Indonesia--they must get the fiscal design right. then assesses the region's revenue performance. Critical to effective fiscal design is the ability of local The ensuing two sections take up the question of governments to adjust budgets and thus respond to revenue administration and address the politics of community preferences regarding the quantity and local taxation, highlighting issues of policy imple- quality of public services. Ideally, and according to mentation. The last section offers final comments theory, subnational governments provide services on the implications of these findings for both pol- to their constituents up to the point at which the icy and administration. cost--in terms of taxes--equals the benefit, in The chapter's overarching conclusion is that East terms of the value of the services. To satisfy this con- Asian countries reveal many inconsistencies and dition, local governments must have the authority contradictions vis-ŕ-vis commonly accepted criteria to exercise own-source taxation, and be in a finan- for a "good" intergovernmental revenue system. cial position to do so. This is the essence of account- However, this newly decentralized region is also see- ability and efficiency under decentralization, and ing rich experimentation and innovation, which that is why decentralized revenue policy matters. bodes well for further reform. The design of a local revenue system includes three central dimensions: the assignment of rev- The Setting and Normative enue sources among types of government, the Framework degree of autonomy with which subnational gov- ernments can exercise their assigned authority, and Except in the Philippines and perhaps Indonesia, the efficiency of the revenue administration system. subnational governments in East Asia make little For a decentralized system to meet expectations, use of own-revenues to finance local services (see policymakers must ensure coordination between table 6.1). That is, the region is not characterized by these dimensions. significant fiscal decentralization. The implications 107 108 East Asia Decentralizes TABLE 6.1 Estimated Own-Source Revenue many of the resulting perspectives place a high prior- of Subnational Governments ity on economic efficiency. The first principle of rev- (as percentage of total subnational enue assignment--summed up by the "finance- government revenue) follows-function" refrain--is that it should be based on assigned expenditures. A second principle, given Country Percentage thematchingof revenuestoexpenditures,isthatlocal Cambodia (2003) 5 taxation should avoid introducing economic dis- China (2003) 5 tortions by inappropriately taxing the factors of pro- Indonesia (2002) 15.4 duction.The third general principle--subsidiarity-- Philippines (2002) 31.1 holds that revenue-raising powers should be Thailand (2002) 10.9 assigned to the lowest-possible level of government, Vietnam (2003) 5 except where such assignment would produce eco- Sources: Indonesia (Ministry of Finance), nomic distortions or negative externalities.2 Philippines (Department of Finance), and These three principles provide a general frame- Thailand (World Bank 2004c). World Bank staff work for assessing revenue assignment, which-- estimates for Cambodia, China, and Vietnam. whenfleshedoutwithmorespecificconsiderations-- Note: Most recent year available. Figures include only official, legal revenues. offers a set of practical guidelines. According to Norregaard, governments should not levy "local taxes" on very mobile factors lest they encourage of this are considerable. The fact that subnational taxpayer migration (though what constitutes "very governments have both limited powers to raise rev- mobile" is largely an empirical issue) (Norregaard enues and limited capacity to collect them poses 1997). Such taxes should not be unevenly distrib- profound questions about the actual benefits of uted among jurisdictions (as in the case of natural decentralization. Can local governments respond to resource taxation), should raise enough revenue to local preferences in taxation and thus service deliv- avoid large vertical fiscal imbalances, should not be ery, thereby achieving greater accountability and exportable to nonresidents (which would weaken efficiency? Are fiscal tools and capacities sufficient the accountability link), and should be based on the to generate sustainable own-source revenues? Has a benefit principle. Taken one step further by Bird, reliance on transfers, in their myriad forms, made these principles suggest a number of major tax subnational governments dependent on national sources usually prescribed for subnational govern- governments and weakened subnational incentives ments, "more or less in order of preference--user to improve own-source collections? charges, property taxes, excises, personal income The three dimensions of revenue policy provide taxes, payroll taxes, general sales taxes, and business a framework for addressing these questions:1 taxes" (Bird 2003b, 4­5). The following sections Which taxes should be authorized or assigned to address each of these options. central governments and which to subnational gov- ernments? The decision should be legal and trans- Autonomy and Policy parent, occurring through both constitutional and legislative processes. However, as discussed below, The decentralization literature can be surprisingly governments sometimes assume "authority" ille- unclear about the fundamental question of what gally. That such illegality and informality are unde- constitutes a subnational tax. However, if terminol- sirable will become clear; nonetheless, in some ogy is not clear as to what constitutes subnational countries, particularly China, these approaches are own-source revenue, policymakers will not have commonplace. Focusing on assignment of revenue the conceptual tools needed to design fiscal systems sources among governments stresses the impor- and appreciate their consequences. Such a situation tance of understanding decentralization as an will also obfuscate debate over the policy changes intergovernmental partnership. needed to allow a jurisdiction to realize the effi- Numerous public finance economists have pro- ciency benefits promised by decentralization. vided guidance on how to think about assigning Subnational revenues may be divided into cate- taxes between national and local governments, and gories of decreasing local autonomy (see table 6.2). Subnational Own-Source Revenue: Getting Policy and Administration Right 109 TABLE 6.2 Classification of Subnational Taxes by Degree of Central vs. Local Control Local autonomy in Subnational government (SNG) Greatest access to own-source revenues. revenue policy sets tax rate and base. These usually include fees and charges. SNG sets tax rate only. Necessary and sufficient condition for "own-revenue." Piggybacking and tax base harmonization permitted. SNG determines the tax base. Refers largely to local authority to grant exemptions that erode the local base. Limited autonomy SNG sets tax rate, but within In this case the center typically specifies a centrally permissible ranges. high/low tax range or caps the top rate. Tax sharing, whereby central/local Can result when a local authority collects revenue split can change only the tax and remits it to the center. with consent of SNG. No local autonomy Revenue sharing, with share 100% control by center; this is a source of determined unilaterally by misspecification of central vs. local central authority. revenue. (For example, the International Monetary Fund's Government Finance Statistics includes this category as a "local tax.") Central government sets rate and May accompany political decentralization. base of "SNG revenue." Sources: Adapted from OECD 2002; Ebel and Yilmaz 2002. If subnational governments have total or significant A typical argument for rate ceilings is that they control over a tax, fee, or charge, as shown by con- prevent local governments from doing egregiously trol over the tax rate (that is, if it is necessary and inappropriate things (though ceilings may simply sufficient), it is a subnational tax. If, in contrast, be a nontransparent mechanism for central con- subnational governments have no control over the trol). The practice, however, counters the accounta- base and rate of a tax, as, for example, when the bility and efficiency arguments for decentraliza- central government determines how to split rev- tion. One might make a case for centrally imposed enues ("tax sharing"), it is not a subnational source rate ceilings during the transitional period from of own-revenue. centralization to decentralization, as they might This taxonomy of taxation serves the very useful allow local governments to develop policy and purpose of setting out the basic definition of what administrative capacity, but such limitations is and is not a source of own-revenue. Thus, for should be short-lived. Subnational governments example, the taxonomy makes clear that although a build capacity by using their assigned powers. shared tax adds to subnational budget receipts, such revenue is not "own" taxation. Tax sharing Autonomy and Administration occurs when the base and rate of a tax are centrally set and then some percent is returned, typically on The third dimension of revenue policy--that sub- a derivation basis, to the "originating locality." To national governments must have some control over be an own-tax or revenue source, the subnational revenue administration--matters for two reasons. unit must, at the very least, legislate the rate. First, control over some aspects of revenue admin- More nuanced is the practice whereby a central istration is instrumental for controlling revenues at government restricts the rate of an own-revenue the margin, as this allows for changes in the effec- source (such as the rate ceiling). This clearly limits tive tax rate (a ratio of actual tax collected com- subnational autonomy, and some would argue that pared with the size of the legal tax base). Local once the cap is reached it fully eliminates local governments can change the effective tax rate by autonomy, as the subnational government loses its boosting compliance through audits and enforce- ability to raise taxes at the margin. Such rate regula- ment, or by lowering compliance costs for tax- tion is particularly common in East Asia. payers through better services (such as more 110 East Asia Decentralizes TABLE 6.3 Tax Assignment by Local Government in the Philippines Revenue source Provinces Cities Municipalities Barangays Real property tax 40% of provincial 25% of collections provincial or 30% of city collections Transfer of real property tax Tax on sand, gravel, and other quarry resources 30% of provincial 40% of provincial collections collections Amusement tax 50% of provincial collections Business taxes Franchise tax Community tax 50% of collections Source: Local Government Code (R.A. 7160), 1991. user-friendly tax forms and payment processing, political with fiscal decentralization promises more and access to information). efficient delivery of public services. This dimension is relevant in all cases, especially The Philippines has the region's longest-running in China and Vietnam, where subnational govern- record of an explicit decentralization policy and ments collect taxes whose rates and bases are deter- is also the most revenue decentralized of all the mined centrally. Second, and conversely, this con- focus countries. Yet, as the Philippines example also sideration suggests that subnational governments attests, the decentralization sorting-out process need not have full control over administration to takes time, and even the Philippines is a "young" call a tax or fee an own-source revenue. Devolved decentralizing state. The country has clear oppor- responsibility does not necessarily imply fully tunities for further reform. devolved administration, especially in the context In the Philippines, primary responsibility for of weak local capacity. National governments, for subnational taxation rests with provinces, munici- example, might take responsibility for certain palities, and cities (see table 6.3). Cities are the most administrative functions, such as assessing prop- autonomous: they are authorized to impose the full erty, or assist local governments with core functions set of local taxes, while provinces and municipali- related to information and communications tech- ties can levy only subsets. In some cases the rev- nology. This opens up a much-needed discussion enues collected by provinces and cities are allocated of the appropriate division of labor between to municipalities and barangays. The latter are also national and subnational jurisdictions, and of the responsible for collecting miscellaneous taxes and options for assisting low-capacity subnational gov- fees and charges. ernments with revenue administration (see below). But again, whereas this assignment suggests a well-designed system of revenue decentralization, the Local Government Code is not fully consistent Review of Current Practice with the autonomy criteria (see table 6.2). The cen- and Initial Evaluation tral government sets tax rate ceilings, leaving local Whereas several East Asian revenue systems rely on governments little control over one of the main central controls, countries are also showing a will- levers for mobilizing revenue, including the prop- ingness to review the twin features of assignment erty tax, which is a large revenue generator. The and autonomy. This represents an important policy code also fixes maximum rates for most other taxes crossroads: if, as shown, assignment and autonomy and nominal per unit amounts (as in the case of the can be reinforcing, the opportunity to combine professional tax). The code further specifies that Subnational Own-Source Revenue: Getting Policy and Administration Right 111 local governments can adjust tax rates only once in enues to communes. However, communes and five years, and not by more than 10 percent.3 sangkats (urban communes) may collect four types In Indonesia, subnational governments have the of own-revenue. These include administrative fees authority within the framework of Law 34 of 2000 for civil registry functions, agency fees for functions on regional revenue to levy a number of important performed on behalf of line ministries and others, but minor revenue sources. The central government contributions to development projects to meet the controls the most potentially productive local matching requirement imposed by transfers from revenues--those on real estate and personal the Commune Sangkat Fund, and user fees and income--and shares the receipt.4 Less revenue- charges to cover the recurrent costs of providing productive taxes are assigned to the provinces services (UNCDF 2004). Data on actual collec- (motor vehicle registration, transfer and fuel taxes, tions are not available, but the amounts are thought and a water user fee, all of which are shared with to be quite small. Provinces--deconcentrated levels cities and regencies), and cities and regencies (excise of the national government--may also collect own- taxes on hotels, entertainment, advertisement, street revenues, which accrue to governors' budgets (the lighting, mining of selected minerals, and parking). Salakhet).6 These revenues, which are not proper The national government restricts this arrange- own-source revenues, accounted for 48 percent of ment even further: subnational governments may the total Salakhet, of which the tax on motor vehi- not impose surcharges on national taxes, and Law cles (17.4 percent), the excise on public lighting 34 of 2000 sets maximum rates on those that are (12.3 percent), the wealth transfer tax (7.0 percent), assigned.5 Evidence suggests that most regions and the business tax (6.8 percent) are the most already charge the maximum rate, and that some significant (World Bank 2003b). wouldraiseitfurtherif allowedtodoso.Forexample, Vietnam, as a transitional country, reveals some the widely assessed hotel and restaurant rate is lim- similarities with China. The Law on State Budget, ited to 10 percent,yet a World Bank report concludes which took effect in January 2004, establishes that some jurisdictions (such as Bali and Jakarta) three types of revenue assignments. These include could gain from rate increases (World Bank 2003a). revenues assigned completely to the central level, China overhauled its subnational revenue sys- those assigned completely to the local level, and tem under the 1994 Tax Sharing System reform. those shared between the central and subnational The tax structure now includes three tiers of taxes: governments. Shared revenues--which include those fully accruing to the national government, the value-added tax, enterprise (corporate) income those shared between the national and subnational tax, personal income tax, special consumption governments, and those fully accruing to subna- taxes, and gasoline and oil fees, among others-- tional units. Allowable subnational own-source constitute the bulk of revenues at all levels. Own- revenues include the urban land use tax, for which source revenues are virtually nonexistent in the local government can set the rate up to a ceiling; Vietnam. The only exception is user fees, such as local option entertainment and slaughterhouse road tolls and select fees for schools and hospitals, taxes; and a local option surtax on collective enter- which are not generally significant sources of prises, for which the subnational government can revenue (World Bank 2000). influence the rate (Bahl 1999). Revenues in Thailand are national, local, or Own-Source Revenue Practices shared. Shared taxes include the value-added tax and Options and sales tax, the special business tax, the natural resource tax, excise taxes, and the vehicle tax, all of Revenue assignment in East Asia is consistent with which accrue to local governments. Own-taxation the framework's principles and guidelines in some is limited to the house and land tax, land develop- ways and inconsistent in others. The next section ment tax, signboard tax, petrol tax, tobacco tax, and further highlights specific regional practices for hotel tax. Local governments may also collect user each revenue source. This discussion serves to stress fees, charges, permits, license fees, and fines. that a well-designed subnational revenue system Cambodia is still at an early stage of decentral- should rely on a mix of taxes, and also suggests fur- ization: it has not yet assigned functions or rev- ther options for reform. 112 East Asia Decentralizes As table 6.4 reveals, there is no own-source the market value, as the country has no formal revenue common to all six countries, though a market for land transactions. number of countries make use of user fees and Indonesia provides another important example charge business taxes, excise taxes, and property of limited local autonomy over the property tax. taxes. As admirable as the user charge and property Subnational governments are not responsible for tax package is, international experience has shown taxing property (or property transfers). While the that such taxes can be both difficult to implement central government shares about 80 percent of prop- and unlikely to provide an adequate fiscal base if erty tax revenues with the originating region, and subnational governments have major social spend- distributes another 10 percent among all regions, ing responsibilities (Bird 2003b). Accordingly, policy and administration are firmly in central several options are available to further mobilize hands. This has led to two negative developments: a own-source revenues. reliance on taxes that essentially substitute for prop- erty taxes, such as the street lighting tax; and the proliferation of nonbenefit taxes--user charges and Property Taxation service fees not linked to the provision of services. Given the role of property taxes as revenue genera- The Philippines is the only country where a tors in the developing world, the fact that the prop- traditional property tax is a source of subnational erty tax is not an own-revenue source in many coun- own-revenue. The property tax accounts for nearly tries raises concerns about sustainable revenue 37 percent of subnational own-revenues. However, flows, subnational autonomy, and thus the promised the central government limits control over tax benefits of decentralization. In the 1990s, property rates and bases. For example, the national govern- taxes accounted for 40 percent of all subnational ment sets the real property tax rate (including taxes in developing countries, but only 12 percent in the Special Education Fund levy) at 2 percent for transition countries (Bird and Slack 2002). Except provinces and 3 percent for cities and municipali- for the Philippines and Indonesia, East Asian coun- ties in metropolitan Manila. All provinces outside tries do not come close to the benchmarks of either the capital region avail themselves of the maximum developing or transition countries, suggesting that rate, while most cities are under the maximum.7 property taxes could yet serve as a greater source of International experience suggests that subna- revenue in the medium term, and an important tional governments are likely to use discretion over source in the longer term. property tax rates, so they vary widely. Bird and Vietnam, Thailand, and Cambodia diverge the Slack report, for example, that the effective rate most from international practice. Vietnam has no of property tax in the United States ranges from property tax in the modern sense of the term. The 0.4 percent to 2.9 percent for residential property country does have taxes on land and housing, land and 0.7 percent to 6.0 percent for commercial rent, and transfers of land use rights, but local gov- property (Bird and Slack 2002). ernments have little or no control over these taxes. Property tax reform can offer a source of rev- Thailand imposes land taxes, but they account for enue, perhaps modest in transition countries, and a only about 5 percent of subnational resources, and source of autonomy and accountability across the subnational governments lack authority over their region. Rationalizing land taxation in Vietnam-- rates and bases. Cambodia's land taxes are insignif- and moving toward a modern property tax with icant in revenue terms. some local discretion over rates and introducing China has multiple taxes on property, often on modest property taxes in Cambodia--would be the same base, which do not rest on market or first steps in those countries. Devolving authority updated property assessments. Thus, these taxes do over rates in China, Indonesia, the Philippines, and not deliver on their revenue potential. Yet unlike in Thailand is also a reform option. Vietnam, local governments can set the tax rate on urban land use subject to legislated maximums and Business Taxation minimums (larger cities can set higher rates, for example). In fact, the tax on urban land use is one Some analysts regard business taxes as a potentially of only a small set of own-source revenues in inefficient means of raising revenue. Often levied at China. The tax base is the physical land area, not high rates, these taxes can distort firms' investment Subnational Own-Source Revenue: Getting Policy and Administration Right 113 TABLE 6.4 Own-Source Revenue Assignment Cambodia China Indonesia Philippines Thailand Vietnam User Yes, provinces No, but "informal" Yes; these are Yes; these are Yes, on trans- Yes, on roads, charges? and or "illegal" local numerous, and numerous, portation, education, communes extrabudgetary some have high and some public and health. can collect fees have administrative have high utilities, and for some proliferated and compliance administra- markets. basic in health, costs. Some are tive and services. education, and "nuisance" compliance roads. Some charges. costs. are "nuisance" Some are charges. "nuisance" charges. Property No, but minor Yes, to an extent, No; revenues Yes; rates set No, for the land No; some taxes taxes? provincial in that SNGs shared with by SNGs development on land and taxes on can set the rate SNGs, but they subject to a tax and the housing, unused for the urban have no control maximum. house and land rent, land. land use tax up over rates and Administered land tax. and transfer to a ceiling. little control over by SNGs with The central of land use There are also administration. little central government rights. But numerous other assistance. sets rates. SNGs have taxes on the no control same property over rates and land base. and little control over administra- tion. No modern property tax. Excise No, but taxes No, but taxes on Yes, on motor No; SNGs are No; central No. taxes? on motor vehicle and vehicles and fuel, prohibited government vehicles vessel use. though assigned from levying controls at the to provinces. excise taxes, motor vehicle provincial Also on minerals, including and other level. raising equity on motor excise taxes. issues. vehicles. Personal No. No; a PIT is No; a PIT is shared Not exactly, No. No, but central income assigned to with SNGs, but though the government taxes SNGs, but they they have no community and SNGs (PIT)? have no control control over tax tax is in the share PIT. over tax policy. policy. form of a poll tax. Payroll No. No. No. No. No, though No. taxes? base of national PIT includes payroll. General No. No, but central No. No. No, but central No, but central sales government govenrment government taxes? and SNGs share and SNGs and SNGs value-added share value- share value- tax. added and added tax. sales taxes. Business No, though Yes; a business Yes, though Yes; relatively Yes, though No, but central taxes? business tax on gross officially limited large revenue limited to a government license receipts, an to a few sectors. source. small number and SNGs charges enterprise The number and of sector- share and market income tax, and type of "business specific corporate taxes are other surcharges taxes" are business-type income tax. assessed and surtaxes growing, raising taxes. at the (e.g., urban concerns about provincial maintenance "nuisance" level. and construction taxation. tax). Source: World Bank staff. Note: SNG subnational governments. Own-source revenue defined as a legal tax or charge over whose rate an SNG has some control. 114 East Asia Decentralizes decisions (such as their debt-equity ratio). These simple low-rate, broad-based "single business tax" taxes might also serve as barriers to new firms and the to replace sectoral and "nuisance" levies and fees.8 expansion of small ones.On the other hand,business Cambodian provinces levy business licenses, though taxes are potentially large revenue generators and it might be more appropriate to transfer business more elastic than other traditional subnational taxes licensing to the commune level and introduce a (such as the property tax), although they may also be standard business tax at the provincial level. more distortionary. Business taxes can be justified according to the benefit principle: firms are consum- Personal Income Taxation ing benefits provided by subnational governments No country in the region makes use of personal and thus should be charged for them. income taxes as a source of subnational own- This principle offers a rationale for differenti- revenue. China, Indonesia, and Vietnam use such ated business taxes at local and regional levels, such taxes as shared revenues, however. Such taxes would as user charges along with some form of business thus appear to be a potential new source of revenue licensing tax, and perhaps a low-rate tax on gross for Cambodia, the Philippines, and Thailand.Allow- receipts, either in place of or in addition to those ing regional (provincial) governments to piggyback options. At the regional level, the benefit case taxes on national personal income taxes is an impor- argues for a broad-based levy that remains neutral tant option for boosting local revenues, and poten- toward the factor mix, such as a value-added tially for increasing autonomy. This assumes that income tax or a business value tax (both options for subnationalgovernmentscouldsettherates,andthat taxing value-added income). One further option is the central government would administer the tax to to levy both a payroll tax and a tax on capital. Pay- avoid burdening local administrative capacity.9 roll taxes are easy to administer and productive at low rates. However, they can act as a barrier to User Fees and Charges employment in the formal sector--a concern that might outweigh their benefits in economies strug- Official use of charges and fees is widespread in the gling to boost rates of formal employment. region.10 In Thailand, subnational governments Table 6.4 shows that China and the Philippines levy user charges on garbage collection, public util- have formal business taxes, which are major sub- ities, mass transportation, and medical and child- national revenue generators. The business tax in care. The Philippines has more than 33 different China--levied on gross receipts not subject to the types of user fees and charges, ranging from animal value-added tax--covers a wide range of sectors, and civil registration to garbage collection fees. including transportation, communications, and Total collections from each major source are rela- construction, and accounts for a large share of tively small, reflecting the dispersion of revenue provincial tax revenues (22.6 percent, on average, in sources (see table 6.5). Moreover, a large number of 2001). The business tax in the Philippines is similar, fees and user charges together generate less than in that it is also levied on gross sales and accounts 0.10 percent of the total operating and miscella- for a significant share of local revenues (29.8 percent, neous revenue of subnational governments. It is on average, in 2002, including business licenses). important to note, however, that the main eco- Giving subnational governments in China and the nomic rationale for levying user charges is to Philippines control over rate setting could be both promote efficient use of public resources through revenue-productive and autonomy-enhancing. the pricing mechanism, not necessarily to raise Cambodia and Indonesia do not have local busi- revenue. Still, eliminating these extremely low- ness taxes per se. Indonesia relies on taxes on specific yield fees would reduce the administrative and sectors, including hotels, restaurants, and advertis- compliance costs arising from the complexity of the ing. A growing number of subnational governments system. in Indonesia are also taxing specific sectors, mimick- In Indonesia, the most significant provincial ing business taxes. There is concern that the prolifer- collections from user fees come from charges for ation of these taxes will result in distortions and health services. Collections from charges for build- inefficiencies at the local level. This suggests the ing licenses are second in importance. At the city and need to rationalize the taxes imposed on businesses district level, 62 percent of fee revenue comes from to minimize distortions, such as by introducing a public service fees, 23 percent from licensing fees, Subnational Own-Source Revenue: Getting Policy and Administration Right 115 TABLE 6.5 Highest-Yield Fees and Sources of Operating Revenues of Subnational Governments in the Philippines, 2000 (pesos) Pesos % of total Market receipts 1,568,806,000 15 Electrical light and power fees 738,108,000 7 Mayors' permit fees 659,004,000 6 Hospital fees 657,528,000 6 Rents 368,598,000 4 Building permit fees 361,172,000 4 Garbage fees 337,992,000 3 Total operating and miscellaneous revenue 10,218,000,000 100 Source: COA 2001. TABLE 6.6 Most Significant Charges by Local and Provincial Governments in Indonesia, 2002 (million rupiah) Share of total Local government Revenue (%) Charges for health services 745,903 33.25 Charges for building permit 240,547 10.72 Charges for market services 194,134 8.66 Charges for printing resident's ID card and birth certificate 128,072 5.71 Charges for use of regional property 96,259 4.29 Garbage disposal/sanitation levies 92,160 4.11 Bus terminal levies 87,353 3.89 Disturbance permit levies 62,824 2.80 Public roadside parking levies 60,387 2.69 Motor vehicle inspection levies 46,347 2.07 Share of total Provincial government Revenue (%) Charges for health services 311,133 45.42 Charges for building permit 134,071 19.57 Charges for use of regional property 46,384 6.77 Motor vehicle inspection levies 36,259 5.29 Regional production sale levies 20,451 2.99 Wholesale market and shopping complex levies 15,607 2.28 Public roadside parking levies 12,603 1.84 Recreation and sports ground levies 11,774 1.72 License allocation of land use 10,711 1.56 Garbage disposal/sanitation levies 8,741 1.28 Source: Ministry of Finance. and 15 percent from business service fees. All other local charges throughout the country in 2002). Many provincial and local charges together contribute a of these low-yielding fees are levied on businesses, very small percentage of total revenue. In fact, creating a heavy administrative burden and sug- provinces levy 14 charges that generate less than gesting the need for rationalization (see table 6.6). 1 percent of charge revenues, and local governments Bird and Tsiopoulos sum up the challenge of levy at least 30 such charges (with at least 93 different user charges as ensuring "that the right prices are 116 East Asia Decentralizes charged for the right services" (1997, 33). They also The Chinese case is particularly acute given the argue that central governments need to provide large number of "unofficial" charges. Because of a guidance--perhaps in the form of an overarching lack of autonomy, local governments have resorted legal framework--to subnational governments on to introducing fees not permitted by law, and these creating and managing user charges. Elements of represent a significant percentage of local budg- such a framework include clear and transparent ets.13 The fees include surcharges on household parameters for setting prices and a process for con- utility bills, hospitals and school charges, road sulting with local stakeholders. National govern- maintenance, advertisement, vehicle purchases and ments in countries with a particular concern in this others (World Bank 2002). While these fees may area, such as China, Indonesia, and Vietnam, might introduce distortions and raise compliance costs be warranted in intervening--at least in the near imposed on taxpayers, the World Bank has argued term--with lists of allowable or prohibited charges. that "fears of run-away local governments arbitrar- ily creating a jungle of local fees and charges do not Excise Taxes appear to be justified." Yet the growth of extrabud- getary financing among local governments has East Asian countries are split on excise taxes. The become a serious concern. The World Bank esti- Philippines, Thailand, and Vietnam do not use mates that extrabudgetary funds and off-budget major excises, which suggests another reform option for these countries.11 Allowing local governments to activities may represent 18 to 22 percent of gross domestic product (GDP) (World Bank 2002). impose excise taxes or fees on motor vehicle regis- The Chinese government has had a policy of tration would give those governments an appropri- converting informal fees into official taxes, but it ate and potentially important source of revenue that would be relatively easy to administer.12 is implementing this policy unevenly. In Gansu province, for example, provincial officials report- Significant Others: Fees, Other edly approve all local fees at the prefecture, county, Taxes, Charges, and the Problem and township levels. But no one knows the extent of Illegal Proliferation to which local governments are staying ahead of provincial authorities by implementing new fees. Illegal Activities Similarly, Cambodia has had serious concerns Lack of control over tax policy has encouraged local about the proliferation of "informal" fees and governments to seek other tax and nontax sources charges at the commune/sangkat level. In fact, this of revenue. In a number of East Asian countries, proliferation seems to be impeding development local governments have the authority to enact new of a system of own-source revenues for newly taxes and fees, though their authority is regulated elected subnational governments. Evaluations of by law, and in some countries is subject to review existing practices have been only exploratory, but by the central government. Central control has there are reports of birth registration fees being encouraged local governments to become entrepre- "unofficially inflated" from US$0.10 to US$10 neurial, with both positive and negative results. In (UNCDF 2003). China, this situation has led to the proliferation of In China, fees and charges undermine the tax "illegal" extrabudgetary fees, some of which have system because they are not officially on the books. distortionary effects. Indonesia has seen the prolif- In the Philippines, a similar argument can be eration of nuisance taxes, which collect very little in made--but with a variation. In the Philippines, revenue yet impose high administrative costs on numerous fees and charges on the books under- local governments and compliance costs on taxpay- mine the system because they are either not col- ers. In the Philippines, local revenue codes yield a lected, or are not collected in accordance with the tax system of great complexity, with a resulting loss revenue code. In the city of Bacolod, for example, of transparency and ability to monitor the system. the mayor's business permit fee has more than 200 Some subnational governments are avoiding col- different rates, which vary by type of establishment. lecting legal taxes and others are collecting "illegal" This complexity adds greatly to administrative taxes, undermining the integrity of local gover- costs. Thus, putting revenues on the books and nance and thus public support. legalizing them does not necessarily solve the Subnational Own-Source Revenue: Getting Policy and Administration Right 117 problem or eliminate the need to develop the local major reform option for at least two reasons. First, capacity to manage user charges. the positive list of taxes provided in Indonesia Overall, the proliferation of illegal, extrabud- (Law 34 of 2000) does not include any taxes that can getary revenues raises concerns about efficiency generate significant revenue. This prompts many losses stemming from distortions and relatively high subnational governments to introduce taxes that do administrative (and possibly compliance) costs. At not necessarily generate much revenue either, as a the same time, citizens of some localities may be way to assert their autonomy. As a result, nuisance willing to pay for off-budget services from local taxes and charges abound. The Ministry of Finance governments (in these cases, efficiency losses would found that more than 200 regulations submitted to presumably be lower). Some subnational govern- the central government between August 2001 and ments collect these off-budget sources in response January 2003 violated Law 34. Many are levied on to limited autonomy, so the practice is somewhat agriculture, mining, and interregional trade. In understandable, if not justifiable. However, these Lombok, for instance, three local governments levies can undermine public trust in the tax system, jointly impose a 5 percent tax on 174 products which cannot be good for long-term compliance. A leaving the island (Ray 2003). Thus, the problem policy of reviewing and converting illegal fees into in Indonesia is not the open list itself but the fact official ones, as in China, is warranted. that it does not include appropriate broad-based taxes. Second, the review mechanism does not seem to Open Lists work very well without adequate monitoring. For One of the issues around the use and abuse of user example, Lewis estimates that subnational govern- charges is the authority to enact new levies. In both ments send just 40 percent of all regional tax- and the Philippines and Indonesia, local governments charge-related regulations to the Ministry of Home have formal authority to introduce some new taxes Affairs for review, so many potentially harmful and fees. In the Philippines, the Local Government taxes and charges remain in effect (Lewis 2003). Code provides a range of tax and fee options for local governments, though the country should con- Revenue Assignment between sidermoreoptions.Thecodegiveslocalgovernments Subnational Levels the option to levy other taxes, fees, and charges, pro- vided that the code does not specifically prohibit Revenue assignment between local governments in them and the National Internal Revenue code does East Asia often seems to create incentive problems. not already include them.14 That all makes sense These occur between provinces and subprovincial within the framework established above. But what is levels in China and Vietnam; between provinces, problematic is if subnational governments abuse cities and municipalities, and barangays in the this"open list"approach by levying taxes and fees in Philippines; and between provinces, cities and a nontransparent, illegal manner. Monitoring and regencies, and villages in Indonesia. In China and enforcement of tax law matter. Vietnam, the lack of formal revenue assignment In Indonesia, besides formally assigned taxes, creates unpredictability and reduces accountability. Law 34 of 2000 allows cities and regencies to levy In the Philippines and Indonesia, the transfer of additional taxes if they follow a number of general significant shares of collections from province to criteria.15 To ensure that governments meet these subprovincial levels may reduce provincial incen- criteria, the central government requires them to tives for collecting own-revenue. submit regional regulations to the Ministry of In China, provincial governments may assign Home Affairs for review. If the ministry, in consul- revenues to local governments within their jurisdic- tation with the Ministry of Finance, finds that a tax tion (see box 6.1). Most, if not all, provinces seem to violates legal provisions, the ministry may cancel follow the traditional hierarchical approach, assign- the regulation, in which case the local government ing revenue between the province and the first layer must rescind the tax.16 of local government (cities and prefectures), and As noted, the "open list" approach has its merits. leaving each layer to work out arrangements with However, countries might not want to seize on it as a the one below it. As mentioned, while the lack of 118 East Asia Decentralizes BOX 6.1 Revenue Assignment across Subnational Governments in China's Gansu Province Gansu province is centralized, in that the provin- · For all other local taxes, prefectures and cial level retains a relatively large share of all cities have full discretion over sharing local taxes and rebates. For example, the arrangements. province retains the entire amount of the value- · For certain local taxes, counties have full dis- added tax shared with subnational governments cretion over sharing arrangements with (that is, 25 percent of total national collections). towns and township governments. These A number of important features characterize taxes include the farmland occupation tax, subprovincial revenue assignment: the land appreciation tax, the housing prop- erty tax, the vehicle license tax, and animal · Revenues from the enterprise income tax and husbandry charges. the value-added tax are allocated to the level · Given discretion at each level of government, of government that "owns" the enterprise. a variety of other arrangements exist. For · The city maintenance tax is shared 30 per- example, in some counties the prefecture cent with the province and 70 percent with government received a share of the county prefectures and cities. value-added tax despite the provincial rule · The resource tax is shared 70 percent with that allocates the tax by ownership. the province and 30 percent with prefectures and cities. Source: World Bank 2002, pp. 57­58. · The urban land tax is shared 40 percent with provinces and 60 percent with prefectures and cities. formal assignment may have some advantages that despite policy, administrative, and political for provinces, the disadvantage for subprovincial challenges, own-source collection shows some pos- governments in terms of uncertainty would seem to itive signs, which augurs well for the future. outweigh the potential advantages. However, in most cases, own-source revenues compose a small percentage of total subnational revenue and their share has either not improved Revenue Performance much or has actually declined. This means that Own-source revenues in Indonesia and the subnational governments have not reduced their Philippines have risen in nominal (and real) terms, dependence on central government transfers. but have not fared as well as a percentage of GNP. The next sections analyze that result for several In the Philippines, own-source revenues rose sig- countries. nificantly right after decentralization, but have been stagnant as a percentage of GNP--at 0.91 in Composition of Own-Source Revenues 2001--ever since. In Indonesia, own-source rev- enues rose from 1.1 percent of GNP in 1995 to From 1992 to 2002, own-source revenues in the 1.4 percent in 2002. In China, in contrast, local Philippines composed 34 percent, on average, of taxes, which are predominantly shared revenues as total resources available to local governments. opposed to own-source revenues (in fact, official However,own-source revenues declined from 38 per- own-source revenues are minimal), grew signifi- cent in 1992 to 31 percent in 2002. Thus, these cantly as a percentage of GNP since the Tax Sharing revenues,althoughnominallygrowing,havenotkept System reform, from 5.0 percent in 1994 to 7.0 per- up with the growth in transfers from the central cent in 1999. Moreover, estimates of buoyancy in government (the Internal Revenue Allotment). the post-reform period--a measure of revenue pro- In the Philippines, as in numerous developing ductivity as a result of economic growth--show countries, property taxes account for the largest that it is quite high.17 Overall, these results indicate single component of local revenues. In 2002, the Subnational Own-Source Revenue: Getting Policy and Administration Right 119 real property tax accounted for 36.5 percent of sub- and regency level declined from 12 percent to 9 per- national own-source revenues, while business taxes cent from 1995 to 2000, and dropped further to and licenses accounted for nearly 30 percent, and 6 percent in 2001, after decentralization. Transfers operating and miscellaneous revenue accounted for rose in relative importance over the same period, 22 percent (see table 6.7).18 These percentages have from 84 percent to 87 percent of total revenues. remained stable over time. In Indonesia, own-source revenues accounted Trends in Own-Source Revenues for about 39 percent of total revenue at the provin- cial level, but only 7 percent at the city and regency Both the Philippines and Indonesia show a clear level (see table 6.8). Moreover, the percentage of trend: own-source revenues have risen slowly own-source revenues to total revenue at the city but steadily. In the Philippines, nominal and real TABLE 6.7 Own-Source Revenue Composition by Type of Local Government in the Philippines, 2001 (percentage distribution)a Sources All SNGs Provinces Cities Municipalities Real property taxes 36.5 47.3 36.8 30.3 Business taxes and licenses 29.8 0.3 36.3 26.3 Other taxes 11.2 22.4 10.0 9.0 Operating and miscellaneous revenue 22.3 29.2 16.9 34.3 Capital revenue 0.1 0.7 0.0 0.1 Source: COA 2002. Note: SNGs subnational governments. Totals may not sum to 100 percent because of rounding. TABLE 6.8 Composition of Regional Revenue in Indonesia, 2002 (millions of rupiah; percentage of total revenue) Provinces Cities/regencies Own-source revenues 14,207,830 7,454,629 38.46% 7.19% Taxes and charges 12,500,929 5,109,501 33.84% 4.93% Other 1,706,901 2,345,128 4.62% 2.26% Grants 7,393,745 64,100,112 20.02% 61.83% Revenue sharing 8,084,119 17,310,428 21.89% 16.70% Carryover 6,307,652 9,752,994 17.08% 9.41% Other 944,849 5,052,425 2.56% 4.87% Total 36,938,196 103,670,588 100.00% 100.00% Source: Ministry of Finance. Note: Figures are projections based on data for 28 provinces (including Jakarta) and 324 cities and regencies. 120 East Asia Decentralizes TABLE 6.9 Annual Nominal Growth in Local Own-Source Revenue in Indonesia, 1996­2002 (percent) Annual increase 1995/1996 1996/1997 1997/1998 1998/1999 1999/2000 2001 2002 Regional tax 31 22 20 48 26 21 23 Regional charges 18 14 11 12 20 32 32 Profits--enterprises 18 33 2 20 26 10 89 Other 40 0 15 75 1 155 93 Total 24 16 14 20 20 40 42 Source: BPS and Ministry of Finance. collections have grown steadily in all major cate- were less than half those from provincial taxes and gories: property taxes increased nearly fourfold charges. from 1993 to 2000, while taxes on goods and serv- Street lighting and hotel and restaurant taxes ices rose fivefold. Taxes on goods and services account for over two-thirds of own-source collec- recorded the highest average annual increase, tions among cities and regencies. "Other" taxes-- followed by the property tax. Growth rates for all which most likely include a large number of nui- categories slowed during the latter part of the sance taxes that local governments have introduced period. since Law 34 of 2000--represent only 6 percent of In Indonesia, decentralization spurred a 40 per- tax revenue, indicating that the new taxes do not cent nominal rise in local own-source revenues generate significant additional revenue. between fiscal years 2000 and 2001.19 While rev- enues from taxes and charges grew by 21 percent Own-Source Revenue and 32 percent, respectively, the largest increase Administration occurred in the category of "other," which rose by 155 percent (see table 6.9). In 2001, locally raised Arrangements for tax administration vary through- taxes contributed 43 percent of own-source rev- out the region, ranging from highly decentralized enues, while charges raised 33 percent and the share in the Philippines to highly centralized in Vietnam, of "other" revenues grew to 21 percent. with Indonesia and China between these two Between 2000 and 2001, provincial tax revenue poles. Yet in all cases the relative roles of local and more than doubled in Indonesia. Several factors national governments have not been well designed, may explain this sudden jump. One factor seems to resulting in both capacity and incentive challenges. be that provinces received additional taxing powers The Philippines, for example, is highly decentral- under Law 34 of 2000 (even though the law abol- ized with respect to tax administration. The Bureau ished some nuisance taxes). Another factor is that of Internal Revenue (BIR) administers national tax collections, especially from the vehicle transfer taxes, while each local government administers its tax, rose considerably. Third, changes in revenue- own-source revenues according to the Local sharing arrangements may have created stronger Government Code and local revenue codes. More- incentives for provincial governments to increase over, the law provides for little formal cooperation: collections. the BIR operates independently of local tax admin- Fuel and motor vehicle­related taxation domi- istrations, and the national government provides nate provincial own-source revenues in Indonesia. little support to local governments.20 Meanwhile, Provinces also appear to levy taxes that are formally local tax administrations usually operate independ- assigned to local governments, such as the hotel ently of one another. and restaurant tax and the street lighting tax. At the other extreme is Vietnam. The General Despite their insignificant expenditure respon- Taxation Department--operating under the sibilities, provinces raise almost twice the own- Ministry of Finance, with offices at the provincial source revenues as cities and regencies. In 2001, for and district levels--is responsible for collecting example, revenues from local taxes and charges all internal revenues. Local governments have no Subnational Own-Source Revenue: Getting Policy and Administration Right 121 tax administrative responsibilities. However, tax national level to administer national and shared administrators operate under a system of dual sub- revenues, and a provincial tax administration ordination, in that they are responsible to their tasked with all subnational revenues. According to ministerial management as well as to representatives the World Bank, "de facto dual subordination" of of local governments.21 Thus, the system is charac- central tax administrators to local governments is terized by built-in tensions. Martínez-Vázquez still a problem, owing to old allegiances and the fact notes that administrative centralization may reduce that local governments provide bonuses and assess incentives for revenue collection, as central officials penalties to stimulate collection, thus creating have fewer incentives to collect local revenues com- potential conflicts of interest (World Bank 2002). pared with local administrators, who would have greater incentives to collect local revenues. On Local Revenue Administration: the other hand, he notes, provincial authorities Models and Options have been known to pay bonuses to tax adminis- trators who improve their collection performance Tax administration can also be assessed from the (Martínez-Vázquez 2003). In this sense, subnational perspectives of autonomy and efficiency. Vehorn governments have some administrative control at and Ahmad (1997) offer four models for tax the margin. administration in decentralized polities. These In Indonesia and China, the central government include central tax administration with revenue administers all shared taxes, while local govern- sharing, central tax administration with assignment ments administer revenues assigned to them. In of taxing powers to different levels of government, Indonesia, local revenue agencies administer the multilevel administration with revenue sharing, taxes for which they are responsible, with little and self-administration by each level of govern- support from the central government. As in the ment. Mikesell (2002) stresses another dimension: Philippines, the result is that administrative capac- the extent to which national and subnational ity and collection costs vary widely by locality. authorities cooperate or operate independently. China's Tax Sharing System reforms of 1994 cre- Table 6.10 reveals a great deal of diversity in the ated two separate tax administrations--one at the region on the administrative side, with transition TABLE 6.10 Tax Administration Models in East Asia Models Countries Observations Central administration with Vietnam Highly centralized, but dual tax sharing Cambodia subordination of tax administrators gives SNGs some control at the margin. Multilevel administration China Formally separate with tax sharing administrations for national and provincial levels, though dual subordination in practice. Indonesia Formally separate administration, though some cooperation between central and SNG tax agencies, including on property tax. Thailand Formally separate administration at the national, municipal/city, and subdistrict levels. Self-administration by each Philippines Separate provincial, municipal/city, level of government and barangay administrative levels; little cooperation between central and SNG tax agencies. Source: World Bank staff. Note: SNG subnational government. 122 East Asia Decentralizes economies closer to the centralized pole and administration is one that simply follows the tax Indonesia, the Philippines, and Thailand closer to code, leaving aside the question of whether the law the decentralized pole. is good policy. In the Philippines, bureaucrats are Guidance from the literature on good practice in employees of the local government and therefore tax administration in decentralized contexts is less agents of local executives. But bureaucratic agents clear. No consensus has developed on the principles of higher-level governments could also be responsi- of devolved administration. ble to the local chief executive. The point is that tax The efficiency criterion would argue for reducing administrators do not need to be local government total administrative and compliance costs by taking employees to ensure accountability to local govern- advantage of economies of scale and scope. The ments. Devolved responsibility does not necessarily Philippines, for example, includes hundreds of imply devolved administration, especially in the small-scale tax administrations collecting revenues context of weak local capacity. throughout the country. Their ability to attract and A number of options would preserve local auton- retain qualified personnel is limited, as is their omy while improving efficiency. These options need access to information technology. This limited not be universal for all subnational governments in capacity has direct consequences for taxpayers in a given country. Rather, subnational governments terms of higher compliance costs. Variations in local could consider the options on a case-by-case basis, capacity also mean that taxpayers do not receive which would imply asymmetrical treatment. uniform treatment throughout the country. Depending on local conditions, asymmetry might The key question is whether it is possible to cen- make sense, and would likely generate useful pilots tralize some administrative functions to reduce for more comprehensive reforms. costs while not curtailing local autonomy. That is, A similar approach would encourage local gov- to what extent is local tax administration a sine qua ernments with greater capacity to perform, for a non of autonomous local governance? One could fee, some functions for other local governments argue that since some functions of tax administra- with less capacity. That occurs with Lima's Tax tion effectively control marginal revenues, subna- Administration Service, which collects property tional governments must have control over these taxes for two other Peruvian municipalities (Ate functions in order to have--by definition--own- and La Victoria) for a 5 percent commission. The source revenues. Looking at administration as a critical issue is whether subnational governments bundled set of functions rather than a homoge- would control administration at the margin. neous process allows one to think about differential Another option would be to establish a tax treatment of administrative functions. For exam- agency that would assist local governments--on a ple, the level of enforcement activity will have a case-by-case basis, for a fee--with core administra- direct bearing on the level of tax arrears collections. tive functions: registration, collection, and compli- Thus, a subnational government that controls ance. Such a subnational tax agency would allow enforcement activities would be able to increase economies of scale and scope, which could lower revenues at the margin. The same holds for tax- administrative and compliance costs. At present payer registries, which can be managed more or less there are no known examples of this approach. aggressively, and taxpayer services. The argument is Yet another approach is for the national govern- less true for other functions. Take property valua- ment to take responsibility for functions such as tion. If valuation relies on market methods, there is property assessment, or to help local governments not much scope for differences in implementation. with core functions. National governments assist The point is that administrative as well as policy with core functions in many countries. In Colombia, levers can affect marginal revenues (though some the central government maintains the property reg- administrative effects might be quite small). ister and updates property valuations. In Cyprus, Local governments do not, in theory, need to Estonia, Jamaica, Malawi, and Pakistan, the central control all tax functions directly if they control the government is responsible for assessing property administration of those functions. The law requires and collecting taxes, even though property taxes are tax administrators, as bureaucrats, to do as their assigned to local governments (Vehorn and Ahmad political principals say. One definition of a good tax 1997; McCluskey and Williams 1999, cited in Subnational Own-Source Revenue: Getting Policy and Administration Right 123 Mikesell 2002).22 To add a dynamic element to this smaller jurisdictions in many countries. For exam- approach, one could envision local governments ple, in Indonesia and the Philippines, provinces col- progressively taking over more functions as they lect and transfer some revenues to lower levels. This develop capacity in these specialized areas. All local option--provided that higher jurisdictions are governments would not necessarily perform all tax given adequate incentives--is worth exploring as a administration functions, as some undoubtedly way to rationalize administrative arrangements make more sense administered centrally. between national and subnational governments,and Some East Asian countries have already taken between subnational governments themselves. advantage of economies of scale and scope. Respon- Though there are some success strories in the sibilities for administration vary among subnational region (see, for example, box 6.2), the administra- levels of government, with larger jurisdictional tive capacity of subnational governments is weak levels having some responsibility for collections for in many cases, and is the binding constraint on BOX 6.2 Improving Tax Administration in Quezon City, Philippines Quezon City--the largest city in metropolitan other taxes, including the business tax and Manila in land area and population--faced a seri- mayor's permit fees. ous budget deficit in 2001. The administration of · Direct withholding of taxes from city contrac- Mayor Feliciano Belmonte, Jr., who assumed office tors and suppliers, which rose from 12 per- in July 2001, inherited outstanding obligations of cent to 75 percent of gross collections. = P1.4 billion and bank debt of =1.2 billion. The city P decided to improve its revenue collection. The "carrots" include the following: The mayor's first step was to convene a search committee, headed by the dean of the College of · An increase in the discount on early payment Public Administration of the University of the of property taxes, from 10 percent to 20 per- Philippines, to recommend candidates for treas- cent for annual payers, and from 5 percent to urer. The mayor settled on Dr. Victor Endriga, who 10 percent for quarterly payers. quickly implemented reforms over the next 18 · Improved taxpayer facilities, including mod- months designed to reverse the city's fiscal course. ern air-conditioned lounges with automatic Treasurer Endriga adopted a "carrot-and-stick queuing systems, free coffee and tea, free approach." The "sticks" include the following: local telephone calls, and televisions. · Plaques from the mayor presented in a public · Property auctions for delinquent property ceremony to the 10 most "outstanding" taxpayers. The city conducted three auctions taxpayers. in 2002--the first postcode auctions in Quezon City. (Although delinquent taxpayers The city also computerized its property and busi- have sued the city, they do not phase the ness tax registries and collection processes, hired treasurer: "it's part of the game.") an outside firm to input all paper records, insti- · Delinquency letters sent to recover the esti- tuted new security features, raised tax rates, and mated =10.7 billion owed the city. (Each staff P reassigned employees within the treasurer's member must send out at least 20 letters office to avoid familiarization. The city now daily, as staff had sent few before.) rewards (such as with overseas trips) and pun- · The use of presumptive minimum levels of ishes revenue examiners based on actual collec- gross sales for the business tax and for markets. tions, and conducts house visits of delinquent · A requirement that all business establish- taxpayers. ments with gross receipts over =500,000 sub- P This reform program has paid off. Own- mit their previous year's financial statement, source revenues rose from =2.3 billion in 2001 to P as well as information on monthly tax pay- = P3.9 billion in 2002, and the city closed the year ments from the Bureau of Internal Revenue. with a surplus of =0.5 billion. P · A requirement that all payments of real estate taxes include confirmation of payment of Source: Endriga 2003. 124 East Asia Decentralizes improving revenue performance throughout the reported that instead of confrontation, mayors used region (see box 6.3). As Bahl has argued regarding their "charisma" to persuade businesspeople to pay China, "Tax administration shortcomings plague taxes. One government reported active opposition Chinese fiscal policy" (1999, 66). The principal from the chamber of commerce to the General problems include the following: Revision of Assessment, and another reported intense lobbying from the private sector against · The prevalence of stop filers, nonfilers, and late increases in tax rates, which resulted in compromise. filers, owing to low local capacity to register tax- Other localities report that businesses have taken payers. This results in delinquent payments and legal action against property auctions and tax the accumulation of arrears, especially in the increases. Taxpayers sued Quezon City, for example, Philippines. over"confiscatory"tax rates. · Infrequent audit and enforcement (temporary While no one knows the precise impact of closures and property auctions), resulting in low political constraints on local taxation, a number compliance. This seems to be a problem in both of hypotheses--focusing on incentives--may be Indonesia and the Philippines. posited. One possibility is that politicians simply · The limited availability of taxpayer services, weigh the net impact, or political tradeoff, of higher although some local governments in Indonesia revenues against greater expenditures. Simply put, and the Philippines offer important examples of officials decide to increase taxes (either through good practice. higher rates or better administration) when the dis- · The low professional qualifications of staff in all counted benefits of greater expenditures are higher three countries. than the discounted costs of higher taxes. Many · Inadequate support from and coordination with officials seem to decide that the political costs of the national government. higher taxation are greater than the expected bene- fits. There are several possible reasons. First, the Overall, the East Asian experience suggests that expected marginal benefit is small compared with the multilevel administration model (with tax shar- the political cost of unhappy constituents. Ineffi- ing) may hold the most short-term promise. The ciencies in delivering services (such as patronage or advantage of this model is the possibility that dif- padded projects) might also outweigh the impact ferent levels of administration could assume differ- of greater revenues, "wasting" the tax increase. ent functions, and it could also facilitate assistance Another possibility is that the incidence of to subnational governments. Assigning complex benefits and costs--in terms of expenditures and tasks with economies of scale (such as property val- revenues--might undercut local elites. Increases uation) to either a central government agency or an in property tax collections would most likely harm agency dedicated to subnational support would the wealthy from a distributive standpoint, while reduce administrative costs and likely improve increases in service delivery would most likely favor service quality. East Asian countries could further lower-income groups. However, politicians may explore the idea of a subnational tax support not weigh these impacts equally, and may be reluc- agency, funded by subnational governments and tant to use higher tax collections to fund pro-poor under their control. Other solutions, such as for service delivery. subnational governments to contract out to other Other explanations focus on the timing of bene- local governments or piggyback on national taxes, fits and costs, the extent of executive control over are also worth consideration. spending, and the electoral strength of the incum- bent. Term limits on local executives--which The Politics of Local Taxation limit them to two terms of three years each in the While policy and administrative constraints on local Philippines, for example--might also make them taxation are critical, political constraints also affect reluctant to invest in tax improvement programs performance. Several local governments in the that would yield fruit over the long term. Philippines, for example, reported that "political Property tax collection by local governments in intervention"underlay their limited use of the power the Philippines may be particularly problematic. As to auction or close businesses. Other governments in most former Iberian colonies, landholding in Subnational Own-Source Revenue: Getting Policy and Administration Right 125 BOX 6.3 Local Revenue Administration in Action in East Asia Evidence from Indonesia: ­ A lack of automated registers. ­ Low-quality record keeping. · Revenues are administered according to a · Business tax registers are known to be incom- multilevel model: plete, given frequent changes in registered ­ The central government administers establishments, which results in low levels of national taxes. control and compliance. ­ Local tax agencies, generally known as · Problems undermining property tax collec- DIPENDAs, play a minor role in administer- tions include: ing the property tax. ­ Collection efficiency for property taxes is ­ Vehicle and vehicle transfer taxes are jointly low, resulting in the hemorrhaging of the administered by the provincial DIPENDA, most important source of local revenues. the national police (as the coordinator), Local governments seem to be in a weak and a state-owned insurance firm. position to collect the taxes. Improving the · Own-source revenues are administered efficiency of collection can raise these rev- directly by DIPENDAs of the cities, regencies, enues significantly. and provinces, except for street lighting and ­ Property assessments have not kept up fuel taxes. However, by issuing permits with changes in market values. Most local and licenses, local departments actually governments have not performed the Gen- collect the user charges coordinated by the eral Revision of Assessment since 1991, DIPENDAs. resulting in significant undercollection of · The administrative performance of the property taxes. DIPENDAs varies widely. ­ Noncompliance with the requirement for · DIPENDAs have few cooperative agreements regular assessment requires urgent atten- or information exchanges with other agen- tion. The country could revise the code to cies within the same government, except for allow the national government to do the property tax field offices of the Directorate assessments for local governments that do General of Taxation. not comply. · DIPENDAs may use a certain percentage of · The compliance function seems inadequate. total tax revenues to pay allowances to staff, Major constraints include: though these bonuses are not usually based ­ Infrequent exercise of local audit authority. on performance. ­ Infrequent exercise of local enforcement · The quality of tax administration varies. Most authority, in the form of temporary clo- DIPENDAs receive taxes directly in their sures of firms and auctions of property. offices, while others use partially government- · The difficulties of local governments in owned regional development banks. enforcing compliance suggest the need for · One of the highest priorities of most DIPENDAs presumptive taxation in some cases, espe- is developing the ability to professionally audit cially for the business tax and some fees and taxpayers, considered the weakest link in the charges. system. The approach to taxpayer auditing · The principal constraint on taxpayer services varies by local government. DIPENDAs tried to is their limited availability. The lack of even introduce information technology in the early the most basic taxpayer services and written 1990s, but few local governments are still materials means high compliance costs are operating the computer systems because of likely to be the norm across the country. lack of training. · One main cause of underperformance is weak administrative capacity, owing partly to Evidence from the Philippines: the low professional qualifications of staff. Inadequate coordination with the Bureau of · Principal constraints on taxpayer registration Internal Revenue, and low levels of support include: from the Bureau of Local Government ­ A lack of regular maintenance and validation Finance, which might establish information- of the property tax and business registers. sharing protocols, are also factors. 126 East Asia Decentralizes that country has been concentrated. This legacy avenues, subnational governments do not appear to pits powerful landowners--some of whom have have exhausted all their options. In some countries diversified their wealth into other assets--against and for some taxes, local governments appear to weak local governments, some of which have been prefer a lax collection strategy (property taxation in run by local elites for generations. Not surprisingly, the Philippines, for example). Moreover, to the the ability of local governments to enforce compli- extent that the quality of local tax administration ance with local revenue codes is weak in the face of reflects both capacity and interest (according to elite resistance. the "revealed-preference" line of thought), then many subnational governments "prefer" weak administration. Conclusions and Implications Incentives thus play a role in determining collec- This review of experiences in Cambodia, China, tion levels of own-source revenues. This chapter Indonesia, the Philippines, Thailand, and Vietnam suggests that two kinds of incentives play a role. finds that local governments have limited authority The first is the systemwide incentive. For example, and ability to raise their own revenues at the mar- inherent in the intergovernmental fiscal transfer gin. As a result, own-source revenues are low as a system are incentives that affect own-source collec- share of total subnational resources, and in some tions. In Indonesia, the fact that the principal inter- cases have been declining as a share of total governmental grant is based on the difference resources. Limited revenue-raising power and between fiscal needs and revenue capacity, rather capacity raise questions about the supposed bene- than actual revenues, gives subnational govern- fits of decentralization in improving accountability ments incentives to raise collections to close the and allocative efficiency. Reliance on transfers in gap. The opposite is the case in Vietnam, which their myriad forms creates dependence on the bases transfers on the difference between expendi- national government and may weaken subnational ture needs and forecasted revenues, which are incentives to improve own-source revenue. Four based on previous collections. This formula could key messages merit emphasis. provide negative incentives, since higher collections First, local governments have limited control result in lower transfer amounts (World Bank over tax policy, including the ability to set rates and 2004b). Second, rational politicians might not pre- define the tax base. Decentralization is thus more fer to increase own-source collections under cer- political than fiscal. Lack of control over taxation tain circumstances. at the margin breaks the tax-accountability link, Fourth, improvements in local tax administra- undermining the expenditure efficiency promised tion would greatly strengthen subnational finance by decentralization. While the general limit on systems. Tax administrations vary throughout the own-source revenues is not the only constraint, region, ranging from highly decentralized in the policy autonomy is essential for significant Philippines to highly centralized in Vietnam, with improvement to occur over the medium term. Indonesia and China between these two poles. Yet Second, the lack of authority over tax policy the relative roles of local and national governments seems to have spurred local governments to seek have not been well designed, resulting in both unofficial tax and nontax sources of revenue, with capacity and incentive challenges. Local govern- deleterious consequences. In a number of coun- ments tend to underperform on own-source collec- tries, local governments have the authority to enact tion and administration, reducing the credibility of new taxes and fees and thus engage in entrepre- the local tax system and contributing to a culture of neurial behavior, yielding both positive and nega- noncompliance by raising compliance costs for tax- tive results. The resort to informal and illegal fees payers. Local administrative capacity is thus quite is even more unfortunate considering that subna- weak in many cases, and the binding constraint on tional governments in the region are unable to avail improving revenue performance. themselves of many taxing options open to govern- This chapter points out that the lack of develop- ments in other regions. ment of significant own-source revenues in many Third, despite the lack of opportunity to raise countries is limiting the extent to which subna- revenues and the apparent interest in "unofficial" tional governments can finance decentralized service Subnational Own-Source Revenue: Getting Policy and Administration Right 127 delivery and make decisions about taxation and Pakistan, where small towns (tehsils) adopted levies on gross service levels. However, it must be noted that local receipts in 2002 (and contracted out collections). 9. It is noteworthy that the region has no cases of piggyback governments in some East Asian countries receive taxation with local rate setting. relatively large shares of national income. Thus 10. User charges are defined here, following Bird and Tsiopoulos any efforts to boost own-source revenues would (1997, 39), as "charges levied on consumers of government goods and services in relation to their consumption," when need to occur in the broader context of matching consumption is voluntary (such as public water charges). total resources--including transfers and shared Fees are defined, in contrast, as cost recovery for a mandated revenues--to expenditure responsibilities. public service (such as automobile licensing). 11. In fact, provincial councils will set the rate for Thailand's Challenges remain for improving local tax policy vehicle tax as of 2005. and administration. The lack of autonomy under- 12. However, the Local Government Code in the Philippines mines the ability of local governments to realize the prohibits local excise taxes (Sec. 133 (h)) as well as levies on benefits of decentralization by tapping significant motor vehicle registration and driving (Sec. 133 (l)). 13. Ahmad et al. (2000), report that extrabudgetary funds rep- revenue sources to satisfy local preferences regarding resented nearly 40 percent of all local revenues in 1999. the level and quality of services. Fiscal sustainability 14. R.A. 7160, Sec. 186. requires improvements in own-source revenue col- 15. For any taxes not listed in Law 34, local governments can decide on appropriate tax bases and rates, as the central lection and administration more generally. Weak government establishes only general principles. administration undermines local tax systems by con- 16. The regional parliament must pass these regulations to tributing to high rates of noncompliance, high com- authorize the introduction of a new tax or charge. 17. The World Bank calculated the buoyancy coefficient of pliance costs for taxpayers, and high administrative local taxes at 1.6 for the 1994­1999 period (2002, 55). costs for local governments. Getting the relationship 18. However, a significant share of property tax revenues is between the national and local governments right-- earmarked for education spending. On average, over the in both policy and administrative terms--is pivotal. period 1992­2000, the Special Education Fund, which is earmarked for education, accounted for 44 percent of total property tax collections, which is quite high considering Endnotes that property taxes are the main source of subnational own-revenue. Earmarking is another form of reduced sub- 1. For further discussion, see Bird 2003b; and Martínez- national autonomy. Vázquez 2003. 19. This is based on annualized data for fiscal year 2000 (which 2. For further discussion, see Mikesell 2002. Many of these lasted nine months). In general, observers agree that the papers and others are available at www.decentralization.org. quality of fiscal year 2000 data is questionable, since-- 3. Local Government Code, Sec. 191. besides being shorter--this year was the transition to decen- 4. The percentage split for the Land and Building Tax (exclud- tralization. Thus, it would not be appropriate to place much ing mining and plantations) is: center (9 percent), which is emphasis on comparisons involving fiscal year 2000 data. intended to cover administrative costs of the deconcen- 20. The only notable exception, in some ways, to this rule is trated revenue offices; provinces (16 percent); originating that subnational treasurers are employees of the Depart- local government (65 percent); 6.5 percent equally across ment of Finance. regions; and 3.5 percent to regions that exceed their previ- 21. Still, the General Taxation Department is responsible for ous year's revenue target. A Land and Building Transfer tax appointing, promoting, and transferring departmental is also shared: provinces (16 percent), originating local gov- staff. To what extent local officials have input into these ernment (64 percent), and the remainder to all local gov- processes is unknown. ernments. The personal income tax is 80 percent central, 22. See also Mikesell 2002 on intergovernmental tax adminis- 8 percent provincial, and 12 percent originating local gov- tration compacts in the United States. ernment. See Law 25/1999, GR 115/2000, GR 104/2000. 5. Reference to Indonesia's"regions"generally means provinces and regencies and cities, while the term "local governments" Bibliography refers to regencies and cities. 6. The Salakhet refers to the portion of the provincial budget Ahmad, Ehtisham, Li Keping, and Thomas Richardson. 2000. that is under the control of the provincial governor, not the "Recentralization in China." Paper for IMF Fiscal Decentral- budgets of the deconcentrated line ministries at the provin- ization Conference, Washington, DC. cial level. Nationally, the Salakhet represents about 20 per- Aoki, Naomi. 2003. Internal draft document on China and cent of the total provincial budget allocation. Vietnam. World Bank, Washington, DC. 7. Only 7 cities charge the maximum 3 percent rate, and only Bacolod City,Philippines.1995.Ordinance no.154.Revenue Code. 21 cities charge rates greater than or equal to 2.5 percent Bahl, Roy. 1999."Fiscal Policy in China: Taxation and Intergovern- (outside the National Capital Region). No cities or provinces mental Fiscal Relations."San Francisco,CA:The 1990 Institute. in the NCR charge the maximum allowed. Bird, Richard M. 2003a. "A New Look at Local Business Taxes." 8. For example, countries could impose the business tax on gross Tax Notes International (May): 695­711. receipts or rely on a subtraction method value-added tax.Note ------. 2003b. "Local and Regional Revenues: Realities that this will require interjurisdictional apportionment of the and Prospects." Internal draft document. World Bank, tax base. However, the process need not be complex as in Washington, DC. 128 East Asia Decentralizes Bird, Richard M., Robert D. Ebel, and Christine I. Wallich. Norregaard, John. 1997. "Tax Assignment." In Fiscal Federalism 1995. Decentralization of the Socialist State: Intergovern- in Theory and Practice, ed. Teresa Ter-Minassian. Washing- mental Finance in Transition Economies. Washington, DC: ton, DC: International Monetary Fund. World Bank. OECD (Organisation of Economic Co-operation and Develop- Bird, Richard M., and Enid Slack. 2002."Land and Property Tax- ment). 2002."Fiscal Design Surveys across Levels of Govern- ation: A Review." Internal draft document. World Bank, ment." Tax Policy Studies no. 7. Paris: OECD. Washington, DC. Ortega, Mary Jane, Mayor, San Fernando City, Philippines. 2003. Bird, Richard M., and Thomas Tsiopoulos. 1997. "User Charges Interview. for Public Services: Potentials and Problems." Canadian Tax Ortega, Victor, Governor, La Union Province, Philippines. 2003. Journal 45 (1): 25­86. Interview. Byrne, Peter. 1995. "Overview of Privatization in the Area of Philippines Center for Investigative Journalism. 2001. Investigat- Tax and Customs Administration." Bulletin for International ing Local Governments. Manila: Philippines Center for Inves- Fiscal Documentation 49: 1. tigative Journalism. CabanatuanCity,Philippines.1992.Ordinanceno.165-92.Revised Ray, D. 2003. Summary paper for conference on Decentraliza- Revenue Code. tion, Regulatory Reform, and the Business Climate, Partner- Chakeri, Jasmin. 2003. "Regional Tax Policy and Administration ship for Economic Growth and U.S. Agency for International in Indonesia." Internal draft document. World Bank, Development, Jakarta. Washington, DC. Republic of the Philippines. Local Government Code, 1991. Commission on Audit. 2001. Annual Audit Report on Local Republic Act 7160. Manila. Government Units. Quezon City: Republic of the Philippines. San Fernando City, Philippines. 1998. Tax Ordinance no. 98-001. ------. 2002. Annual Audit Report on Local Government Units. San Fernando Municipality, Philippines. 1993. Tax Ordinance Quezon City: Republic of the Philippines. no. 92-002. Revenue Code. Devas, Nick, and Roy Kelly. 2001. "Regulation or Revenues? An Sugana, Rubi. 2003. "Local Own-Source Revenue Administra- Analysis of Local Business Licenses, with a Case Study of the tion in Indonesia." Internal draft document. World Bank, Single Business Permit Reform in Kenya." Public Administra- Washington, DC. tion and Development 21: 381­391. Tabunda, Manuel S., and Mario M. Galang. 1992. A Guide to the Ebel, Robert, and Serdar Yilmaz. 2002. "On the Measurement Local Government Code of 1991. Manila: Mary Jo Publishing and Impact of Fiscal Decentralization, Part 2." In Public House. Finance in Developing and Traditional Countries: Essays in Taliercio, Robert R. 2003. Internal draft document on the Honor of Richard Bird, eds. Jorge Martínez-Vázquez and Philippines. World Bank, Washington, DC. James Alm. London: Edwin Elgar Press. United Nations Capital Development Fund (UNCDF). 2003. Endriga, Victor B., Treasurer, Quezon City, Philippines. 2003. "Supporting Cambodia Fiscal Decentralization Priorities: Interview. 2004 and Beyond." Phnom Penh. ------. 2001. "Manila, Makati, Pasig, and Quezón City: Local ------. 2004. "Draft Material for UNCDF Project Document Government Taxation." Manila: Quezon City. for Support to the Cambodian Ministry of Economy and ------. 2003."The Financial Recovery of Quezon City." Manila: Finance on Fiscal Decentralization." Phnom Penh. Quezon City. USAID (U.S. Agency for International Development). n.d. Guevara, Milwida. 2003. "Feeling the Pulse of the Public." Inter- "Geographic Information Systems Applications in Local nal draft document. World Bank, Washington, DC. Governance: Selected GIS Outputs Produced by Philippine ------. 2003. "Taming the Deficit." Internal draft document. Local Government Units, 1999­2000." Washington, DC: World Bank, Washington, DC. USAID. La Union Province, Philippines. 1996. Provincial Tax Ordinance Vehorn, Charles L., and Ehtisham Ahmad. 1997. "Tax Adminis- no. 001-96, tration." In Fiscal Federalism in Theory and Practice, ed. Lewis, Blane. 2003. "Tax and Charge Creation by Regional Gov- Teresa Ter-Minassian. Washington, DC: International Mone- ernments under Fiscal Decentralization: Estimates and tary Fund. Explanations." Bulletin of Indonesian Economic Studies 39 Vergara,Jay,Mayor,CabanatuanCity,Philippines.2003.Interview. (2): 177­192. World Bank. 2000. "Vietnam: Managing Public Resources Bet- Makati Municipality, Philippines. 1993. Municipal Ordinance ter." Joint report of the Donor Working Group on Public no. 92-072. Makati Revenue Code. Expenditure Review. Hanoi: World Bank. Martínez-Vázquez, Jorge. 2003. "Improving the Design of Fiscal ------. 2002. "China: National Development and Subna- Decentralization inVietnam."Internal draft document.World tional Finance--A Review of Provincial Expenditures." Bank,Washington, DC. Washington, DC: World Bank. McCluskey, William, and Brendan Williams. 1999. "Introduc- ------. 2003a. "Decentralizing Indonesia: A Regional Public tion: A Comparative Evaluation." In Property Tax: An Inter- Expenditure Review--Overview Report." Washington, DC: national Comparative Review, ed. William McCluskey. World Bank. Brookfield, VT: Ashgate Publishing. ------. 2003b. "Enhancing Service Delivery through Improved McLure, Charles E., and Jorge Martínez-Vázquez. 2002. "The Resource Allocation and Institutional Reform." Washington, Assignment of Revenues and Expenditures in Intergovern- DC: World Bank. mental Fiscal Relations." Internal draft document. World ------. 2004a. "Decentralization in East Asia: Country Bank, Washington, DC. Review--The People's Republic of China." Background Mikesell, John L. 2002."International Experience with Adminis- paper. Washington, DC: World Bank. tration of Local Taxes: A Review of Practices and Issues." ------. 2004b. "Decentralization in East Asia: Country Internal draft document. World Bank, Washington, DC. Review--Vietnam." Washington, DC: World Bank. Musgrave, Richard. 1959. The Theory of Public Finance: A Study ------. 2004c. "Fiscal Decentralization in Thailand." Internal of Public Economy. New York: McGraw-Hill. draft document. World Bank, Washington, DC. 7 Managing Human Resources in a Decentralized Context Amanda E. Green A central rationale for decentralization is that, management is a two-way process. The behavior of by moving government closer to the people, it civil servants has important consequences for gov- brings public sector activities and decisions in line ernment performance in a decentralized setting. with popular preferences. Yet, in practice, the civil Conversely, decentralization alters both the incen- service--a critical component of government-- tives of and the demands on the civil service. rarely enters the decision calculus of decentraliza- Managing this behavior is critical to realizing the tion design. The civil service is often recognized as benefits of decentralization. an afterthought rather than seen as essential to suc- Accordingly, this chapter argues that civil cessful decentralization. This is lamentable but not service management--or more broadly human surprising, as decentralization is quintessentially a resource management--should be seen as an political process. In East Asia, whether it was the essential component in the design of decentraliza- desire to quell the forces of regional disintegration tion rather than a separate, stand-alone process. in the Philippines and Indonesia, the urgent need The discussion marshals evidence from across East to meet the demands of economic transition in Asia and around the world. The following section China and Vietnam, or pressure to improve service presents a framework for exploring the interaction delivery and citizen participation in Cambodia and between administrative decentralization and civil Thailand, the primary motivation for decentraliza- service management--in terms of both the theo- tion has been political--notwithstanding varied retical goals of decentralization and their realiza- proximate causes. tion in practice. The third section delves into the The common failure to address the details of realities of administrative decentralization in East civil service management as an integral part of the Asia through the experiences of six countries: decentralization package has significant implica- Cambodia, China, Indonesia, the Philippines, tions. Civil servants form a crucial link between the Thailand, and Vietnam. The varied approaches delivery of financial resources to the government and responses of these countries offer insights and the delivery of essential public services to into the process of managing civil servants in a the people (World Bank 2003a). The relation- decentralized setting. The final section examines ship between decentralization and civil service the central dilemmas that arise in designing 129 130 East Asia Decentralizes administrative decentralization and draws lessons menting decentralization policy has slowed con- on how countries in East Asia and beyond can siderably. China and Vietnam have taken a more maximize the benefits and minimize the risks of opportunistic approach, experimenting with dif- decentralization. ferent degrees of local autonomy. However, the central government and Communist Party con- tinue to influence how decentralization plays out. The Interplay between In Thailand, a legal framework defines an interme- Decentralization and Human diate form of administrative decentralization, but Resource Management most of this framework has yet to materialize in Decentralization is a spectrum rather than a single practice. Finally, Cambodia can best be described state, ranging from deconcentration to delegation as a deconcentrated system, with a high propor- to devolution (see box 1.1). As table 7.1 shows, tion of staff based in the field but working on key features of the civil service vary with the behalf of the central government. degree of administrative decentralization. Overall, Movement along the spectrum of administrative East Asian countries fall into the intermediate cat- decentralization depends partly on interplay with egory, with local managers enjoying some freedom the political and fiscal aspects of decentralization. to recruit and allocate staff subject to central For example, a local government that has full guidelines on pay levels and total employment. authority over the size of its civil service can Indonesia and the Philippines took the biggest nonetheless face restrictions in the use of funds leaps along the decentralization continuum in transferred from the center. Similarly, the strength both law and practice, but central governments of a country's accountability framework is influ- retained considerable control over civil service enced by whether political decentralization has cre- wages at the local level. In Indonesia, recent revi- ated institutions for locally elected politicians to sions to the Law on Regional Administration have oversee the activities of local governments. The case rolled back decentralization in hiring decisions. studies illustrate the results of a mismatch in the In the Philippines, the momentum behind imple- dimensions of decentralization. TABLE 7.1 Key Features of Administrative Decentralization Administrative Decentralization Deconcentration Delegation Devolution (minimal change) (intermediate change) (substantial change) · Provider staff working at local · Providers could be employees · Providers are employees of level are employees of and of central or local government, local government. accountable to the center, but the center typically defines · Local government has full usually through their pay and employment. discretion over salary levels, respective ministries; central · Local government has some allocation, and numbers of employees compensate for authority over hiring and staff, as well as the authority weak local capacity. location of staff, but is less to hire and fire. · Accountability remains likely to have authority over · An overarching civil service distant: the short route of firing. framework covering local accountability may be weak if · Both long and short routes of governments may still provider monitoring is weak, accountability are potentially establish standards and and citizens may have to stronger; greater local procedures for hiring and rely on a weak long route knowledge can allow better managing staff. stretching to politicians at the matching of supply with local · Potentially strongest long and center; a strong compact preferences and better short routes to accountability, between policymakers and monitoring, strengthening but influenced by local social providers can compensate to both the compact and client norms and vulnerable to local some extent. power. capacity constraints and politics. Source: Adapted from World Bank 2003h, 189. Managing Human Resources in a Decentralized Context 131 What Makes a Decentralized country. Otherwise, local governments in poorer Civil Service Work?1 areas will find it difficult to compete against richer areas, and national coverage of public services may To better understand the opportunities and obsta- suffer. Second, central government involvement in cles that a country may encounter on the road to staffing may expand civil servants' career paths by administrative decentralization, it is first useful to opening channels between local and central consider the destination. A functioning system of employment. Third, the center may wish to retain decentralized civil service management--as dis- control over hiring and pay to shield subnational tinct from a centrally directed model--has several governments from local pressure to overspend on key characteristics: wages and salaries. Fourth, centralized rules can be · Local government functions are clearly defined important in sectors that need to enforce minimum so staff know what is expected of them and man- professional qualifications, such as with teachers, agers can adapt the local civil service to reflect doctors, and nurses. Fifth, where ethnic or other what needs to be done, without inefficient gaps tensions threaten stability, the central government or overlap with other levels of government. may use the civil service as a tool for national · Local government can allocate staff across func- integration. Finally, the central government's reluc- tions as needed. This requires that civil service tance to decentralize may stem simply from a desire managers have autonomy--or at least influ- to retain control. ence--in operating the local establishment as well as deploying staff across different depart- The Implications of Decentralization ments and facilities. How a country resolves the tension between the · Local government is able to attract and retain motivations for decentralizing the civil service on qualified individuals, and to build a team with a the one hand, and the reasons for caution on the diverse set of skills. This requires that local other, will shape the design of decentralization and, government have something to offer, through ultimately, how the process plays out. The implica- competitive pay, career opportunities, prestige, tions of decentralizing civil service management or other incentives. can be examined along four critical dimensions: · Local government has flexibility in managing capacity, incentives, autonomy, and accountability. financial resources. Managing civil servants Each factor plays a significant role in the success of requires managing their cost, either directly decentralization and is, in turn, heavily influenced through pay levels or indirectly through staff by the decentralization process. numbers. These four dimensions are closely linked and · Local government can hold staff accountable for entail important tradeoffs. For example, civil serv- their performance. This requires the capacity ice training programs are not likely to strengthen to supervise and monitor civil servants, the abil- capacity in a sustainable way unless incentives moti- ity to reward good performance through pay vate civil servants to use what they learn. Similarly, increases, promotions, and other benefits, and improvements in accountability at the local level the authority to punish deficient performance require that civil servants have the capacity-- through disciplinary measures or dismissals. through accounts and records--to render that Countries may not meet some or all of these accountability effectively. Finally, holding local civil criteria for numerous reasons. In some cases, the service managers accountable for their decisions is failure is one of design. For example, as mentioned, difficult when they do not have autonomy in mak- comparatively limited independence among local ing those decisions. governments on fiscal and political matters may Capacity. For civil servants to deliver the constrain their administrative autonomy. higher-quality local services envisioned under In other cases, it is by design that the criteria are decentralization, they need to have the capacity to not met. First, keeping local civil services under do so. This involves both individual and institu- the umbrella of central direction can ensure some tional elements. First, the success of decentraliza- standardization in working conditions across the tion depends on the ability of individual civil 132 East Asia Decentralizes servants to take on new tasks, at both central and dismiss ineffective staff, and to trim numbers to local levels. Second, smaller size and budgets can keep costs down. Though less common at the local constrain the institutional capacity of local govern- level, direct financial autonomy--such as the ability ments. The process of decentralization itself can to set pay levels and charge user fees--can improve have important implications for capacity needs at staff performance and thereby enhance the benefits the local level. The devolution of public service of decentralization. However, to achieve these bene- responsibilities requires both a broader variety of fits, performance incentives and accountability skills and a greater depth of knowledge in specific frameworks must be robust enough to prevent areas, such as financial management and perfor- inefficiency and mismanagement. By definition, mance monitoring. Moreover, local leaders need to administrative decentralization would seem to aug- learn to supervise staff, mobilize more own-source ment local autonomy, but this is not always the case. revenues, interact with local constituents and The center often retains a significant degree of elected officials, and develop local institutions. For- control, particularly in financially sensitive areas merly central civil servants who move to local levels such as the size and wage levels of the civil service, can transfer some of these skills; local governments especially when local capacity and accountability are need to gain the rest through experience. At the in question. Table 7.2 provides a stylized but useful same time, central employees need to shift from comparison of East Asian countries in these areas. "doing" to facilitating and supervising. Accountability. The potential for improving Incentives. In some cases, what appears to be a service delivery through decentralization depends, lack of capacity to carry out the functions of decen- among other factors, on accountability relation- tralized government is instead a lack of motivation ships--the degree to which civil servants are held to act in the public interest. The structure and man- to account for their performance and integrity, and agement of the civil service influence the outcome to whom they are accountable. Without a strong of decentralization reforms by affecting how local system of local accountability, devolving authority civil servants behave. The level of pay and benefits, and financial resources to local governments can options for career mobility, and degree to which lead to waste or misuse of public funds, and the merit is recognized or unsatisfactory performance potential for political capture at the local level can penalized can determine the dedication with distort the benefits of decentralization. On the which a civil servant works, as well as the type of other hand, where corruption is systemic at the individuals who choose to become civil servants. central level, devolution may enhance service deliv- Decentralizing functional and management ery. Where political decentralization allows for responsibilities to the local level, in turn, modifies oversight by locally elected bodies, the need for the incentive structures of local civil servants. The reelection offers strong incentives for better per- proximity of local government to the recipients of formance. If civil society monitoring mechanisms public services can tighten the link between efforts are strong, the downward accountability of local and results. However, some local civil services are staff will tend to encourage a closer connection too small to offer significant opportunities for between public services and citizen demands. If career advancement, and poorer local governments accountability is only to the center, however, decen- may be unable to pay salaries high enough to tralization may not deliver the potential benefits of attract talent. In remote areas, a combination of bringing government closer to the people. low pay and difficult conditions creates a vicious Decentralization can, in turn, affect local circle whereby an inability to attract high-quality accountability. In shifting responsibility for moni- staff leads to further deterioration in conditions. toring civil servants to local managers, decentral- Autonomy. The argument that decentralization ization can make it more difficult for civil servants boosts the responsiveness of civil servants assumes to get away with laziness and corruption. The that local managers have the authority to respond to closer proximity of local citizens to government the demands of their constituents. Local autonomy decision makers can enhance their ability to hold in allocating human resources can improve effi- those decision makers accountable. However, if ciency by allowing managers to hire staff whose critical checks and balances to protect the neutral- skills align with planned activities, to discipline or ity and independence of civil servants are not in Managing Human Resources in a Decentralized Context 133 TABLE 7.2 Central Authority over the Subnational Civil Service Cambodia China Indonesia Philippines Thailand Vietnam Legislation and regulations: Designation 1 2 1 2­3 2 1 Recruitment 1 4 2 3 2 2­3 Structure and career management: Establishment control 1 3 2 2­3 2 1 Appointment and mobility 2 3 3 4 2 3 Employment framework 1 1 1 3 1 1 Performance management: Standard setting and rewards 2 2 2 3 2 2 Training and development 2 3 3 4 3 3 Accountability 2 4 3 4 2 2 Sources: World Bank 2003d, 16; staff estimates. Note: 1 total central authority; 2 central dominance; 3 central guidance; 4 central leadership; 5 autonomous. place, decentralization can lead to nepotism, with whom they consider civil servants. Some countries local managers rewarding family members and treat teachers, health workers, and police as part of supporters with coveted positions, and can facili- the civil service,while others consider them separate tate political capture by bringing civil servants categories. The distinction between ministry staff within reach of local power bases. on the one hand, and employees of public service Taken together, the capacity, incentives, auton- agencies and state-owned enterprises on the other, omy, and accountability of civil service manage- may also blur the definition of a civil servant.Finally, ment provide both a lens through which to evaluate the determination of which civil servants are subna- the design of decentralization and a picture of how tional employees can reflect either their physical administrative decentralization has played out in location or the level of government that pays them. practice. The following section shows how these Even in the absence of agreed standards and four dimensions have influenced decentralization definitions, broadly characterizing civil service in East Asian countries, and how policy makers can decentralization in the region is worthwhile. leverage those dimensions to get the most out of Figure 7.1 shows the wide variation in civil service decentralization. structures across the region. The share of employ- ees at the subnational level ranges from around 19 percent in Thailand to just over 90 percent Civil Service Management in China. The share of personnel expenditures in on the Ground in East Asia total subnational spending also reflects the extent The structure of government and, by extension, the of administrative decentralization. As figure 7.2 civil service varies greatly across East Asia. Some shows, average subnational spending on personnel countries have focused decentralization efforts on ranges from roughly 41 percent of total spending the lowest levels of government, while others have in Thailand to around 60 percent in China. It is emphasized deconcentration to an intermediate important to note, however, that these figures do level, such as provinces. Countries also differ in not necessarily point to China as the region's most 134 East Asia Decentralizes FIGURE 7.1 Subnational Employees as a FIGURE 7.2 Personnel Spending as a Share Share of Total Government Employment of Total Subnational Government Expenditures 100 90 70 80 60 70 60 50 50 40 percent 40 30 30 percent 20 20 10 10 0 0 China IndonesiaVietnam PhilippinesCambodia Thailand China Cambodia Philippines Indonesia Thailand Sources: Cambodia--World Bank 2003a (includes political appointees and nonpermanent staff); China--World Bank 2002a, Sources: Cambodia--World Bank 2003a; China--World Bank annex 2 (includes Public Service Units outside health and 2003b; Indonesia--World Bank 2003d; Philippines--World Bank education); Indonesia--World Bank 2003d; Philippines--World 2003f; Thailand--Weist 2003. Bank 2003f; Thailand--Kingdom of Thailand, Office of the Civil Note: Data are for the most recent available year, ranging from Service Commission; Vietnam--Government of Vietnam, General 2000­3. Statistical Office, 2003. Note: Data are for the most recent available year, ranging from 2000­3, and exclude staff in the education, health, military, and police sectors, unless otherwise noted. decentralized country. The degree of authority and innovative responses, the cases attempt to glean exercised by subnational civil service managers in important lessons about the interplay between determining the amount, structure, and allocation capacity, incentives, autonomy, and accountability of resources plays an important role. Though dif- in decentralized civil service management. ferent countries have focused their decentralization policies on different levels of government, table 7.3 The Big Bang and Beyond: The Philippines provides a general picture of how civil service man- and Indonesia agement practices have taken shape in East Asia. The remainder of this section outlines the expe- In both the Philippines and Indonesia, decentral- riences of East Asian countries in dealing with the ization was an integral part of a political open- human resource implications of decentralization. In ing following the overthrow of an authoritarian Indonesia and the Philippines, extensive decentral- regime. In these sprawling archipelagos, rising ization of the administrative apparatus has brought social tensions added a note of urgency to the deci- to light both the opportunities and the risks of sion to decentralize, and local autonomy came to be devolving authority to local managers. In China seen as the key to quelling threats to national unity. and Vietnam, where decentralization has proceeded Neither government felt it had the luxury of per- in stages, the roles of state and party offer additional fecting the design of intergovernmental fiscal and insights into the relationship between decentraliza- administrative arrangements before decentralizing. tion and civil service management. Finally, in Cam- Following the logic that a chasm cannot be bodia and Thailand, the focus on building capacity crossed in two leaps has worked well in many before rather than through decentralization has ways.2 Neither country experienced major service meant limited implementation of decentralization disruptions during the transition, and the rela- policies, and this experience offers an opportunity tively open-ended design of decentralization has to consider how best to balance caution and encouraged innovative local responses to new progress. Together these six case studies explore the responsibilities. However, once the dust settled, realities of human resource management in a momentum slowed, and important challenges decentralized setting. In addressing both pitfalls remain in correcting some of the imperfections Managing Human Resources in a Decentralized Context 135 TABLE 7.3 Staffing Authority among Subnational Governments Enabling mechanisms Cambodia China Indonesia Philippines Thailand Vietnam Budget control · Determine the wage envelope · Dismiss surplus staff Establishment control · Control overall staffing numbers · Control staffing numbers in individual offices and facilities Recruitment · Formal employer · Have authority to hire · Have independent merit-based recruitment mechanism (local civil service commission) Career management · Promotion is available · Transfers within local government are possible · Horizontal mobility within local civil service cadre Performance management · Direct and supervise activities and tasks · Conduct evaluations · Offer financial rewards · Discipline and fire underperforming staff Pay policy · Set overall wage rates · Set local incentives/ salary top-ups Source: Author's estimates. Note: yes; partial; no. Data are for most recent available year, ranging from 2000 to 2003. Ratings refer to the subnational level prioritized by the country's decentralization policy and to de facto practices as well as de jure authority. brought to light by the decentralization process. devolving substantial government functions, and As East Asia's two most rapid decentralizers, the attendant financial and human resources, from the Philippines and Indonesia provide a useful starting national to the local level.3 Over 70,000 staff were point for addressing administrative decentraliza- transferred to local governments. tion in the region. In each tier of government, the local chief executive--governor, mayor, or barangay captain-- Central Rules and Local Responses in the can hire, fire, and promote staff, subject to guide- Philippines. The 1991 Local Government Code lines from the central Civil Service Commission (LGC) provided the major impetus for"genuine and (CSC).4 Civil servants are bound to a code of con- meaningful local autonomy" in the Philippines, duct and required to declare their net worth and 136 East Asia Decentralizes financial and business interests. Government The 1989 Compensation and Position Classifi- employees are also supposed to disclose a list of cation Act set salaries in all but first-class local gov- close relatives in the public service, but this does ernments lower than those at the center.8 However, not in practice seem to limit family connections in because the salaries of national staff transferred to the civil service. Indeed, despite the constitutional local governments remained the same, this created requirement that "appointments in the civil service a wage gap between local and devolved staff. In shall be made according to merit and fitness," fact, some civil servants who transferred to lower- several mechanisms cause merit and fitness to lose income municipalities earned more than the out to patronage and nepotism at the local level.5 mayor. This influx of better-paid staff imposed For example, although candidates on the shortlists a budget crunch on local governments. The 1993 for civil service appointments generally meet Salary Standardization Law, which called for unify- the minimum qualifications set out by the CSC, the ing pay regimes across all levels of government, local chief executive is not bound to choose the exacerbated this problem by raising salaries at top-ranked applicant. lower levels. The law has also affected the incentives Moreover, recruitment for confidential, highly of civil servants by compressing salaries and thus technical, and noncareer staff positions is exempt lowering the increases that employees can expect as from the prescribed appointment process. Perhaps they move up the ranks. The compact organiza- because these workers are relatively easy to hire, tional structure of local governments also restricts noncareer employees composed over one-third the career mobility of civil servants. (38.6 percent) of total local government staff in Local governments have responded in different 2001--significantly more than in national govern- ways to these central directives. Some cash-strapped ment agencies (4.67 percent) (World Bank 2003e). governments are forced to simply ignore the man- Furthermore, local governments can retain emer- dated salary scales and pay their employees less. gency and casual staff for up to six months without Some local chief executives have elected not to fill CSC approval. Local officials often use this provision mandatory positions to leave room for other staff or to circumvent central controls and delays and salary supplements. Others have attempted to lay avoid contributing to the national employee benefit off workers, though this is rare owing to the politi- scheme.Repeated extensions of temporary contracts cal sensitivity of retrenchments. Local governments lead to de facto permanent employment, and not all with greater access to own-source revenues supple- local governments advertise nonpermanent posi- ment staff incomes even for positions covered by tions. This lack of transparency undermines merit- the national government, such as police. Local gov- based hiring, as temporary positions are often ernments have also responded to rising personnel handed out as patronage for loyal supporters. costs by charging some expenditures, such as pay- Central regulations on local establishment size ments for contract workers, to other budget lines. and expenditure allocations constrain the chief This crowds out nonpersonnel expenditures while executive's flexibility in appointing staff. A uniform masking the true extent of staff costs. As the central system of job classification imposes rigidities on government does not have the capacity to enforce small local civil services. The central government its regulations, local governments have much more limits local personal services expenditures to 45 to flexibility than the legal framework implies. 55 percent of the previous year's income, depend- ing on the income class of the local government.6 Balancing National Unity and Local Autonomy in The center also determines salary scales and benefit Indonesia. Indonesia's Big Bang decentralization-- entitlements. Base wages tend to be low, but are ambitious and swift as it was--risked disrupting supplemented by a complex system of both mone- public services, inciting local unrest, and leaving tary and in-kind allowances and benefits. From civil servants unpaid. In actuality, the transition 1992 to 2001, aggregate local spending on personal went quite smoothly. Delivery of services contin- services averaged 56.8 percent of the previous year's ued for the most part unscathed, and expected total regular income, exceeding the nationally man- upheavals among reassigned civil servants failed to dated cap on the wage bill. In municipalities, the materialize. Over 2.1 million staff--nearly 60 per- average was 64.4 percent (World Bank 2003f).7 cent of central employees--were transferred to Managing Human Resources in a Decentralized Context 137 provincial and (primarily) district civil services with central government introduced this transitional relatively little difficulty, although few people had to mechanism with good intentions, it created incen- physically move because the previous system had tives for overstaffing at the local level, as regional been highly deconcentrated. The central govern- governments began to assume that excess person- ment greatly aided the process by continuing to pay nel costs would continue to be covered by supple- devolved civil servants from national coffers for five mentary funding. months. The structure of the DAU itself discourages pru- In the wake of the Big Bang, several legal dis- dent local management. Currently, the DAU is crepancies came to light. Considerable confusion divided into three parts: a "lump sum" distributed ensued in applying various laws and regulations, in equal amounts to every region, a "formula" leading to overlapping authority and eroding effi- amount determined by the extent to which a ciency. According to the original administrative region's financial need outstrips its fiscal capacity, framework (Law 22 of 1999 on Regional Adminis- and a "balancing factor" based largely on the tration), the head of region had the right to hire, region's wage bill as a share of total subnational fire, promote, transfer, and discipline staff. The wage expenditures in the previous year. The balanc- Civil Service Law, however, maintained that ing factor deters local managers from cutting costs authority for civil service appointments rested first because lowering the wage bill lowers next year's and foremost with the central government, which DAU allocation. This disincentive will become could delegate that authority to heads of region. stronger with the implementation of the revised Government Regulation 97 of 2000 allowed the Law on Fiscal Balance (Law 33 of 2004). The equal- head of the regional administration to determine izing lump sum component will be eliminated in the size of the regional establishment. However, 2006, and the "hold harmless" clause that keeps central regulations requiring nationwide advertis- regional DAU allocations from decreasing over ing for some posts--although perhaps intended to time applies only until 2008. Moreover, in the inter- spark interregional mobility--limited the flexibil- est of national unity, the central government has ity of regional managers. resolved to fully fund regional salaries through the Recent changes to the framework laws have DAU, which will create strong incentives for subna- moderated the initial thrust toward devolution and tional managers to increase staffing. introduced even more uncertainty about the roles Until recently, a zero-growth policy prevented of different levels of government in managing the formal recruitment to the Indonesian civil service. civil service. The revised Law on Regional Adminis- New staff appointments in November 2004 were tration (Law 32 of 2004) has rolled back local the first of their kind since decentralization was authority to manage human resources, stipulating implemented. At the same time, Government Reg- that the central government will supervise regional ulation 8 of 2003 on Local Government Structure civil servants within a national framework, and that and Organization defines the maximum number of regional heads can hire, fire, and relocate only staff in the regions, though it is not clear how lower-level staff at the same level of government.9 strictly this regulation is enforced. Both the central Regional managers no longer have explicit author- and regional governments have taken to hiring ity to manage their civil services. contract workers to circumvent these formal The central government also determines pay restrictions. At the local level, this strategy has levels and increases regardless of their affordability allowed managers to hire and fire temporary staff at the local level, limiting the autonomy of regional and consultants as desired, without interference managers.10 Local governments must further pay from the center. Contract workers, who now the pensions of civil servants transferred from the account for about 10 percent of government center, which can create an unsustainable burden, employees, are also attractive because their employ- particularly as the cost of pensions rises along with ment is not subject to qualification standards or salaries. In 2001, the dana alokasi umum (DAU)-- pension obligations. or general allocation grant--was supplemented by In theory, the government must base hiring and a "contingency fund" to cover the 14­30 percent promotion decisions on merit, but anecdotal pay increase mandated by the center. Though the evidence reveals that this needs strengthening in 138 East Asia Decentralizes practice, especially in promoting staff. The head of becomes deeply entrenched will be important, as region has the right to discipline and dismiss lower- interregional mobility of civil servants can help level staff who are underperforming, subject to due ensure national unity and acceptable levels of serv- process and approval from the provincial governor. ice delivery in more remote regions. Possible mea- Dismissals are rare in practice, however, and staff sures to address this issue include joint training reassignment seems to be the preferred penalty. across regions and a secondment system, in which The civil service agency, BKN, has begun to fixed-term placements in local areas improve the define job classifications and qualifications for the career prospects of central civil servants. civil service as a whole, including regional govern- ments. The Indonesian government has also been Ironing Out the Wrinkles. Both the Philippines developing minimum standards for service deliv- and Indonesia considered decentralization neces- ery, including broad directives on supervising min- sary to keep the nation together and regain the trust imum wage levels as well as highly specific technical of disaffected local governments. There was a guidelines. Heads of region have generally been strong sense that decentralization plans could not enthusiastic about this initiative, both because they await the relatively long process of institutional prefer to be held accountable against established development and capacity building, and thus that criteria rather than arbitrary standards set by local human and institutional capabilities should be parliaments, and because they expect to receive expanded during the decentralization process. more funding in exchange for taking on greater This approach succeeded in focusing the policy responsibilities (World Bank 2003d). In principle, agenda on the actual transfer of functions and minimum service standards can improve civil serv- management authority to the local level, but left in ice management by offering some basis for unity its wake some unfinished business that has among public servants, ensuring a minimum level slowed the momentum behind decentralization. of performance across the country, and enhancing In Indonesia, concern over the loose ends has local accountability through improved understand- prompted the central government to roll back ing of what is expected of local civil servants. In devolved authority. Indonesia, however, most line ministries are issuing Capacity. Decentralization policies in the standards without giving enough thought to their Philippines and Indonesia have focused on the low- feasibility and affordability at the local level. To be est tier of local government, but implementation of effective, minimum standards must be designed devolved functions is running up against the carefully, so they are specific enough to provide limited capacity of staff in these administrations, clear direction on what is expected of the local civil particularly in planning, budgeting, and financial service but not so detailed as to hamstring regional management. In response, the Indonesian govern- leaders. ment has adopted a National Framework for The design of decentralization in Indonesia con- Capacity Building, but will now have to work to stricts the interregional mobility of civil servants, finance and implement this plan. It is also unclear and therefore their career paths. No streamlined whether the top-down application of this frame- system exists for relocating subnational civil ser- work will adequately address the capacity-building vants to other areas. Discriminatory hiring prac- needs of local governments. In the Philippines, tices at the regional level seem to further limit the continuing mismatch between required and mobility. Despite regulations designed to prevent available skills at the local level points to a need for this, some heads of region are reportedly engaging better coordination in training local civil servants. in preferential treatment of "sons of the soil," or Most activities are now provided in isolation, either discriminating based on ethnicity, religion, or other by individual central agencies or by local govern- special interests. Many regions have resisted taking ments themselves. on staff transferred from the center, owing largely Incentives. The decentralization process has also to ethnic tensions (central civil servants are prima- affected incentive structures at the local level. Both rily Javanese) and the association of staff in Jakarta the Philippines and Indonesia have encountered with the authoritarian rule of Soeharto (World difficulties in designing intergovernmental financ- Bank n.d.). Curbing this tendency before it ing mechanisms. The Internal Revenue Allotment Managing Human Resources in a Decentralized Context 139 in the Philippines has not followed the assignment managers to do what needs to be done, but the of functions as closely as it should have, flooding resulting lack of clarity and transparency poses a some localities with excess resources while saddling considerable risk in reduced accountability and vul- others with unfunded mandates. The former situa- nerability to corruption. In the Philippines, weak tion skews incentives toward overspending and controls on staff appointments open the door to overstaffing, while the latter discourages long-term patronage and nepotism in local government. In planning and spending on maintenance. In Indonesia, though the speed of decentralization Indonesia, the central government's historical use probably prevented vested interests from influenc- of the DAU to cover local salary shortfalls and, now, ing decentralization policies, there is evidence that to fully fund subnational wages creates a perverse local positions are bought and sold in return for the incentive for overstaffing regional governments. In promised rewards of graft in the public service both countries, performance incentives for civil (World Bank 2003c). The accountability of local servants are dampened by compressed salary struc- civil servants is further threatened by a lack of tures, ineffective monitoring of performance, and clarity--in Indonesia because of imprecise roles, weak links between performance and pay. These and in the Philippines owing to the "double subor- effects have been compounded by a lack of mobility dination" of local staff to the local executive and the within the civil service--vertically in the case of the relevant central agency. Internal accountability Philippines, and horizontally in Indonesia. mechanisms remain weak in both countries, though Autonomy. In its initial thrust, decentralization external accountability institutions--such as in the Philippines and Indonesia significantly ombudsman institutions, complaint bureaus, and boosted the autonomy of local managers. Indone- citizen monitoring--are growing in importance. sia, however, has recently repealed local staffing Overall, the experience of East Asia's rapid authority. In both countries the central government decentralizers shows that, although establishing the has retained control in defining pay levels for local right rules is important to administrative decen- civil servants. Subnational governments are subject tralization, good design on paper is not enough. to national wage scales that impose significant Countries must pay attention to the realities of fiscal burdens, especially in poorer localities. implementation, particularly to creating incentives Centrally imposed salary increases have further for local governments to work within the rules and strained local budgets. In Indonesia, the DAU to maintaining the momentum to make needed covered these, but in the Philippines, the Salary policy adjustments. Standardization Law is not yet fully implemented because many local governments find it simply Decentralization in Transition: China and Vietnam unaffordable. Another unfunded mandate that impinges on the autonomy of local civil service Administrative decentralization in China and managers in the Philippines is the 1993 Magna Vietnam was born of the transition to a market Carta of Public Health Workers Act. The central economy. During the economic opening, central government extended this generous package of governments relinquished control over some of wage and benefit guarantees to health employees their activities and offered greater administrative as an inducement to accept devolution. This has and, to a lesser extent, political independence to widened an already noticeable gap in the remuner- subnational entities in return for greater central ation of local versus devolved staff, and has authority over revenue assignments. This system imposed significant costs on local governments has confused the roles of subnational civil servants while reducing the ability of local managers to allo- and limited the autonomy of local leaders in man- cate resources according to need. In both Indonesia aging their activities. and the Philippines, limits on local authority have On the other hand, in "crossing the river by feel- led managers to bypass established regulations by ing the rocks,"11 decentralization in China and hiring staff on a temporary basis and topping up Vietnam has encouraged governments to experi- salaries with nontransparent allowances. ment with local autonomy and address problems Accountability. The tendency to circumvent the before moving on to the next stage of decentraliza- rules may reflect an adaptive response by local tion. And local governments in both countries have 140 East Asia Decentralizes formulated innovative responses along the way. as substitute teachers, are exempt from the national The decentralization experiences of these two guidelines. countries--different in size yet emerging from sim- The number of posts authorized for local juris- ilar bureaucratic traditions--provide insights into dictions is determined by local branches of the the impacts of administrative decentralization on State Commission for Post and Establishment, a civil service management and local service delivery joint government-party organization, at the next- in transition countries. highest level of government. The commission uses a weighted formula to calculate each jurisdiction's The Ebb and Flow of China's Decentralization. In overall staff complement, including a quota for the a country as large and varied as China, some form number of support staff.13 Staffing numbers are of administrative decentralization is a necessity, not necessarily efficient, however, and inefficiencies yet the statutory basis for decentralized human can be preserved over time because responsibilities resource management is fairly limited. The 1993 for budget preparation and staffing policy are Provisional Regulations on Civil Servants address separate--a holdover from the days of central the nationally unified core civil service, which planning. includes only white-collar workers such as man- Civil servants are paid according to a national agers and professional staff. Teachers, doctors, sup- salary scale, which is benchmarked to the wages of port staff, research institute employees, and mem- staff with similar responsibilities in state-owned bers of the military are not considered part of the enterprises. This restriction pegs salaries closer to core service, but of separate Public Service Units those of staff in larger, urban localities. Bonuses (see below). Political appointees are considered and benefits are determined locally and often com- civil servants, however, as no distinction is drawn pose a hefty proportion of overall remuneration. between political and bureaucratic personnel. Well-off localities use these perks liberally to Though the Chinese Communist Party (CCP) is supplement the pay of civil service cadres. Poorer formally separate from the government, it does areas have difficulty meeting nationally mandated influence staffing and other aspects of public wage increases, and some have been forced to management. Central and subnational civil ser- ignore these instructions or supplement personnel vants alike are subject to national guidelines; budgets with funds originally allotted to capital although local jurisdictions prepare their own reg- expenditures. The central pay scale does not, there- ulations, these usually conform to the national fore, ensure nationally consistent pay. The fact that rules (Wong 2003). local governments must cover civil service pen- The People's Congress usually makes the deci- sions, unemployment benefits, and other safety net sion to hire or fire a local civil servant at the corre- expenditures intensifies the fiscal pressures caused sponding level; however, the CCP controls the by central control over wage levels. When the appointments of senior staff through the nomen- national government raises pay, these salary-based klatura system of bureaucratic patronage common expenditures rise as well, tightening local budgets in many communist countries. Since 1984, high- beyond the control of local managers. level officials in each administrative tier have been Relatively high salary scales further burden less appointed by the party committee at the next- prosperous local governments, where the civil serv- highest level.12 Recruitment is supposed to be ice may be one of few sources of wage employment. determined by open, competitive examinations, Indeed, the tendency for local administrations and the selection and promotion of employees to serve as the employer of last resort is exacerbated based on merit and performance. It is important to by higher-level insistence that they absorb streams note, however, that these criteria are generally of recent graduates (Wong 2003). These forces defined to include "political integrity," or commit- together inflate staffing levels, especially at the ment to party policies. Local staff are accountable county level, which accounts for over half of the to the central government rather than to local excess staff (World Bank 2002a). Despite several administrations. Civil servants must follow codes attempts to downsize the civil service and streamline of conduct set out in both the Provisional Regula- the structure of government, redundant agencies tions and party writings. Temporary workers, such remain and have tended to staff up again over time. Managing Human Resources in a Decentralized Context 141 Wage levels appear to be high enough to allow counties, PSU staff complements expanded by 21, governments to recruit and retain qualified 21, and 75 percent, respectively, between 1991 and employees. The number of higher-level staff with 1999 (World Bank 2002a). university degrees has risen from about 20 percent PSUs are subject to the authority of and in 1981 to more than 80 percent in 1994. Though accountable to the central government. The bound- pay increases and bonuses are designed to reflect aries between these two forms of public employ- performance evaluations, this does not seem to ment often blur. Throughout various attempts to motivate staff to perform well. The majority receive downsize the civil service in the late 1980s and positive appraisals, and managers generally hand 1990s, PSUs often absorbed retrenched employees, out bonus payments across the board rather than limiting cost savings. There is an expectation that to reward exceptional work. Overall pay levels are staff can move freely between PSUs and core gov- highly compressed. The ratio of highest to lowest ernment positions. In essence, PSUs are a reservoir salary is 5.6 to 1 for base wages, but flattens to of extrabudgetary human resources upon which something on the order of 3 to 1 when subsidies the government can draw (World Bank 2002a).15 and bonuses are included (World Bank 2002a).14 Moreover, the functions of core agencies and PSUs The relatively small increments between pay cate- often overlap, creating inefficiencies in administer- gories substantially reduce the payoff gained by an ing and delivering public services. employee who ascends to the next rung of the bureaucratic ladder. There are no explicit restric- Traditional and Innovative Approaches to tions on the transfer of local civil servants from one Decentralization in Vietnam. Along with the locality to another, or up to the central level, but the transition to market-oriented economic principles, structure of the Chinese labor market--in which a major outgrowth of the doi moi reforms begun in the hukou system of household registration assigns 1986 has been a shift in the balance of state power, workers to a designated area--limits mobility. first from the Communist Party of Vietnam (CPV) Some movements occur among higher-ranked to the government, and then from central to subna- staff, but rarely in the lower echelons. In the late tional levels. The impetus for administrative decen- 1990s, the central government initiated a program tralization is growing with the Master Program on to relocate mid-career officials to local jurisdictions Public Administration Reform, which strongly for six months to one year as part of a career path to advocates realigning the management of human senior public service. and financial resources to address the fact that The core civil service, as defined by the Provi- "administrative machinery at local levels is not sional Regulations, does not encompass a substan- really responsive to people" (Government of tial proportion of government-financed workers Vietnam, 2001, 2). employed by Public Service Units (PSU), such as Most civil servants in Vietnam are hired and schools, hospitals, and research institutes. PSUs fig- fired by the People's Committee at the next-highest ure prominently in the public employment land- level of government. In the provinces, the Provin- scape, accounting for 96 percent of the civilian cial People's Committee makes these decisions, non-state-enterprise workforce at the central level, though high-level appointments require the ap- and 73 percent at the subnational level. Human proval of the prime minister. Openings must be resource management is slightly more flexible in advertised within the locality, usually by way of a PSUs than in core government entities. Local PSUs public posting outside the Office of the People's can be run at the central or local level or managed Committee. While job classifications have been jointly. As in the core service, the next-highest defined centrally, they are not accompanied by pre- branch of the State Commission for Post and Estab- cise job descriptions, and thus offer local govern- lishment determines the size of PSU staff. These ments significant discretion in determining their restrictions do not seem to be heeded in practice, staffing profile. however, as local PSU employment averaged 110 All civil service appointments and promotions percent of the authorized staff size in 1999. And are to be based on merit, as determined by exami- PSU employment is growing, particularly at the nations conducted by provincial-level political subnational level. In provinces, prefectures, and schools. There are indications that, in practice, 142 East Asia Decentralizes seniority is accorded more weight than perfor- Salaries and allowances are specified centrally mance in promotion decisions. Selection to higher- and apply uniformly to all levels of government. level positions also requires ranking membership in Wage levels have traditionally been considered low the CPV. No performance management system compared with alternative sources of employment, links high-quality work to high return on a civil though new analysis casts doubt on the validity of servant's career path. Two 1998 regulations--the these comparisons. Bales and Rama argue that, Ordinance on Complaints and Denunciations and while professional and technical employees may the Ordinance on Procedures for Resolving Admin- be underpaid compared with the private sector, the istrative Disputes--offer mechanisms for disciplin- alternative employer for most civil servants-- ing staff, but managers do not exercise either of particularly those based in rural areas--is most these very often. This is due in the former case to a likely the informal sector, where salaries tend to be lack of specificity in the provisions, and in the latter lower (Bales and Rama 2002). In poorer regions, in case to the limited effectiveness of Administrative fact, the comparatively high wages fixed by the Courts. No formal procedures are in place to allow center--along with recent salary increases--are dismissed civil servants to address grievances or tightening budgets. This pressure will only build appeal personnel actions. with the planned pay reform program, which Though officially no geographical constraints promises to raise salaries by 30 percent, on average. hinder the recruitment and transfer of staff, incen- Because staff pay represents a considerable share of tives for interregional mobility are generally inade- spending in service sectors, wage costs can freeze quate, as local residents fill most local positions. out other important spending. Though the crowd- Furthermore, remote areas tend to have difficulty ing out of operations and maintenance expendi- attracting high-quality employees. There is also tures is not a major problem overall in Vietnam, it room to boost recruitment of regional minorities, is an important issue in poorer areas. particularly above the commune level. In theory, local civil servants are not barred from moving into Turning Transition into Forward Momentum. the central civil service,but in practice,only People's Decentralization in China and Vietnam has been Committee chairs move up to service in Hanoi. incremental, with the central government reacting Instead, central civil servants seem to be spend- to the effects of each policy before taking the next ing more time away from the capital. In 2003, the step.While this approach allows experimentation to government launched a new program of staff rota- address unexpected results before they become tions aimed at deepening the hands-on experience entrenched, the lack of a predefined plan can create of top central bureaucrats. The intent is that time inconsistencies in decentralized governance. For spent at the grassroots level will serve as a critical example, sufficient fiscal resources have not accom- step on the path to high-level government posts panied administrative mandates--indeed,in China, (Cohen 2003). Whether this promising initiative the recent trend has been toward recentralization of will bear fruit, however, remains to be seen. There finances. In Vietnam, the PAR Master Program, is concern that the rotation scheme has been used which sees the capacity and accountability of civil not as a tool for instilling an improved public servants as critical to improving the effectiveness service ethos but instead to banish political and efficiency of public service, is an important step opponents. toward redressing imbalances in fiscal and human The central Ministry of Home Affairs manages resources. the personnel establishment through staff quotas at Capacity. The PAR Master Program calls for each level of government. Until 2004, these staff developing a capacity-building plan to provide bet- quotas also determined the budget allocations for ter training opportunities at local as well as central personnel expenditures at the local level, but the levels, refresh the curriculum and methods, and one-to-one relationship between a larger staff size align content more closely with needed on-the-job and a larger budget allocation created incentives for skills. Enhancing the reach of educational opportu- overstaffing and penalized leaner, more efficient nities will be important in addressing the low administrations. Budgets now reflect a province's capacity of many local governments in Vietnam, population rather than its staffing levels. particularly in remote areas. China has also made Managing Human Resources in a Decentralized Context 143 training civil servants a priority: local governments Vietnamese government has had significant success must upgrade the skills of civil servants in line with in improving incentives for prudent management the needs of the market economy. In both coun- of administrative expenditures through an innova- tries, the central government provides resources tive pilot program for block-grant budgeting in Ho for locally organized training programs. In China, Chi Minh City (see box 5.3). local governments are taking advantage of these Autonomy. Controlling administrative expendi- resources to send their employees to Chinese uni- tures has indeed proved difficult for local managers versities as well as training programs outside the in these countries because of their restricted auton- country. However, local administrations in both omy in setting wage and staffing levels and, in China and Vietnam continue to lack expertise in China, associated safety net expenditures. The high accounting and financial management. In Vietnam, percentage of local budgets preempted by person- a recent audit found an absence of record keeping nel costs, in turn, limits local discretion regarding in the communes, which keeps local managers from other expenditures. In the meantime, local leaders effectively overseeing the civil service and prevents have focused on a few areas of personnel manage- the center from tracking the use of funds. ment in which they do have autonomy. In China, Incentives. Improving the ability of local govern- local governments determine bonus pay, and have ments to manage financial resources addresses only used it to supplement civil service wages. In both part of the problem when the incentives of civil serv- countries, local governments have the right to sup- ice managers are skewed away from prudent super- plement their budgets by collecting user fees for vision of local budgets. Until recently, Vietnam's certain services. While this has eased the strain on system for allocating administrative budgets-- some local budgets, many of these fees fall dispro- calculated per staff member per year--encouraged portionately on the poor. Moreover, because these local managers to inflate staffing levels to obtain resources are external to the budget, governments more resources from the center. In China, local offi- often spend them in nontransparent ways that cials have little motivation to reduce personnel costs, do not reflect budget priorities and undermine as they are budgeted separately from other expendi- accountability. tures. Indeed, managers who save on wages lose part Accountability. The accountability of local civil of their budget allocation. In poor regions, adding servants in China and Vietnam is blurred by a lack staff is often seen as the only way to increase the local of clarity on the roles and responsibilities of each budget allocation. Coupled with the government's level of government, though Vietnam's new Bud- position as employer of last resort, this system has get Law has helped to address this problem. In encouraged overstaffing in local governments. Vietnam, and at the township level in China, the The ability of decentralization to improve serv- lack of clarity is intensified by dual subordination, ice delivery also depends on the incentives of civil in which local staff are beholden to both the direc- servants themselves. In both China and Vietnam, tives of the central government and the demands of despite official rules defining merit as the basis for local elected assemblies. Audits of local expendi- promotion, the career paths of local staff appear to tures in both countries have found that accounta- be more limited than those of their central govern- bility mechanisms are weak. In China, the internal ment counterparts. Seniority and party loyalty audit function is strong, focusing mainly on ensur- receive as much or more consideration than per- ing compliance and preventing fraud. External formance, and salary compression limits the bene- ex post audits are less effective, as local audit fits gained from hard work. China's attempt to bureaus report to the central government and thus enhance incentives for high-quality work through lack needed independence. The effectiveness of performance-based pay has not succeeded because audits in Vietnam is constrained by a lack of capac- managers have applied bonuses across the board. ity and low pay. In both countries, efforts are under Staff rotation schemes in both countries focus on way to improve accounting and financial manage- developing the careers of central civil servants ment at central as well as local levels. These mea- through time spent in regions, rather than on rotat- sures will help the central government improve and ing local staff to other areas to facilitate sharing of monitor local service delivery. The presence of the experiences and information (see box 7.1). The Communist Party in China and Vietnam holds 144 East Asia Decentralizes BOX 7.1 Personnel Exchange in India and Japan: Whose Capacity Is Being Strengthened? In many civil services around the world, systems collaboration between the center and subna- enable staff to transfer across levels of govern- tional governments. ment. The motivations behind these personnel In Japan, on the other hand, the impetus for transfers are varied. In some cases, the intent is personnel exchange arose from the need to to groom central civil servants for high-level boost technical capacity in newly established postings by using hands-on work at the local local governments. Many local postings have level to ground and broaden their policy skills. In since become "hereditary," in the sense that other cases, central government staff are trans- staff members returning to central government ferred to subnational jurisdictions to fill gaps in are typically replaced by other central staff from local capacity to implement national programs. the same ministry. Though it could be argued Where civil servants are transferred across that this practice has prevented qualified local subnational governments, the aim may be to staff from gaining access to senior positions keep them at a safe distance from local vested in subnational governments, the fact that interests. autonomous local administrations continue to A comparison of civil servant rotation pay for relocated central staff implies that the schemes in India and Japan demonstrates how transfer of information and knowledge from the design of these human resource manage- the center still plays a useful role. Moreover, the ment tools can shape their impact in a decentral- Japanese system of personnel exchanges is ized setting. In India, the principal focus is on much broader than that in India. Staff not only preparing members of the elite Indian Adminis- transfer from the center to prefectural govern- trative Service (IAS) for senior positions in central ments, but from prefectural governments to and state governments. Those who make the cut municipalities, from local governments to cen- in the highly competitive recruitment process tral postings, and across prefectures as well. and rigorous training program are dispatched to While in both countries placing centrally state governments, where they spend several recruited officials in local administrations can years working in field and secretariat assign- strengthen the capacity of local staff as well as ments in different sectors. After reaching senior transferred officials, the design of the Japanese positions in the field, IAS officers typically rotate system is more conducive to developing the between central and state governments, as skills of local staff. repeated exposure to field conditions is key to keeping abreast of changing on-the-ground Sources: Schiavo-Campo et al. 1997; Inoki 2001; Iqbal realities. The IAS system also fosters active 2001. important consequences for the accountability of media and citizens are paying increasing attention civil servants in a decentralized setting. The party to corruption issues. In Vietnam, after peasant exerts considerable influence over staff appoint- protests in Thai Binh and Dong Nai provinces ments and promotions, particularly at senior levels. brought to light the extent of corruption in the local As such, civil servants most likely feel more civil service, the central government introduced a accountable to the party than to the local popula- Grassroots Democracy Decree to enhance local con- tion. While the party can be seen as a source of dis- sultation with residents on decisions that affect cipline in an otherwise weak system, its traditional them. While this could in principle provide a strong control over the flow of information can reduce check on abuse of power and strengthen accounta- transparency and undercut efforts to hold civil ser- bility by increasing the demand for good gover- vants accountable (Wong 2003). nance, implementation has been weak because of Government transparency is beginning to limited management responsibility at the commune improve in China andVietnam,both within the civil level,insufficient training of civil servants in sharing service and with respect to the public. The state-run information with local citizens, and inadequate Managing Human Resources in a Decentralized Context 145 mechanisms for tackling corruption in social assis- the institutions and developing the legal basis tance funds. for devolving responsibilities to subnational administrations. So far, few civil servants have been transferred The Cautious Decentralizers: Cambodia to local administrations. Only about 4,000 staff and Thailand (less than 1 percent of the 1.2 million national The implementation of decentralization has been civil servants) in five departments--principally the relatively limited in Thailand and Cambodia. In Department of Public Works and the Ministry response to growing demand for improved service of Interior's Accelerated Rural Development delivery at the local level, both countries have rec- Department--have moved to local service. The ognized the importance of devolving some man- transfer of staff has thus not followed the transfer agement responsibilities--and the rights that of functions, as stipulated in the decentralization accompany them--to local governments. Yet con- legislation.16 The strategy of the Office of the Civil cerns about the limited capacity of subnational Service Commission for transferring remaining governments have engendered a cautious approach staff prioritizes voluntary transfers, though manda- that has concentrated on developing the legislative tory transfers and compensated retrenchment are framework and building local capacity to manage still options. The commission has set up a Public and monitor public services. In Thailand, a weak- Sector Personnel Development and Deployment ening political imperative to decentralize has also Center as a hub for training and deploying central stalled implementation of outlined policies. staff to positions in local government. Provincial In low-capacity countries, a gradual transfer of personnel transfer centers are evaluating the responsibility can be a pragmatic way to ensure staffing implications of devolved authority for local continuity in service provision and limit the poten- governments. tial for political capture of inexperienced local civil Still, the low number of relocations thus far servants. However, gradualism also has its disad- points to a lack of motivation on the part of both vantages. Piecemeal approaches may lack an overall central agencies and public employees. Ministries in strategic focus, leading to a mismatch between Bangkok are understandably unenthusiastic about responsibilities and resources, and between author- letting their staff go.Poorer local governments,espe- ity and accountability. Moreover, caution can serve cially the small tambons, may be reluctant to accept as an excuse for inaction on the part of central offi- more staff because of the burden they may pose to cials who are reluctant to relinquish their authority. tight budgets.Employees hesitate to move owing to a Thailand and Cambodia have faced some under- lack of clarity about comparable pay, benefits standable reasons for delay. Yet in taking such a (including pension fund provisions), recruitment long time to establish the building blocks of decen- and dismissal procedures, career mobility, and local tralization, these countries have deferred potential quality of life. Some staff worry that relocation to improvements in service delivery. In the meantime, smaller local communities will leave them--and the success of several small-scale experiments with their jobs--at the mercy of local politics (Wegelin local autonomy indicates that, with careful design, 2002). Civil servants who transfer to the regions are decentralization can work in these countries. The entitled to the same pay as in the center, but the sys- following discussion of guarded decentralization in tem for classifying local positions does not offer the Thailand and Cambodia explores how countries same range of job levels (and accompanying pay may resolve the tension between exercising caution grades).17 The central government has offered a and achieving tangible results. financial incentive to public servants who volunteer to serve in local governments for a three-year term, Building the Foundations for Decentralizing but this does not address the longer-term issue of Personnel Management in Thailand. Thailand's permanent personnel transfers.18 1997 Constitution represents a significant step in The central government exercises general bringing government closer to the people in a authority over personnel decisions at each level of traditionally centralized country. Building on this subnational government through the Civil Service solid foundation, efforts have shifted to creating Commission. Local administrations may hire 146 East Asia Decentralizes low-level staff and contract workers but are not Cambodia has established the legal framework authorized to recruit higher-skilled or career for devolution, but its impact on staffing has been employees. Vacancies are not formally advertised. limited. While recognizing that communes should Recruitment is based on competitive examinations have their own financial resources, the Law on administered by the commission, although it can Administrative Management is much less ambitious waive the exam requirement in certain circum- about decentralizing personnel and human stances. All civil servants are subject to a code of resource management. The law requires only two professional ethics, which requires those in higher- technical staff positions at the commune level: a level positions to publicly declare their assets. commune clerk appointed and paid by the Ministry Disciplinary procedures are rigorous and provide of Interior to assist the commune council, and an for appeal. accountant from the provincial treasury. These The center rigidly prescribes the number and individuals wield central influence over the other- grade level of authorized staff positions at each wise relatively autonomous commune councils, level of local government. Fifth-class tambons may although the commune chief can request the employ no more than 3 staff members, and first- appointment of a new clerk based on the council's class tambons no more than 21. Limiting staff so decision. If needed, councils may directly employ strictly may undercut remote rural areas that need other staff, whom the law deems outside the "State the most intensive work on service delivery and framework."These employees are retained on a tem- public outreach. Yet keeping staff lean can help porary basis for the duration of the council's man- localities manage the fiscal burden of civil service date, but the next council can extend their terms. salaries. Local governments must adhere to a cen- At the commune level, general administration tral pay scale and a 40 percent cap on personnel expenditures are restricted to 30 percent or less of expenditures as a proportion of local spending. the allocation from the center. This limit reduces This is a Catch-22 for poorer tambons, where a the temptation for nepotism and overstaffing while sheer lack of staff constrains the capacity to per- giving communes considerable discretion in how form local functions, while paying more people funds are spent within the threshold. There is con- would leave little money for anything else (Wegelin cern that the centrally determined allowances paid 2002). To address local personnel issues and redress to commune councilors are high relative to civil disincentives to staff transfers, the Local Civil servant salaries, especially as the former are only Service Administration Working Group has pro- part-time functionaries. This can foster resentment posed creating a Local Government Civil Service among local staff and further restrict administra- Commission--as distinct from the provincial and tive expenditures. municipal commissions that already exist. The The legal framework for deconcentration is less Ministry of Interior is also looking into a personnel well defined, but the number of staff affected is management system to track the pay, benefits, and high. As of 2003, some 78 percent of staff were career streams of subnational civil servants. Finally, based in provinces, including health and education the Thai government's recent acceleration of employees. The relevant central ministry deter- broader civil service reforms may reinvigorate the mines staffing needs and positions at the provincial decentralization process. level, but provincial administrations have discre- tion in whom they hire (World Bank 2002b). Top-Down and Bottom-Up Approaches to Though recruitment is supposed to be based on Decentralization in Cambodia. According to competitive examinations administered by the its National Poverty Reduction Strategy for State Secretariat for Civil Service, anecdotal evi- 2003­2005, Cambodia is pursuing a two-pronged dence indicates that provinces do not strictly follow approach to the devolution of power, tackling both this requirement (World Bank 2003a). Similarly, political decentralization to the communes and promotions within each grading category are to be functional deconcentration to provincial-level out- based on merit, but seniority appears to play a posts of the central government. The government stronger role in practice. Performance evaluations has recently initiated a process to develop an inte- should be conducted annually, but this is not grated policy framework. common. Disciplinary procedures exist but are Managing Human Resources in a Decentralized Context 147 rarely used. Incentives are therefore not structured Ministry of Education, Youth, and Sports has tried to encourage civil servants to perform well and to improve the efficiency of education services improve over time. through financial incentives, including pay supple- Incentives are further dampened by extremely ments for work in remote areas, double shifts or low and highly compressed pay. Wages are so low multigrade teaching, and performance incentives to that many civil servants are forced to seek addi- heads of provincial and district education offices. tional employment in the private and informal sec- tors, and the resulting absenteeism undermines Balancing Caution and Action. The measured public sector performance. While ministries are approaches to decentralization in Thailand and responsible for monitoring attendance, punish- Cambodia reflect expressed concerns on the part of ment for absenteeism is rare (World Bank 2002b). both governments that local administrations and Low pay also affects how civil servants spend their their staff do not yet have the capacity to take time when they are at work. For example, the pay- responsibility for service delivery. In Cambodia ment of per diems for field visits encourages staff to these concerns have some basis in reality; in Thai- focus on activities that require travel, regardless of land, however, they may reflect a weakened political whether this is the most efficient use of their time drive toward decentralization. In both cases, policy or is indeed necessary. makers have sought to avoid the potential dangers of The central government determines salaries, decentralization by putting capacity building first. allowances, and social benefits, which local govern- Capacity. Low capacity is certainly an issue in ments must pay on a priority basis. This limits the Cambodia, where decades of civil war and the flexibility of provincial budget managers and may atrocities of the Khmer Rouge regime took a crowd out other important activities. At the provin- devastating toll on the country's pool of human cial level, personnel expenditures averaged 48 per- resources. Even in Thailand, a middle-income cent of total expenditures in fiscal year 2002, country with a bustling capital city, most local gov- though this figure ranged from 18.6 percent in ernments lack the technical skills and institutional Mondul Kiri to 57.4 percent in Kandal (World resources to deliver even basic services. Both coun- Bank 2002b). This share is much higher than the tries have created decentralized structures without 33 percent that the central government spends on devolving significant authority or responsibility. its wage bill (World Bank 2003a).19 Throughout the Developing the skills of local staff is seen as Cambodian civil service, a lack of systematic estab- the bridge between these initial steps and more sig- lishment control has allowed the proliferation of nificant decentralization. In Thailand, municipal contract workers and high levels of overtime pay, staff are regularly rotated to enhance the sharing of and these problems are likely even more noticeable knowledge and experience among subnational civil at the local level. servants. Various training institutions offer a wide The combination of low pay, low staff motiva- array of courses, but these have not yet been incor- tion, and low mobility has left line ministries des- porated into a focused strategy for developing local perate to skew incentives back toward improved capacity in a sustainable manner. Training and service delivery, and several ad hoc responses have technical assistance for decentralization are also evolved. The Ministry of Health has allowed health growing rapidly in Cambodia, building on the suc- centers to collect user fees and to apply 49 percent cess of the donor-supported Seila Program. A criti- of receipts to supplement staff salaries.20 This cal challenge for Cambodia is to leverage this posi- approach has shown tangible benefits, as the use of tive experience to design coherent and sustainable health centers has risen by up to 60 percent. Clients capacity-building initiatives that extend to all local are evidently willing to pay extra for better service. staff. Thailand and Cambodia have also turned to As Turner notes, "The salary supplements mean deconcentration as on-the-job training for civil that staff are likely to devote more time to their offi- servants--in Thailand, through Local Education cial duties. When salaries might be only US$15 per Authorities and Area Health Boards, and in month, an extra US$10 makes a considerable differ- Cambodia, through Health Center Management ence" (2002, 357). However, these supplements Committees, which have the authority to supple- may lead to pay inequalities across regions. The ment staff salaries with proceeds from user fees. 148 East Asia Decentralizes Capacity is not just about technical and bureau- service delivery.Though local managers in these two cratic skills, however. The institutional and finan- countries do enjoy discretion in whom they hire, cial capacity of local governments is also crucial. In rural communities do not have access to the highly both Cambodia and Thailand, many local govern- skilled staff available in the center. In Cambodia, a ments are simply too small--in geographical lack of predictable financing further constrains the size, population, staff numbers, and budgets--to autonomy of local leaders. Unsure of when and in execute their functions. Most of Thailand's tambons what quantity resources will arrive, field-based employ only a handful of staff and cannot be con- managers are reluctant to make long-term invest- sidered viable units of government, as many are ments, including in human resources. The govern- unable to support even a primary school let alone ment has taken some steps to alleviate this problem more technical services such as infrastructure. In by piloting mechanisms for streamlined execution Cambodia, a maximum of 11 people sit on the of budgets in priority sectors. One example is the commune councils, supported by two centrally Priority Action Program, which allows local educa- appointed civil servants. tion authorities to supplement the low pay of their Incentives. The small size of local governments staff. Still, in these traditionally hierarchical soci- has influenced the incentives of civil servants to eties, local managerial autonomy remains limited. take local employment, which has, in turn, affected Accountability. The limited autonomy of subna- service delivery. Few local civil service positions are tional governments in Cambodia and Thailand available, and those that are sit at lower grade levels poses important challenges for the accountability of with low salaries and limited career paths. Attract- local staff. In both countries, local governments ing highly skilled professionals is difficult, as local at the lowest tier are officially accountable to governments do not have enough work to sustain popularly elected assemblies. Yet, so far, they are full-time technical experts. As a result, central offi- responsible only for a limited set of tasks, and most cials in Thailand have strongly resisted transfer to decisions are handed down from higher levels. As a local service. The Thai government has attempted result, it is difficult for local citizens to hold civil ser- to address this by offering financial incentives to vants responsible for the quality and timeliness of those who volunteer, but unclear rules and their services. Community monitoring is also lim- benefits have prevented this program from having ited by incomplete dissemination of information any significant effect. In Cambodia, where over on decentralization plans and transferred responsi- three-quarters of staff are based in the field, local bilities. The flow of information on public prefer- staff are reportedly migrating to provincial centers ences is interrupted by the "missing middle" of in search of opportunities to supplement very low decentralized politics, as provincial authorities in public sector wages. Local public servants are in Cambodia and Thailand are not elected and thus effect being paid full time to work for part of the report only to the central government. Despite the day. This is a particular problem in health clinics, time taken by both governments to lay a solid foun- which despite a 24-hour mandate often remain dation for decentralization, many local officials are closed for most of the day. The government may still not fully aware of their own responsibilities-- want to consider explicitly allowing part-time work and thus the outcomes for which they are account- in public health facilities so they can ensure contin- able. Local standards and benchmarks are not yet in uous availability of care.21 place to help the Thai and Cambodian central gov- Autonomy.In both Thailand and Cambodia,lim- ernments evaluate local performance. In both its on personnel expenditures have prevented local countries, vote buying, corruption, and elite cap- governments from hiring enough staff to get the job ture may be significant at local levels, posing an done--and even from filling all the positions to important risk to decentralization plans. which they are entitled. In Cambodia's Kampong Cham province, however, low use of health services Successfully Managing the Civil means that even the small numbers of local staff Service: The Way Forward employed in provincial and district offices are exces- sive compared with the workload (WHO 2003). East Asia's experience with administrative decen- This hinders the efficiency and effectiveness of tralization highlights both the opportunities and Managing Human Resources in a Decentralized Context 149 the challenges associated with devolving authority The experiences of East Asian countries also for managing the civil service to lower levels of gov- highlight several critical dilemmas that govern- ernment. As the case studies show, decentralizing ments must address in designing policies and insti- the management of human resources can improve tutions for administrative decentralization: the responsiveness and resourcefulness of local Centralized Control vs. Decentralized Management governments. Yet, without careful design, devolu- tion can also bring fiscal imbalances, negative Implicit in the notion of decentralization is the incentives, and confused accountability at the local devolution of some responsibility and authority to level. There is no single formula for successful civil local levels, which brings decisions on public serv- service management in a decentralized context. ice delivery closer to clients while freeing central Success can be found in various forms, and government of the day-to-day details of local tactics that work in one context may produce the administration. On the other hand, retaining some opposite result in a different environment. The control at the central level can yield substantial diverse approaches of the six cases offer lessons for benefits, including national minimum standards countries considering decentralizing civil service for service delivery and wider mobility within the management. The relatively smooth start to Big civil service. Bang decentralization in Indonesia shows the It is critical for decentralization policy to achieve importance of carefully managing administrative a rational equilibrium between these opposing but elements of the transition, as when the central gov- valid considerations. Decentralization policies ernment continued to pay the salaries of trans- should not aim to devolve all authority to the local ferred civil servants on an interim basis. The lim- level,but rather to redefine the responsibilities of the ited implementation of decentralization policies in center and local governments. Central authorities Thailand shows that having the approach right on must recast themselves as guides, providing local paper does not necessarily lead to getting it right administrations with a comprehensive, feasible in practice. Nonetheless, the latter is important, national policy framework and enhancing their abil- as evidenced by the legal ambiguities and rever- ity to evaluate local performance. At the same time, sals arising in the wake of Indonesia's rapid policies must align the responsibilities and decentralization. autonomy of local managers. The roles of local gov- The experience of East Asian countries in man- ernments and their civil servants must be clearly aging the capacity, incentives, autonomy, and specified and accompanied by the resources and accountability of the subnational civil service can flexibility to tailor activities to the needs of residents, provide valuable lessons for other countries. as well as by mechanisms to ensure accountability The relatively stable transitions to decentralization and prevent proximity from opening the door to in the Philippines and Indonesia show that coun- elite capture. Clear, locally feasible standards for tries can strengthen local capacity on-the-job. minimum performance can mitigate the risks of Cambodia's positive experience with improving the devolving autonomy to subnational governments, use and quality of health services validates the role as can better monitoring of performance through of positive incentives--such as salary supplements the collection, maintenance, and transmission from user fees--in motivating civil servants to of information on performance across levels of perform better. Vietnam's experiment with block- government. grant budgeting in Ho Chi Minh City has reduced Uniformity vs. Unification of the Civil Service administrative costs and overstaffing, showing that more local autonomy can improve civil service effi- The degree to which countries devolve authority ciency.22 Finally, Vietnam's recent passage of a Bud- for civil service management has implications for get Law that clarifies the responsibilities of various uniformity across the civil service. If the center levels of government, as well as the experience of totally circumscribes the conditions of local public the Philippines in improving performance through employment, administrative decentralization will community report cards, shows how efforts both remain a myth, and its potential benefits unreal- within and outside the government can improve ized. On the other hand, if local managers are the accountability of civil servants. entirely free to adjust salary and staffing levels, the 150 East Asia Decentralizes BOX 7.2 Lessons from Uganda's District Service Commissions Uganda's decentralization experience is gener- enforce disciplinary decisions is limited. More- ally considered a success story in terms of its over, because the central Public Service Com- extent and impact. As in Indonesia and the mission defines the size of local governments Philippines, decentralization in Uganda took on and institutions, vets DSC membership, and the characteristics of a Big Bang following an era approves recruiting standards, the central gov- of political opening. Along with the launch of ernment retains influence over subnational per- regular local elections and the transfer of broad sonnel decisions. Finally, the existence of a sep- service responsibilities and attendant fiscal arate DSC in each district--and the associated resources to local governments, the country rise in fiscal transfers and local employment-- took steps to decentralize the administrative has created political pressure to add new apparatus. Staff posted to districts formally districts. This can lead to the proliferation of transferred to local governments, and separate smaller and smaller local administrations and District Service Commissions (DSCs) were set up eventually to a loss of scale economies and to manage human resources in district and local operating efficiency. administrations. As East Asian countries grapple with how best The right of the DSCs to hire, fire, and over- to manage civil servants under decentralization, see subnational staff was enshrined in the new the Ugandan experience with DSCs--both posi- Constitution and further detailed in the Local tive and negative--may provide a useful guide. Government Act. DSCs were meant to provide If designed well, subnational public service com- institutional protection for the relative auton- missions can promote the qualifications and pro- omy of district governments in managing fessionalism of local civil services throughout the subnational civil servants. However, the system country. has encountered several obstacles. Recruitment procedures are slow, performance evalua- Sources: Ndegwa and Levy 2003; Evans and Manning tions rarely occur, and the authority of DSCs to 2003. size and composition of local governments--and commissions could manage the careers and interre- their paychecks--are likely to vary considerably gional mobility of subnational civil servants. These and may lead to interregional inequalities in commissions must have the authority to recruit, employment conditions. transfer, and properly oversee staff, however (see In its ideal form, a country's civil service will be box 7.2). uniform but not unified. Public servants will share a distinct national identity without necessarily being Economies of Scale vs. Client Responsiveness subject to identical rules on pay and employment. Countries can foster servicewide identity through Administrative decentralization also raises ques- joint training of civil servants working in different tions about the appropriate level of government to localities, or by creating subnational associations which responsibilities should be devolved. If the such as the League of Mayors in the Philippines or goal is to bring government closer to the people, professional groups that bring together civil ser- then countries should hand off public service pro- vants who confront similar issues across the coun- vision to the lowest tier of government, where civil try. At the same time, some degree of local variation servants are best placed to understand and respond is important. Solutions to inequitable salary differ- to the needs of local residents. However, improving entials include allowing local governments to top the responsiveness of local governments through up centrally defined base wages through user fees proximity can disrupt the potential for economies and other own-source revenues, as in Cambodia, of scale. Running a government, no matter how big China, and Vietnam, or having the center set a or small, requires a minimum number of staff-- salary range but allowing local managers to accountants, managers, and secretaries, for establish exact salaries. If working conditions vary example--and thus a minimum of human and across subnational governments, public service financial resources. Managing Human Resources in a Decentralized Context 151 In devolving responsibilities to the local level, cen- to take on new tasks. Thailand, for example, has tral governments must consider the capacity of small seen some success with ad hoc forms that devolve civil services to undertake complex technical and autonomy to social sector agencies based on "readi- managerial tasks, and avoid the proliferation of unvi- ness" criteria. However, it is important not to allow able local administrations. Small local governments overly targeted approaches to delay the realization generally have difficulty paying the salaries of of decentralization's potential benefits. The experi- highly skilled staff, and those staff are unlikely to ence of East Asia's rapid decentralizers--Indonesia have enough technical work to keep them busy full and the Philippines--shows that countries can time. Diseconomies of scale in providing public build capacity through decentralization. In this sce- services are a particular issue in Cambodia and nario, careful planning for coordinated training Thailand, where local budgets and staffs are too within a national framework that is sensitive to small to viably provide such services. In these local considerations is essential. countries, ongoing boundary reviews may boost local technical capacity by consolidating local units, but they have not led to any decisions thus far. Civil Service Reform before or after Decentrali- Alternatively, local governments could consider zation? Though a full discussion is outside the joining forces to provide public services that require scope of this chapter, the nexus between decentral- more sophisticated technical equipment and skills, ization and civil service reform is worth mention- or that create spillover benefits for other jurisdic- ing. In principle, it makes more sense to decentral- tions. In China and Indonesia, on the other hand, ize the structure and management of the civil the larger scale of subnational governments limits service after ensuring that it is efficient and effective their ability to effectively reach local populations. at the national level. Yet given the political drive for In addition to striking a balance in decentraliza- decentralization and the difficulty of civil service tion design, countries also face important decisions reform, many countries cannot wait for these to in sequencing reforms. Administrative decentral- occur before decentralizing. On the other hand, the ization entails two main temporal tradeoffs: decision to delay civil service reform until after decentralization can allow the central government Capacity First or Capacity through Decentrali- to transfer the political burden of difficult policy zation? How best to sequence decentralization measures. Countries should explicitly consider the reforms and capacity-building initiatives is a sub- links between these two key reforms to avoid repli- ject of considerable debate. On the one hand, ana- cating national flaws--such as duplication of effort lysts see devolution of responsibility for manage- or overstaffing--at the local level. ment and public services as a necessary step in building individual and institutional capacity. Skills Endnotes development, performance incentives, and innova- 1. This section draws on Evans and Manning 2003. tion are encouraged through "learning by doing," 2. Vaclav Havel, former president of the Czech Republic, used which is generally a more sustainable and individu- this phrase to describe the Czech reform experience. alized approach than occasional courses developed 3. The Local Government Code of the Philippines, Section 2, Sections 76 to 97, address human resource management without close attention to the specific tasks per- and development at the local level. formed by civil servants. On the other hand, 4. The exception to this general rule is the local treasurer, who Prud'homme and others warn of the "dangers of is appointed by the Department of Finance but paid by the local government. Manasan n.d., 12. decentralization": the low capacity and inexperi- 5. Philippine Constitution, Article IX, Section 2 (2). ence of local civil servants can mean a sharp deteri- 6. The 45 percent cap applies in first-class through third-class oration in the quality and efficiency of public serv- local governments, and the 55 percent cap in fourth-class ices (Prud'homme 1995). through sixth-class local governments. 7. Owing to the penchant among local governments to use The critical challenge is to maintain forward alternative sources to supplement wages,these official figures momentum while balancing capacity considerations. likely underestimate the degree to which local governments Asymmetric approaches to decentralization can have exceeded restrictions on personnel expenditures. 8. Pay scales were set at a fixed percentage of national levels, allow gradual devolution of responsibilities to local ranging from 75 percent in sixth-class local governments to governments that have demonstrated the capacity 95 percent in second-class local governments. 152 East Asia Decentralizes 9. Under this revised law, provincial governors will have Cohen, Margot. 2003. "Cadres Get a Field Lesson." Far Eastern authority to hire, fire, and relocate provincial staff only at Economic Review 166 (45): 24. echelon 2 and below. Similarly, heads of districts and cities Evans, Anne, with Nick Manning. 2003. "Decentralization: A will be able to hire, fire, and relocate staff only at echelon 2 Review of Staffing Practices in Eight Countries." Internal and below, and only after consulting with the provincial draft document. World Bank, Washington, DC. governor. Fritzen, Scott. 2002. "The `Foundation of Public Administra- 10. Some local governments, though, have begun to informally tion'? Decentralization and Its Discontents in Transitional supplement salaries. Also, several regional allowances are Vietnam." Paper presented at the Asia Conference on Gover- offered that are imperfectly recorded. nance in Asia: Culture, Ethics, Institutional Reform, and 11. At the 11th Party Congress in 1978, Deng Xiaoping used Policy Change, City University of Hong Kong. this analogy to urge Chinese policy makers to take a grad- Government of Vietnam. 2001. "Master Programme on Public ual approach to economic reform. Administration Reform for the Period 2001­2010." 12. Before 1984, these appointments were made by party Attachment to the Prime Minister's Approval Decision committees two levels up in the territorial hierarchy. Burns No. 136/2001/QD-TTg, September 17. n.d., 8. Government of Vietnam, General Statistical Office. 2003. Statis- 13. At all levels, the criteria used are population, land area, and tical Yearbook 2002. Hanoi: Government of Vietnam. value of industrial and agricultural production; at the city Government of Vietnam, Ministry of Finance. 2000. "Vietnam: level, additional factors include the number of component Improving the Budget-Norm System for Recurrent Expendi- administrative units, local budget income, and the amount tures." VIE/96/208, Public Expenditure Review Phase II. of developed land. www.undp.org.vn/projects/vie96028/norms.pdf. 14. Including rank-based benefits, such as housing and car Inoki, Takenori. 2001. "Personnel Exchange among Central and allowances, does widen the gap. Local Governments in Japan." Working Paper. Washington, 15. In Hebei province, for example, the director of budget DC: World Bank Institute. preparation had kept his post as director of the Budget Iqbal, Farrukh. 2001. "Evolution and Salient Characteristics of Department in the Hebei finance bureau. the Japanese Local Government System." Working Paper. 16. Webster (2002) argues that functional devolution does not Washington, DC: World Bank Institute. require the transfer of all corresponding civil servants. He Kingdom of Thailand. 1997. Constitution of the Kingdom of argues that if decentralization is meant to improve the effi- Thailand. www.oefre.unibe.ch/law/icl/th00t___.html. ciency of public services by transferring authority to units Litvack, Jennie, Junaid Ahmad, and Richard Bird. 1998. closest to the beneficiaries, then fewer people should be "Rethinking Decentralization in Developing Countries." able to perform the same task just as well as or better than Sector Studies Series. Washington, DC: World Bank. at the center. Litvack, Jennie, and Jessica Seddon. 1999. "Decentralization 17. Teacher pay, for example, could be much lower after Briefing Notes." Working Paper. Washington, DC: World employees transfer to the local level. See Wegelin 2002, 23. Bank Institute. 18. The amount of this incentive is B 6,000 per month, or Manasan, Rosario G. "Fiscal Decentralization: The Case of the roughly US$150. Philippines." World Bank. http://www.worldbank.org/wbi/ 19. The national wage bill includes defense and security expen- publicfinance/documents/ASEM/manasan.pdf. ditures. Ndegwa, Stephen, and Brian Levy. 2003. "The Politics of Decen- 20. The other 51 percent is to be spent on operational tralization in Africa: A Comparative Analysis." Paper pre- expenses. See Turner 2002, 357. sented at the Regional Workshop on Governance and Public 21. Conversation with Pamela Messervy, World Health Organi- Management, Johannesburg. zation, Phnom Penh, Cambodia, October 2003. OECD (Organisation for Economic Co-operation and Develop- 22. See box 5.3. Most surplus staff were transferred to other ment). 1997. "Managing across Levels of Government." localities rather than retired or retrenched. Paris: OECD. People's Republic of China. 1982. Constitution of the People's Republic of China. http://english.peopledaily.com.cn/ Bibliography constitution/constitution.html. Prud'homme, Rémy. 1995. "The Dangers of Decentralization." Asia Foundation. 2002. First Indonesia Rapid Decentralization World Bank Research Observer 10 (2): 201­20. Appraisal (IRDA) Synopsis of Findings. College Park, MD: Republic of the Philippines. 1987. 1987 Constitution of the University of Maryland, IRIS Center. Philippines. www.chanrobles.com/philsupremelaw1.htm. Asian Development Bank. 1999."Governance in Thailand: Chal- ------. 1991. The Local Government Code of the Philippines lenges, Issues, and Prospects." Manila: Asian Development (Republic Act No. 7160). www.chanrobles.com/localgov. Bank. htm. Azfar, Omar, Tugrul Gurgur, Satu Kähkönen, Anthony Lanyi, Royal Government of Cambodia. 2002. National Poverty Reduc- and Patrick Meagher. 2000. "Decentralization and Gover- tion Strategy, 2003­2005. Phnom Penh: Royal Government nance: An Empirical Investigation of Public Service Delivery of Cambodia. in the Philippines." College Park, MD: University of Mary- Schiavo-Campo, Salvatore, Giulio de Tommaso, and Amitabha land, IRIS Center; and Washington, DC: World Bank. Mukherjee. 1997. "Government Employment and Pay: A Bales, Sarah, and Martin Rama. 2002. "Are Public Sector Work- Global and Regional Perspective." Policy Research Working ers Underpaid? Appropriate Comparators in a Developing Paper No. 1771. Washington, DC: World Bank. Country." World Bank. http://econ.worldbank.org/files/ Steedman, David W., and Lloyd R. Kenward. 2005. "Civil Service 3358_wps2747.pdf. Reforms at the Regional Level: Opportunities and Con- Burns, John. "Country Reform Summaries: China." www1. straints." Internal draft document. World Bank, Washington, worldbank.org/publicsector/civilservice/rsChina.pdf. DC. Managing Human Resources in a Decentralized Context 153 Turner, Mark. 2002. "Whatever Happened to Deconcentration? ------. 2003a."Cambodia: Enhancing Service Delivery through Recent Initiatives in Cambodia." Public Administration and Improved Resource Allocation and Institutional Reform." Development 22: 353­64. Integrated Fiduciary Assessment and Public Expenditure Webster, Douglas. 2002. "Implementing Decentralization in Review. Washington, DC: World Bank. Thailand: The Road Forward." Internal draft document. ------. 2003b. "China Country Economic Memorandum: Pro- World Bank. http://www.worldbank.org/wbi/publicfinance/ moting Growth with Equity." Washington, DC: World Bank. documents/East%20Asian%20Decentralization%20 ------. 2003c. "Combating Corruption in Indonesia: Enhanc- Dialogues%20II/Douglas.pdf. ing Accountability for Development." Washington, DC: Wegelin, Emiel. 2002. "Thailand: Decentralization Capacity World Bank. Assessment, Findings, and Recommendations." Internal ------. 2003d. "Decentralizing Indonesia. A Regional Public draft document. World Bank. http://www1.worldbank. Expenditure Review Overview Report." Washington, DC: org/wbiep/decentralization/eaplib/wegelin.pdf. World Bank. Weist, Dana. 2003. "Thailand's Decentralization: Progress and ------. 2003e. "Human Resource Management." Internal draft Prospects." Paper prepared for the Asian Development Con- document for Philippines Decentralization and Service ference on Decentralization in Asia, Kita-kyushu, Japan. Delivery Study. World Bank, Washington, DC. Wong, Christine. 2003. "Country Review: The People's ------. 2003f. "Philippines Decentralization and Service Republic of China." Internal draft document. World Bank, Delivery Study." Internal draft document. World Bank, Washington, DC. Washington, DC. World Bank. 2000. "Vietnam: Managing Public Resources ------. 2003g. "Thailand Country Development Partnership: Better." Public Expenditure Review. Washington, DC: World Governance and Public Sector Reform," Phase I Monitoring Bank. Workshop. Washington, DC: World Bank. ------. 2001. "Indonesia: Priorities for Civil Service Reform." ------. 2003h. World Development Report 2004: Making Services Internal draft document. World Bank, Washington, DC. Work for Poor People. Washington, DC: World Bank; and ------. 2002a. "China: National Development and Subnational New York: Oxford University Press. Finance--A Review of Provincial Expenditures." Washing- ------. n.d. "Subnational Government." www1.worldbank.org/ ton, DC: World Bank. publicsector/civilservice/subnational.htm. ------. 2002b. "Decentralization and Deconcentration in WHO (World Health Organization). 2003. "Functional Analysis Cambodia." Internal draft document for the Cambodia for Human Resources: Kampong Cham Province." Geneva: Public Expenditure Review. World Bank, Washington, DC. World Health Organization. ------. 2002c. "Vietnam: Delivering on Its Promise." Vietnam Development Report. Washington, DC: World Bank. 8 Decentralizing Health: Lessons from Indonesia, the Philippines, and Vietnam Samuel S. Lieberman, Joseph J. Capuno, and Hoang Van Minh This chapter examines the decentralization experi- no widely accepted, best-practice health policy ence of three East Asian countries from the per- framework. This is partly because decentralization spective of how well they have addressed the special in many low-income countries is a recent develop- features and requirements of the health sector. ment, while affluent countries historically moved These features include the substantial role of exter- in a centralizing direction as constituent states nalities, the high degree of specialization, the criti- came together to form federal unions. cal role of quality and timeliness, and the high level Another constraining factor has been the top- of knowledge required to participate in the health down, centrist bias in the influential Health for All care system at all levels. These characteristics have (HFA) paradigm, used to build dominating but important implications for the design of health difficult-to-manage and ineffective health ministries policy in general, and especially for a decentralized in many countries. HFA's main sponsor, the World system of service delivery and sector management. Health Organization (WHO), has been traditionally This chapter outlines the decentralization health uncomfortable with decentralization. The literature policies and programs of Indonesia, the Philippines, on government roles in health systems is relevant to and Vietnam, focusing on the period 1985­2003, the extent that privatization is a form of decentral- spanning the years before and after significant ization, but this literature misses the critical decen- decentralization began in these countries. The tralization issue: the allocation of roles among levels chapter also points to areas where reforms may of government.The literature also lacks a connection facilitate more effective health care delivery. between options for decentralization and health financing. Meanwhile, most approaches that focus on health challenges--including the Millennium The Health Care Context Development Goals of the United Nations and the of Decentralization World Bank­sponsored Poverty Reduction Strategy Experience with decentralizing health in develop- Papers in developing countries--assume a strong ing economies is limited, and the literature reveals central role for ministries of health. 155 156 East Asia Decentralizes Nevertheless, a debate has developed around health services starting in 1992, 1996, and 2001, decentralization design issues, with contributors respectively.1 Evidence suggests that health ministries dividing into two camps. Proponents see decen- in these countries initially were not prepared to tralization, if handled well, leading to systematic articulate and assume a new specialized role of sys- citizen involvement in setting the goals, design, tem manager rather than main provider. Inconsis- and financing for health policy, and in monitoring tent policies further indicated that expediency service provision and other functions. In this view, rather than strategy guided official responses to decentralization can also spur providers to obtain transition problems. the skills, material support, and authority they At the same time, the faltering performance of need to offer high-quality services. Decentraliza- the health systems in these countries before tion can further enable clients to secure informa- decentralization signaled a need for significant tion, financing, and bargaining power, and offer changes in health policy. In the Philippines, health ministries a chance to jettison impractical improvements in infant survival rates and other obligations and carve out a new role and image. health status indicators in the 1980s were begin- Detractors warn, however, that (badly designed) ning to plateau, indicating decreasing returns decentralization heightens vulnerability to near- from health expenditures that were higher than in term crises and longer-term risks. Typical start-up other developing countries in terms of gross problems include staff opposition, leading to domestic product (GDP) (Solon et al. 1992). breakdown of deployment and other personnel In Vietnam, the collapse of the agricultural mechanisms; mismatches between health care financing system and economic reform in the 1980s funding and spending requirements; ambiguity in undermined funding for primary health care serv- responsibilities and premature delegation of func- ices and produced shortages of drugs and skills, tions, leading to deteriorating service quality; and deteriorating quality of care, and a decline by a half disruptions in reporting, accountability, and qual- or more in use of government facilities. Funding ity control. Medium-term concerns include rising gaps also led to higher user fees, which became a system costs. Specifically, downsizing administra- financial barrier and reduced access to care by the tive units may yield designs for key health func- poor. tions that are neither technically efficient nor cost- In Indonesia, the 1997 financial crisis brought effective because of diseconomies of scale. Such funding cuts that confirmed the susceptibility of "transitional" problems may be difficult to correct. the government network to drug shortages and Because of these risks, public health commenta- other breakdowns. But performance problems had tors have called for careful introduction and man- existed earlier. The country did not sustain favor- agement of decentralization. Most analysts support able trends in survival and nutrition rates from the WHO's recommendation that countries phase in 1980s in the 1990s, despite large-scale intervention. devolution under central guidance, subject to strin- Use of public services also faltered. After rising to gent criteria, with health ministries continuing to nearly a third, the share of people who sought take responsibility for specialized services, medical outpatient care from public providers fell below supplies, basic education and training, and other 30 percent by 1995, and below 20 percent by 1998. key functions (WHO 1995). This advice illustrates Meanwhile, households in the top expenditure two recurring themes: that the overriding rationale quintile were far more likely than the poor to use for health decentralization is improved effective- public facilities as inpatients, and nearly as likely as ness and efficiency, and that the timing of the to use such facilities as outpatients. process is subject to ex ante design. Such premises are usually not valid, since the Objectives of the Chapter impetus for decentralization is generally political. Improved health is only a second-order objective, This chapter examines decentralization experiences with imperatives such as preserving national unity in Indonesia, the Philippines, and Vietnam with an usually driving the process and shaping the deci- eye toward three sets of questions. First, how can sion to devolve to particular levels of government. developing countries design decentralization to That was arguably the case in the Philippines, provide an appropriate framework for a public Vietnam, and Indonesia, which decentralized their health system, and what policies and instruments Decentralizing Health: Lessons from Indonesia, the Philippines, and Vietnam 157 promise to be effective in improving the efficiency decentralized arrangements have not worked as and equity of a decentralized health system? Second, well as hoped, especially regarding access to high- how should countries handle transition problems quality health services for the poor. and other risks? Third, what lessons do the experi- ences of these three countries reveal? The Origins of Decentralization In addressing these questions, the chapter exam- ines the emerging role of central health ministries. The fact that these countries have experienced only Critical functions include monitoring and control- modest health gains from decentralization reflects ling communicable diseases, setting standards and circumstances outside the control of policy makers. assuring quality for devolved health services and In particular, these countries introduced decentral- pharmaceuticals, ensuring access of the poor to ization in less than favorable economic and politi- health services, and sustaining health financing. cal environments. For example, the 1997 Asian The three countries have broad features that financial crisis underscored the direct link between facilitate comparison. All have tropical or semi- macroeconomic performance and health expendi- tropical climates, and all are highly populated tures. Before 1998, the Philippine economy--much developing countries composed mostly of rural- like that of Indonesia and Vietnam--was growing based agricultural households, with significant steadily. Per capita income in the Philippines rose numbers of poor. Each country also has a colonial from US$2,310 in 1985 to US$3,870 in 1997. Fol- history in which the struggle for independence led lowing the outbreak of the crisis in late 1997, per to a unitary form of government with a strong capita income fell to US$3,730. The impact of the center. Communicable diseases are the main cause crisis in Indonesia was graver. Its per capita income of morbidity and mortality in all these countries, declined from US$3,030 in 1997 to US$2,580 although each is now experiencing an epidemio- in 1998; by 2001 per capita income was US$2,900, logical transition that brings growth of non- still lower than before the crisis. Vietnam was less communicable, lifestyle-related diseases. Yet each affected by the crisis, although per capita govern- country also possesses unique features that provide ment health spending leveled off and may have interesting contrasts to the other two. Unlike the fallen after the crisis. Philippines and Indonesia, for example, Vietnam In the Philippines, the crisis prompted the has adopted a market-oriented economic policy national government to invoke "an unmanageable only recently while retaining socialist features in its public sector deficit"--a provision under the Local government structure. Unlike Vietnam, Indonesia Government Code of 1991 that allows a 10 percent and the Philippines are archipelagos that are insu- cut in the Internal Revenue Allotment (IRA) dis- lated to a degree from disease transmission across tributed to lower levels of government.2 Because land borders. most local governments depended heavily on the The available data limit comparison between IRA, the reduction further reduced local health these countries. Official statistics on health, demo- spending, especially among provinces and munici- graphics, government finances, and other socio- palities, which had absorbed the bulk of devolved economic indicators vary in scope, detail, and qual- health functions. ity. The same applies to secondary sources of Uneven regional growth aggravated the situa- information. Data constraints also add to the tion. In the Philippines, provinces in the Eastern methodological challenges of tracing the impact of Visayas and Northern Mindanao continued to lag decentralization on health amid other socioeco- behind other provinces, especially those in Metro- nomic factors, external conditions, and policy inter- politan Manila and surrounding provinces. In ventions. Thus, the analysis draws only broad con- Vietnam, the cities of Hanoi and Ho Chi Minh were clusions and policy guidelines. developing faster than other areas, and similar The analysis suggests that decentralization divi- unevenness existed in Indonesia. Utilization rates dends so far have proved modest and concentrate and other indicators of health access therefore var- in some areas of each country. Decentralization ied widely across regions in all three countries. may have helped sustain overall improvements in Indonesia and the Philippines also implemented health status and spurred local initiatives in health decentralization amid considerable uncertainty fol- planning, service delivery, and financing. However, lowing political crises. After the fall of the Marcos 158 East Asia Decentralizes regime, the Philippines ratified a new Constitution decentralized health care systems in these three in 1987 and further articulated strong decentralist countries. The final section summarizes findings provisions in the Local Government Code of 1991. and draws lessons regarding the role of the central In Indonesia, the overthrow of the Soeharto regime health ministry in managing the health sector. in 1998, and then de facto secession of the erstwhile province of East Timor in 2000, contributed to the Health Policy under clamor for decentralization. The Philippines expe- Decentralization rienced several military uprisings after 1986, the most recent in late 2003, and has had four presi- Consistent with their respective constitutions, the dents and nine secretaries of the Ministry of Health three countries passed legislation that enabled,if not under the present Constitution. These frequent mandated, the decentralization of health services. musical chairs in the health ministry have disrupted Besides added administrative powers and responsi- policy priorities and the ministry's momentum in bilities, local governments attained greater fiscal adapting to a decentralized setting. Persistent rural autonomy through higher shares of national gov- insurgencies and kidnappings in areas such as ernment revenues and expanded taxation powers. Southern Mindanao have made it difficult for both In Indonesia, the principal enabling legislative the private sector to pursue investments and the acts were Law 22 and Law 25 of 1999, while Regula- public sector to reach out to the poor. tion 25 of 1999 facilitated implementation. The Weak governance in the Philippines, including Philippines promulgated decentralization through corruption in key branches of government, has also the Local Government Code of 1991, implemented led to loss of revenues and waste of limited the following year. resources. Mechanisms like Health Boards and In Vietnam, the doi moi economic reforms that other local consultative bodies have seldom been began in 1986 and the Public Administration convened for counsel or feedback, contrary to the Reform of 1995 shaped health decentralization, intent of the Local Government Code (World Bank with implementation based largely on the 1996 and 2000a). However, the proliferation of nongovern- 2002 State Budget laws. The latter two measures mental organizations (NGOs) and other civil soci- brought fundamental changes in the preparation, ety groups has been a major positive development. approval, and execution of budgets for all govern- Many such organizations now work side by side ment agencies, from the central to local levels. Since with key national agencies in agrarian reform, 2004, province-level People's Councils have had health advocacy, local capacity building, livelihood more authority to prioritize expenditures and projects, community mobilization, and governance determine sectoral allocations and transfers to reform. lower tiers, and stronger means of mobilizing A second set of reasons for the modest gains resources. Transfers from the center for stable peri- from decentralization in these three countries ods of three years will promote local planning, relates to weaknesses in policy itself. While external while provinces must produce forward-looking factors limited the potential benefits, better man- expenditure plans in return. agement by the central ministry of health, espe- Decree 10, another element in Vietnam's legal cially in critical health functions, would have underpinning for decentralization, took effect in helped. Experiences in the three countries suggest July 2002. When fully implemented, this decree how to define and pursue an effective role for the will give managers of facilities much greater con- central health ministry. trol over their budgets, and more (though still lim- The next section analyzes the features and ited) discretion regarding pay and employment, implementation of each country's decentralization user charges for nonbasic services, and domestic policy. The following section examines the impact borrowing. of decentralized health services on health status, In each country, later laws further articulated, service coverage, overall efficiency, and equity in directly supported, or affected the decentralization these three countries. Ensuing sections examine of health services. In the Philippines, for example, intergovernmental fiscal challenges, personnel these laws included the Magna Carta for Public management, and service delivery under the Health Workers of 1992, the Barangay Health Decentralizing Health: Lessons from Indonesia, the Philippines, and Vietnam 159 Workers' Benefit and Incentives Act of 1995, and assigned additional health tasks to provinces and the National Health Insurance Act of 1995. In districts (Fritzen 1999). The law established finan- Vietnam, the Seventh Communist Party Congress cial links underpinning a unitary system in which passed a resolution to broaden the"scope of respon- national authority is delegated to lower levels. At sibilities and power of the sectors and localities," each level, budget preparation and implementation and passed the Grassroots Democracy Decree in are the responsibility of the People's Council. 1999 (Communist Party of Vietnam 1993; Govern- In contrast, implementation in the Philippines ment of Vietnam 1999). and Indonesia occurred in Big Bang fashion. The former completed the transfer of 45,896 health personnel, along with hospitals, clinics, and other Main Design Features and Implementation facilities, in 1993, two years after passing the Local At first glance, the division of responsibility for Government Code. Indonesia completed a similar critical health functions between the national and transfer in 2001, less than two years after enacting local governments in these countries broadly Laws 22 and 25. reflects efficiency principles. That is, local govern- The Big Bang approach has its merits, but expe- ments have assumed responsibility for health func- riences in Indonesia and the Philippines reveal its tions that are simple to administer or confer local- disadvantages. In Indonesia, decentralization laws ized benefits. The central government or higher and rules and regulations do not provide enough local governments have assumed responsibility for detail on functional and operational responsibili- health functions with significant economies of ties, resulting in confusion and divergence between scale or interjurisdictional spillovers. For example, provinces and districts. For instance, provinces are basic, primary health care services are assigned to supposed to handle cross-district tasks, but no communes in Vietnam, including the network of definitive finding tells them how to apply that rule. village health workers, to villages in Indonesia, and The laws and regulations governing decentraliza- to barangays (villages) in the Philippines. Primary- tion are also often inconsistent with other laws, level health facilities are assigned to cities and especially civil service rules. This inconsistency has municipalities in the Philippines and to districts in limited the ability of local governments to right- Vietnam. Secondary-level hospitals are assigned size inherited health bureaucracies and anticipate to provinces in the Philippines and Vietnam. personnel matters. Tertiary-level and specialty hospitals, on the other Moreover, administrative preparation was inad- hand, are mainly the responsibility of the central equate. For example, many local officials in the government--that is, the central heath ministry-- Philippines were unaware of the precise nature and in all three countries. extent of their new expenditure responsibilities and Central governments continue to provide cer- powers, and the central Department of Health tain public goods such as health research and devel- (DOH) was slow to transform itself structurally opment, and merit goods such as maternal, child and operationally.3 Lack of personnel severely care, and family planning services. Local govern- hampered the Local Government Assistance and ments are often involved in and sometimes cofi- Monitoring Service, created to troubleshoot transi- nance these programs. However, overall, the devo- tion problems, and the service lacked clout, as dif- lution of health functions and corrective measures ferent DOH divisions managed public health pro- reveal flaws. grams as before. With DOH looking uncertain, Decentralization occurred gradually in Vietnam many local governments seemed to adopt a wait- and not without setbacks. Local mobilization was and-see strategy, apparently hoping that the agency seen as a key element in the country's impressive would be blamed for the breakdown in the public achievements by the mid-1980s in delivering pri- health system and be forced to recentralize health mary health care. As mentioned, the combined functions. province and commune share of government In different degrees, these three countries also health outlays was already significant in the early introduced local governance mechanisms to pro- 1990s. Thus, local officers had experience with mote transparency, accountability, and participa- decentralization when the 1996 State Budget Law tion as they devolved health services. This, of 160 East Asia Decentralizes course, complicated the transition, as local govern- cancer and heart diseases began to match--if not ments initially had to adopt these mechanisms on overtake--that of communicable diseases such as their own without much guidance or experience, tuberculosis and malaria (Solon et al. 1999). leading to delays, perfunctory compliance, or fail- In Vietnam, disparities in survival rates between ure to convene the mandated consultative bodies. regions appear to have widened in the late 1990s, captured in the rising ratio of highest to lowest regional IMRs by region. After growing from 1.7 in Health Dividends under 1989 to 2.3 in 1994, this ratio rose to 3.6 in 2002. Decentralization This is not to imply that rates and underlying con- On the whole, each country sustained favorable ditions were static. On the contrary, IMRs them- trends in overall health status after decentralization selves fell by at least half in every region between (see table 8.1). In the Philippines, gaps in health roughly the early 1990s and 2003. However, the status across regions continued to close during the decline in these changes was extraordinary, drop- 1990s. Measured as the difference in infant mortal- ping to a third or less of the early 1990s figure in the ity rates (IMR) between the poorest region (Eastern Mekong, Central Highlands, Southeast, and Cen- Visayas) and the richest region (Metropolitan tral Coast regions. Manila), the gap narrowed from 15 in 1980 to 9.8 in 1985. By 1990 the gap was almost zero, with the Some Progress in Health Outputs and Access IMRs of Metropolitan Manila and Eastern Visayas 27.4 and 27.1, respectively. In 1995, the gap The favorable trend in overall health status was remained near zero, although the IMRs of Metro- arguably due partly to progress in health outputs politan Manila and Eastern Visayas improved to and service coverage. In the Philippines, for exam- 21.3 and 21.6, respectively. Following the 1997 ple, the proportion of births attended by trained Asian financial crisis, however, the disparity health workers, and of the population with access widened. By 2000 the IMR of Metropolitan Manila to clean water source and sanitation services, rose was 19.4--worse than the 10.7 of Eastern Visayas.4 in the 1985­2000 period. Each study country also experienced an epi- Similar developments in health status, outputs, demiological transition in the 1990s, in which the and access indicators occurred in Indonesia over incidence of chronic, lifestyle-related diseases like the same period. The 2002 Demographic Survey TABLE 8.1 Selected Health Status Indicators Year Indicators 1980 1985 1990 1995 2000 2001 2002 Infant mortality rate (per 1,000 live births) Indonesia 79 70 60 46 35 33 1.9 Philippines 65 55 45 36 30 29 1.9 Vietnam 50 43 36 32/30 28/18 30/18 1.9 Under-5 mortality rate (per 1,000 live births) Indonesia 125 108 91 66 48 45 1.9 Philippines 81 74 66 51 40 38 1.9 Vietnam 70 60 50 43 34 38 1.9 Life expectancy at birth Indonesia 55 59 62 64 66 66.3 66.7 Philippines 61 63 66 68 69 69.5 69.8 Vietnam 60 63 65 67 69 69.4 69.7 Sources: World Bank 2002; WHO 2002. Decentralizing Health: Lessons from Indonesia, the Philippines, and Vietnam 161 and Health Survey pointed to a continuation or public sector share fell slightly from 31.5 percent in even an acceleration of favorable trends in fertility, 1997 to 28.5 percent in 2001. In the Philippines, contraceptive use, malnutrition, and trained mater- the public share rose from 43 percent in 1997 to nal care. Some indicators worsened: immunization 45 percent in 2001. rates fell between 1997 and 2002­3 for children A closer look at public sector outlays reveals a under age two, for example, while the prevalence shift in the financing burden from the central to of childhood illness remained the same as in 1997. local governments. Most local governments In Vietnam, on the other hand, output and access devoted health spending to hospital and personal measures all pointed in a positive direction care services, much like the pattern before devolu- between the mid-1990s and 2002. For example, the tion. This is understandable, as local governments country reported a significant increase in child- absorbed many hospitals under decentralization. hood vaccination coverage, and in the proportions However, this orientation may be inappropriate of women receiving prenatal care and giving birth given the high prevalence of communicable attended by skilled health personnel (Committee diseases and high relative cost of hospital-based for Population, Family, and Children 2003). interventions. The Philippines certainly saw such a shift in financial burden. The annual share of local govern- Health Expenditures ments in public health expenditures climbed up Decentralization may have more than sustained from less than 5 percent before 1992 to 12.5 percent momentum in improving health status and even in 1993. By 2001, the local share reached 20.9 per- reversed worsening trends. Unfortunately, available cent, exceeding the 16.6 percent share of the data do not allow us to verify these two supposi- national government. Moreover, personal care serv- tions, nor do input measures such as health expen- ices constitute the bulk of public expenditures for ditures enable definitive conclusions. health in the Philippines, and, ominously, a grow- According to the World Development Report ing portion of the health outlays of local govern- 2004, annual health expenditures remained more ments as well (Solon et al. 1992). or less a constant proportion of GDP throughout In Indonesia, regional governments now account 1997­2001 in the three countries. The average for most routine spending, while development annual proportion was 2.5 percent in Indonesia, spending at the regional level grew fourfold. How- 3.5 percent in the Philippines, and 4.9 percent in ever, central development outlays rose almost Vietnam. In per capita terms, however, total health threefold, and nearly half of development expendi- spending fell in Indonesia from US$26 in 1997 to tures still come from the central budget. Under- US$16 in 2001, and in the Philippines from US$41 standably, local governments continue to regard the in 1997 to US$30 in 2001 (World Bank 2004).5 Ministry of Health as a key supplier of financial Asia's financial crisis led to a steep decline in 1998 resources as well as personnel, equipment, drugs, in health spending in these two countries: 50 percent and vaccines.7 in Indonesia and 24 percent in the Philippines. In Vietnam in the early 1990s, subnational gov- Seemingly immune to the financial crisis, Vietnam's ernments, including those at the commune level, per capita health spending rose from US$16 in 1997 were already spending more on health than the to US$21 in 2001.6 In general, health expenditures central government (Knowles et al. 2003). Provin- as a percentage of GDP in these three countries cial spending accounted for 68 percent of govern- were similar to those of most of their neighbors. ment health expenditures in 1991 (not including For example, the average percentage share of health revenues from user fees and donor support), while expenditures in GDP in Thailand and Malaysia was central spending accounted for 13 percent. How- 3.7 and 3.2, respectively. ever, five years later, after the country passed the Meanwhile, the public sector share of total Law on State Budget, provinces accounted for health expenditures in each of the three countries 53 percent and central units 26 percent of all did not change much between 1997 and 2001. government outlays. By 2000 the latter figure had Indonesia's public sector accounted for roughly a fallen to 17 percent, while the province-level share fourth of total health expenditures. In Vietnam, the (narrowly construed) had dropped to 44 percent. 162 East Asia Decentralizes However, the thrust of the 1996 legislation would Indonesia before decentralization. Like their coun- seem to suggest including revenue from health terparts in the Philippines, the poorest families insurance and user fees in the provincial total. In had infant and under-five mortality rates several thatcase,provincially"controlled"outlaysaccounted times higher than those of the richest families. for 76 percent of total government health spend- Discrepancies in health status between the ing, up from 70 percent in 1996. (Donor outlays poorest and richest households appeared to be are treated as a separate category influenced by worse in Indonesia than in Bangladesh, India, particular agendas and criteria.) the Philippines, and Vietnam. In light of these financing and spending pat- Perhaps because of the inferior quality of public terns, it is doubtful that decentralization has health services, the poor--like well-off fellow widened access by the poor to quality health care. A Filipinos--continue to self-finance their access to national client survey confirmed that Filipinos in private health services. Private sources, including general were more satisfied with private hospitals direct out-of-pocket payments, accounted for an and clinics than with government health facilities. annual average share of 57 percent of total health Filipinos also tended to rate traditional healers as expenditures in the Philippines from 1991­2001. more satisfactory than any public providers (World Bank 2001a). The low regard for public health serv- Local Initiatives in Health Services and Financing ices prevailed even among the poor, an indication that the public health system does not serve its tar- Decentralization has given local authorities and get clients well. other stakeholders greater leeway to adapt or even A World Bank study of socioeconomic differ- replace once-standard methods for delivering and ences in health, nutrition, and population in financing health services. And these greater discre- selected developing countries corroborated these tionary powers have led to numerous local innova- observations (see table 8.2). In the Philippines, tions in health planning, service delivery, and children born in 1998 to the poorest families were financing. Most notably in the Philippines, there twice as likely to die within a year as children born are the provincial health insurance programs of to the richest families. The infant mortality rate for Bukidnon and Guimaras, as well as other "text- the poorest families (48.8) was 1.7 times that of the book" cases, such as the health card system of richest families (28.8). The life chances of these Paranaque City, the City in the Pink of Health unfortunate Filipinos did not seem to improve with program of Marikina City, and the Community age: under-five mortality rates were 79.8 and 29.2 Primary Hospital Program of Negros Oriental for the poorest and richest families, respectively. (Pineda 1998; Bautista et al. 1999; Quimpo 1996; Vietnam also had disparities in health status and Legaspi 2001). Several of these programs have across economic groups, resembling those in received formal recognition from government TABLE 8.2 Health Status of the Poorest and Richest Population Groups in Selected East Asian Countriesa Infant mortality rate Under-5 mortality rate (per 1,000 live births) (per 1,000 births) Country Poorest/ Poorest/ (year) Poorest Richest richest Poorest Richest richest Bangladesh (1996­7) 96.3 56.6 1.701 141.1 76.0 1.857 India (1992­3) 109.2 44.0 2.482 154.7 54.3 2.849 Indonesia (1997) 78.1 23.3 3.352 109.0 29.2 3.733 Nepal (1996) 96.3 63.9 1.507 156.3 82.7 1.890 Philippines (1998) 48.8 20.9 2.335 79.8 29.2 2.733 Vietnam (1997) 42.8 16.9 2.533 63.3 23.0 2.752 Source: www.worldbank.org/hnp. a. Economic groups are based on asset (wealth) quintiles. Decentralizing Health: Lessons from Indonesia, the Philippines, and Vietnam 163 BOX 8.1 Local Innovations in Health Service Delivery in the Philippines Charging User Fees for Health Services Transforming a Rural Health Center into a in Malalag, Davao del Sur Community Clinic in Sebaste, Antique In December 1993, the local Sangguniang Bayan Under Mayor Juanita de la Cruz, Sebaste in (municipal council) of Malalag, in the province Antique--a remote sixth-class municipality-- of Davao del Sur, enacted the Malalag Revenue became a prime example of how to transform a Code. This code established a socialized fee basic rural health center into a full-service com- schedule for health services, among other provi- munity clinic despite limited resources. With only sions. The graduated payment scheme reflects =800,000 in IRA funds from the center, the gov- P users' annual family income: those earning ernment tapped foreign donors, local people, and =15,000, from =15,000 to =50,000, and more P P P former residents living abroad for support for its than =50,000 pay 25 percent, 50 percent, and P health goals while also appealing to the sense of 100 percent, respectively, of fixed service mission of health personnel. After creating a trust charges. The code also gave low-income families fund, the government infused =3.085 million into P priority in receiving health services. Public con- the project from 1994 to 1998. By 1997, the sultations, hearings, and an information and community clinic employed 16 people, including education campaign overcame initial resistance two physicians, and remained open 24 hours a to the scheme. Partly as a result, the local gov- day, providing primary health care, laboratory ernment earned about =1 million worth of fees P and pharmacy services, and minor surgery. The on an outlay of =688,888. This enabled the gov- P clinic has reduced the cost of these services to the ernment to provide additional health services, local clientele while also serving the medical including surgical, medical, and dental services. needs of residents of neighboring municipalities. With these improvements, the local clientele have become more demanding of the quality of Source: Galing 2001. health services and the performance of health personnel. agencies and private bodies such as the Galing chaired by the vice-chair of the People's Commit- Pook Foundation (see box 8.1) and the Philippine tee. District and communes adopted the same Human Development Network.8 structures. Vietnam has seen numerous instances of In Indonesia, Yogyakarta province showed how sponsored and spontaneous innovation at the to use the country's still immature decentralized province level. An example of the former was the framework to introduce health sector reforms and health ministry's effort to encourage local responses elements of a health insurance system. The to childhood diseases, including community- province used donor funds to secure technical determined indicators (Fritzen 1999). Reactions to assistance and conduct assessments, trials, bench- HIV/AIDS illustrate the spontaneous case. As in marking, workshops, training, and coordination several other provinces, the epidemic spurred the meetings with districts, as well as advocacy events. Thanh Hoa government to pursue preventive The province established a board of trustees and activities such as harm reduction and 100 percent new fund-holder institutions, as well as a benefit condom use. These initiatives resulted from strong package and an insurance premium. commitment by the People's Committee. Besides The province has created a plan for a quality ensuring the participation of the police (Depart- council to accredit facilities and license practition- ment of Public Security), the committee allocated ers based on local standards. Meanwhile, the an annual budget to fight HIV/AIDS. The com- province increased user fees under local control to mittee also organized a provincial Steering Com- reflect the actual unit costs of providing services. mittee on HIV/AIDS headed by the Department Task forces developed strategies for improving of Health, under an umbrella Steering Committee service quality based on consumer surveys; pur- on HIV/AIDS, Drugs, and Prostitution Control chasing equipment; developing accountability 164 East Asia Decentralizes mechanisms based on focus groups and the missing models--including political dividends and complaint resolution system established during other signals--appears to be holding back needed the financial crisis; and improving health work- policy interventions by local decision makers. force management, and submitted a tighter organi- zation structure to the provincial government for Health Care and Intergovernmental consideration. Fiscal Challenges The province launched the new system in 2003 by paying the premium to enable the poor to use A country's intergovernmental fiscal system should public facilities. The program became available to meet the complex goals of its health system, as with the nonpoor in 2004, competing with private other public services. The key to an effective fiscal providers in providing a benefit package. The system is "finance follows function." The intergov- approach in Yogyakarta gives districts a key role, as ernmental fiscal system must also usually address specified in Law 22, but also responds to the loss of horizontal as well as vertical equity, key relation- economies of scale that may make health services ships between levels of government and jurisdic- ineffective and inefficient. The province encourages tions, and incentives for collaboration. Weaknesses cross-district collaboration, especially in upgrading in the design of each country's fiscal system have technical support, sharing medical and technical had important consequences for the delivery of specialists and trainers, and organizing communi- health services. cable disease control, quality assurance, and health In Vietnam, the budgeting and financing system education and advocacy. The Joint Health Council formalized in the 1996 Budget Law revealed such facilitates such activities, with task forces making challenges: recommendations. But the provincially staffed Technical Review Team plays the largest role by · Norms emerged during the 1990s to determine reviewing district proposals and providing feed- almost every kind of input into the system. For back and guidance. example, gaps between provinces in per capita A number of other provinces are closely watch- health spending reflect a complex fund alloca- ing and applying this approach. Central agencies tion based on population norms and allocations have not been deeply involved, although Yogyakarta to and within sectors. These take into account sought their guidance on establishing standards for differing geographical conditions between its regulatory framework. provinces but do not offset revenue and cost dis- Overall, the health impacts of decentralization advantages and variations in need. The Ministry are not easy to estimate. Few data indicate signifi- of Finance may also rely on other criteria during cant windfalls in health benefits linked to decen- the budget negotiating process. tralization. The early phases were not incompatible · Provinces have substantial discretion in allocat- with sustaining impressive overall improvements in ing resources to districts and communes, and health status, and decentralized governance opened the methods they use vary considerably. Dis- the way to promising local initiatives in health tricts have little autonomy, and interdistrict planning, service delivery, and financing. However, variations are significantly greater than at the much better results would seem to be within reach province level. through policy adjustments. · The norm-based system relies on flows of accu- Identifying appropriate policies--though not rate data. For example, user fees and insurance easy--is critical. In the Philippines, central agencies, reimbursement rates reflect costs and expendi- often in partnership with NGOs, have documented, tures in different settings. Besides the high cost disseminated, and advocated best practices in local of regularly updating such information, these public services through the media, educational trips figures are unlikely to capture variations linked for local officials, and various training programs. to scale and quality as well as discretionary ele- Despite these initiatives, however, the speed of ments. What's more, because the steps to create innovative practices has been limited, and the over- fee schedules are cumbersome, time consuming, all level and quality of local health services have and costly, the schedules remain in effect for barely improved. A lack of incentives rather than years; that in use today dates from 1995. Decentralizing Health: Lessons from Indonesia, the Philippines, and Vietnam 165 · Despite formal autonomy at lower levels of gov- problems, and to provide financial assistance to ernment, the norm-based system constrains local governments unable to maintain health serv- flexibility. Civil service salaries have first call on ices or meet their Magna Carta obligations because funds and absorb most expenditures. Provinces of inadequate resources. DOH also implemented a with fewer local revenues have less flexibility. conditional matching grant program, the Compre- · The norm-based process limits sectoral inter- hensive Health Care Agreements, intended to ventions. The Ministry of Health did not fully secure local funding for devolved functions and participate in budget discussions and lacked core public health programs. This reflected an detailed information on expenditures by important part of the country's strategy of using provinces and lower levels of government. The incentives and disincentives to achieve national ministry also could not assess whether actual objectives in a decentralized system. spending by lower levels was consistent with sec- However, the relationship between service deliv- toral policies. ery and financing arrangements entailed significant · Limited investment in local facilities, such as weaknesses. For example, devolution of public upgrades to community health centers, also af- facilities led to fragmentation of the hospital refer- fected quality of care,and more patients bypassed ral system. Under the new regime, each hospital or such facilities as a result. clinic primarily serves the constituency of a local · The Ministry of Health focused on allocating government. Several provinces therefore reduced funds to national programs aiming at combat- budget appropriations to urban hospitals and ing high-priority diseases such as tuberculosis. channeled resources to less-well-off municipalities, Some of these programs were particularly in the process raising the average cost of urban important to disadvantaged groups and imple- services. Instead of cofinancing these facilities with mented mainly through local governments. the provinces, many cities opted to refurbish their However, no mechanisms ensure that once own clinics or build enclave hospitals. Further, national goals are met, those programs are weak monitoring of local compliance with Com- discontinued. prehensive Health Care Agreements did not help · Reliance on norms has also discouraged creation ensure financing of the devolved services. of medium-term planning frameworks that facilitate recognition of tradeoffs and set priori- Ensuring Equity ties between and within sectors. In Vietnam, two factors have undermined the dis- The December 2002 budget law, which took tribution of heath services to the poor. First, the effect in January 2004, gives more discretion to sub- central government has not targeted resource flows national governments. Province-level People's to poorer provinces, concentrating instead on the Councils have more power and a greater obligation supply side by improving multitiered service deliv- to prioritize health spending, determine allocations ery. The government has taken demand for services and transfers to lower tiers, implement policy, and largely for granted and has not weighed it heavily in mobilize resources. The fact that the central gov- policy making, at least until recently. ernment establishes three-year transfers once it Second, longstanding funding shortfalls contin- reaches agreement with provinces on expenditure ued through the decentralization process. Starting plans may allow the Ministry of Health to influence in 1989, hospitals in Vietnam were allowed to col- allocations across functions and service levels. lect user fees and mark up drug prices, and the In the Philippines, the primary fiscal vehicle resulting revenues became, and still are, a sizable supporting decentralization is the Internal Revenue source of health financing. However, user fees were Allotment (IRA), which transfers funds to local a disincentive to enhanced utilization by the poor. governments. As noted, most local governments And with user fees only partially offsetting funding depend heavily on this source, as do devolved gaps, lower quality followed. All this led to reduced health services. The central Department of Health use of health services from the late 1980s, with (DOH) also created the Local Government Assis- demand often shifting to "private" providers rang- tance and Monitoring Service to manage transition ing from retired government doctors to informal 166 East Asia Decentralizes drug vendors. These developments likely exacer- regions, and only cities have generally robust bated variations in health indicators by region and economies. Most provinces and municipalities rely income group, with poorer areas such as the north- heavily on fiscal transfers, principally the IRA. ern uplands recently falling further behind. However, the IRA formula favors highly populated Some cities and provinces reportedly reduced local governments and those with large land areas, user fees charged to the poor and other groups. and so does not ensure an overall propoor bias in Recent findings show that distribution of central health services. Studies have also shown that other and local budget funding, official development fiscal transfers, including those administered by the assistance (ODA), and health insurance reimburse- central Department of Health, correlate only ments among provinces benefit the poor dispro- weakly with poverty, with poor regions appearing portionately more than do other sources of to have received lower DOH budget allocations in province-level funding. However, only the ODA is 1994 and 1997. On the other hand, one of the rich- strongly propoor. Neither central and local budget est regions in the country receives a disproportion- funding nor ODA relates significantly to province- ate amount of the DOH budget (Mercado 1999; level measures of health needs, household poverty Capuno 2002). rates, or the percentage of minorities. The distribu- As in Vietnam, government hospitals in the tion of public health expenditures among provinces Philippines may collect user fees and impose up to is weakly propoor, thanks largely to the state budget a 30 percent markup on drugs. However, cost- and ODA (Knowles et al. 2003). recovery rates remain low because of the inordinate Decision 139, issued in October 2002, further volume of charity and subsidized patients. In the requires each province to set up a Health Care Fund case of provinces, for instance, the combined share for the Poor to finance free health care for disad- of hospital fees never reached 13 percent of total vantaged groups, with budgetary support from the hospital outlays from 1992 to 2000. Municipalities central government. Decision 139 entails a major fared better, with the share of fees in total hospital increase in health spending in Vietnam amounting outlays rising from 9 percent in 1992 to 29 percent to D 700 billion (some US$0.5 billion) per year. in 2000 (Capuno 2002). But because most hospitals This program is starting slowly to allow the country were devolved to provinces, the unintended result to overcome difficulties in identifying the poor and of low cost-recovery rates is that many hospitals are channeling funds to poorer provinces. The pro- poorly maintained, understaffed, and ill-equipped. gram may improve health access in remote areas as Vietnam and the Philippines instituted health it does not cover the indirect costs of care. Along insurance schemes in 1993 and 1995 that target the with Decree 10, which gives hospital managers poor. Health insurance in Vietnam has become a much greater control over their pay and employ- significant financing source, more important than ment, user charges, and use of surplus funds, Deci- in the Philippines (Knowles et al. 2003). Neverthe- sion 139 represents a shift in that the Ministry of less, in Vietnam coverage is still low and mainly Health is moving from direct service provider to includes civil servants and others employed in the sectoral steward, directing central resources to the formal sector. Decision 139 represents a potentially poor and other vulnerable groups based on clear significant scaling up of the number of people with definitions of eligibility. The directives also imply insurance. differentiation of government roles, with provincial Records from the Philippine Health Insurance health departments organizing delivery of care and Corporation (PHIC) show that among those Vietnam's insurance agency responsible for collect- enrolled in the national health insurance program, ing contributions and purchasing services. the number of paying members, from both the This change is important because while the public and private sectors but excluding the insured main clients of devolved health services in all these indigent families, grew from about 5.57 million in countries are the rural poor, their access to quality 1999 to 7.62 million in 2001. This suggests that health services is highly uneven owing to wide vari- nearly four in eight Filipinos have social insurance ation in local revenues and the flawed design of coverage, but that the program is still far from fiscal transfer programs. In the Philippines, for achieving its target of universal coverage. However, example, economic growth remains uneven across since 2000, the PHIC has been aggressive in Decentralizing Health: Lessons from Indonesia, the Philippines, and Vietnam 167 enrolling indigent families under its Medicare para which supplied temporary, contractual, and better- sa Masa (indigent program). As a consequence, the paid doctors to remote areas. In May 1993 the total number of indigent families enrolled has program began to deploy physicians to 271 munic- grown from 2,904 in 1997 to 1,762,116 in 2003. By ipalities lacking doctors, and by December 2003, June 2004, the total ballooned to 6,175,651 indigent 198 of these municipalities had received doctors. families. Whereas in 2001 about 37.8 percent of They receive an attractive package of salary and these indigent members were concentrated in the benefits for serving two years, and some also receive richest regions, by 2003 the same regions accounted honoraria and material support, such as free board for only about 19.91 percent of the total member- and lodging, from local governments. However, ship. The fact that other regions have gained signif- only about a third remain after their two-year tour icance suggests that a wider set of poor households of duty, discouraged by the lower pay and fewer now enjoys coverage. However, this trend is likely to privileges that accompany local employment. The slow as more local governments must copay with number of applicants to the program is dwindling the national government the insurance premium of owing to a surge in foreign demand, and conflict poor constituents. Most local governments see this areas remain underserved because of a lack of contribution as another unfunded mandate. incentives. To supplement the local health force, the Barangay Health Workers' Benefit and Incentives Health Care Personnel and Civil Act of 1995 provided for training volunteer work- Service Management ers and providing minimal incentives to convince Many local governments find it difficult to hire them to join barangay health stations. These volun- physicians, nurses, and medical technicians, who teers assist in clerical tasks and minor health proce- are in great demand in foreign markets. In the dures, such as weighing and measuring patients. Philippines, for example, local governments in However, these workers do not effectively cater to many areas where tuberculosis is epidemic have the health needs of the population. found it difficult to hire medical technologists and In Indonesia, the central government estab- rural physicians. Indeed, staff anxiety and opposi- lished the contract doctor (PTT) scheme in the tion were major problems during the transition to early 1990s to ensure a flow of doctors to remote decentralization in Indonesia and the Philippines. locations. Doctors hired after completing their ini- Though usually temporary, staff discontent can tial medical degree received substantial monetary affect the quality and quantity of personnel avail- incentives for practicing in more remote areas for able under decentralization. three years, as a condition of advancement. Special- In the Philippines, health workers were perhaps ists also had to complete compulsory assignments the largest group opposing decentralization. Many for one to four years or two to three years as PTT initially feared loss of job security, "politicization" staff. of their functions and positions, limited career Discontent grew over the obligatory nature of prospects, and lower pay.9 assignments, relatively low salaries, and poor To appease devolved workers, the central administration of program benefits. In 1999, regu- Department of Health pushed for the Magna Carta lations were eased to permit alternatives such as for Health Workers in 1992. Among other features, teaching in a medical school, working as a PNS this law provides for higher compensation and (civil servant) in designated areas, or working in extra benefits and allowances to all health workers, private practice as a clinic employee in remote including those devolved to local governments, and areas. Service requirements for very remote areas requires the latter to pay the additional compensa- were reduced to two years, and new graduates tion. This unfunded mandate would have demoral- could postpone mandatory service if they wanted ized other staff and made some rural physicians the to start specialist training. These changes did not highest-paid local public employees, earning more satisfy the PTT lobby, and doctors continued to than mayors, which they considered unacceptable. press to scrap the regulations. As a stopgap measure, the Department of Health PTT issues remained unresolved as decision instituted the Doctor to the Barrios Program, makers launched decentralization. The Ministry of 168 East Asia Decentralizes Health has been exploring new ideas such as allow- doctors (nearly all at commune health centers are ing medical personnel to serve in the military and upgraded former assistant doctors), and often lack the police. Districts, meanwhile, have stayed with enough staff with other training as well, except in the residual national system despite its flaws, as densely populated delta areas and near cities. For without central funds and guidance they might not example, in 1997, 26 percent of communes lacked have been able to integrate the numerous centrally an obstetric-pediatric assistant doctor or a midwife. assigned, locally based staff transferred overnight The Ministry of Health requires all communes to via Law 22. Moreover, few districts can turn down retain such an employee, reflecting the high priority central offers to recruit and assign PTT doctors accorded to local maternal and child care. using central funds. Still, district officials are con- Average monthly salaries of health staff have cerned that staffing policies that reflect local priori- remained essentially unchanged in real terms since ties and conditions have not been established, 1994. In 1998, the average monthly salary of a gov- including options to "right-size" staffs within each ernment health worker was merely US$29, even district. This issue arises especially in districts obli- though user fees supplement salaries somewhat. gated to handle staff oversupply left behind by This low pay, compared with the education sector, flawed centralized-era policies. has induced many health staff to seek additional Strong political and administrative leaders in sources of income, reducing the time, attention, some provinces have created master plans to and dedication they devote to their work (Dung reshape the bureaucracy to fit local conditions. et al. 2001). These include using downsizing mechanisms such as redeployment of staff, early retirement, volun- Service Delivery Mechanisms tary resignation with severance payments, and retraining to encourage entrepreneurship. How- Health programs are prime examples of the need to ever, implementation of these plans awaits full design institutional arrangements carefully to political commitment, facilitating legal steps, and ensure that parties in the service delivery process an injection of cash. The inability to proceed high- have the understanding, ability, and incentives to lights concerns voiced by district and province fulfill their roles. Coordination is invariably a cru- decision makers about dependence on central cial requirement of system effectiveness. While a government for salary payments and methods for country's central health ministry should take key "right-sizing." responsibility for controlling communicable dis- Vietnam confronted personnel issues under eases, it cannot do this efficiently and effectively decentralization as well, as the distribution of without the cooperation of local governments, as health personnel did not occur exactly as planned. the latter are at the forefront of service delivery. Enough doctors and other higher-level staff are In the Philippines, programs to combat commu- generally available in cities, but numbers are inade- nicable disease depend on devolved health person- quate in rural areas. Provinces with medical schools nel and local counterpart funds, which are in short have about the right number of staff, but poorer supply. Local governments see their participation as provinces do not, especially newer ones with no another unfunded mandate, and program coordi- secondary medical schools. One study showed only nation has suffered as a result. To elicit local sup- 1.7 doctors per commune in the North Highlands port, the central Department of Health under the and the North Central Coast, while a commune in Comprehensive Health Care Agreements (CHCAs) the Southeast Region averages 6.8 doctors (World matches each peso that a local government com- Bank 2001b). mits with a higher amount. However, the local gov- The number of health workers at the provincial ernment must first commit a minimum amount level is generally adequate (in relation to Ministry for its devolved health functions. This requirement of Health guidelines), although some provinces do has proved stringent, as many local governments not have enough specialists. But districts generally initially lacked the resources to finance these func- lack enough doctors who specialize in priority tions, much less meet their contractual obligations areas, such as obstetrics and gynecology and emer- for vertical programs. DOH also did not develop a gency surgery. Communes do not have enough monitoring and enforcement mechanism to track Decentralizing Health: Lessons from Indonesia, the Philippines, and Vietnam 169 compliance. Worse, many local officials believed To ensure drug quality in all public health facili- that strict compliance with the program was not ties, the central health ministry in all three coun- necessary, as DOH would always take the blame for tries has adopted drug formularies and drawn up public health failures (Esguerra 1997; Medalla an essential drug list. In Indonesia and the Philip- 1996). Hence, only after two rounds of implemen- pines, the central ministry even advocates and pro- tation, CHCAs were discontinued in 1997. motes generic drugs. However, these regulatory In the three study countries, integration between measures have not ensured the overall quality of programs within provinces as well as across drugs, owing to weak information campaigns and provinces remains poor. Subnational governments enforcement mechanisms. In the Philippines, for implement national programs separately, leading to example, many local governments, with support overlap and overload of grassroots health facilities. even from their own health officials, routinely buy Vietnam partly solved this problem by giving branded drugs because of their supposed proven provinces a greater role in setting goals, developing efficacy. Further, the Bureau of Food and Drugs, plans, and using funds for national targeted pro- which lacks laboratory and regulatory capacity, has grams. This is appropriate given variations in not convinced doctors of the supposed equivalence disease profiles across regions, and is said to have of generic drugs (Lim and Pascual 2003). In raised immunization rates and lowered fatality Indonesia, hospitals buy drugs and unbranded rates. products outside the essential drug list. Unlike in the Philippines, in Vietnam and Indonesia state-owned enterprises dominate the Low Quality and Unsteady Supply drug supply, as they can assure quality more easily of Drugs at the Local Level than a private drug market. In Vietnam, the state- The supply and quality of drugs at the local level owned VINAPHARM, which includes central and have become a concern owing to limited funds, provincial trading and manufacturing enterprises, deficient drug management systems, and loopholes is responsible for supplying drugs countrywide. in procurement rules. In Indonesia, provinces have The Drug Administration Department within the not been aware of or prone to intervene in drug Ministry of Health is responsible for overall drug supply, stocks, and use at the district level under management, supported by the Drug Quality Con- decentralization. Districts have been able to plan trol Institute and the Drug Inspectorate. In each for and purchase their own drugs based on stan- province a Drug Quality Control Department falls dard procurement practice. However, compliance under the Provincial Health Bureau, while a Drug with quality assurance procedures has been poor, Testing Center and Inspection Department moni- partly because responsibilities have not been clear tor drug quality in the local market. and districts do not have the technical capacity to In Indonesia, four state-owned enterprises pro- handle the task. duce generic drugs and vaccines. Regulatory func- In the Philippines, each local government simi- tions, including enforcement, are the responsibility larly manages its own system of drug procurement, of the Directorate General of Food and Drug Con- inventory, dispensing, and financing. The quality of trol, a unit of the Ministry of Health. Quality assur- locally procured drugs is generally poor, the pur- ance efforts include establishing the essential drug chase price is often higher than in private pharma- list; enforcing standards in the development, test- cies, stock shortages are frequent, and irrational ing, registration, manufacture, and distribution of drug use occurs. A principal reason is that local drugs; and overseeing health professionals. Work- therapeutic committees are not constituted, not ing through 26 province-level branches, the direc- functioning, or not well trained in modern drug torate monitors drug quality and safety through management. Local drug procurement is also follow-up visits and testing programs. The direc- corrupt in many places: bids are rigged, qualified torate bases inspection of manufacturers on criteria bidders are "preidentified," and bidders connive. for good practice adopted in 1971. Moreover, the supply chain extends only to urban State-owned enterprises impose their own inef- centers; poor outlying municipalities rely on itiner- ficiencies on the market. In Indonesia, these units ant drug peddlers who arrive infrequently. are protected by tariffs and limits on final-product 170 East Asia Decentralizes imports, constraints on foreign investment, and The change from central to district procurement restrictions on registering new drugs, opening new may have also increased drug prices because of lower pharmacies, and the nonpharmaceutical activities procurement volumes.This would likely widen vari- of retailers. Reforms adopted in the 1990s relaxed ations in drug prices, reducing equity and lowering some restrictions on foreign drug companies, the availability of orphan drugs. encouraged generic drug prescriptions in public Vietnam represents an interesting comparator. health centers, and enforced good manufacturing As in Indonesia, deregulation of pharmaceutical practice. Hospitals could also keep drug revenues to production and distribution brought heightened secure supplies at the facility level. activity among informal drug vendors and phar- However, inconsistencies and missteps weak- macy shops and greater availability of drugs ened or negated these pre-2001 reforms, and the throughout the country. Consumer purchases of outcomes were unsatisfactory. Moreover, districts drugs, especially for self-medication, grew as well, inherited a flawed and incomplete reform agenda, from 2.1 annual contacts per capita with drug ven- with impacts on government stewardship obliga- dors and pharmacy shops in 1993 to 6.8 by 1998. tions. Since Law 22 has taken effect, deviation from However, the two countries differ in their experi- established standards, patterns, and procedures has ences with drug prices. In Vietnam, deregulation grown. For example, complaints about the physical brought a 30 percent fall in the real price of medi- appearance or expiration date of drugs are wide- cines in the 1993­98 period, while in Indonesia spread, suggesting that longstanding quality assur- price reductions do not seem to have followed pol- ance procedures are not being observed. Reporting icy reforms. of drug quality problems is not formalized, and Moreover, the risks facing Indonesia are compa- procedures for addressing quality concerns are rable to those of Vietnam, where drug vendors unclear and complicated by multiple sources of account for roughly two-thirds of health service drugs and funding. Limited skills surely play a role contacts, and antibiotic resistance has reached epi- as well,as many service units and district warehouses demic levels thanks to excessive and otherwise depend on unqualified staff. Nor can provinces step inappropriate use.10 The resistance problem is in, as they lack the authority to monitor, yet alone compounded by the limited training of pharma- supervise, district drug procurement. cists and the low average education level of drug More generally, laws and regulations provide vendors and the public. Even when doctors little detail on operating responsibilities and prescribe drugs, compliance with appropriate have brought confusion and divergence between treatment guidelines is low. Oversight of health provinces and districts. For instance, some pro- providers is weak as well. Enforcement of the many curement procedures have spurred small purchases regulations and decrees governing minimum qual- from 15 or more suppliers. Districts usually reject ity standards and the protocols expected of health pooled procurement despite possible cost savings. providers through regular inspections of health Nor is there a definitive view on which drugs facilities is less than satisfactory. belong in categories defined by the Ministry of In the Philippines, on the other hand, a few drug Health in pre-2001 preparations. Central involve- manufacturers and importers, which are mostly ment appears to be limited, and provinces reveal no multinationals, dominate the upstream segment, common pattern of procurement. Some are not while a single drugstore chain effectively controls supplying any drugs, and plan to reduce future drug the retail segment of the domestic drug industry. A supply. Some still buy drugs to cover emergencies parallel drug importation policy has not helped and temporary district shortfalls. Districts are using bring down the overall price of drugs because the their own funds to buy drugs from all three classes of government chose to maintain an import monop- the essential drug list. Meanwhile, the drug supply oly with capital of just P=50 million--not enough to and regulatory system in hospitals is different from affect the multibillion-peso domestic drug trade. that at the primary care level. Hospitals, which have long been allowed to procure and dispense drugs Health Information Systems outside the essential drug list, are buying mainly branded drugs, funded through self-financing Decentralization in Indonesia, the Philippines, and revolving funds and using spot buying methods. Vietnam has fragmented the health information Decentralizing Health: Lessons from Indonesia, the Philippines, and Vietnam 171 system and undermined coordination among vari- Philippines and Indonesia continue to rely heavily ous sectors, and thus effective and efficient control on input indicators such as the number of beds, of communicable diseases. Subnational govern- floor area, and medical instruments, with only ments are unaware of their roles, and, more criti- infrequent verification of such information by local cally, lack the incentives and technical capacity to governments. assume those roles. Subnational governments need In the Philippines, although the Department of support for activities from collecting health infor- Health deploys its own representative to provinces, mation to providing further inputs to performing cities, and municipalities to help monitor disease overall health planning to actually implementing outbreaks and coordinate vertical programs, the programs. flow of health information remains slow. DOH From 1992 to 1995, the Philippine Department representatives must often double-up as service of Health implemented the German-funded Health providers, as many local governments lack the and Management Information System, whose main needed personnel. objective was to institutionalize a "need-responsive and cost-effective health information system" at the Performance Standards and Incentives national and local levels. The initiative introduced software modules and processes to fortify the pro- In the Philippines, efforts to improve the quality of duction and use of information. Besides developing health services have relied on both incentives, district-level health indicators, the system sup- including awards and accreditation measures, and ported innovations in community health care disincentives. One example of an approach that financing and service delivery. However, the initia- encourages local governments to upgrade their tive stopped short of a nationally integrated but health services is the Sentrong Sigla accreditation. locally operated health information system and was Aside from providing a mark of quality, this accred- not sustained. itation originally conveyed a P=1 million grant to In Indonesia and the Philippines, the central local health centers. Based mostly on input indica- health ministry relies on local governments to tors of a health facility's "readiness to provide serv- report information voluntarily. This has resulted in ices," the program had certified some 48 percent of erratic or delayed submission and poor-quality health centers, 14 percent of district and provincial data. The devolved staff members who were hospitals, and 3 percent of barangay health stations responsible for such data under the old regime now by October 2003. Though these numbers are supply information on health expenditures and encouraging, they represent only a minority of the input indicators to provincial and lower-level country's facilities. Moreover, most of the cash elected officials, who are less concerned with out- awardees are better-off municipalities that do not puts and outcome indicators. need the funds as urgently as localities that do not In Indonesia, only 36 percent of health centers qualify (Lamberte 2003). reported infectious disease surveillance data in In the Philippines, the Department of Health 2002.11 Those that do report do so irregularly or issued Administrative Order no.100 in 2003, which late. Thus, the limited data that flow through the established new guidelines to improve the Sentrong system may not be reliable enough for use in plan- Sigla program. Instead of cash awards, the new ning, policy analysis, or evaluation. guidelines specify a matching grant for new quali- In these countries, the central health ministry, fiers, and also make certification a prerequisite for lacking information, is less able to monitor the other DOH grants and the Capitation Fund pro- quality of laboratory services, hospitals, and other gram, recently introduced by the Philippine Health devolved services. This is worrisome as, for exam- Insurance Corp. Under the latter program, an ple, local health centers in the Philippines are accredited local government can claim reimburse- responsible for both finding cases of tuberculosis, ment for services extended to PHIC-insured indi- which requires sputum examination, and monitor- gent families in their localities. However, these ing cases. Indonesia discontinued some programs accreditation schemes are voluntary. such as those tracking leprosy because districts did Vietnam and Indonesia, in contrast, have not monitor the number of cases. Quality assur- adopted norm setting as their approach. Vietnam's ance systems for provincial hospitals in both the Ministry of Health sets province-level norms, but 172 East Asia Decentralizes quality remains uneven because province and dis- health status are therefore greater in well-off trict health officials introduce norms and guide- provinces, and service innovations have failed to lines as well. The Ministry of Health has used spread beyond areas where the local economy is decrees and circulars to define the quality of robust and the political situation relatively stable. human resources and equipment and performance In these provinces, local governments have had standards. The ministry also has issued more than the resources to meet growing demand for health 100 treatment guidelines, though a survey con- services. ducted in 20 district hospitals in 2000 on acute res- Weakness in decentralization policy also con- piratory infections showed that compliance was 25 tributed to lower-than-expected health payoffs. to 40 percent, probably reflecting weak support and These include ambiguities in goals, lack of detailed supervision at the local level. Meanwhile, overuse design, inconsistency with other policies, and and overprescription of injection drugs were com- poorly thought-out implementation strategy. mon, with representatives of drug companies influ- These follow from the fact that health was not the encing physicians' prescriptions (Dung et al. 2001). main--much less the sole--driver of decentraliza- Indonesia established minimum health service tion. The Philippines, for example, included health standards by ministerial decree in 2003. Districts services in its decentralization strategy only when must deliver services according to local needs in resistance from the education lobby forced legisla- 32 areas, including immunization, nutrition serv- tors to look at other national expenditures. ices, prevention of communicable diseases, and Inconsistent priorities have translated into curative care. Such standards will help define the inconsistencies in policies and poor design of pol- service levels that districts are accountable for icy instruments, especially the intergovernmental delivering. Whether these standards will be require- fiscal transfer system. Local governments are also ments or targets needs further consideration, along typically unaware of the types and timing of with measures for dealing with districts that do not national interventions--information that is crucial meet the standards. to their own budget and investment planning. With prior knowledge of available grants, technical assis- tance, and other support from national agencies, Toward a New Role for Ministries local governments can use information on local of Health needs, and proximity and direct accountability to Decentralization in Indonesia, the Philippines, and beneficiaries, more effectively. Vietnam may help sustain overall improvements in In Indonesia, the slow and arduous emergence health that have occurred during the last two of a consensus on a health decentralization frame- decades. Decentralization has appeared to spur work partly reflects a governmentwide determina- local initiative in planning, delivering, and financ- tion to avoid service interruptions. But this focus ing services. Users are now participating in plan- has also allowed the central government to post- ning in many regions, leading to more appropriate pone difficult decisions over the role and scale of and better-targeted health services. Volunteers key central ministries. supplement limited local financial and technical The Indonesian Ministry of Health also tends to resources. More important, perhaps, citizenship view the public as passive service recipients rather and trust in local government have deepened. The than discriminating customers, owners, and poten- resulting efficiency gains and social capital support tial allies, and to present itself as a policing and the decentralization of health services. standards-upholding authority rather than a tech- Still, experience in these three countries reveals nical agency. What's more, the government has that decentralization dividends have been modest not yet developed a clear concept of the role of for two reasons. First, these countries decentralized provinces in the health system. Decision makers health services in less-than-favorable environ- know that districts are typically too small to sup- ments. Inequitable economic growth, population port cost-effective programs, but they have not pressures that brought epidemiological changes, designed institutional solutions to encourage joint and political uncertainties have limited the poten- service areas, or to make provinces agents of public tial gains from decentralization. Improvements in health and related programs. Decentralizing Health: Lessons from Indonesia, the Philippines, and Vietnam 173 Policy weaknesses also stem from laws and regu- diseases but also workforce training, recruitment, lations, introduced in Big Bang fashion, that lacked pay and benefits, and supervision. detail on functional and operational responsibili- Other core public health functions include ties and brought confusion and divergence between ensuring that the poor have access to affordable provinces and other local governments. The Min- care, overcoming micronutrient shortfalls, creating istry of Health also failed to coordinate with local sustainable funding arrangements, acting as a governments and other actors in performing criti- source of ideas and best practices from the cal health functions. provinces, and providing technical assistance on a In these countries, reactive responses to transi- selective basis. As the steward, a Ministry of Health tion problems did not necessarily resolve fundamen- would build consensus on national health objec- tal design issues. For instance, in the Philippines, the tives and standards, and coordinate rather than Magna Carta for Public Health Workers--which require local governments and civil society groups provided supplemental funds to only a few local to meet these goals. Instead of relying on some- governments--temporarily appeased disgruntled times heavy-handed regulation, the ministry health workers but upset local governments by should align incentives to elicit the cooperation and imposing unfunded mandates. Similarly, the coun- participation of all sectors. Rather than impose try "resolved" the mismatch between the distribu- high standards, the ministry should perhaps pro- tion of the IRA and devolved expenditures after 1992 mote them through advocacy and by strengthening by providing grants to cities for hospitals they were local governance mechanisms. already financing before 1991. Thus, compensating The stewardship role also entails pushing for cities for their supposed "losses" due to the adjust- greater consistency among goals, programs, and ments in the IRA was a politically necessary but policies of different national agencies to support costly way of ensuring adequate funding for the local governments. Finally, stewardship includes health functions devolved to provinces and munici- more than content: it clearly entails leadership and palities (Capuno 2001). a flexible, opportunistic mode of building partner- However, efforts to correct these weaknesses ships and exploiting opportunities. could expand the gains from decentralizing health, Central interventions are warranted partly even within a less-than-favorable environment. because these functions may not convey the Toward that end, central health ministries must urgency or tangible appeal of disease-specific pro- focus on specific tasks such as setting up quality grams, and thus districts may neglect them. Subna- assurance mechanisms for drug supplies, safe- tional governments have little incentive to pursue guarding access to medicines by the poor, and dis- core public health functions because they cannot mantling state monopolies on drugs. Responsibili- fully capture the returns, and because some func- ties for communicable diseases include monitoring tions are difficult to perform well because of national and regional trends, supporting laboratory limited resources or lack of scale economies. The capacity and quality control and assurance, alerting impacts of core public health functions are also provinces to outbreaks elsewhere, and advocating hard to measure: gauging the effects of a strong dis- for emergency financing. ease surveillance and reporting system is difficult, while the direct distribution and use of drugs by infected patients has obvious benefits. With the Stewardship of the Ministry of Health These examples illustrate the contributions needed Has the Philippines Turned the Corner? from central government during decentralization of health services. Above all, central agencies The central health ministries in all three countries should concentrate on activities that go beyond the have taken steps to fulfill this new role, but no direct provision of preventive and curative services, definitive transformation into stewardship has yet focusing on core public health functions, respond- occurred. Furthest along is the Philippines, whose ing to overall imperatives, and preventing potential Department of Health has examined whether failures.12 These efforts include not only tasks decentralization has paid off as a health reform related to pharmaceuticals and communicable vehicle. 174 East Asia Decentralizes In 1999, under the Estrada administration, guided by HSRA. For example, the country can DOH formulated a comprehensive decentralization do more to prepare the intended beneficiaries, such strategy called the Health Sector Reform Agenda as local governments and health care users, and (HSRA). The HSRA noted a slight resurgence of diffuse political resistance. At the same time, the certain diseases and persistent inequities in service central Department of Health needs to extend its access. To counter these problems, DOH posi- partnership with health NGOs and civil society tioned itself as a health leader, enabler, and capac- organizations with whom it is already working. ity builder, administering only certain services Local health finances must rest on a firmer footing, (Department of Health of the Philippines 1999b). including through greater reliance on local funds. As a leader, it would primarily be responsible for Carefully designed user charges would not only setting national health policy and regulations and make service delivery more efficient but also make strengthening regulatory agencies. As an enabler local health programs sustainable and help subsi- and a capacity builder, it would seek to promote dize the health needs of the poor. But to justify innovations and standards in health services, espe- higher user fees, local governments must improve cially at the local level. And as an administrator, it service quality and require up-front financing would confine itself to pushing hospitals toward for facility improvements, personnel training and fiscal autonomy, securing funds for priority public hiring, and drugs and medical equipment. DOH health programs, and promoting universal cover- matching grants could support enhanced services if age under the National Health Insurance Program. local governments introduced new fee schedules. The novel aspects of this strategy are DOH Grants would also make local public employment reengineering and the convergence of all DOH more attractive to health workers. interventions in each province under the HSRA framework. DOH reengineering meant streamlin- Waiting for Indonesia... ing operations, finances, and bureaucracy, and deploying 1,638 personnel from the central office Indonesia has not clarified the health roles and to regional health offices, retained hospitals, and responsibilities of central and lower governments other DOH agencies. Under the convergence strat- after three years of decentralization. Nor has the egy, some provincial officials have drawn up health country moved to emphasize core public health development plans and interlocal health zones, functions, or seen marked improvements in spe- with the DOH providing technical input and other cific areas such as infectious disease control, phar- assistance. These zones bring together contiguous maceuticals, and human resources. Sectors besides local governments around a district hospital to find health also have indeterminate policies, prompting ways of improving the hospital referral system, advice to clarify assignments across levels of gov- exploiting economies of scale, and containing ernment and sectors (World Bank 2003a). spillovers. Two strands of thinking on decentralization are In December 2004, for example, Capiz province evident within the Ministry of Health. The first is devised a five-year development plan for enrolling distilled in a 2003 decree that lists 29 strategic issues indigent families in the National Health Insurance related to core public health functions and adds key Program, upgrading selected hospitals, and adopt- steps to address them, such as minimum service ing revolving drug funds and new drug manage- standards (MSS), partnerships with NGOs, and ment systems, with specific targets and activities services for the poor. The decree points to account- at provincial and zone levels. With initial DOH ability mechanisms and traditional command- support of P=10 million, the Capiz plan is expected and-control instruments to limit the risks of service to yield gains from economies of scale in hospital disruption. The former include the use of MSS to operations, pooled procurement of drugs, and elicit district commitment, including assistance in control of epidemics. Similar arrangements are funding core public health functions. The Ministry expected in 2005 in other sites such as Pangasinan, of Health has been relatively assertive in exerting Agusan del Sur, and Misamis Occidental. its authority in responding to infectious disease As a dynamic process, decentralization in the outbreaks such as severe acute respiratory sys- Philippines will continue to require adjustment tem (SARS), and overseeing surveillance of and Decentralizing Health: Lessons from Indonesia, the Philippines, and Vietnam 175 programs to combat diseases of national impor- provinces rationalize staffing numbers. The min- tance and involving international obligations, such istry could also sponsor trials of approaches to as tuberculosis and HIV. The Ministry of Health attract doctors, especially specialists, to remote and depends on central and donor funding to achieve undesirable locations. The ministry also needs to these ends, though each is unreliable, and has develop standards that provinces and districts can looked for district support, seeing MSS as targets use to license service providers; work with profes- for district spending.13 The decree assigns key sional associations to strengthen quality improve- responsibility to district chief executives, and states ment efforts and establish partnerships for profes- that efforts to attain MSS should rely entirely on sional development; consult with consumer groups district budgets, with central and provincial gov- and hospitals on workforce quality; focus on trends ernments providing technical help, supervision, in medical education; and fund and deploy special- and oversight. purpose health teams. This approach is risky. Detailed, extensive MSS Opportunities, imperatives, and stakeholder could undermine decentralization, and poorer dis- pressures could support the Ministry of Health's tricts could reject them because of limited fiscal impetus toward devolution and health reform. capacity. Most MSS have been set at high levels, Budget constraints may force the ministry to look imposed on rather than owned by local govern- to districts and provinces as sources of funds and ments, with the means of enforcement and penal- jobs for the health workforce. Demands for better ties for noncompliance undisclosed. MSS would be service quality and other public pressures have better seen as medium-term goals rather than per- begun to register with local political leaders and formance requirements that trigger funding and within the national ministry. A medium-term sce- require enforcement. The Ministry of Health needs nario resembles that in the Philippines: contested to develop ways of boosting district ownership of decentralization followed by a faster pace and efforts to prevent and control infectious diseases. major adjustments, leading to consolidation and The second strand of thinking within the mid-course correction. The Filipino experience Indonesian Ministry of Health takes a more benign most relevant for Indonesia is arguably the change and constructive view of decentralization. This in outlook within the Department of Health regard- approach is embodied in initiatives under way in ing its role in a decentralized system. Yogyakarta and three other provinces (Lampung, North Sumatera, and West Java), with twenty-one Stewardship by Vietnam's Ministry of Health other provinces due to come on stream later. Min- istry officials who support this approach are trying The tempo of change in Vietnam is quickening, to use the momentum of decentralization as a cata- with several distinct drivers governing the pace. lyst for sector reform, with provinces playing an The first is the reappearance of deadly public health important mid-level role. However, the above-cited threats at the top of the policy agenda. Vietnam is decree limits provinces to backstopping central and clearly vulnerable to new and more virulent strains district-level initiatives. of diseases such as SARS, HIV/AIDS, and influenza, The province-based approach remains new and and reemerging diseases such as tuberculosis and under trial. It has already survived early bureau- dengue fever. Malaria remains a major public cratic and other challenges, but the Ministry of health problem in mountainous and ethnic minor- Health needs to carefully assess experiences, impli- ity areas. cations, and lessons emerging from the province- Successive crises have spawned rapid-reaction based framework and disseminate them to key structures and shown the importance of timely and stakeholders. The ministry can also help implement well-targeted responses guided by updated disease new and existing provincial programs, including surveillance data. The Ministry of Health appears interventions that widen and deepen the decentral- to be developing expertise in explaining disease ized approach. The ministry can also support pilot challenges and engaging the public and political work and research aimed at helping provinces leaders while soliciting various sorts of assistance. respond to the diverse challenges of managing and The ministry has also gained credibility and built developing the health workforce, such as by helping stronger ties to decision makers at the provincial 176 East Asia Decentralizes level and in key central ministries, including the replacing allocation norms and hospital payment Ministry of Finance, the Ministry of Public Infor- mechanisms with instruments based on the price of mation, and the Ministry of Labor, Invalids and health care services. The ministry would like to pre- Social Affairs. These drivers of change could inter- pare expenditure norms to support management, sect if, as seems likely, the revised agenda on com- monitoring, supervision, and control functions, municable diseases leads to requirements for more and explore the use of norms that reflect population spending. This effort should involve a review of needs and improve equity in service access and use. arrangements for funding disease control and The Ministry of Health also recognizes that it the possibility of consolidating such outlays. A needs other policies with significant near-term related issue is the need to avoid substituting for impacts to address disparities in health outcomes local expenditures; the ministry could make a and per capita health expenditures across provinces. case for requiring matching contributions from Per capita spending in the richest seven provinces local governments. is over three times that in the poorest seven The 2002 law requires some acceleration in provinces. Central and donor transfers do not pro- efforts to adjust government roles. These efforts vide a counterweight, as the richest provinces will entail significant shifts, and the experience and receive the largest per capita amount, and because credibility accumulated in fighting SARS and other the resources involved are relatively small. diseases could prove helpful. In particular, the advent of provincially man- Endnotes aged service delivery suggests the need for formal 1. No consensus has developed on the starting point for recognition backed by real authority and resources, health decentralization in Vietnam. This chapter treats the 1996 Law on State Budget as a path-breaking measure. with the Ministry of Health focusing on key stew- 2. On December 27, 1997, then-President Ramos issued ardship functions. Implementation of the 2002 law Administrative Order 372, which effected the withholding will thus enable the ministry to get out from under of the "amount equivalent to 10 percent of the IRA." Local second- and third-best aspects of the de facto governments challenged this order before the Supreme Court and won in June 2000. health decentralization system that took hold in the 3. This is a common observation by local officials interviewed 1990s: for the Rapid Field Appraisal of Decentralization (Associates in Rural Development 1993a, 1993b, 1994). 4. The standard deviation in regional IMR also fell from 8.51 · Provinces are supposed to provide updates on in 1980 to 4.84 in 1990, and then to 3.34 in 2000. how they are allocating their recurrent budget. 5. Figures from the Philippine National Health Accounts also However, this requirement appears to be show that total per capita health spending rose from US$20.82 in 1991 to US$29.79 in 2001. In real terms, largely a formality. The Ministry of Health has however, the country recorded a per capita decline from little information on health budgets, and it is US$12.15 in 1991 to US$8.84 in 2001. not clear to what extent, if at all, it can influ- 6. Vietnam also achieved a hefty threefold nominal rise in ence provincial spending of budgets already total health expenditures over a five-year period, from US$0.68 billion in 1993 to US$2.17 billion in 1998 approved. (Knowles et al. 2003). · The ministry lacks a clear role in formulating 7. This occurs through the JPS-BK (the health component and assessing policy and determining central of the social safety net); the fuel subsidy; and the DAK (a special-purpose grant) channels. allocations to health; the ministries of Finance 8. The Galing Pook Foundation presents awards to local gov- and Public Information are the key agencies in ernments for the best innovations in public service deliv- this process. Central recurrent health spending ery. The Philippine Human Development Network uses the Human Development Index to identify and honor reflects projected revenue growth and recurrent provinces that made the greatest strides in promoting expenditures. health, wealth, and education. · Such incremental budgeting is not sensitive to 9. Moreover, the Department of Health itself did not antici- the goals and priorities set by the Ministry of pate these issues, as the initial plan was to decentralize edu- cation first. Opposition from that sector led the govern- Health. ment to consider decentralizing health instead (Diokno 2003). The ministry can respond to the 2002 law partly 10. Data from the 1998 Vietnam Living Standards Survey indi- cate that 93 percent of all drug vendor contacts entailed by strengthening budgeting procedures as well as efforts to obtain medicines without a prescription, with improving allocation. These efforts may include little variation across economic groups. Decentralizing Health: Lessons from Indonesia, the Philippines, and Vietnam 177 11. A survey conducted by the National Family Planning Coor- Working Paper No. 94.11. Cambridge, MA: Harvard dination Board in 2002 found that less than 10 percent of University, Center for Population and Development Studies. health centers followed the manual on preventing infec- Committee for Population, Family, and Children. 2003. "Demo- tions that may result from the use of contraceptives. The graphic and Health Survey 2002." Hanoi: Population and survey also discovered that counseling in the family plan- Family Health Project. ning program was poor, and that 20 percent of public facil- Communist Party of Vietnam. 1993. 7th Congress's Resolution ities had never been supervised. of Vietnam's Communist Party. Hanoi: Communist Party of 12. The Pan American Health Organization established 11 Vietnam. essential public health functions through international Department of Health of the Philippines. 1997. "An Assessment consensus. These have been field-tested and implemented of the Devolved Health System in the Philippines, 1996." in 43 countries of Latin America, the Caribbean, and Manila: Department of Health, Local Government Assis- Europe. tance and Monitoring Service. 13. Central spending is limited and subject to strong competi- ------. 1999a. National Objectives for Health, 1999­2004. HSRA tion from outside and inside the Ministry of Health. For Monograph Series No. 1. Manila: Department of Health. example, during 2001 and 2002, the central budget was ------. 1999b. Health Sector Reform Agenda, 1999­2004. HSRA highly constrained, and the ministry's main funding initia- Monograph Series No. 2. Manila: Department of Health. tive aimed not to control disease per se but to hire doctors ------. 2000. Department of Health Strategic Plan 2000­2010. to fill vacancies. The ministry allocated some funds for Manila: Department of Health. combating tuberculosis and HIV and a few other pro- Diokno, B. E. 2003. "Decentralization in the Philippines after grams, leaving insufficient central resources for tackling Ten Years: What Have We Learned? What Have I Learned?" other diseases. Discussion Paper No. 0308. Quezon City: University of the Philippines, School of Economics. Bibliography Dollar, D., P. Glewwe, and J. Litvack, eds. 1998. Household Welfare and Vietnam's Transition. Washington, DC: World Associates in Rural Development. 1993a. Synopsis of Findings: Bank. Second Rapid Field Appraisal of Decentralization. Reports Dung, P. H., et al. 2001. "Improving Physicians' Prescription at submitted to the USAID-Governance and Local Democracy Outpatient Department of District Hospitals." Hanoi: Project. Manila: Associates in Rural Development. Health Strategy and Policy Institute. ------. 1993b. Synopsis of Findings: Third Rapid Field Appraisal Esguerra, E. F. 1997. "Pursuing a National Health Strategy in a of Decentralization. Reports submitted to the USAID- Decentralized Fiscal Regime." Philippine Review of Econom- Governance and Local Democracy Project. Manila: Associates ics and Business 24 (2). in Rural Development. Fritzen, S. 1999. "Decentralization, Disparities, and Innovation ------. 1994. Synopsis of Findings: Fourth Rapid Field Appraisal in Vietnam's Health Sector." of Decentralization. Reports submitted to the USAID- ------. 2002. "The `Foundation of Public Administration'? Governance and Local Democracy Project. Manila: Associ- Decentralization and Its Discontents in Transitional ates in Rural Development. Vietnam." Paper presented at the Asia Conference on ------. 1995. Synopsis of Findings: Fifth Rapid Field Appraisal Governance in Asia: Culture, Ethics, Institutional Reform, of Decentralization. Reports submitted to the USAID- and Policy Change, City University of Hong Kong. Governance and Local Democracy Project. Manila: Associates Galing, Kaban. 2001. The Philippine Case Bank on Innovation in Rural Development. and Exemplary Practices in Local Governance. Galing Pook ------. 1996. Synopsis of Findings: Sixth Rapid Field Appraisal Foundation. of Decentralization. Reports submitted to the USAID- Government of Socialist Republic of Vietnam. 1999. Decree Governance and Local Democracy Project. Manila: Associates 29/CP. in Rural Development. Gwatkin, D., et al. 2000. Socioeconomic Differences in Health, ------. 1998. Synopsis of Findings: Eighth Rapid Field Appraisal Nutrition, and Population in the Philippines. Washington, of Decentralization. Reports submitted to the USAID- DC: World Bank. Governance and Local Democracy Project. Manila: Associates Knowles, J. C., N. T. Hong Hoa, D. B. Huong, N. Khang, T. T. Mai in Rural Development. Oanh, N. K. Phuong, and V. Ngoc Uyen. 2003. Making Bautista, M. C. G., M. E. C. Yap, and E. S. Soriano. 1999. "Local Health Care More Affordable for the Poor: Health Financing in Government's Health Financing Initiatives: Evaluation, Vietnam. Hanoi: Vietnam Ministry of Health and Asian Synthesis, and Prospects for the National Health Program in Development Bank. the Philippines." Bethesda, MD: Abt Associates. Lamberte, E. E. 2003. Sentrong Sigla: The Challenge Ahead and Brinkerhoff, D., and C. Leighton. 2002. "Insights for Imple- Way Forward--Formative Assessment and Implementation menters: Decentralization and Health System Reform." Part- Review Study. Manila: Philippines Department of Health, ners for Health Reform, No. 1. Bethesda, MD: Abt Associates. Bureau of Local Health Development. Capuno, J. J. 2001. "Policy Reform under Decentralization: Legaspi, P. E. 2001. The Changing Role of the Local Governments Financing of Health Services in the Philippines." Regional under a Decentralized State: Some Cases in Philippine Local Development Studies 7: 109­28. Governance. Quezon City: University of the Philippines, ------. 2002. "An Analysis of Health Care Financing and Ser- National College of Public Administration and Governance. vice Delivery in the Philippines under Decentralization." Lieberman, S. 2001. "Decentralization and Health in the Paper prepared for the Philippine Public Expenditure, Philippines and Indonesia: An Interim Report." Procurement, and Financial Management Review. Manila: www.worldbank.org/eap/eap.nsf. World Bank. Lieberman, S., and P. Marzoeki. 2000. Indonesia Health Strategy Chen, Lincoln, and Linda Hiebert. 1994. "From Socialism to in a Post-Crisis: Decentralizing Indonesia. Washington, DC: Private Markets: Vietnam's Health in Rapid Transition." World Bank. 178 East Asia Decentralizes Lim, J., and C. Pascual. 2003. Toward a Pro-Poor Drug Policy in Finance Development Project, Monograph No. 2. Quezon the Philippines. Manila: World Bank. City: University of Philippines School of Economics. Litvack, J., and D. Rondinelli, eds. 1999."Fiscal Decentralization, Solon, O., A. N. Herrin, R. H. Racelis, M. G. Manalo,V. N. Ganac, Disparities, and Innovation in Viet Nam's Health Sector." In and G. V. Amoranto. 1999. "Health Care Expenditure Market Reform in Vietnam: Building Institutions for Develop- Patterns in the Philippines: Analysis of National Health ment. Westport, CT: Quorum Books. Accounts, 1991­1997." Discussion Paper No. 9908. Quezon Manasan, R. G. 1997. "Local Government Financing of Social City: University of Philippines, School of Economics. Service Sectors in a Decentralized Regime: Special Focus on Ter-Minassian, T., ed. 1997. Fiscal Federalism in Theory and Provincial Governments in 1993 and 1994." Discussion Practice. Washington, DC: International Monetary Fund. Paper No. 97-04. Makati City: Philippine Institute for Devel- Tuan Tran. 1995. "Historical Development of Primary Health opment Studies. Care in Vietnam: Lessons for the Future." Takemi Program in Medalla, F. 1996. "Co-Financing of Devolved Health Services: International Health, Research Paper No. 100. Cambridge, Lessons from the CHCA Experience." Paper submitted to the MA: Harvard University, School of Public Health. Health Policy Development Project. Quezon City: University WHO (World Health Organization). 1995. Decentralization and of the Philippines, School of Economics. Health System Change: A Framework for Analysis. Geneva: Mercado, R. G. 1999. "Regional Budget Allocation: A Policy WHO, Division of Strengthening of Health Services, Revisit." Discussion Paper No. 99-29. Makati City: Philip- National Health System and Policies Unit. pine Institute for Development Studies. ------. 2002. World Health Report: Reducing Risks, Promoting Ministry of Health of Vietnam. 1999. Health Statistic Year Book Healthy Life. Geneva: WHO. 1999. Hanoi: Ministry of Health. World Bank. 1992. "Vietnam: Population, Health, and Nutrition ------. 2002. Annual Activity Report. Hanoi: Ministry of Sector Review." Washington, DC: World Bank. Health. ------. 1995. "Vietnam: Poverty Assessment and Strategy." Pham Tat Dong, Pham Trong Thanh, Dam Viet Cuong, Duong Washington, DC: World Bank. Huy Lieu, and Nguyen Hoang Long. 2002. "Impact of User ------. 2000a. "Combating Corruption in the Philippines." Fees and Health Insurance on Equity in Financing, Access Washington, DC: World Bank. to, and Utilization of Health Services." Hanoi: Ministry of ------. 2000b. "Vietnam: Public Expenditure Review." Health. Washington, DC: World Bank. Philippine Health Insurance Corp. 1998. Annual Report 1998. ------. 2001. "Vietnam Growing Healthy: A Review of Quezon City: Philippine Health Insurance Corp. Vietnam's Health Sector." Hanoi: World Bank. Pineda, V. S. 1998. "Health Management: Strategies from ------. 2001a. "Philippines Filipino Report Card on Pro-Poor Selected Cities." Discussion Paper No. 98-36. Makati City: Services." Washington, DC: World Bank. Philippine Institute for Development Studies. ------. 2002. "Philippines: Public Expenditure, Procurement, Quimpo, B. A. 1996. "Devolution Matters. A Documentation and Financial Management Review." Washington, DC: of Post-Devolution Experiences in the Delivery of Health World Bank. Services." Manila: Philippines Department of Health, ------. 2003a. "Decentralizing Indonesia: A Regional Public Local Government Assistance and Monitoring Service. Expenditure Review Overview Report." Washington, DC: Rood, S. 1998. "An Assessment of the State of Knowledge Con- World Bank. cerning Decentralized Governance under the Philippines' ------. 2003b. "Poverty." Vietnam Development Report 2004. 1991 Local Government Code." Philippine Journal of Public Joint Donor Report to the Vietnam Consultative Group Administration 42 (1, 2): 58­82. Meeting, Hanoi. Smithson, P. 1993. "Health Financing and Sustainability in ------. 2004. World Development Report 2004. Washington, Vietnam." Hanoi: Save the Children. DC: World Bank. Solon, O., R. M. Gamboa, Brad Schwartz, and A. N. Herrin. 1992. "Health Sector Financing in the Philippines." Health 9 Education Reforms in East Asia: Policy, Process, and Impact Elizabeth M. King and Susana Cordeiro Guerra In most developing countries, responsibility for personnel, who presumably know more about local providing primary and secondary education has educational problems than national officials, and resided with the central government. However, a who have an incentive to lobby for more resources growing number of countries throughout the and to innovate. Indeed, as the broader decentraliza- world, including those in East Asia, are transfer- tion literature suggests, the benefits of decentraliza- ring this responsibility away from the center, typi- tion lie in reinforcing accountability among those cally as part of a broader reform to decentralize responsible for delivering services--between the cen- government functions. This transfer has taken var- tral government and local governments, between ious forms, including devolving fiscal responsibil- governments and school personnel, and between ity and management to lower levels of govern- school personnel and the communities they serve ment, making public schools autonomous, (Ahmad et al.1998).1 In countries as large and diverse requiring the participation of communities in as China and Indonesia, generating local solutions to operating schools, expanding community financ- educational problems and mobilizing local energies ing, allowing families to choose their schools, and and resources can yield dividends for all. stimulating private provision of education. The Despite its promises, however, decentralization impetus for decentralization has often been politi- is not a policy panacea. As this chapter shows, cal or financial rather than educational, yet sup- choosing an appropriate design for transforming porters of decentralization would argue that it can an education system is difficult. What's more, the address difficult problems confronting education reform process is never smooth. It is likely to be systems, especially those relating to performance punctuated by bursts of progress and frequent set- and accountability. Education systems are backs, which may lead to rising frustration and extremely demanding of the managerial, technical, growing mistrust among stakeholders who see and financial capacity of governments, so the themselves as losers under the reform process. potential returns to making such systems more This chapter reviews the experiences of East efficient and effective are great. Asian countries in decentralizing their education The promise of decentralization lies in giving systems, with the goal of understanding the more voice and power to local leaders and school challenges of designing reforms, distilling lessons 179 180 East Asia Decentralizes on implementation, and examining the impact on The fourth section reviews evidence on the educational development. Decentralization laws impact of decentralization and the factors that have typically stipulate dramatic reallocations of influenced its effectiveness. Because educational authority and responsibility among levels of gov- development is rarely the rationale for decentral- ernment and also transfers of resources. However, ization, there is no guarantee that the reform will, the experience in East Asia--and, indeed, in in fact, improve education outcomes. With the nearly all countries that have decentralized--sug- exception of China, East Asia's experience with gests a lack of congruence between design and decentralizing education is fairly recent and re- implementation, or between de jure and de facto search on its impact nascent, so the review focuses decentralization. on shifts in education expenditures and on inequal- Several factors have given rise to this incongru- ity, and then relies on lessons from around the ence: incomplete design and implementation lags, world to evaluate the impact of decentralization on which may be due to weak technical and adminis- learning. The final section summarizes key findings trative capacity, and lack of broad political support and lessons about decentralization given experi- for reform. For example, central agencies are not ences in the East Asian countries. shy about transferring responsibilities for financing and delivering education services to local govern- The Impetus for Decentralizing ments but are not as eager to share corresponding Education authority and resources, and so find ways of reasserting control. Local governments that are Educational achievement in parts of East Asia is supposed to yield some decision-making authority much admired. Japan, Korea, Singapore, Taiwan, to schools may also hold back from doing so. and Hong Kong have achieved high enrollment Indeed, two common challenges are to align func- rates and high-quality education, with their stu- tions, powers, and resources among levels of gov- dents consistently topping international tests (see ernments, and to define an appropriate role for the table 9.1) (Martin et al. 2004a and 2004b). Other central authority within a decentralized system. East Asian countries have not done as well, but they, Achieving a better alignment of functions, powers, too, have achieved high enrollment rates (see figure and resources is primarily a matter of improving 9.1). These countries face other educational chal- design in some countries, and of improving imple- lenges: The emerging economies of China, Indone- mentation in others. sia, Malaysia, the Philippines, and Thailand must The next section examines the rationale for better manage their education systems to reduce decentralization in East Asian countries. The disparities between wealthier and poorer regions, principal motives rarely relate to expanding or and to improve the overall quality of education. The improving public services, so the allocation of poorer countries--Laos, Cambodia, and Papua New functions and resources often does not provide a Guinea--must expand the number of children who coordinated framework for managing services enter school, cut the number who drop out at the more effectively. The third section reviews the primary level, ensure that the system produces nature and design of education reforms in these enough talent to support economic growth and countries, as well as their implementation. This sec- development, and address difficult problems in tion focuses on the overall legislative framework: financing and managing their education systems. how decentralization has changed governance and These challenges, however, have not been the pri- management; which responsibilities and functions mary rationale and main driving force behind efforts countries have devolved; whether resources are to decentralize the education systems in these coun- adequate to act on these; whether the structure of tries. Rather, political factors and fiscal concerns have the system is aligned with the changes; and what been the impetus.2 Key design aspects of reform-- functions the central agency has retained. East including central-local transfers, local tax authority, Asian countries reveal common design features but and civil service rules--may therefore ignore legal, also important differences, emerging partly from financial, and administrative issues that are critical for differences in motivation for reform, initial achieving national education goals, and may establish conditions, and the political milieu. structures and incentives that imperil those goals. Education Reforms in East Asia: Policy, Process, and Impact 181 TABLE 9.1 Student Performance on Mathematics and Science Tests (ranking among 38 countries) Mathematics score and rank Science score and rank Country 1999 2003 1999 2003 Singapore 604 (1) 605 (1) 568 (2) 578 (1) Korea 587 (2) 589 (2) 549 (5) 558 (3) Taiwan 585 (3) -- 569 (1) -- Hong Kong 582 (4) 586 (3) 530 (15) 556 (4) Japan 579 (5) 570 (5) 550 (4) 552 (6) Malaysia 519 (16) 508 (13) 492 (22) 510 (21) Thailand 467 (27) -- 482 (24) -- Indonesia 403 (34) 411(35) 435 (32) 420 (27) Philippines 348 (36) 378 (42) 345 (36) 377 (43) Sources: Trends in International Mathematics and Science Study 1999 and Martin et al. 2004a and 2004b. Note: -- not available. Scores reported are for eighth grade. Ranking is among 38 countries (1999) and 46 countries (2003). FIGURE 9.1 Net Enrollment Rates in East Asia, 2000 100 90 80 70 60 50 percent 40 30 20 10 0 of PDR China Lao Rep. Cambodia Mongolia Vietnam Indonesia Thailand Malaysia Philippines Korea, primary level secondary level Note: Countries are listed according to their gross domestic product per capita for 2002--from Cambodia to Republic of Korea. Data on Thailand's and China's net enrollment rates at the secondary level are not available. Sources: UNESCO 2002/2003; World Bank 2003b. In China, decentralization of education can be all but a few sectors in 1999 (World Bank 2003a). In traced to the decollectivization and economic liber- the Philippines, the 1987 Constitution mandated alization reforms of the 1970s, which laid the decentralization, and the 1991 Local Government groundwork for transferring responsibility for Code provided legal guidelines for transferring social services to local governments. Fiscal con- responsibility for providing services to subnational straints on the central government were also seen as governments. Except for the transfer of construc- a primary motive for that transfer (Hawkins 2000; tion and maintenance of school buildings to local Bray 1999; Cheng 1997). In Indonesia, political governments, however, the Philippines has not factors--a national call for democracy, the end of formally decentralized governance of elementary the Soeharto regime, the failures of the highly cen- education. Political considerations underlie this tralistic government, intensified by the financial exception. One often-cited reason is that public crisis of 1997--drove the decision to decentralize schoolteachers have traditionally counted votes 182 East Asia Decentralizes during elections, so decentralization would make financing and managing basic education in rural them vulnerable to local politics, possibly compro- areas was transferred from township to the county mising election results (Loehr and Manasan 1999).3 level in 2001.4 In 2002, the People's Congress passed In Thailand, decentralization is said to result the Private Education Promotion Law, which from the groundswell of support for greater democ- defined the legal status as well as the rights and racy, shared powers and resources between the cen- responsibilities of the private sector, further open- tral government and local levels, and greater ing the door for diversified provision and multiple accountability, culminating in the 1997 Constitu- sources of funding for education (Wang 2004). tion (Mutebi 2003; Weist 2001). The motivation for In Indonesia, Laws 22 and 25 of 1999 transferred decentralization in Cambodia was also predomi- governance and management of primary and jun- nantly political: building democratic governance in ior secondary education to district governments, a country ruled by centralized power for most of its and the upper secondary level to provincial govern- modern history. Because the regime that emerged ments, while the central government retains con- from a long civil war was marked by rigid organiza- trol of the tertiary level. The Education Law 20 of tion, inefficiency, leakage of funds, budget allocation 2003 takes decentralization a step further, moving difficulties, and little community participation, civil control of basic levels of education from districts to society and the development community pushed to schools (World Bank 2004a). In Cambodia, recent deconcentrate government functions to improve laws have transferred functions and powers-- service delivery, especially for the poor (Royal including the provision of public services--to com- Kingdom of Cambodia 2001). munes, and the country plans to boost accountabil- ity further by increasing the"operational autonomy" of schools and postsecondary institutions (Royal The Design and Practice Kingdom of Cambodia 2001).5 of Education Decentralization A second common feature of decentralized The design of decentralized education across East education across East Asia is that, at the deepest Asian countries reflects common features. One is level, the vehicles for governance and manage- that devolved education systems rest on multilay- ment are typically community councils and ered governance and management structures, with school committees involving local officials, civic the result that forging a coherent national policy leaders, and parents. In Thailand, each school is requires a much larger effort. Central and interme- supposed to have a board composed of represen- diate (provincial, state, municipality, and district) tatives of parents, teachers, community organiza- levels of government generally continue to govern tions, alumni, and students. Parents' organizations post-basic education, but the lowest level of with jurisdiction over schools are to establish a government, and even schools themselves, govern quality assurance system, and communities are basic education. urged to "participate in educational provision by China's policy stipulates multiple layers of edu- contributing their experience, knowledge, expert- cational supervision involving the National Educa- ise and local wisdom for educational benefits" tional Supervision Agency as well as corresponding (Kingdom of Thailand 1999). In Indonesia, each agencies in local governments (Hawkins 2000; school is supposed to have a School Committee-- Wang 2004). The provincial level takes responsibil- declared an independent body by the 2003 Educa- ity for developing specific local policies and regula- tion Law--to provide advice, direction, and sup- tions in line with national education objectives. port for managing schools (Government of The local government--the township level in rural Indonesia 2003). In China, school principals are areas (the lowest level of the bureaucracy without charged with greater responsibility than in the education offices), and the district level in urban past but also enjoy more autonomy. They are settings (with education offices)--has responsibil- expected to generate additional resources for the ity for ensuring that all children receive nine years school and ensure teaching quality, because they of compulsory education. Earlier implementation can choose teachers without much intervention revealed inadequate capacity of township govern- from the district or county, as well as determine ments to manage schools, so local responsibility for incentives for teachers (Wang 2004). Education Reforms in East Asia: Policy, Process, and Impact 183 Such deep decentralization is common outside tion to lower levels, it may retain control of peda- the region, too, a means not only of mobilizing gogical matters, personnel management, and local resources but also of fostering greater financing and resource allocation, or it may decen- accountability and better performance. In Brazil, tralize those functions to school councils. How reform in several states has entailed establishing each country assigns these specific decision areas is, school councils, allowing the direct transfer of in many respects, a more accurate measure of its resources to schools, and giving communities the degree of decentralization. power to elect the local principal. El Salvador's In 1998, the World Education Indicators survey, Community-Managed Schools Program--better conducted under the auspices of the Organisation known as EDUCO (Educación con Participación de of Economic Co-operation and Development la Comunidad)--transferred management of each (OECD), collected information on the locus of preprimary and primary school to an elected Com- decision making in lower secondary education in a munity Education Association composed mostly of small sample of developing countries, including a parents and other community members (Jimenez few East Asian countries (OECD 1998). The survey and Sawada 1999). These councils are legally examined 38 decisions pertaining to instructional responsible for running the schools, raising funds, content, personnel management, and resources and and hiring and firing teachers, with the goal of financing. To update the resulting data for this improving accountability, attendance, and achieve- chapter, we undertook a similar, though more ment. Nicaragua's school autonomy reform gave modest, information-gathering effort in East Asian school councils--composed of principals, parents, countries. (See the chapter annex for details on the teachers, and students--the authority to hire and method we used and a list of functions comparing fire teachers, veto power over principals' decisions, the two sets of data.) and discretion over the sanctions of the Ministry of Both databases reveal de facto rather than de Education and the obligations of teachers and stu- jure decentralization in East Asia.6 However, these dents (King and Ozler 1998). In Australia, under assessments of decision-making authority are sub- site-based management reform, school councils jective for at least two reasons: First, practice can develop a school charter, which is a contract vary widely within each country, so country-level between the school and the government, while par- information is impressionistic rather than a ents play a supervisory role through the council "weighted average" of practice across areas. Second, (Pascoe and Pascoe 1998). Overall, international periodic assessments are likely to reflect variation experience suggests that deep decentralization with in implementation of legislation, a change in legis- empowered, accountable schools presents the best lation, or both. These factors affect these two data opportunity for improving schools. sets. We present the results briefly, nonetheless, as a One important difference between experience in rough indication of the change in the degree of Latin America and East Asia, however, is that school decentralization of each country relative to other councils in Latin America elect their members from countries. the community, whereas this is not yet the practice In 1998, the proportion of decisions related to in East Asian countries. Elected council members secondary education made by the central govern- can truly represent the interests of the community ment varied widely in East Asia--from one-fifth in and provide built-in accountability. In East Asia, China to three-fifths in Indonesia, indicating that the duties and legal powers of school committees-- China's education system was then much more in general and relative to school principals, who highly decentralized than Indonesia's (see table 9.2). tend to sit on the committees--are often unclear. At that time, China's provincial and local govern- ments were making one-third of such decisions and schools about one-half, while Indonesia's The Locus of Decision Making provinces made less than one-tenth of such deci- Countries do not devolve responsibility and power sions and schools about one-third. By 2003, the to lower levels of government and schools whole- allocation of decision making in Indonesia and sale. While a central government may transfer gov- China had shifted considerably. China seems to ernance and overall management of basic educa- have retracted powers from schools but widened 184 East Asia Decentralizes TABLE 9.2 Percentage of Decisions Related to Lower Secondary Education at Each Level of Government, 1998 and 2003 Central State/provincial/ government local government School Country 1998 2003 1998 2003 1998 2003 Cambodia -- 75 -- 11 -- 14 China 21 3 33 77 46 20 Indonesia 63 36 7 28 30 35 Philippines 37 62 24 20 39 18 Thailand 55 75 0 6 45 20 Sources: OECD 1998; World Bank survey for this study, 2003. Note: -- not available. Percentages may not add up to 100 because of rounding. provincial and local power (among counties and ernment in these countries, decision making on townships)--not uncommon in the country's specific functions actually occurs at different levels decentralization history. In Indonesia, the 1999 of government and in schools.The next section asks decentralization reform assigned powers and whether these countries reveal a pattern in allocat- responsibilities to district governments, quadru- ing specific decision areas, and whether such allo- pling the proportion of education-related decisions cation is likely to improve the way the education by these governments. systems operate and thus promote better outcomes. The numbers from 1998 and 2003 for the Philippines and Thailand are puzzling: they imply Who Makes Which Decisions? that these countries have recentralized rather than decentralized during this period. The 1998 data With different levels of government involved in mul- suggest that the central government was making tiple areas of decision making, the goal is to ensure only about one-third of decisions in secondary delineation and alignment of responsibilities, coor- education in the Philippines, and about one-half of dination, and information sharing. Although these the decisions in Thailand. These countries seem to challenges may have existed before decentralization, have pushed back decentralization in 2003, with pressure to address them has intensified. Decentral- three-fifths and three-fourths of decision making ization is not likely to improve the education system lying with the central government and the role of if local governments have the authority to hire and schools greatly reduced. Historical background, fire teachers but not to influence their promotion, however, suggests that decentralization to the compensation, and development, or if schools have degree suggested by the 1998 numbers did not the authority to choose teaching methods but not exist, and that the 2003 assessment more accurately textbooks. reflects reality. In Thailand, following the 1997 Patterns emerge in the locus and mode of 22 of Constitution and the Decentralization Act of 1999, 38 decisions related to secondary education in the a Decentralization Master Plan approved in 2000 East Asian countries (see table 9.3). Setting curricu- stipulated details for transferring responsibilities. lum content, instruction time, and teachers' However, implementation has been slow. In the salaries, and allocating resources to schools, remain Philippines, the legal framework for decentraliza- the domain of the national or state and provincial tion has not transferred overall management of governments. In contrast, all five countries leave secondary education to local government, although the choice of teaching methods and support activi- strong local governments nevertheless use their ties for students (such as remedial classes) entirely substantial autonomy under the Local Government to schools; four of five countries also assign deci- Code to supplement their administrative authority. sion making on teachers' careers to local entities. In sum, despite the transfer of governance of Cambodia is the most centralistic with respect to lower secondary education to lower levels of gov- the 22 decisions: Education Reforms in East Asia: Policy, Process, and Impact 185 TABLE 9.3 Locus and Mode of Key Decisions in Lower Secondary Education, 1998 and 2003 Cambodia China Indonesia Philippines Thailand 1998 2003 1998 2003 1998 2003 1998 2003 1998 2003 Instructional matters Instruction time -- Designing programs of study -- Defining course content -- Choosing textbooks -- Teaching methods -- Mode of grouping students -- Support activities for students -- Creation/closure of schools -- Creation/abolition of grades -- Setting qualifying exams -- Credentialing -- -- Methods for assessing students' regular work -- Personnel management Hiring teachers -- Hiring principals -- Fixing teacher salaries -- Fixing principal salaries -- Career of teachers -- Career of principals -- Resources Allocation to school for teaching staff -- Allocation to school for nonsalary current expenditure -- Allocation to school for capital expenditure -- Use in school for capital expenditure -- Sources: OECD 1998; World Bank survey for this study, 2003. Note: -- missing data. Symbols indicate locus and mode of decision according to this legend: Decision made in full autonomy: , Central government; , Intermediate government; , Local government; , School. Decision made in consultation or within framework: , Central government; , Intermediate government; , Local government; , School. Instructional matters. All these countries are system helps promote a national identity as well as reluctant to delegate standard setting and decisions shared values and culture. The management and on core curricula to local governments and schools, quality assurance functions of local education reflecting the widely held belief that the education councils and school committees remain fairly 186 East Asia Decentralizes limited, bounded by a national framework. China's Countries outside East Asia show a similar reluc- central government continues to keep a close watch tance to surrender control over the substance and on curriculum, selection of textbooks, school-leav- quality of education to subnational governments ing qualifications, and teacher education, and also and schools. For example, in Chile, the central min- retains control over core subject areas such as istry maintains curriculum-setting, regulatory, and moral-political education (Bray 1999; Shen 1994; quality assurance functions (Delannoy 2000). In Hawkins 2000). A national curriculum frame- Australia, the Curriculum Standard Framework work--developed primarily by the central defines eight key learning areas, incorporating both government with some consultation with local content and process standards (Pascoe and Pascoe governments and adopted in 1992--specifies 1998). Likewise, the British school reform estab- compulsory courses. Local autonomy in education lished a national curriculum with learning objec- content appears to be limited to art, music, and tives for core subjects each year and at each key sports. Continuing to take control over the national stage (Rodríguez and Hovde 2002). In Spain, which core curriculum, in 1999 the central government is less centralized, the decentralization law estab- developed new curriculum standards for 18 subject lished that the Ministry of Education defines areas for the nine-year compulsory education level. 65 percent of the instructional material taught in These standards emphasized the need for the cur- all schools, while autonomous communities may riculum to respond to rapid changes in technology define 35­45 percent of domains that reflect and China's economy (Wang 2004). The new core regional interests (Hanson 2000). But in other curriculum also allows for local and school OECD countries, schools choose teaching methods, curricula, however. textbooks, and techniques for assessing students According to Indonesia's Education Law 20, the day-to-day, although usually within a framework central government still determines the curricu- established by a higher level of government (OECD lum framework and structure for basic and sec- 1998). This is similar to the approach among some ondary education (Article 38). The central govern- states in Brazil, such as Paraná, Pernambuco, and ment is also establishing minimum service Rio de Janeiro, where most schools are responsible standards for education. However, district govern- for elaborating pedagogical proposals and inte- ments, given constrained financial and technical grating them into the core curriculum process resources, may have trouble meeting these stan- (Machado 2002).7 dards. Likewise, in Thailand, the Commission of Teacher management. Different levels of govern- Basic Education--a pillar of the central education ment make decisions regarding teachers, often agency--is responsible for proposing standards leading to confusion and inefficiencies. These deci- and the core curricula for basic education in line sions range from teacher training to recruitment, with the National Scheme of Education, Religion, deployment, performance evaluation, human Art, and Culture. Parents' groups will provide resources databases, payroll, and redeployment. internal oversight of each school, while a central Some countries decentralize some functions, such agency will develop criteria and methods for as hiring and firing of teachers, while keeping oth- assessing student performance and school quality. ers effectively centralized, such as setting compen- In the Philippines, the central government also sation levels. retains responsibility for policy, curriculum, per- Indonesia illustrates a mixed--and confus- sonnel, and operations. ing--policy regarding teacher management. The Decisions on creating or closing a school are 2003 Education Law stipulates that the central and made at the national level in the Philippines and district governments share responsibility for "get- Thailand but at the local level in Cambodia, China, ting educators and education personnel to ensure and Indonesia. These countries usually decentralize the implementation of good quality education decisions on textbooks and teaching and assessment programs" (Article 41), and that these govern- methods to provincial and local governments. All ments will "supervise and develop educational countries except Cambodia allow schools to make personnel in education units" (Article 44). How- autonomous decisions regarding grouping students ever, many aspects of teacher management remain and providing extracurricular activities. centralized, including managing the personnel Education Reforms in East Asia: Policy, Process, and Impact 187 database, registering personnel actions, and trans- Fiscaldecentralizationwasakeyfeatureof China's mitting this information to the payroll system. reform, with the central government reducing its While districts manage personnel and payroll, the subsidies to local schools, and local governments recording of such actions--necessary to trigger intensifying their efforts to find alternative funding the payroll--is still centralized, and, according to for basic education through taxes, community con- civil service law, the central government retains tributions, and income from enterprises (Hawkins much authority over teacher wages, position 2000; Tsang 2002). As the central government cut allowances, family and rice allowances, and even school subsidies, the share of nongovernmental honoraria. In focus-group discussions, teachers sources rose from 19 percent in 1993 to 24 percent in reported that while they support decentralization, 2000 (Hawkins 2000). Reform documents suggested they prefer central management of their employ- six sources of funding: urban and rural surcharges ment (World Bank 2004a). According to teachers, levied by local governments, contributions from given that processes such as promotion still industry and social organizations,donations by indi- require the center's approval, decentralization has viduals and community organizations, tuition fees, slowed action on personnel matters because it has income from school-run enterprises, and central added a bureaucratic layer. Teachers also claim authorities. In 1994, however, the central govern- that management processes are neither more ment reversed itself and removed certain tax author- transparent nor better monitored, even though ity from local governments, and has continued to they occur at the district level. Without authority fund teachers' salaries and certain capital expendi- or significant influence over teacher-related mat- tures, citing growing disparities across regions ters, local governments and schools lack the single (Tsang 2002; West and Wong 1995). The practice of most important tool to influence the quality of charging fees is prohibited by the central government education. but encouraged by local governments, which use In countries outside East Asia, the approach someof theseadditionalresourcestofundacompen- to managing teachers is also mixed and reveals a satory mechanism.The local government defines the willingness to experiment. In Chile and Mexico, fee scale and collects a certain percentage from fee control over contracts is centralized, and a national revenues. For example, district governments receive salary scale standardizes teachers' pay. Other coun- 25­50 percent of fee revenues collected by schools. tries have encouraged greater local participation. In The revenues remitted to local governments are then El Salvador, community education associations are used to help other schools repair their school build- legally responsible for hiring and firing teachers. In ings and improve their facilities (Wang 2004). the United Kingdom, while the national level sets a Financing for education in Indonesia is also minimum pay scale and qualifications for educa- meant to be a "shared responsibility" of the central tors, public schools are responsible for hiring and government, district governments, and communi- paying their own teachers. In Brazil, communities ties (Article 46), but legislation has sent mixed mes- across an increasing number of municipalities rely sages about how autonomous local governments on direct elections to select school directors (Namo actually are in raising funds (Government of de Mello 2005). Indonesia 2003; World Bank 2004a). On the one Financing and resource allocation. This decision hand, laws have expanded the revenue-raising abil- area is the most decentralized, as countries have ity of district governments and allowed them to sought to mobilize local funds for schools--but not determine their own financial management, without second thoughts. Initial enthusiasm for accounting, and procurement systems within broad granting revenue-raising authority to local govern- guidelines. On the other hand, three design features ments has been dampened by inequalities, followed of the reform limit local autonomy. First, a suppos- by attempts to rein in the tendency of local govern- edly temporary hold-harmless component assigns ments to impose many new taxes. Nonetheless, part of the block grant to districts to cover the declines in subsidies from the central government salaries of existing teachers. Second, the reform ear- and emerging fiscal gaps have forced communities marks at least 20 percent of the national budget and and schools to seek supplementary funding, often a similar percentage of the regional budget (net of by raising user charges. salaries) for education (Article 49 of the Education 188 East Asia Decentralizes Law). Third, the funding mechanism for education of financing from local sources can lead to a is still too diverse and fragmented. Given these 0.14 percent decline in total costs (Jimenez and features, some regional education officials have Sawada 1996). In Indonesia, local government expressed frustration at not knowing the total level spending and parental contributions boost school of resources actually available to them (World Bank efficiency: cost per student falls as the local share of 2004a). Without information or transparency, they funding rises, though at a diminishing rate (James find it difficult to plan ahead, to develop coherent et al. 1996). In sum, if used with an eye toward and effective educational programs, and to monitor equity concerns, local funding can improve effi- and assess the flow of funds through the system. ciency without worsening inequality. In the Philippines, education financing is more The role of the central education agency. Under centralized because public education is not for- decentralization, the central education authority mally decentralized, but local governments spend needs to redefine its role and reform its structures their own resources for education nonetheless.8 and processes so it can fulfill its new core functions. The sources of local government financing are the Lower levels of governments simply have no incen- Internal Revenue Allotment (IRA), which the cen- tive to carry out some policies and programs tral government sends to each municipality and because they cannot fully capture the returns, they city, and the Special Education Fund (SEF), which are unable to perform them well because of a lack is a 1 percent tax on assessed values of real proper- of economies of scale, or they do not have enough ties owned by a municipality or city. One-half of resources. These policies and programs include the SEF is spent at the municipality or city level and setting goals and standards for service provision, the other half is remitted to provinces for education experimenting when needed, rewarding innovation projects. The provincial Local School Board deter- from other parts of the system, disseminating mines the allocation of this fund among munici- information widely and regularly, establishing and palities.9 Because the Local Government Code enforcing a transparent regulatory framework, and devolved construction and maintenance of elemen- ensuring more equitable education spending. tary and secondary school buildings to municipali- East Asian countries recognize these roles. The ties and cities, the SEF also sometimes finances such 1985 decision by the Central Committee of China's construction and repair, as well as equipment, edu- Communist Party on reforming the education cational research, books and periodicals, and sports system retains a guiding and monitoring role for the development. Many local governments have also central government on major policies, principles, shown initiative in using the fund to establish new and the general plan. In 1993, the State Council secondary schools and hire more teachers, or to top issued the Program for China's Educational Reform off the salaries of the centrally hired public teachers and Development to address weaknesses in the edu- (Azfar et al. 2001; Manasan 2002).10 User charges cation system and to emphasize the link between have also boosted local funding: the share of school the country's economic development and education. fees in education spending by households rose to This policy guideline provided for local govern- 17 percent in 1997 (Manasan 2002).11 Fearing that ments to assume more responsibility for managing fees might reverse gains in enrollments, in 2001 the and financing basic education, and encouraged the central Department of Education prohibited ele- gradual establishment of community-sponsored mentary schools from collecting user charges. schools, while retaining the central government's If the central government adopts a strong com- role as the arbiter of rules and regulations (Hawkins pensatory policy in distributing funds across 2000). This mandate of local responsibility and dif- regions, then local financing and modest user ferentiated levels of management was reinforced by charges can boost performance by allowing parents the State Education Commission in 1995 through and the community to exert greater control over the Education Law of the People's Republic of China school operations. In the Philippines, schools that (Wang 2004). rely more heavily on local sources--including Thailand's National Education Act of 1999 contributions from the local school board, munici- assigned administration and management functions pal government, and parent-teacher associations-- related to academic matters, budget, and personnel are more efficient. A 1 percent increase in the share to "educational service areas, educational insti- Education Reforms in East Asia: Policy, Process, and Impact 189 tutions, and local administration organizations" presents two measures of such impact: the level of (Section 9) (Kingdom of Thailand 1999). However, spending on education, and geographical dispari- this legislation maintains a large role for the central ties in enrollment and literacy rates. The section authority in designating standards and defining focuses on three of the five countries--China, the procedures while supporting local governance Philippines, and Indonesia, which have had slightly through boards and committees. longer experience with decentralization. The evi- International experience illustrates the role of dence is suggestive rather than conclusive, because the central government in reducing education it reflects not only the effects of decentralization inequalities within a decentralized setting. In both but also those of other reforms and developments. Spain and Chile, the central government used Even if we could capture the direct impact of revenue-sharing and transfer mechanisms to decentralization, the results would reflect its design, implement this goal. Besides a block grant budget procedural and implementation capacity, and transfer, Spain created the Inter-Territorial Com- political maturity more than its inherent flaws and pensation Fund (FCI).As a result, in 1996 Andalucia benefits. This section also presents the impact of received 38 percent of state redistributed income decentralization on student performance, but and 39 percent of the FCI, while Madrid received focuses on lessons from international experience less than 1 percent of state redistributed income because of lack of data on East Asia. (Hanson 2000). In Chile, the P-900 program designed pedagogic support initiatives for rural Greater but More Unequal Education Spending students and the least-advantaged 10 percent of the primary school population. Chile used several Has total spending on education grown as a result measures to improve equity in the 1990s: a capita- of decentralization? Are funding levels appropriate? tion grant to rural schools; scholarships for indige- Answers to these questions need to consider nous, low-income, and distinguished students; whether a country has created an appropriate bal- school feeding programs; and an expansion of pre- ance between assigned expenditure responsibilities school education (Delannoy 2000). In 1998, the at various levels and allocated revenues ("vertical Mexican government adopted a formula-driven equalization"). In other words, local governments system for allocating transfers to states. Under should control resources commensurate with their the new formula, states receive at least the same assigned responsibilities, and transfers from the amount as the previous year, as well as budgetary national government should supplement what they increments based on the number of needed schools lack. Has the spending share of local governments and teachers (Lopez-Acevedo et al. 2003). In Brazil expanded relative to the share of the central gov- in the mid-1990s, the Lei de Diretrizes e Bases da ernment? Has the type of government spending for Educaçăo Nacional assigned the federal government education changed? Have central governments the role of narrowing inequalities in access and tended to spend more on, say, capital investments finance, and launched the Fund for the Mainte- than recurrent items such as salaries and opera- nance and Development of Basic Education and tional expenditures? Teacher Appreciation (FUNDEF) to equalize China. This country's share of education expen- financing for basic education. This fund guarantees ditures in total fiscal spending more than doubled-- a minimum per pupil expenditure in primary from 6.7 to 18 percent--from 1978 to 2001. The schools throughout the country and partially central government devoted 16.3 percent of its equalizes per pupil funding within states. budget to education in 2001 (People's Republic of China 2002). From 1986 to 1992, the budget alloca- tion and out-of-budget funds grew annually by The Impact of Decentralization: 3.5 percent and 19.7 percent in real terms, respec- Educational Dividends tively, while per-student budgeted spending rose by East Asia's experience with decentralization has 9.6 percent at the primary level and 5.1 percent at been relatively brief, so it is too early to assess the the secondary level. Yet because China's economy real impact of decentralization reforms on many grew rapidly, the share of education spending in measures of educational development. This section gross domestic product (GDP) fell from 2.9 percent 190 East Asia Decentralizes TABLE 9.4 China: Education as a Share of ernments provide a financial subsidy to poor areas, Total Fiscal Expenditures the subsidy is small and an ad hoc instrument rather than a regular part of financing for compul- 1997 1998 1999 sory education. The result is that teachers are not Hezheng County 17.3 14.3 13.3 paid on time, many schools are in poor physical Jinshishan County 19.0 18.4 18.3 condition, and the goals of the Universal Compul- Linxia Prefecture 16.0 17.2 16.1 sory Education program have been delayed. Despite Gansu Province 14.9 15.5 16.0 the program's efforts to raise the minimum provi- CHINA 16.7 16.0 sion of education in poor regions, they provide less education in terms of quantity and quality and pass Source: World Bank 2000. more costs along to families. And, contrary to the law, some county governments are borrowing from in 1991 to 2.5 percent in 1997, and to 2.2 percent in the private sector to finance their schools (Tsang 1999 (Tsang 2002). Even as a share of total govern- 2002). In addition, although tuition fees in compul- ment spending, education expenditures fell some- sory education are forbidden by law, fees in public what from 16.7 percent in 1997 to 16.0 in 1999, schools are often collected in the form of a "joint such as other fiscal expenditures rose faster (see construction fee" or as voluntary donations (Wang table 9.4). Trends varied across counties, however. 2004). In some, as Hezheng and Jinshishan, education's Philippines. In this country, total public spend- share of total spending declined. In Gansu province, ing on education as a percent of GDP rose at an which began the period with a lower share for edu- average annual rate of 6 percent from 1987 to 2001 cation, that share rose. (see figure 9.2). During this 15-year period, spend- Interprovincial disparities in per-student spend- ing declined only during times of fiscal adjustment ing have also widened. The highest-spending (1990­94) and financial crisis (1998­99), mainly provinces spend many times more for primary and because of dips in central government spending. lower secondary education than the lowest-spend- GDP grew by 4.1 percent in real terms, so this larger ing provinces, and these gaps have grown, especially share of education meant a substantial rise in real for primary education (see table 9.5). Observers spending for education. As a share of total spending conclude that the lack of a clear equalization by local governments, education spending rose scheme is a fundamental weakness of the system's from 3.8 percent in 1990­91 to 7.8 percent in financing (Hawkins 2000; Tsang 2002; West and 1998­2000, with the Special Education Fund (SEF) Wong 1997). While the central and provincial gov- fueling a growing part of this local spending, rising TABLE 9.5 China: Per Student Educational Expenditure, 1989, 1997, and 2000 Primary level Lower secondary level 1989 1997 2000 1989 1997 2000 Highest-spending provinces 393 2,351 2,756 788 3,425 2,788 Lowest-spending provinces 75 255 261 174 491 420 Ratio of highest-to- lowest-spending provinces 5.2 9.2 10.6 4.5 7.0 6.6 Mean 166 593 492 353 1,096 680 Source: Tsang 2002. Note: Data for 1989 and 1997 pertain to total education expenditures by provinces; 2000 data pertain only to recurrent expenditures. In 1997, for which both total and recurrent expenditure data are available, the ratios of recurrent spending by the highest- and lowest-spending provinces are similar (9.4 and 7.1, respectively). Education Reforms in East Asia: Policy, Process, and Impact 191 FIGURE 9.2 Central and Local Government TABLE 9.6 Philippines: Median Values Education Spending in the Philippines of SEF Resources per Pupil 4.5 Income 40th 4.0 classification Median percentile 3.5 City class 1 590 450 3.0 City class 2 382 270 2.5 GDP City class 3 341 190 %2.0 City class 4a 140 1.5 City class 5a 120 1.0 Class 1 132 101 Class 2 65.7 51 0.5 Class 3 50.2 40 0 Class 4 46.7 37 Class 5 32.5 26 198719881989199019911992199319941995199619971998199920002001 Class 6 17.6 11 total government spending central government spending Source: Manasan and Atkins 2004. local government spending Note: The classes pertain to income classifica- tions: 1 to 5 for cities and 1 to 6 for other Source: Manasan 2002. municipalities. a. These numbers represent few observations. from 57 percent in 1992 to 79 percent in 1999. The result has been a shift in the shares of central and On average, district governments in Indonesia local governments in education spending. In 1991, do have more resources at their disposal than in the the local government share was only 2.5 percent, past, and the allocation formula stipulates that rising to 7.4 percent in 2001. Meanwhile, the shares poorer districts should receive larger transfers. of the central government in both capital expendi- However, the central government expects districts tures and operating expenditures for education to mobilize more of their own resources to supple- have declined.12 ment the transfers. Herein lies the risk that inequal- Data on average SEF spending per pupil indicate ities among districts will grow, as in China and the wide variation across income classes of municipali- Philippines.13 Decentralization laws have given ties and cities (see table 9.6). The poorest munici- taxing authority to district governments if the cen- palities spend only 13 percent of the SEF per pupil tral government authorizes the taxes and districts resources of the richest municipalities, and only 3 abide by principles in Law 34 of 2000.14 The reality, percent of the SEF resources of the richest cities. however, is large inequalities in local revenue bases. These numbers plainly show that transferring Many district governments have limited capacity responsibility for funding basic education to local to raise taxes from land, buildings, and natural governments leads to wide regional disparities in resources, which constitute only about 5 percent school inputs. of their revenues. Provincial governments have a Indonesia. It is still too early to tell whether larger own-resource base but must share these rev- decentralization will raise overall public spending enues with district governments. For example, the on education in Indonesia, but early signs have per capita GDP (excluding oil and gas) of the rich- been positive. Before decentralization, Indonesia est province, Jakarta, is almost nine times larger was spending the smallest share of GDP on educa- than that of the poorest province, East Nusa tion among East Asian countries: only 1.4 percent. Tenggara (Akita and Alisjahbana 2002). Partly as a This share rose to more than 4 percent in 2002--a result, per student allocations for recurrent and significant expansion in resources for the sector. capital expenditures vary widely, with districts at In 2001, district governments accounted for about the lower end of the range surely not meeting any two-thirds of total education spending, whereas kind of education standard (see table 9.7). the share of provincial governments was only Summary. The trends in education spending in 4 percent. China, the Philippines, and Indonesia show that total 192 East Asia Decentralizes TABLE 9.7 Indonesia: Per Capita Education Spending, 2001­2 (in rupiah) 2001 (actual) 2002 (planned) Total 134,000 175,058 (1,586/463,753) (1,193/540,479) Recurrent 126,118 159,460 (998/450,789) (1,013/539,287) Capital 16,185 21,692 (177/205,044) (402/415,463) Source: Sistem Informasi Keuangan Daerah (SIKD), Ministry of Finance. Note: Numbers in parentheses indicate minimum and maximum values. resources for education have grown under decentral- education. Enrollment rates at the primary level are ization; yet whether these increases are larger than high, and rates are rising even for post-basic educa- they would have been without reform is difficult to tion. These numbers hide large inequalities within say. It is also clear that the share of education spend- each country, however. These inequalities predate ing by local governments has grown, partly because decentralization reforms, but those reforms could the central government has devolved resources and exacerbate them. Large inequalities in the distribu- responsibilities for spending those resources to local tion of resources among geographical regions can governments, and partly because local governments produce large disparities in education outcomes. are expected to generate their own resources to meet Transferring fiscal responsibilities to local areas and those expenditures. However, local governments are relying on local resources and expertise is likely to far from equal in their ability to mobilize their own widen educational gaps between areas with a strong resources, and thus the gap in education expendi- revenue base and those that are less prosperous tures per student between wealthier and poorer areas while weakening the central government's ability to can only widen. Central governments clearly need to close these gaps. establish a mechanism for equalizing education China. So how large are within-country dif- resources across municipalities and cities. ferences in education outcomes, and have they The block grant system--which gives local deci- increased or decreased since decentralization? sion makers latitude to act on local goals with gen- China's overall enrollment rates in basic education erally unrestricted funds--does not guarantee that are high, but provinces differ widely in literacy officials will spend enough resources on education. rates and in enrollment rates at the secondary On the one hand, local decision makers may choose level. Literacy rates (for those aged 10 and above) to finance budget items that promise a quicker and rose significantly from 1982 to 1999--in a few more stimulating effect on the local economy. On provinces by as much as 25 percentage points-- the other hand, they may respond to the desire of and inequality fell (see figure 9.3). The coefficient local voters for more and better schools, or to the of variation for literacy rates declined from 0.19 in fact that schools provide local employment and can 1982 to 0.13 in 1999. With the sole exception of be a source of prestige for the community and its Tibet, where literacy was only 35 percent in 1999, leaders. Greater local funding is expected to create literacy rates across China exceeded 70 percent. public pressure to spend resources more wisely and Despite this progress toward equalization, thus make the sector more efficient, given an undoubtedly the result of the national policy appropriate system of accountability. of universal basic education, provinces such as Qinghai, Gansu, Guizhou, and Yunnan lag by nearly 20 percentage points behind the most liter- Reducing Educational Disparities within Countries ate provinces of Liaoning, Jilin, and Tianjin. Tibet's As mentioned, developing countries in East Asia literacy rate also improved, but its gain was one of have generally made important progress in basic the smallest. Education Reforms in East Asia: Policy, Process, and Impact 193 FIGURE 9.3 Literacy and Enrollment Rates Education of minority ethnic groups in in China, by Province poorer, interior rural regions has been a concern. Literacy rates, 1982 and 1999 Enrollment rates are 15 and 10 percentage points lower for minority girls and boys than for Han girls Beijing Jilin and boys, respectively (Hannum 2002). Secondary Liaoning Tianjin enrollment rates are also unequal among provinces. Shanghai In 2000, Shanghai and Beijing had enrollment rates Shanxi Guangdong close to 90 percent, while Tibet, Guizhou, Guangxi, Xinjiang Heilongjiang and Hainan had enrollment rates one-third lower. Hunan But while the coefficient of variation of enrollment Hebei Guangxi rates is smaller (at 0.10) than that for literacy rates, Jiangxi Hainan the tendency is for provinces that had higher Chongqing literacy rates in 1982 to have higher secondary Hubei Zhejang enrollment rates in 2000, suggesting that lagging Henan Inner Mongolia provinces will continue to fall behind. Sichuan Philippines. Similarly, in the Philippines, literacy Jiangsu Shaanxi rates and enrollment rates vary widely across Fujian Shandong provinces. Literacy rates (for the population age Anhui Ningxia 10 years and above) increased substantially over the Yunnan five-year period from 1989 to 1994--by as much as Guizhou Gansu 1982 nearly 20 percentage points in Western Mindanao, Qinghai 1999 Tibet and by almost 15 percentage points in three 20 30 40 50 60 70 80 90 100 other regions (see figure 9.4). With these gains, the percent literacy gap narrowed among the regions, but by 1994 literacy rates still ranged from 61 percent in Secondary enrollment, 2000 the Autonomous Region of Muslim Mindanao Shanghai (ARMM) to 92 percent in the national capital Beijing region (metropolitan Manila).15 These trends do Tianjin Inner Mongolia not indicate that decentralization has helped Zhejang Shaanxi reduce education gaps. Jiangsu Enrollment rates at the secondary level have also Qinghai Jilin shown significant gains since decentralization, with Xinjiang Hubei increases fairly equal across regions.16 The notable Liaoning Chongqing exceptions are the regions in Mindanao: Northern Sichuan Mindanao is the only region in which secondary Hebei Shandong enrollment rates declined, and the increases in the Hunan Shanxi other Mindanao regions are smaller than in any Fujian other region. In 2002, enrollment rates varied from Gansu Jiangxi just 32 percent in ARMM to 94 percent in the Henan Guangdong Ilocos Region. Heilongjiang The experience of ARMM is noteworthy because Ningxia Yunnan it is the region with the fullest autonomy, including Anhui Hainan in managing its education system.17 Legislation Guangxi passed in 2001 contains detailed provisions that the Guizhou Tibet region's schools will adopt the basic core courses, 50 60 70 80 90 100 minimum curriculum, and textbooks required by percent the national government, but will have the preroga- Source: People's Republic of China 2001. tive and responsibility to add other courses and Note: Provinces are arranged in descending order according to instructional materials that reflect Islamic values. values in the more recent year. With respect to the two indicators considered above, 194 East Asia Decentralizes FIGURE 9.4 Literacy and Enrollment Rates in is particularly difficult. As a result, fewer than 2 of the Philippines, by Region 10 children who enter grade one complete high Literacy rates, 1989 and 1994 school. In sum, while at least two other factors might explain the region's poor education indicators-- Metro Manila high poverty and protracted armed conflict--it is Southern Tagalog Central Luzon reasonable to conclude that greater autonomy has Cagayan Valley not produced better outcomes. Ilocos Region Indonesia. Indonesia's enrollment rates,especially North Mindanao Bicol Region at the primary level, compare favorably with those Western Visayas of East Asian countries with higher per capita Central Visayas income. With decentralization relatively nascent in Eastern Visayas South Mindanao Indonesia,these overall gains cannot be attributed to Cordillera Admin the reform. But differences among regions reveal Central Mindanao the equity challenge for decentralization.In 2002,the West Mindanao 1989 enrollment rate of youths aged 13 to 15 (roughly ARMM 1994 the junior secondary level) ranged from 68 percent 50 60 70 80 90 100 in South Sulawesi to 94 percent in Yogjakarta (see percent figure 9.5).18 Increases since 1998 have been modest. Secondary enrollment rates, 1991 and 2002 Ilocos Region Metro Manila FIGURE 9.5 Enrollment Rates in Indonesia, Southern Tagalog by Province Cordillera Admin Central Luzon Enrollment rates of Western Visayas 13- to 15-year-olds, 1998 and 2002 Cagayan Valley Central Visayas Yogyakarta Bicol Region Jakarta Eastern Visayas North Sumatera Central Mindanao East Kalimantan South Mindanao West Sumatera West Mindanao Riau North Mindanao 1991 Bali ARMM 2002 Lampung Central Java 10 20 30 40 50 60 70 80 90 100 Jambi percent East Java Source: Department of Education Statistical Yearbook, Bengkulu several years; National Statistical Coordination Board. Central Kalimantan Note: Provinces are arranged in descending order according to SE Sulawesi values in the more recent year. West Java West Kalimantan North Sulawesi South Sumatera however, ARMM ranked dead last, with changes N.T. Barat not large enough to allow it to catch up with other South Kalimantan regions. Although the enrollment gap in primary N.T. Timur 1998 Central Sulawesi education between ARMM and the rest of the coun- South Sulawesi 2002 try has narrowed since the 1990s, it remains sub- 60 70 80 90 100 stantial: nearly 20 percentage points separate percent school-age children in the poorest quintile in Sources: Susenas 1998 and 2002. ARMM and those in other regions (World Bank Note: Figure reflects provincial divisions in 1998; the provinces 2003b). ARMM children begin school later and are of Aceh, Irian Jaya, Maluku, and North Maluku have been only about half as likely to continue through the excluded because their sample sizes in 2002 were much smaller and limited than in 1998. Provinces are arranged in descending elementary grades, and the transition to high school order according to values in the more recent year. Education Reforms in East Asia: Policy, Process, and Impact 195 Disparities within provinces are even larger than at large is difficult. Student-level data on national variation between provinces. Only about one- exam results are usually not available even to fourth to one-third of inequality in enrollment researchers. Summary test results do not allow rates in primary and secondary education is due to study of whether differences in test scores result differences among provinces--the rest reflects dif- from changes in students' economic conditions or ferences among districts within a province. This from aspects of the decentralization reform. Lastly, means that equalization among districts within most East Asian countries have relatively brief each province is a greater challenge than equaliza- experience with decentralization. However, evalua- tion among provinces. tions of the experiences of countries outside East Overall, these changes in the levels and inter- Asia provide lessons on the potential impact of provincial distribution of education expenditures, decentralization on student performance. and in basic indicators, in China, the Philippines, In the United States, two examples are illustra- and Indonesia point to both positive and negative tive. The 1995 Chicago School Reform Amendatory effects of reform. However, the analysis is still at an Act modified a 1988 autonomy reform by estab- aggregate level. Ultimately, the test of decentraliza- lishing stronger central support functions and tion's effectiveness is whether schools are better and requiring external accountability mechanisms.20 students are learning better. Decentralization laws Although attributing causality to either the 1988 or encourage greater local and community participa- the 1995 reform is difficult because of their com- tion in providing and financing education, but this plementary nature, student scores in elementary feature exposes inequalities between prosperous reading and math have improved consistently since and poor areas, and the inability of poor areas to 1995. The percentage of students scoring at or mobilize adequate resources. Central transfer above the mean in elementary reading tests rose mechanisms need to equalize resources across from 26.5 percent in 1995 to 36.1 percent in 1999.21 areas. Indeed, the push for greater local mobiliza- In 1995, Memphis introduced a similar set of tion of resources in decentralizing countries risks reforms that differed in one important feature: city widening disparities between regions with a strong schools received a menu of eight different restruc- revenue base and those that are less prosperous. But turing models from which to choose. Before 1995, implementing an equalization scheme is a consid- the schools that later became autonomous had erable political challenge. On the one hand is the lower student achievement; two years later, their issue of how much inequality in educational out- scores were substantially higher than those in a comes the political system can tolerate; on the other control group (Ross et al. 1998). hand is the question of to what extent the system In Chile, two phases of reform appear to have can redistribute from richer areas to poorer areas, produced significantly different results. One evalu- and from urban areas to rural areas. ation concluded that the first phase had either a negative effect on student performance or no effect. A confounding factor was that education expendi- Improving the Quality of Education tures declined during the same period. A later eval- Good education is not only about quantitative uation concluded that the reform did not improve targets, such as boosting the supply of classrooms, the quality of public schools, and that test scores for textbooks, and teachers, but also about incentives the majority of students declined. Another evalua- that lead to better instruction and learning. The tion found that test scores were higher in private level and use of public spending for education is schools, but concluded that this was largely because only part of the educational story in East Asia.19 those schools chose better students. Yet another East Asian countries do not yet have a record of the study found that teacher autonomy exerts greater impact of decentralization on student outcomes, positive effects on student performance when especially on learning. There are several reasons for decision-making authority is also decentralized this: Decentralization is a wide-ranging reform, (Winkler and Gershberg 2000; Prawda 1992; encompassing and influencing school functions in McEwan and Carnoy 1999; Hsieh and Urquiola different ways, so isolating its impact from other 2001; Vegas 2002). Evaluations of the second phase changes in the education system and the economy show more positive results, with an 18 percent rise 196 East Asia Decentralizes in language and math test scores on standardized owing to teacher absences are significantly lower in tests. But again, isolating the effects of decentraliza- EDUCO schools (Jimenez and Sawada 1999). tion from other influences, such as the substantial In sum, international experience yields mixed rise in education expenditures throughout the evidence on the impact of decentralization on stu- decade, is difficult. dent performance. U.S. experience provides posi- An evaluation of Nicaragua's 1991 school auton- tive evidence, but the experience in Latin America omy reform shows that schools took some time to yields ambiguous results. Reform-oriented schools exercise the new functions and powers given to in the cities of Chicago and Memphis have substan- school councils. Controlling for de jure and de tially higher test scores than schools in the control facto autonomy, the results indicate that school group. In Latin America, whereas decentralization autonomy--especially in decisions related to staffing appears to have improved student performance in and monitoring of teacher activities--improves stu- Nicaragua and Brazil, and to have reduced teacher dent performance (King and Ozler 1998). Moreover, and student absenteeism in El Salvador, evaluations math and language scores were significantly higher of Chile's long-running reform are inconclusive in schools where teachers felt more empowered and concerning the impact of greater local participation influential in decision making. and school choice on student performance. Brazil's reform remains to be evaluated at the student level, but researchers have already used Lessons about Decentralizing state-level measures of educational performance to Education assess progress.While school councils and the direct Experiences in the five East Asian countries are transfer of resources are not significantly related to beginning to provide lessons for implementing better student performance, the election of school decentralization--the factors that have affected directors is positively associated with higher test scores.22 Rather than testing the reform as a whole, their experiences, the sources of resistance or sup- port they have encountered, and the risks and chal- the researchers decided to break it down into three lenges that have emerged. Actual practice often components and analyze the impact of each innovation on educational performance.23 Among deviates from formal rules on decentralization, and it is important to understand why. As mentioned, reform-minded Brazilian states, the most promi- China has had the longest experience with decen- nent is Minas Gerais, whose reform included school tralization and thus offers valuable lessons. While autonomy and dramatically changed schools' inter- Indonesia and Thailand previously tried to decen- nal structure of accountability (Guedes et al. 1997).24 As in Nicaragua, however, qualitative stud- tralize some functions, their formal decentraliza- tion reforms are recent. Cambodia's reform is more ies have shown that while de jure autonomy rarely accurately characterized as deconcentration, while exerts any influence in most schools, de facto auton- reform in the Philippines is largely a side effect of omy appears to boost teacher motivation, and thus a broader decentralization that has formally the potential for improving student learning and bypassed education. Indonesia has chosen the rapid participation in the classroom (Cordeiro Guerra Big Bang approach, while Cambodia and Thailand 2003). are proceeding at a much more measured pace. An evaluation of El Salvador's EDUCO reform Lessons from these experiences include: found that parents in EDUCO schools participate more actively in school affairs, feel they have more Strive to clarify the assignment of functions, sim- influence over decision making in the school, and plify new processes and structures, and provide mech- have a more direct relationship with teachers than anisms to coordinate and foster a shared understand- parents in traditional public schools. This greater ing of reform at different levels of government, as well local participation has had a positive effect on edu- as to adjudicate disagreements. cation outcomes. Controlling for school and Decentralizing education systems requires har- student characteristics, a study found that students monizing a complex set of functions at each level in EDUCO schools do not perform worse on and type of education, and is a difficult reform to achievement tests despite the fact that they come design and implement. Central governments tend to from poorer families, and that student absences devolve management of education to different levels Education Reforms in East Asia: Policy, Process, and Impact 197 of government, but devolution does not happen 2003d).26 Ironically, even this compromise suffers wholesale. The central government may retain indi- from vagueness on a time frame. vidual decision areas that may or may not cut across In Cambodia, the initial framework for decon- levels of education, or may devolve them even more centrating and decentralizing education is clear deeply. This complexity leads to confusion, redun- about delegating authority to provincial and district dant bureaucracies, and weak implementation. authorities, but is much less clear about the roles of Mixed signals from legislation and policies make it the school cluster system and commune councils. difficult for subnational governments and schools to The Cluster School Policy, created in 1996, encour- fulfill their functions efficiently and effectively. ages decentralized management of resources, but is Despite two decades of implementing decentral- ambiguous about what functions school clusters ization in China, for example, the division of need to perform. Similarly, commune councils, responsibilities between county and townships is though endowed with new financing mechanisms, still unclear. The 1994 Guidelines for the Reform lack clearly defined roles and responsibilities. This and Development of Education state that both lack of clarity weakens the institutional structures county and township governments are responsible that are closest to the community, ultimately weak- for delivering compulsory education, although the ening accountability as well. former manages education revenues and the latter safeguards the right to compulsory education for The central government plays important but school-age children and adolescents. The legisla- different roles in a decentralized context, and needs to tion is unclear on how these responsibilities differ, transform both its structure and skills. and also seems to conflict with the Budget Law, Though stripped of some traditional functions which clearly states that each level of government under decentralization, the central authority needs should budget separately for its own jurisdiction to step into its new roles quickly to facilitate (Hawkins 2000). successful reform. These roles include setting In Thailand, implementation difficulties stem standards and performance measures for use from vague compromises on the overall decentral- throughout the education system; ensuring that ization plan and time frame. Local governments are decentralized units can meet the standards through poorly informed about their roles and responsibili- systems development, training, and funding; and ties as well as the decentralization plan and time brokering links between local governments to cap- frame. Legislation is itself unclear about the decen- ture economies of scale. The central government tralization process. For example, the 1999 National also needs to perform overall system planning and Education Act transfers authority from the central forecasting and prioritize investment, including of government's provincial and district offices to 175 teacher supply and demand; design and implement school districts or local education areas (LEAs), an equalization scheme, fostering support from each with its own committee and office. This aspect wealthier regions; and stimulate experiments and of the reform, if implemented, would require spread lessons learned. staffing cuts and redeployment totaling half of all The central government clearly has a critical education administrators in various provinces role to play in designing and implementing equal- (Mutebi 2003).25 In response, implementers of the ization schemes. Decentralization laws encourage National Decentralization Act argue that they have greater local and community participation in pro- the authority to transfer power to local govern- viding and financing education, but this feature ments only after they fulfill a set of readiness crite- exposes inequalities between prosperous and poor ria. The latter piece of legislation would seem to areas and the inability of poor areas to mobilize retrieve some of the autonomy delegated by the adequate resources for education, and risks widen- National Education Act. Parties have reached a ing those disparities. Appropriate transfer mecha- compromise to merge both acts, which entails nisms can equalize resources across regions. How- administrative deconcentration of central power to ever, equalization efforts are not only about infusing LEAs in the short to medium term, and gradual more money into local systems but also--and decentralization of responsibilities from LEAs to more importantly--about changing incentives. local governments in the long term (World Bank Implementing an equalization scheme can be a 198 East Asia Decentralizes considerable political challenge, as efforts to redis- of local governments to manage compulsory edu- tribute from richer areas to poorer areas, and from cation seems to have improved substantially, in urban areas to rural areas, may meet resistance. areas such as training personnel, collecting and Furthermore, the relative importance of the using information, expanding the use of technol- roles of the central government is likely to change ogy, and incorporating research findings and con- as decentralization matures, requiring the central sulting experts in the decision-making process government to boost its involvement in certain (World Bank 2004c; Kerr 1999; Tsang 2002). functions at certain times. For example, it makes more sense for the central government to help For decentralization to exert a positive impact on build capacity in countries like Cambodia than in student performance, information and evaluation those like China, which have a larger national sup- systems--as accountability mechanisms--must not ply of experts. Ultimately, redefining the central only be in place but must also function iteratively government's roles means distinguishing between through participation. functions requiring critical involvement through- Information problems become much more out the process, those requiring periodic involve- acute in a decentralized context. Dramatic shifts in ment at different stages, and those requiring one- responsibilities and powers often lead to the break- time initiatives with frequent follow-ups. down of information and evaluation systems, Decentralization puts the complex architecture which typically depend on the central government needed to operate an education system--personnel, to extract information from lower levels of govern- finances, procurement, student assessment, and infor- ment and schools. In Spain, decentralizing admin- mation management systems--under pressure. This istrative functions to regions greatly undermined system needs to be reorganized to reflect the new the country's capacity to collect and disseminate intergovernmental relationships and decentralized national statistics. Newly autonomous regions functions, and capacity needs to be strengthened. began to produce their own statistics using incom- patible methods, and although the Spanish govern- In decentralized education systems, replacing ment took a strong stance on cooperation, it still inappropriate structures and building the capacity met resistance. The Education Law in 1985 tried to to work within new arrangements are key chal- overcome the lack of coordination across regions lenges. A principal bottleneck is a lack of adequate through a Conference of Counselors which included technical and managerial experience among respon- the Minister of Education and autonomous com- sible parties. A lack of viable and coordinated man- munity counselors. agement systems linking central agencies to local Information on performance at all levels is key governments and schools exacerbates problems to accountability. Countries can use sample testing, resulting from weak local capacity. national surveys, and the census to assess the Countries may be tempted to slow the pace of impact of programs, allocate resources, and iden- decentralization because of fears that district and tify geographic areas requiring special attention provincial governments do not have enough (Asian Development Bank 1999). Local govern- capacity to fulfill their newly assigned responsibil- ments, communities, and schools also need infor- ities effectively. Indeed, the traditional approach mation and diagnostic tools to evaluate perfor- has been to build local capacity before transfer- mance in specific subject areas, define learning ring responsibility and authority because of con- challenges in different communities, and compare cerns about irresponsible spending, local corrup- different pedagogical approaches and teacher tion, regional inequities, and service collapse. training mechanisms. School-level data can also However, although capacity building is important, communicate results to parents and the larger com- local governments may have more capacity than munity. At the deepest level, information is instru- most central policy makers assume. As decentral- mental for greater accountability and control. Citi- ization proceeds, local talents and capacities are zens rely mostly on frequent reports, magazines, likely to emerge and to improve with practice, as is and participatory workshops to gain information becoming clear in Cambodia, Indonesia, and the on student performance. However, linking teacher Philippines. Furthermore, in China, the capacity promotion to predetermined inputs and outputs Education Reforms in East Asia: Policy, Process, and Impact 199 BOX 9.1 School Report Cards in Paraná: A New Incentive System In the state of Paraná in Brazil, a new system of comparisons in the state and region. This wealth incentives at the school level entails producing of information enhances competition across school report cards. These report cards--known schools, thus strengthening incentives and over- as Boletim da Escola--include three main sets all accountability at the state level. If applied of information. These are results from the consistently across time, this system will also Statewide Student Learning Testing (covering help schools assess the impact of their policies Portuguese, math, and science), school census on student performance. One lesson from these data (on student promotion, retention, dropout experiences is that such systems are not sustain- rates, enrollment, teacher-pupil ratios, and able if they lack local ownership, and if schools teacher profiles), and surveys of school life (from do not have the technical capacity to maintain students, parents, and school managers). The and use them. report cards therefore focus on individual school performance while allowing for cross-school Source: Vasconcelos-Saliba 2004. could create perverse incentives for transmitting policy evaluation, and to ensure that they reflect information. Some analysts, including the Educa- existing or desired curriculum content.27 tion Commission in the Philippines, argue that the To advance education outcomes, school stakehold- government should use incentives to make teachers ers must have greater voice and exercise some control and school heads directly accountable for student over school operations. outcomes (see box 9.1 for an account of the infor- mation system of the Brazilian state of Paraná). Decentralization is not just about increasing One area that deserves monitoring and evalua- local management and technical skills, but also tion is the flow of funds through the system. In about strengthening the voice of the community in Cambodia, the Department of Finance designed the delivery of public services. In many centralistic an improved system for monitoring financial per- systems, local communities are not used to govern- formance in 2001. Supported by training and tech- ing themselves, electing politicians who represent nical assistance, new budget management forms for their interests, and using their right to vote to make schools, districts, provinces, and central depart- their will known. Breaking out of this mold imme- ments, along with provincial and program reports, diately is difficult. Despite commune councils in will feed into a computerized system. A fund track- China, lack of popular representation persists at the ing system will monitor inputs and outputs and local level. In Cambodia, the central government relate them to strategic outcomes, and will include still appoints provincial governors and district incentives for transmitting information. School heads. And in Indonesia, although the majority of inspection reports will now focus more on school schools now have school committees, they hardly performance indicators, such as development plan- meet and do not yet fulfill their designated func- ning, financial management, community partner- tions (World Bank 2004a). ships, teaching and learning processes, the learning Beyond strengthening voice, experiences in and school environment, and promotion rates. countries outside East Asia suggest that giving par- Accurate and timely information on enrollments, ents, teachers, and other stakeholders decision- teachers, and school inputs is essential, especially for making authority in key areas such as curriculum, assessing the needs of remote areas and underserved training, and pedagogical approaches leads to bet- populations. Widely conflicting statistics on these ter student performance. This occurs through variables are still too common to support robust greater commitment from teachers, more focus on planning and policy making. Most problematic of learning, stronger school leadership, and a sense of all are student testing systems: the challenge is to responsibility for results (King and Ozler 1998; make national tests comparable over time to allow Winkler and Gershberg 2000). 200 East Asia Decentralizes Top-down decentralization has largely precluded ity of teaching and student learning. Research on the involvement of teachers and teachers' unions in teacher incentives has shown that teachers are not designing education reform.28 Such limited agency only sensitive to incentives but also responsive. reflects not only the impetus for decentralization Designing targeted incentives that translate into but also political and unions' inability to mobilize. improved classroom performance and student Unions have been less politically active (China) and learning is extremely difficult because teacher effort have had fewer opportunities to play a significant is both difficult and costly to measure (Murnane role (Cambodia and Philippines) than in Latin et al. 1991; Hanushek 1986; Waterreus 2003). American countries, where teachers' unions have Several countries in Latin America, in parallel with wielded more political power and played critical decentralization, have implemented policy reforms roles in shaping education reform.29 The Latin affecting teacher incentives, such as Mexico's American experience shows that these responses Carrera Magisterial, Chile's Sistema Nacional de can be diverse, including resistance (Mexico), Evaluación del Desempeńo, and Brazil's FUNDEF. negotiation (Bolivia and Chile in the 1990s), cooperation (Brazil), and inaction (Chile in Annex: Research and Data the 1980s). Decentralization inevitably draws Collection Methods many stakeholders--namely, local authorities and associations--into the decision-making process in The 2003 World Bank study on which this chapter government and schools. Fearing a loss of negotiat- is based followed a method similar to that used in ing power on critical issues such as salaries and the the 1998 World Education Indicators survey by hiring and firing of personnel, unions have strongly OECD-INES (Indicators of Education Systems). opposed efforts to decentralize (Gaynor 1998). Yet the Bank study also differs in significant ways. These experiences foreshadow the extent to which What follows is a description and comparison of unions' political influence might affect both the our methodology and research approach, including pace and depth of decentralization in East Asia. the conceptual framework of the questionnaire, A review of 83 studies of school-based manage- data collection procedure, and calculation of indi- ment in North America and among members of the cators. This annex also outlines the methods we OECD revealed greater teacher commitment, more used to make the two studies comparable. collaboration and information sharing, and a change in classroom instruction. In India, student Conceptual Framework of the Questionnaire dropout rates declined and teacher attendance improved--from 33 percent to 78­86 percent-- Decentralization focuses on the distribution of after village education committees began participat- power between levels of government. The OECD- ing in schools and teacher monitoring (Leithwood INES survey examined two dimensions of decen- and Menzies 1998; Pandey 2000). In the Philippines, tralization: the locus of decision making--that is, about 2,000 schools adopted school-based manage- the level of government with authority--and the ment on a trial basis in connection with a World mode of decision making, or the degrees to which Bank­supported education project. Preliminary levels of government are autonomous or share results show that greater involvement among teach- authority. The World Bank study examined the for- ers in planning and managing schools has greatly mer: the locus of decision making. improved their motivation and enthusiasm, while While the OECD-INES survey distinguished six principals' efforts to fully involve teachers in identi- levels of government (central, state, provincial/ fying problems and needs, and to improve commu- regional, subregional/intermunicipal, local, and nication, have helped yield solutions. school), the World Bank study focused on three Countries vary in the extent to which they have main levels of government: decentralized functions directly affecting teach- National/central: The central government consists ers, such as recruitment, deployment, promotion, of all bodies at the national level that make deci- and salary scales. Nonetheless, underlying these sions or participate in different aspects of arrangements are incentives that influence the decision making, including both administrative behavior of teachers and ultimately affect the qual- (government bureaucracy) and legislative bodies. Education Reforms in East Asia: Policy, Process, and Impact 201 State/provincial/local: The state is the first ter- (53), both studies organized those indicators into ritorial unit below the nation in federal countries, four broad categories. The main items within those or in countries with similar types of governmen- categories included: tal structures. The province or region is the first Organization of instruction: Decisions regarding territorial unit below the national level in coun- which school students will attend, school tries that do not have a federal or similar type of careers, instruction time, choice of textbooks, government structure, and the second territorial grouping of pupils, assistance to pupils, teaching unit below the national level in countries with methods, and methods for assessing pupils' a federal or similar types of governmental regular work. structures. The municipality or community is Personnel management: Decisions regarding the smallest territorial unit with governmental the hiring and firing of the principal and teaching authority; the local authority may be the educa- and nonteaching staff, their duties and conditions tion department within a general-purpose local of service, their salary scales, and their careers. government, or a special-purpose government Planning and structures: Decisions on creating whose sole area of authority is education. and abolishing schools and grade levels, selecting School: This level applies to individual schools and designing programs of study and subjects and includes school administrators and teachers taught at a particular school, defining course or a school board or committee established exclu- content, setting qualifying exams for certificates sively for that individual school. The decision- or diplomas, and credentializing students. making body or bodies for this school may be an Resource allocation and use: Decisions on allo- external school board, which includes residents cating resources to a school for teaching staff,non- from the larger community; an internal school teaching staff, capital expenditures, and operating board, which could include headmasters, teach- expenditures, and on using resources for staff, ers, other school staff, parents, and students; and capital expenditures, and operating expenditures. both an external and an internal school board. The study considered parents and teachers as an element of the school level. Data Collection Procedure In practice,however,the decision-making process The OECD-INES approach to collecting data was is not that simple. In determining at what level deci- similar to that of the World Bank, although the two sions are made,numerous unclear situations arise.In studies differed fundamentally in the composition of some cases, a higher level of government may have expert panels formed to assess decision making formal or legal responsibility for decision making, at different levels. In the OECD-INES approach, but in practice that level delegates its authority to a researchers created a panel for each level of educa- lower level of government. In describing the actual tion, composed of one member from each of the decision-making process, we identified the lower three decision-making levels. These groups com- level of government as the decision maker. Similarly, pleted the questionnaire and arrived at a consensus a higher level of government may provide a lower on all questions. The researchers also composed level of government with choices in a particular area, a second panel for each level of education, again such as the selection of textbooks, even though the composed of one member from each of the three higher level establishes the framework for the deci- decision-making levels, and repeated the process. sion. In that case, too, we designated the lower level The INES Network C representative or national of government as the actual decision maker. Finally, coordinator for the World Education Indicators sur- one level of government may have responsibility for vey then compared the results of the two surveys. an individual decision, but inaction results in a deci- Where the responses differed, the INES Network C sion by a lower level. If a decision is left to the discre- representative used source documents and consulted tion of a lower level through lack of determination at the national coordinator to reconcile disagreements. higher levels, then we chose the level that actually We conducted the World Bank study in two makes the decision. phases, relying on intermittent consultation Although the OECD-INES survey included between two different panels of experts. The first fewer indicators (35) than the World Bank study panel was composed of World Bank education 202 East Asia Decentralizes experts with knowledge and work experience Calculating the Indicators within each of the countries. This group completed The OECD-INES study gave equal importance to the questionnaire and arrived at a consensus on all the indicators within each of the four domains. questions. A second expert panel included World Bank education experts located in the field in Each domain contributes 25 percent to the results. the respective countries. This group reviewed the Because each domain includes a different number results from the first round of surveys. Where of items, each item is weighted by the inverse of the the responses differed, this panel reconciled differ- number of items in its domain. The World Bank ences with the first panel. study followed the same approach. TABLE 9A.1 Cross-Study Comparison of Functions World Bank Study (2003) OECD-INES (1998) Pedagogical Organization of instruction Admissions criteria to enter school Decision on what school a child should attend Mode of grouping pupils/class size Mode of grouping pupils No. of periods of instruction/classroom hours Number of periods of instruction Selection of textbooks/teaching material Choice of textbooks Design of programs Design of program subjects Choice of subject matters Definition of curriculum Teaching methods Teaching methods Provision for extra help Assistance to pupils Extracurricular activities Evaluation methods Methods of assessing pupils' regular work Student promotion Decisions affecting pupils, streaming Setting of equivalencies Setting of goals/targets for the school Administrative Planning and structure Opening/closure of school Creation and closure of a school School calendar Creation/abolition of grade(s) Creation and abolition of grades Distribution of textbooks/teaching material Setting of qualifying examinations Setting of qualifying examinations Collection of student data (enrollment/exams) Community outreach Awarding of credentials Awarding of credentials Accreditation of new schools Design of programs for a specific school type Definition of course content Personnel management Personnel management Hiring and firing of staff Hiring and firing of staff Head of school Principals Teaching staff Teachers General staff Nonteaching posts Terms of service and duties Duties and conditions of service of staff Head of school Principals Teaching staff Teachers Nonteaching posts Fixing of salary levels/benefits/incentives Fixing of salary scales for staff Head of school Principals Teaching staff Teachers Nonteaching posts Education Reforms in East Asia: Policy, Process, and Impact 203 World Bank Study (2003) OECD-INES (1998) Evaluation methods/supervision Head of school Teaching staff Promotion Influence over staff careers Head of school Principals Teaching staff Teachers Nonteaching posts Training Head of school Teaching staff Certifying staff Head of school Teaching staff Allocation and use of resources Allocation and use of resources Allocation of resources to the school Allocation of resources to the school Salaries and benefits Teaching staff; nonteaching staff Administrative/operating costs Operating expenditures Capital expenditure Capital expenditures (infrastructure/maintenance) Scholarship/awards School projects/activities Budget/finances User charges/monthly fees/voluntary fees Fundraising activities Allocation from state/subsidies Procurement and disbursement Use of resources in the school Salaries and benefits Staff Administrative/operating costs Operating expenditures Capital expenditures (infrastructure) Capital expenditures Scholarship/awards School projects/activities Relationship with unions Sources: OECD 1998; World Bank study 2003. If a single item produced multiple answers, and Chile (political reasons) and the United Kingdom and although the instructions for the questionnaire did Australia (concerns about state finances). After Franco died in 1975, a new Spanish Constitution promoting a transition not specifically allow this, each answer received half to democracy devolved central functions to regional gov- of the original weight of that item. ernments (Hanson 2000). Through an opposite political The following table compares the indicators used transition in Chile--from a democratic to a military government--newly empowered neoliberal economists in the two studies: and social planners also pushed for more decentralization. In contrast, Australia's fiscal situation provided the princi- Endnotes pal motives for public service reform. The Victorian Commission of Audit assessed public expenditures in key 1. Accountability means more than "answerability" or sectors as too high and pushed for incremental change "enforceability." It implies a precise set of relationships (Pascoe and Pascoe 1998). between principals and agents encompassing five main 3. One impediment to further devolution of education func- features: delegation, finance, performance, information tions is increased politicization of elections at the local about performance, and enforceability (World Bank 2004c). level. In the Philippines, teachers serve on the Board of 2. This, however, is not different from what had occurred in Election Inspectors and thus play an important role in other countries that had decentralized earlier, such as Spain counting ballots (Manasan 2002). 204 East Asia Decentralizes 4. This was undertaken with the 2001 Decision on the Reform 12. The share of central government spending in capital expen- and Development of Basic Education by the State Council ditures fell from 10.5 percent in 1990­91 to 4.6 percent (Wang 2004). in 1996­2000, and the central share of maintenance and 5. Cambodia's policy goals for education reform are con- operating expenditures fell from 18.7 percent in 1990­91 to tained in the Education Strategic Plan (ESP) for 2001­5 10.4 percent in 1996­2000 (Manasan 2002). and implemented through the Education Sector Support 13. With the transfer of authority and management to regional Program and the Priority Action Program rather than governments in Indonesia, "It is hoped that the local gov- through a single piece of legislation. The ESP reiterates the ernments are capable of obtaining more funds in their central government's vision: "The Ministry's vision of an regions and in managing them more efficiently and effec- inclusive education system also includes broad-based par- tively. It is expected that the local governments are able to ticipation al all levels of Government and civil society in use or exploit the sources of funds in the regions, such as taking responsibility for planning and implementation of the private sector and other education stakeholders, in education services . . . An associated goal would be to funding education" (Government of Indonesia 2001). incrementally delegate greater decision-making and spend- 14. Until recently, the national government retained 10 percent ing authority to districts, possibly communes, and schools. of total tax revenues for its own use and provided 9 percent In this way, all national stakeholders would also have to to local tax offices to assist with collections, 16 percent to assume responsibility for frank and open evaluation of how provincial governments, and 65 percent to local govern- the system is performing and in taking steps to put things ments (Lewis 2002). right" (Royal Kingdom of Cambodia 2001, p. 1). 15. The coefficient of variation in literacy rates declined from 6. The OECD-INES method clearly distinguishes between de 0.11 in 1989 to 0.08 in 1994. jure and de facto decision-making power: "The descrip- 16. There was no change in the coefficient of variation of tions of `at what level' and `how' educational decisions are enrollment rates across the regions. made reflect the actual decision-making process. In some 17. The creation of ARMM was a direct result of the 1996 cases, a higher level of government may have formal or peace treaty between the national government and the legal responsibility for decision-making, but in practice, Moro National Liberation Front. Republic Act No. 9054-- that level of government delegates its decision-making the law creating ARMM--contains detailed provisions that authority to a lower level of government. In describing the pertain to the governance, regulation, and funding of actual decision-making process, the lower level of govern- human development sectors in the region. However, the ment is identified as the decision-maker. Similarly, a higher 1991 Local Government Code had already mandated devo- level of government may provide a lower level of govern- lution of many functions and responsibilities related to ment with choices in a particular area of decision-making human development--notably health and social welfare (e.g., the selection of textbooks for particular course). In services--to provinces, cities, municipalities, and barangays that case too, the lower level of government is the actual (World Bank 2003b). decision-maker, but within a framework established by a 18. These comparisons exclude a few provinces where the higher level of government. Finally, there are cases in which samples tended to be in urban areas. According to the one level of government may have the responsibility for an SUSENAS Work Manual (Government of Indonesia 2002), individual decision, but inaction by the higher level results "Because of the unfavorable security situation, in the fol- in a decision being made by a lower level within the educa- lowing provinces/regencies SUSENAS 2002 is only con- tional system" (OECD 1998, p. 407). ducted in Banda Aceh (Aceh), Ambon (Maluku), Ternate 7. Although references are frequently made to state-level (North Maluku), Sorong (West Irian Jaya), Timica (Central decentralization in Brazil, the process has progressed at Irian Jaya), and Jayapura (East Irian Jaya)." Because the such a rapid pace with FUNDEF's implementation that, as samples were not representative, enrollment rates in these municipalities and mayorships are given responsibility for provinces appear to be surprisingly higher than in other the first four years of basic education, discussing decentral- provinces. ization at the state level does not fully reflect the national 19. Education systems suffer from ineffective and substandard context. schools, persistent shortages of good textbooks, and unpre- 8. Local government expenditures on education rose nearly pared and absent teachers. The nondelivery of publicly 14-fold, from = 0.8 billion in 1991 to = 11.6 billion, in 2001 P P supplied textbooks and chairs at the beginning of each (Manasan 2002). school year in several of these countries deprives millions 9. The Local School Board is cochaired by the local chief exec- of children of the chance to do better in school. In the utive and the division superintendent. Other members Philippines, this problem was estimated at 30­60 percent include the chair of the education committee of the local of total contracts with the education central agency legislature, the local treasurer, a representative of the fed- (Chua 1999). eration of local youth councils, the president of the local The proportion of teachers who are absent from the federation of teacher-parent associations, a representative classroom is too high, according to a recent survey of teach- of the local teachers' organization, and a representative of ers in several countries. For example, in Indonesia, 17 per- the nonacademic personnel of the local public schools cent of teachers in primary schools were not on the school (Manasan 2002). premises during school hours. And too many teachers do 10. Because of a shortage of teachers at the local level, many not know their subject matter better than their students, as local governments hire and pay supplementary teachers indicated by the performance of teachers on tests con- despite the fact that this function is one of the primary ducted for a recent study of the quality of primary schools responsibilities of the central Department of Education. in Vietnam (World Bank 2004b). These problems are also 11. This high percentage may be due to the unclear distinction found in a much wider set of countries for which policy between fees and voluntary contributions. and program evaluations do not point conclusively to gains Education Reforms in East Asia: Policy, Process, and Impact 205 from higher per pupil spending or from investments in use exam results to hold providers accountable. However, specific school inputs (Glewwe 2002). the new exam system does not include primary school, and 20. The newly established Academic Accountability Council, does not provide information about the performance of along with the Office of Accountability, was jointly respon- children below the ninth grade. Confronted with this, sible for overseeing a system of review and analysis of school provinces and districts can create their own complementary performance, while local school councils lost some of their testing system using guidance available from the National independence. Evaluation Center. 21. See the Chicago Public Schools website: www.cps.k12.il.us/. 28. Some have argued, however, that even when reforms 22. Paes de Barros and Silva Pinto de Mendonça (1998) con- involve quality issues, unions still oppose them, given that ducted a broad evaluation of decentralization across such reforms require more effort and political sacrifices most Brazilian states. Their research base includes all (Corrales 1999). Similar debates stem from the joint func- geographical units in the country except the Federal Dis- tion of unions as both professional organizations aiming to trict, northern states, and the state of Alagoas. The study promote efficacy and public knowledge and agents of col- examined data between 1981 and 1993 and included 220 lective bargaining (World Bank 2004c). observations. 29. In Mexico, given strong leadership by the teachers' union, 23. The measures of educational performance used included the government quickly realized that it could neither con- the repetition rate (school census); the proportion of front nor ignore it, and so openly included it in the process children outside school and two measures of grade-level lag of negotiation on greater decentralization. Unions in Chile (National Household Survey Sample, or Pesquisa Nacional are also politically powerful and "have systematically stood por Amostra de Domicílios); and student achievement in the way of true curriculum reform and teacher account- (National Basic Education Evaluation System, or Sistema de ability, and the political class has been unwilling to con- Avaliaçăo do Ensino Básico). front them" (Edwards 2003). Less contentious than in 24. The 10-year-old Basic Education Quality Improvement Mexico and Chile, reform in Bolivia has shown progress in Project (Próqualidade) aimed to strengthen school auton- negotiating with teachers. Nonetheless, teachers' unions omy by providing managers in central and regional offices are also extremely powerful and reform has stalled at sev- with tools designed for information-based decision mak- eral points. In contrast, unions in Brazil did not stall ing; and help school directors assume their new roles as reform; union representatives were active stakeholders both leaders and managers in a more decentralized state in reform negotiations, and decentralization progressed to education system. The project also aimed to increase teach- the school level. And, finally, in El Salvador, teachers' ers' access to training opportunities; deliver packages of unions took a strong stance against the EDUCO model, instructional materials to public schools; and upgrade but government dissuaded them with evidence of innova- facility management to ensure equitable access of pupils to tion in teaching practices (Marques and Bannon 2003; classroom time as well as rational use of school space. Stavenhagen 1999). 25. The LEAs were created according to population distribu- tion and density, number of institutions, geographic char- acteristics, sociocultural considerations, and the extent to Bibliography which planned areas overlapped with existing district boundaries. Ahmad, Junaid, Richard Bird, and Jennie Litvack. 1998. 26. This act was supported by the National Decentralization "Rethinking Decentralization in Developing Countries." Committee and part of the Decentralization Action Plan PREM Sector Studies Series. Washington, DC: World Bank. passed by Parliament in early 2002. Akita, Takahiro, and Armida S. Alisjahbana. 2002."Regional 27. In China, the monitoring and evaluation system is quite Income Inequality in Indonesia and the Initial Impact of the comprehensive. Throughout compulsory education, stu- Economic Crisis." Bulletin for Indonesian Economic Studies dents must take exams and tests following each semester 38 (2): 201­22. and school year to graduate. In primary schools, students Asian Development Bank. 1999. Philippine Education for the 21st must pass tests in Chinese and mathematics, while tests in Century. 1998 Philippines Education Sector Study. Manila: the remaining subjects are usually used as checkup. In sec- Asian Development Bank. ondary schools, exam subjects reflect the general subjects Azfar, Omar, Satu Kahkonen, and Patrick Meagher. 2001. taught to a given graduating class, while exams in the "Conditions for Effective Decentralization Governance: A remaining subjects are again used for checkup (China Edu- Synthesis of Research Findings." College Park, MD: Univer- cation and Research Network 2003). Junior secondary sity of Maryland, IRIS Center. school graduates seeking promotion to senior secondary Bray, Mark. 1999. "The Private Costs of Public Schooling: schools must pass locally organized entrance exams prior Household and Community Financing in Primary Educa- to full admission. tion in Cambodia." Paris and Buenos Aires: International In Indonesia, final exams are set at both the national Institute for Educational Planning. and school levels. While the national portion is based on Cheng, Kai-ming. 1997. "The Meaning of Decentralization: multiple choice, the school portion is based on laboratory Looking at the Case of China." In International Handbook of work, written portfolios, or demonstration of a given set of Education and Development: Preparing Schools, Students, and skills. Students are tested on core knowledge in three main Nations for the Twenty-First Century, eds. William Cum- subject areas: mathematics, Indonesian, and English. At the mings and Noel F. McGinn. New York: Pergamon Press. central level, test results play a strong role in resource alloca- China Education and Research Network. 2003. www.edu.cn/ tion. At the school level, they provide comparable informa- HomePage/english/index.shtml. tion on the relative strengths and weaknesses of individual Chicago Public School System. n.d. "Chicago Public Schools: schools. At the client level, parents and the community can Every Child, Every School." www.cps.k12.il.us/. 206 East Asia Decentralizes Chua, Yvonne T. 1999. Robbed: An Investigation on Corruption in King, Elizabeth, and Berk Ozler. 1998. "What's Decentralization Philippine Education. Manila: Philippine Center for Inves- Got to Do with Learning? The Case of Nicaragua's School tigative Journalism. Autonomy Reform." Washington, DC: World Bank, Devel- Cordeiro Guerra, Susana. 2003. "A Proposal for Teacher opment Research Group. Empowerment in Minas Gerais." Cambridge, MA: Harvard Kingdom of Thailand. 1999. National Education Act, B.E. 2542. University. Bangkok: Office of the National Education Commission. Corrales, Javier. 1999. "The Politics of Education Reform: Bol- Leithwood, Kenneth A., and Teresa Menzies. 1998. "Forms and stering the Supply and Demand--Overcoming Institutional Effects of School-Based Management: A Review." Educa- Blocks." Education Reform and Management Series 2 (1). tional Policy 12 (3): 325­46. Washington, DC: World Bank. Lewis, Blane D. 2002. "Revenue-Sharing and Grant-Making in Delannoy, Françoise. 2000. "Education Reforms in Chile, Indonesia: The First Two Years of Fiscal Decentralization." In 1980­98: A Lesson in Pragmatism." Education Reform and Intergovernmental Transfers in Asia, ed. Paul Smoke. Manila: Management Series 1 (1). Washington, DC: World Bank. Asian Development Bank. Edwards, Sebastian. 2003. "How Chile Can Make the Most of Its Loehr, W., and Rosario Manasan. 1999. "Fiscal Decentralization U.S. Trade Deal." Wall Street Journal, January 3. and Economic Efficiency: Measurement and Evaluation." Gaynor, Cathy. 1998. Decentralization of Education: Teacher IMCC Consulting Assistance for Economic Reform II. Management. Directions in Development Series. Washing- Cambridge, MA: Harvard Institute for International Devel- ton, DC: World Bank. opment and U.S. Agency for International Development. Glewwe, Paul. 2002."Schools and Skills in Developing Coun- Lopez-Acevedo, Gladys, Vicente Paqueo, and Suhas Parandekar. tries: Education Policies and Socioeconomic Outcomes." 2003. "On the Use of Transparent Formulae to Allocate Journal of Economic Literature 40 (2): 436­82. Federal Education Transfers." Washington, DC: World Bank, Government of Indonesia. 1998 and 2002."Survei Sosial Latin America and the Caribbean Region. Ekonomi Nasional" [SUSENAS]. Jakarta: Bureau of Central Machado, Maria Aglaę de Medeiros. 2002. "School Autonomy in Statistics. Brazil: General Overview and Selected States." Brasília: ------. 2000. Statistik Indonesia [Statistical Yearbook of Indone- World Bank. sia]. Jakarta: Badan Pusat Statistik. Manasan, Rosario. 2002. "Education Financing and Service ------. 2001. "Decentralization of Education." National Com- Delivery: Selected Issues." Background Paper for Philippines mittee for Education, Subcommittee III, draft report. Trans- Public Expenditure: Procurement & Financial Management lated into English. Jakarta: Government of Indonesia. Review 3 (annex VI). ------. 2003. Act of the Republic of Indonesia Number 20, Year Manasan, Rosario, and John M. Atkins. 2004. "Normative 2003, on National Education System. Jakarta: National Financing in Basic Education." Manila: Philippine Institute Gazette. of Development Studies. Guedes, Andrea, Thereza Lobo, Robert Walker, and Ana Lúcia Marques, José, and Ian Bannon. 2003. "Central America: Educa- Amaral. 1997. Gestión Descentralizada de la Educación en el tion Reform in a Post-Conflict Setting--Opportunities and Estado de Minas Gerais, Brasil. Grupo de Desarrollo Challenges." Conflict Prevention and Reconstruction Work- Humano América Latina y La Región del Caribe, informe ing Paper. Washington, DC: World Bank, Social Develop- no. 11. Washington, DC: World Bank. ment Department. Hannum, Emily. 2002."Educational Stratification by Ethnicity Martin, Michael O., Ina V. S. Mullis, Eugenio J. Gonzalez, Steven in China." Demography 39 (1): 91­117. J. Chrostowski. 2004a. TIMSS 2003 International Mathemat- Hanson, Mark E. 2000. Democratization and Educational Decen- ics Report: Findings from IEA's Trends in Mathematics and tralization in Spain. Education Reform and Management Science Study at the Fourth and Eighth Grades. Boston: Inter- Series. Washington, DC: World Bank. national Association for the Evaluation of Education Hanushek, Erik A. 1986. "Production and Efficiency in Public Achievement (IEA). Schools." Journal of Economic Literature 24 (3): 1141­77. ------. 2004b. TIMSS 2003 International Science Report: Find- Hawkins, John N. 2000."Centralization, Decentralization, Recen- ings from IEA's Trends in Mathematics and Science Study at tralization: Educational Reform in China." Journal of Educa- the Fourth and Eighth Grades. Boston: International Associa- tional Administration 38 (5): 442­55. tion for the Evaluation of Education Achievement (IEA). Hsieh, Chang-Tai, and Miguel Urquiola. 2001. "When Schools McEwan, Patrick J., and Martin Carnoy. 1999. The Effectiveness Compete, How Do They Compete? An Assessment of Chile's and Efficiency of Private Schools in Chile's Voucher System. Nationwide School Voucher Program." elsa.berkeley. edu/ Stanford, CA: Stanford University Press. ~chsieh/c10.pdf. McMahon, W., with N. Suwaryani, Boediono, and E. Appiah. James, Estelle, Elizabeth King, and Ace Suryadi. 1996. "Finance, 2001. Improving Education Finance in Indonesia. Jakarta: Management, and Costs of Public and Private Schools Ministry of National Education, Institute for Research in Indonesia." Economics of Education Review 15 (4): 387­98. and Development, Policy Research Center; UNICEF; and Jimenez, Emmanuel, and Vicente Paqueo. 1996. "Public for Pri- UNESCO. vate: The Relationship between Public and Private School Murnane, Richard, Judith Singer, John Willett, James Kemple, Enrollment in the Philippines." Economics of Education and Randall Olsen. 1991. Who Will Teach? Policies That Review 15 (4): 389­99. Matter. Cambridge, MA: Harvard University Press. Jimenez, Emmanuel, and Yasuyuki Sawada. 1999. "Do Mutebi, Alex M. 2003."Thailand's Decentralization Experiment: Community-Managed Schools Work? An Evaluation of El Evolution: Dimensions and Challenges." Singapore: National Salvador's EDUCO Program." World Bank Economic Review University of Singapore, Lee Kuan Yew School of Public 13 (3): 415­42. Policy. Kerr, Graham. 1999."Local Technical and Managerial Capacity." Namo de Mello, Guiomar. Forthcoming. "Introduçăo: Breve In Decentralization Briefing Notes, eds. Jennie Litvack and Histórico." Background paper for Lemann Foundation. Jennica Seddon. Washington, DC: World Bank Institute. http://www.fundacaolemann.org.br. Education Reforms in East Asia: Policy, Process, and Impact 207 National Statistical Coordination Board. 2003. Philippine Statis- Waterreus Ib. 2003. "Lessons in Teacher Pay: Studies on Incen- tical Yearbook. Manila: National Economic and Development tives and the Labor Market for Teachers." Dissertation. Authority. Amsterdam: University of Amsterdam. OECD (Organisation for Economic Co-operation and Develop- Weist, Dana. 2001. "Thailand's Decentralization: Progress and ment). 1998. Education at a Glance: OECD Indicators. Paris: Prospects." Paper presented at the KPI Annual Congress IIII OECD, Centre for Educational Research and Innovation. on Decentralization and Local Government in Thailand, Paes de Barros, Ricardo, and Rosane Silva Pinto de Mendonça. November. 1998. "The Impact of Three Institutional Innovations in West, Loraine A., and Christine P. W. Wong. 1995."Fiscal Decen- Brazilian Education." In Organization Matters: Agency Prob- tralization and Growing Regional Disparities in Rural lems in Health and Education in Latin America, ed. William China: Some Evidence in the Provision of Social Services." D. Savedoff. Baltimore, MD: Johns Hopkins University Press. Oxford Review of Economic Policy 11 (4): 70­84. Pandey, Raghaw. 2000. "Going to Scale with Education Reform: ------. 1997. "Equalization Issues." In Financing Local Govern- India's District Primary Education Program, 1995­99." Edu- ment in the People's Republic of China, ed. Christine P. W. cation Reform and Management Publication Series 1 (4). Wong. Hong Kong: Oxford University Press, pp. 283­311. Washington, DC: World Bank. Winkler, Donald, and Alec Gershberg. 2000."Education Pascoe, Susan, and Robert Pascoe. 1998. "Education Reform in Decentralization in Latin America: The Effects on the Australia, 1992­97." Education Reform and Management Quality of Schooling." In Decentralization and Accountabil- Publication Series 1 (2). Washington, DC: World Bank. ity of the Public Sector, eds. Shahid Javed Burki et al. Pro- People's Republic of China. 2001. Essential Statistics of Education ceedings of the 1999 Annual World Bank Conference on in China, Vol.1. Beijing: Ministry of Education, Department Development in Latin America. Washington, DC: World of Development and Planning. Bank. ------. 2002. China Statistical Yearbook, Vol. 21. Beijing: World Bank. 2000. "China National Development and Subna- National Bureau of Statistics, China Statistics Press. tional Finance: A Review of Provincial Expenditures." Prawda, Juan. 1992. "Educational Decentralization in Latin Washington, DC: World Bank, East Asia and Pacific Region, America: Lessons Learned." Washington, DC: World Bank, Poverty Reduction and Economic Management Unit. Latin America and the Caribbean Region, Human Resources ------. 2003a."Decentralizing Indonesia." Regional Public Division. Expenditure Review overview report. Washington, DC: Rodríguez, Alberto, and Kate Hovde. 2002. "The Challenge of World Bank, East Asia and Pacific Region, Poverty Reduc- School Autonomy: Supporting Principals." Washington, DC: tion and Economic Management Unit. World Bank, Latin America and the Caribbean Region, ------. 2003b. "From Cyclical Recovery to Long-Run Growth: Department of Human Development. Regional Overview."Washington, DC: World Bank, East Asia Ross, Steven M., William L. Sanders, S. Paul Wright, and Sam and Pacific Region, Poverty Reduction and Economic Stringfield. 1998. "The Memphis Restructuring Initiative: Management Unit, East Asia Update. Achievement Results for Years 1 and 2 on the Tennessee Value- ------. 2003c."Human Development for Peace and Prosperity Added Assessment System (TVAAS)." Memphis, TN: Univer- in the Autonomous Region in Muslim Mindanao." Washing- sity of Memphis, Center for Research in Educational Policy. ton, DC: World Bank, East Asia and Pacific Region, Human Royal Kingdom of Cambodia. 2001. Education Sector Plan, section Development Sector Unit. I (1). Phnom Penh: Ministry of Education, Youth, and Sport. ------. 2003d. Thailand Country Development Partnership: Shen, A. 1994."Teacher Education and National Development in Governance and Public Sector Reform. Phase I monitoring China." Journal of Education 176 (2): 57­71. workshop. Stavenhagen, Rodolfo. 1999. "Latin America's Challenge for the ------. 2003e."Philippines: Decentralization and Service 21st Century." Cambridge, MA: Harvard University, David Delivery--From Promise to Performance. Cross-Cutting Rockefeller Center for Latin American Studies. Themes in Local Government Functioning." Washington, Trends in International Mathematics and Science Study. 1999. DC: World Bank, East Asia and Pacific Region, Poverty http://isc.bc.edu/timss1999.html. Reduction and Economic Management Unit. Tsang, Mun C. 2002."Intergovernmental Grants and the Financ- ------. 2003f. "Vietnam: Delivering on Its Promise." Develop- ing of Compulsory Education in China." New York: Colum- ment Report. Washington, DC: World Bank, East Asia and bia University, Teachers College. Pacific Region, Poverty Reduction and Economic Manage- Turner, Mark. 2002. "Whatever Happened to Deconcentration? ment Unit. Recent Initiatives in Cambodia." Public Administration and ------. 2004a. "Education in Indonesia: Managing the Transi- Development 22 (4): 353­64. tion to Decentralization."Indonesia Education Sector Review. UNESCO (U.N. Education, Scientific, and Cultural Organiza- Prepared in partnership with the Government of Indonesia, tion). 2002/2003. "Education Statistics." Paris: UNESCO, Ministry of National Education; and the Indonesia Education Institute for Statistics. Donor Partnership. Jakarta: World Bank. Vasconcelos-Saliba, Alcyone. 2004. "Governance and Account- ------ . 2004b. "The Quality of Education at the End of ability Tools for Accountability: School Report Cards-- Primary School: Achievement in Grade 5--Baseline Data for The Case of Paraná (Brazil)." Presentation at the Interna- the Nation and Each Province," Vol. 2. Washington, DC, and tional Conference on Governance Accountability in Social Hanoi: World Bank. Sector Decentralization. Washington, DC: World Bank. ------. 2004c. World Development Report: Making Services Vegas, Emiliana. 2002. "School Choice, Student Performance, Work for Poor People. Washington, DC: World Bank; and and Teacher and School Characteristics: The Chilean Case." New York: Oxford University Press. Washington, DC: World Bank. Zhang, Tiedao, Zhao Minxia, Zhao Xueqin, Zhang Xi, and Wang Wang, Yidan. 2004. "Governance of Basic Education: Service Yan. 2003."Universalizing Nine-Year Compulsory Education Provision and Quality Assurance in China." Washington, for Poverty Reduction in Rural China." Case study on scaling DC: World Bank. up poverty reduction for Shanghai Poverty Conference. 10 Decentralizing Basic Infrastructure Services George E. Peterson and Elisa Muzzini Basic infrastructure services are those that house- Each of these decision-making points presents holds and businesses tend to use every day. The an opportunity for decentralization, and different primary components are roads and transporta- countries have responded by decentralizing different tion services, water supply and distribution, and stages of the decision-making process. The issue is sanitation--wastewater collection and removal, and further complicated because infrastructure projects solid waste collection and removal. These services are subject to spillover effects and economies of are often bundled together under the heading of scale. When this is the case, there are often substan- "urban services." However, parallel infrastructure tial benefits to be achieved by coordinating projects systems are found in rural areas, where roads, irriga- across subnational governments. Accounting for tion networks, and latrines are critical to life. Irriga- economies of scale and externalities is especially tion systems, in particular, are sometimes underesti- important when decentralizing decision making mated as infrastructure networks.In the Philippines, and responsibility to low levels, and when infra- irrigation systems account for 80 percent of national structure projects cover multiple jurisdictions, such water consumption. as in managing water resources across large water- Efforts to decentralize infrastructure services sheds or trunk roads that connect regions. Promot- raise distinctive issues regarding policy design and ing equity, harmonizing standards, and ensuring implementation. Because of their capital intensive- efficient revenue collection may also argue for lim- ness, these systems require decision making at sev- iting decentralization. eral different stages: Infrastructure services are distinctive in another important respect. Because they are used so widely · Preparing capital investment plans and setting and often, citizens are familiar with their benefits priorities for individual capital projects. and typically have strong opinions as to which · Operating a network system to provide services types of projects and service improvements should and maintaining facilities to sustain the physical have highest priority in their area. Choices among capital. priorities for infrastructure investment often · Financing the system by both mobilizing capital provide local citizens with their first opportunity to to pay for the initial investment and generating participate in public decision making. Participatory revenues to cover operations and maintenance-- choice at the local level comes more naturally with that is, ensuring financial sustainability. small investment projects--which often quickly 209 210 East Asia Decentralizes yield local benefits--than with school curricula and This chapter has two purposes. The first is to health services, which often require professional review the status of efforts to decentralize infra- expertise and longer waiting times to produce structure services in East Asia, with a focus on benefits. Thus, decentralization strategies often countries that are more advanced in the process, include participatory choice in (small-scale) infra- namely, China, Indonesia, and the Philippines. structure investments, not simply to better respond The second purpose is to highlight the efficiency to local service needs but as a deliberate seedbed for gains in providing infrastructure services achieved democratic participation in governance, with the in decentralized settings, and to underline the intention of strengthening civic commitment to main challenges to realizing the full benefits of the entire decentralization program. The payoff is decentralization. judged only partly by whether infrastructure serv- The next section examines the key features of ices improve; equally important is whether mecha- strategies for decentralizing infrastructure adopted nisms for public participation in decision making by China, Indonesia, and the Philippines, and out- strengthen citizen involvement in governance. lines the impact of decentralization on the level of This chapter views decentralization in the infra- infrastructure investment. The third section ana- structure sector as a work in progress. The transfer lyzes the available evidence from East Asia on the of service and investment responsibilities to the efficiency gains from more decentralized forms local level evokes an immediate response based on of providing infrastructure services. The fourth sec- the capacities and institutional practices of local tion reviews East Asian experience in enhancing governments relative to those of the central govern- community-level participation in managing infra- ment. More important, however, this first round of structure projects, and investigates the scope for adjustment will reveal shortcomings in the new scaling up the benefits of community participation arrangements and stimulate responses by national within a decentralized government structure. The agencies, local officials, local citizens, and interna- fifth section compares the approach to financing tional organizations supporting the decentraliza- infrastructure adopted by China, Indonesia, and the tion process. The fact that East Asian countries have Philippines,given their different strategies for decen- experience with both different strategies and tralizing infrastructure. The sixth section discusses implementation periods for decentralizing infra- the critical role of higher tiers of government in structure enhances the opportunity to learn from decentralized infrastructure, drawing on the experi- comparative results. ences of China and of Indonesia and the Philippines, A final introductory observation is appropriate. with the drawbacks of a "missing middle" in the Decentralization tends to be seen as a transfer of architecture of decentralization being particularly responsibilities and revenues from higher-level to manifest in the latter two countries. The last section lower-level governments. Many decisions about highlights key policy issues that have emerged from infrastructure services, however, occur in a broader the analysis of decentralization in the region. context that consists of unbundling the vast powers of previously centralized agencies responsible for Decentralization Strategies investment and service provision. Some of this in the Infrastructure Sector unbundling entails transferring authority to lower levels of government. Other elements may include East Asian countries have followed two broad transferring investment or responsibility for service strategies in decentralizing the infrastructure sec- delivery to the private sector or to public-private tor. China exemplifies a principal-agent approach. partnerships, including partnerships with non- The central government as principal has retained governmental organizations (NGOs); and restruc- and even strengthened its role in setting investment turing the public enterprises traditionally responsi- priorities across and within sectors, and has rein- ble for larger-scale infrastructure services, such as forced this role by setting highly specific targets urban water supply and wastewater removal.Decen- and timetables for infrastructure coverage in differ- tralization typically entails restructuring public ent classes of cities. These targets and timetables enterprises by making them accountable to local have extended beyond physical investments to governments rather than central line ministries. include, for certain services, mandatory adoption Decentralizing Basic Infrastructure Services 211 by all urban governments of specific guidelines for seen as a way of engaging the citizenry in participa- service charges, and, for all services, mandatory tory governance. In a country as poor and rural as separation of asset ownership from service deliv- Cambodia, Seila now accounts for the bulk of local ery. In this respect, the infrastructure sector infrastructure investment.Vietnam has passed legis- remains highly centralized. At the same time, as lation on grassroots participation as part of its public agents for implementing centrally established poli- administration reform. However, the national level cies, local governments have full responsibility continues to set investment priorities for most infra- for actually executing investments and providing structure services and certainly for urban services, services. They also have significant latitude in decid- with provincial and local authorities viewed prima- ing how to mobilize funds to pay for capital rily as agents implementing national choices. investment, which they now must finance entirely without central grant support. Local governments Investment Levels and National Standards further have significant latitude in framing devel- opment plans, including where they will site major How has infrastructure investment fared during capital projects and how they will sequence invest- decentralization, and what is the role of national ments to meet nationally imposed standards most standards and investment targets in sustaining cost effectively. investments? Both China and Vietnam, which have Indonesia and the Philippines represent an employed the principal-agent model, have experi- alternative approach. As part of the Big Bang initia- enced extremely high--almost unprecedented-- tives launched in those countries, the central gov- infrastructure expansion within the areas given top ernments assigned virtually complete responsibility priority. During the latter half of the 1990s, China for urban and rural infrastructure services to local assigned top infrastructure priority to road build- governments almost overnight. At the heart of this ing, implemented primarily by provincial govern- transfer was local choice in investment priorities. ments for national and provincial highway net- Decentralization laws emphasize the importance of works, and by local governments for urban civic participation in making investment choices, networks (see tables 10.1 and 10.2). Both levels and specify elaborate procedures designed to attracted some private investment under the direc- ensure that citizens, as well as collective groups like tion of provincial and local authorities. Almost NGOs and civil society organizations, are repre- 85 percent of China's road and highway investment sented in the priority-setting process for capital over the two decades ending in 2000 occurred dur- projects. In fact, decentralization is clearly intended ing 1996­2000. Although national and provincial to serve a dual purpose: to make investment choices highway networks--that is, expressways and class 1 in the infrastructure sector more responsive to highways--expanded most rapidly, all classes grew locally perceived needs and thus more efficient; and at high rates, including major urban roads and to become a vehicle for introducing ordinary citi- class 2 and other local roads. zens to participation in governance. Vietnam has assigned priority to piped water dis- Cambodia and Vietnam are in incipient stages tribution in urban areas, with implementation in of decentralization, with broader policy driven the hands of provincial water authorities and, in the largely by the design of infrastructure programs. Cambodia's Seila Program, introduced in 1996, has created commune development committees in more TABLE 10.1 Road and Highway Investment in China than 1,000 villages and 100 communes, with the program expected to reach three-fourths of all com- Billions of Share of munes by the end of 2004 and the rest shortly there- Period yuan total (%) after (Royal Government of Cambodia 2003b). The program provides government and donor funds for 1981­1989 19 2 1990­1995 153 14 small infrastructure projects selected by citizens at 1996­2000 881 84 the most grassroots level, with mechanisms for transmitting their priorities for slightly larger Source: Mitchell Stanfield & Associates, as projects up to the commune level. Seila is expressly reported in Bellier and Zhou 2003. 212 East Asia Decentralizes TABLE 10.2 Road and Highway Investment in China, by Type 1990 2000 Road type Billions of yuan Share Billions of yuan Share Express-ways 0.01 0.1% 56 28% Class 1 0.02 0.3% 30 15% Class 2 0.20 4.0% 60 30% Class 3 0.84 16.5% 12 6% Class 4 2.60 51% 27 13.5% Unclassified roads 1.43 28.1% 15 7.5% Source: Mitchell Stanfield & Associates, as reported in Bellier and Zhou 2003. TABLE 10.3 Piped Water Coverage within treatment of any kind, the targets imply a massive the Urban Population, Vietnam infrastructure investment program.1 Recognizing the need to mobilize capital for such an effort, the Region Development Planning Commission in October Year North Central South 2002 stated that "cities with existing wastewater and garbage treatment facilities shall start to immedi- 1997 42.0% 30.4% 40.4% ately charge a treatment tariff," and that all other 2000 52.1% 38.9% 42.7% cities should do so by the end of 2003. The tariff Source: World Bank 2002b. "shall cover operations cost and a reasonable invest- ment return" for wastewater treatment plants, to generate revenue for raising commercial investment largest cities, water enterprises attached to local gov- funds. Cities in better economic conditions were ernments. Despite high rates of urban population urged to set tariffs high enough to cover the cost of growth, coverage rates have expanded remarkably constructing wastewater collection networks. in a short time (see table 10.3). In 2000, 41.5 percent In a system with strong upward accountability of the average urban water system was five years old such as China's (and Vietnam's), local authorities or younger. take national investment targets seriously.2 Each China's government has recently given especially municipality incorporates specific targets for infra- high priority to urban wastewater collection and structure coverage into its five-year development treatment, reflecting heightened concern over the plan, approved by the provincial government and condition of urban water bodies. Higher standards ultimately by the Development Planning Commis- imposed on local governments show how the sion. The political careers of local officials in the center transmits signals to local authorities in a Communist Party hinge on meeting or surpassing principal-agent relationship. the goals. As a result of either conscientious plan- In May 2000, China's Development Planning ning or competitive zeal, local officials often set tar- Commission--under the Environment Protection gets that exceed national standards. This is true in Agency in the Ministry of Construction--circulated the wastewater area. Analysis conducted for City "City Dirty Water Treatment," which specified that Development Strategies has found that the majority the wastewater treatment rate in all towns and cities of covered cities--although located in the poorer would be at least 50 percent by 2010. The treatment western provinces--are on their way to meeting rate for cities would be at least 60 percent, and investment targets that surpass state-mandated lev- that for provincial capitals and other major cities at els (Chreod Ltd. 2003). Local development plans least 70 percent. The circular also defined treatment emphasize these ambitions throughout urban quality standards for different types of cities infrastructure: municipalities have set targets even (Government of China 2000). Given that the vast for square meters of green space per capita, trigger- majority of China's cities then had no wastewater ing local investment in parks and other green areas. Decentralizing Basic Infrastructure Services 213 Although driving higher infrastructure investment of infrastructure performance standards in steering levels, standards that emphasize the capacity of budget choices at the local level is clear. capital facilities--whether or not they actually are operating or doing so economically--have often led to significant inefficiencies in operations and main- "Autonomous" Decentralization tenance. The next section discusses these impacts. Countries such as the Philippines and Indonesia China's intergovernmental system is gradually have opted for political decentralization, with local moving toward more sophisticated and meaningful authorities formally recognized as autonomous measures of infrastructure performance, such as bodies. Inherent in their powers is setting priorities outcome measures. In the wastewater area, for for local budgets, including capital budgets. Con- example, the national government now requires cern has arisen in both countries as to whether this localities to test the quality of receiving bodies of type of decentralization can sustain capital invest- water, which are subject to quality standards meas- ment and maintenance. In particular, the transfer of ured along seven dimensions. Some cities voluntar- large numbers of central government employees-- ily sample discharge quality from wastewater treat- subject to wage protection--to local rolls, and the ment plants and have included locally defined legal and political difficulties of raising local rev- targets in their five-year development plans. Thus, enues, subject subnational governments to budget the principal-agent relationship has proved to be pressures. In the face of such pressures, local gov- more than a one-way street. The implementing ernments are thought more likely to maintain agent not only incorporates mandated standards employment levels rather than adjust their budgets into its planning but may also set higher standards to sustain investment. Within capital programs, that become the basis for upward accountability. spending on maintenance and repair is believed to be particularly vulnerable. Displacement of local Lessons Learned. The strong investment perfor- investment has potentially serious consequences. mance of subnational governments in China reflects The World Bank has estimated that, in Indonesia many factors. The same national investment priori- for example, some 60 percent of total development ties communicated to local governments have been expenditures are now a local responsibility (World transmitted to the state-controlled banking system, Bank and Asian Development Bank 2003; World clearing the way for lending that has financed much Bank 2003b, 2003c). of the expansion in infrastructure coverage. Com- The Philippines has the longest experience with mercial banks have lent these funds for three- to five- local budget allocations following decentralization. year periods, creating the need for municipalities to The share of capital spending in city and municipal roll over short-term debt. Local governments are budgets surged in 1993, shortly after new revenue therefore beginning to face high debt service bur- and expenditure assignments took effect (see dens, which may exacerbate the uncertain credit- table 10.4). That occurred because initial revenue worthiness of loan portfolios in the banking system. allocations exceeded the cost of transferred func- China also has a tradition of strong policy direction tions, leading to sizable local surpluses, which were from the center, coupled with a high degree of de then drawn down by capital investment. Since facto freedom in local budget management not 1993, however, the share of local budgets devoted to found elsewhere. capital spending has fallen steadily, with declines One lesson that can be generalized, however, is totaling more than 50 percent. These declines accel- the power of performance measurement and erated in 1998 with the Asian financial crisis and accountability in China. Measurable performance the consequent loss of public revenues before reviv- targets tied to upward accountability have driven ing briefly the following year. Such spending does China's infrastructure investment. The quantified not tell the entire story, as central line ministries standards have sometimes proved unduly rigid, continue to pay for some capital projects at the upward accountability has substituted for account- local level, as do some congressional allocations ability to clients, and national standards have cur- that are treated off-budget. However, the overall tailed local investment choice. But the effectiveness trend in local investment is clear. 214 East Asia Decentralizes TABLE 10.4 Capital Spending as a Share of Local Government Expenditure, Philippines 1992 1993 1994 1995 1996 1997 1998 1999 2000 Cities 10.7% 17.3% 17.6% 16.4% 11.1% 8.8% 8.7% 8.9% 8.2% Municipalities 8.7% 9.8% 8.9% 7.5% 6.5% 6.8% 4.2% 5.1% 4.7% Source: Orial 2002. Indonesia's decentralization process is too young The first step is agreement on performance meas- to draw comparable conclusions about its impact on ures in the infrastructure sector. National guidelines investment spending. Development spending on can establish a handful of basic measures reported roads and mainland transportation fell sharply dur- by all local authorities and monitored by central ing the Asian financial crisis--from 15 percent of institutions. Such measures would include funda- total development spending in 1994­97 to a little mentals such as: more than 5 percent in 1999­2001--before recover- ing somewhat. Another potential factor affecting · Hours per day or week of water provision. this slowdown is significant deconcentrated spend- · Coverage of road networks (kilometers per ing by line ministries, and the difficulty of adapting 1,000 persons, kilometers of road per square the DAK (dana alokasi khusus, a conditional equal- kilometer of territory); quality of roads (percent ization grant) and providing funds within a decen- in good condition, percent of all-weather roads). tralized environment.TheWorld Bank has expressed · Wastewater removal rates. concern that road maintenance at the Kabupaten · Expenditure on road maintenance as share level has suffered from underfunding, and that of gross domestic product (GDP) or kilometers decentralization may exacerbate neglect of road of roads. maintenance, with consequences that are not imme- · Affordability of transport services (freight rate diately visible (World Bank 2003d). Whether local per ton per kilometer, average bus fare per pas- maintenance and repair budgets have actually suf- senger per kilometer). fered disproportionately from local budget adjust- ments under decentralization is unknown. However, The absence of such measures handicaps both protecting maintenance funds may require a greater national decisions about infrastructure priorities role for professional planners and engineers, as and local attempts to measure and improve invest- opposedtolocalpoliticalofficialsandcivicgroups,in ment efficiency. allocating expenditures. Decentralizationframeworksnowprovideample The apparent decline in local infrastructure opportunities for incorporating these measures into spending in the Philippines, and concern over local an accountability system. At the national level, cate- investment levels in Indonesia, have generated gorical grants such as Indonesia's DAK already debate about the role of national standards and per- recognize performance measures, although so far formance measures. This debate is occurring most they do not actually take them into account. Where vigorously in Indonesia, where decentralization leg- national authorities have clear investment priorities islation calls for national agencies to develop guide- not adequately captured by local decision making, lines rather than mandatory standards promulgated they can introduce standards--either absolute levels via the provinces, whose capacity for oversight has or improvement from a baseline--as a condition of greatly weakened. capital grants. The ratio of matching grants can Upward accountability for complying with reflect both national priorities and local perfor- mandatory infrastructure standards seems funda- mance. Simple benchmarking of infrastructure per- mentally at odds with strategies that place primary formance and transparent reporting of performance importance on local choice in investment projects measures can enhance efficiency. Individual local and priorities. The challenge is to capture the power authorities, for example, can cite the range of actual of performance measurement and accountability costs per kilometer of standardized road construc- within a framework for political decentralization. tion elsewhere in conducting competitive bidding Decentralizing Basic Infrastructure Services 215 for local construction.3 Civic groups can compare will yield efficiency gains. These gains can come basic output levels with those of other local authori- from two sources. First, production efficiency ties in setting up their own reporting systems. implies that local entities can build and operate a For other types of infrastructure projects given infrastructure package less expensively. Cost and service delivery, formal accountability to the savings may derive from cheaper local building citizenry through contractual agreements and per- materials, less expensive local labor, more efficient formance monitoring appears to be the most effec- project design, fewer layers of bureaucratic over- tive way to use standards under autonomous decen- sight, and less corruption, among other sources. tralization. Experience with 24 water utilities in Sustainability is an important aspect of production Indonesia, known as PDAMs, shows that NGOs efficiency. Second, allocative efficiency implies that can monitor business plans that specify improve- local investment priorities will reflect the prefer- ments in water coverage, quality, and reliability-- ences of citizens more than those of the central established after public debate as part of a social government, and that households will therefore contract. Fulfilling these goals then becomes the value each unit of infrastructure spending more basis for raising service tariffs and supporting highly. As this chapter later shows, governments financial sustainability (Urban Institute 2003). Sev- need to balance such efficiencies with possible eral local governments in the Philippines have sim- economies of scale (which for some infrastructure ilarly published performance goals and invited sectors can be significant) and externalities (both monitoring by the nongovernmental sector after negative and positive) across local jurisdictions. public participation in setting priorities for infra- Analysts have made several attempts to test, in structure services (World Bank and Asian Develop- East Asia, the hypothesis that decentralization ment Bank 2003). enhances the efficiency of infrastructure services. Rigorously evaluating the gains from production Lessons Learned. In the politically decentralized and allocative efficiency is a demanding exercise. systems of East Asia, a lack of standardized perfor- Evidence so far suggests efficiency gains, but it is far mance monitoring severely hampers understanding from conclusive, as it is drawn largely from case of local infrastructure. No country now has a rou- studies (this differs from the health and education tine monitoring and reporting system for the sector. sectors,where systematic monitoring and household Such systems can be built gradually and improved surveys are far more common). As a result, we know over time. However, national guidelines should a good deal more about how to extract efficiency quickly establish a rudimentary system of reporting gains from decentralized infrastructure through that can be built into budgeting at all levels. Upward well-designed implementation strategies than we do accountability is more difficult to establish in politi- about broad generalizations as to whether decentral- cally decentralized systems than in systems that ized approaches, on average, are more efficient. remain centralized. Decentralized systems need to incorporate performance goals into local budgets, Production Efficiency local corporate plans for water utilities, and local development planning--in a form that can be mon- The World Bank has reported that village infra- itored by civil society organizations. Quantitative structure constructed under Indonesia's Kecamatan accountability to informed civic monitors can sub- Development Program, which involves community- stitute for traditional upward accountability, but level planning and implementation, "cost signifi- only if public agencies provide specific and verifi- cantly less--on average about one-third but in able information. many cases more than half less--than equivalent works built through Ministry of Public Works con- tracts." Maintenance costs were also reportedly Efficiency Gains from lower because communities provided the labor. Decentralizing Infrastructure Unfortunately, the analysis supporting this conclu- A fundamental argument in favor of decentralizing sion has not been published, leaving open the ques- infrastructure is that moving decision making on tion of how the study compared costs. In particular, investment and implementation closer to clients such comparisons often do not take into account 216 East Asia Decentralizes the social infrastructure costs of supporting local central Department of Public Works and Highways project development (World Bank 2003e).4 as the most corrupt organization in the Philippines In collaborating with 24 PDAMs in Indonesia, (Azfar et al. 2000). the Urban Institute reported that the PDAMs Because a reduction in corruption costs is an believed they could substantially reduce investment express rationale for decentralizing infrastructure costs when they handled construction, land acqui- services, the findings of household and other sur- sition, project design, and scheduling, as opposed veys asking respondents to compare the severity of to complying with central specifications. Savings corruption at local and central levels therefore hold on projects that produced a given increment in the special interest. In a survey of 468 respondents in capacity of the daily water supply reportedly run as 13 randomly selected Kabupaten/Kotamadya in West high as 50 percent. Some PDAMs have expressed Java, Indonesia, Azfar (2002) found that 29 percent reluctance to accept low-cost loans from central paid fewer bribes following decentralization, while sources, should these become available, if they must only 5 percent reported that they paid more. adopt centrally imposed project standards, believ- Respondents attested to similar changes in the cost ing that the costs of complying with those rules of bribes, or "unofficial payments" (see table 10.5). would more than offset the savings from low-cost At least in the eyes of citizens, corruption becomes credit (Urban Institute 2003). more widespread and costly the further removed Loehr and Manasan (1999) conducted one of the government agencies are from the local level. more ambitious attempts to compare production Although this evidence supports the hypothesis costs for standardized projects in the Philippines. that decentralization reduces corruption, much Drawing on World Bank and other data, the authors depends on how reform is implemented. In the found costs in the range of = 180,000­235,000 when P Philippines, infrastructure projects selected by Local local governments built their own schools, com- Development Councils but built by the Department pared with P 305,000 per classroom for the central = of Public Works and Highways or other central Department of Public Works and Highways.Alonzo agencies tended to trigger a cascading effect of (1998) has reported comparable differentials in the unofficial payments at each layer of government costs of locally versus centrally built classrooms in (Hofman and Kaiser 2002). Analysts have also the Philippines, as well as local savings per kilome- reported examples of corruption and rent seeking ter of road construction. among local councils, including cases where legisla- Corruption is a major source of cost escalation tors have voted themselves large salary increases and in infrastructure projects throughout East Asia. automobiles, and where a local council has voted to Azfar et al. (2000) have estimated that it adds simultaneously approve an investment project and 20­40 percent to the cost of infrastructure projects name the party to be awarded the contract. Thus, in the Philippines. Respondents to a 1999 survey blanket generalizations about comparative corrup- conducted by Social Weather Stations ranked the tion are suspect. TABLE 10.5 Citizen Perceptions of Corruption in Different Layers of Government in Indonesia Rare/ Quite/very Corruption Never infrequent Common common Local government 17.7% 25.4% 23.3% 11.3% Kabupaten/Kotamadya 1.7% 35.7% 28.2% 18.0% Provincial government 0% 19.5% 29.9% 33.3% Central government 0.2% 4.5% 37.0% 48.1% Source: Azfar 2002, p. 7. Note: Percentages do not sum to 100 because the figures exclude "don't know" responses. Decentralizing Basic Infrastructure Services 217 Probably the strongest evidence for gains in pro- completed wastewater treatment plants. Under this duction efficiency from decentralizing infrastruc- yardstick, localities met coverage targets whether ture comes from studies designed to determine or not treatment plants were actually operating, as whether latrines and small-scale water distribution no one measured the volume of treated discharge projects were still functioning and actually used by or the quality of receiving water bodies. Yet in villagers several years after installation. Studies in Changsha, the capital of Hunan province, waste- Cambodia, Indonesia, Lao People's Democratic water treatment plants operated at 50 percent Republic, and Vietnam all concluded that projects capacity for two years because the city simply shut that relied on community consultation on design, off intake valves and diverted incoming flows and community organizations for maintenance and directly into the river. Meanwhile, the municipality oversight, were significantly more likely to be sus- was planning large-scale construction of new tained than projects built from a central design plants. This experience reveals the power of per- without such consultation, or that relied on outside formance measures tied to accountability proce- expertise for maintenance and management (Gross dures: poorly selected performance measures can 2003; Chanthaphone and Lahiri 2003; World fail to capture the intended effects. Similarly, cen- Bank 2002a). Production efficiency overlaps with trally imposed cost targets may jeopardize produc- efficiency in allocative choice and community par- tive efficiency gains if they fail to account for local ticipation (discussed later in the chapter). Rural variations in exogenous costs that are not under Indonesians, for example, were found to have a the control of municipalities. strong preference for pour-flush latrines, which When wastewater treatment facilities feed into were far more likely to remain in operation than the same body of water,coordinating projects across alternative designs. Women assigned a high priority local jurisdictions can yield better water quality at to small-scale water and sanitation projects, and lower cost. The top-down system in China, com- community-scale water management organizations bined with strong decentralized authorities, how- with strong participation by women proved more ever, does not provide enough incentives for local sustainable than organizations operated solely or governments to consider economies of scale and dominated by men. coordination. In the Guangdong River delta, for Such evidence tends to confirm that decentraliz- instance, each of 51 administrative districts has a ing projects to the village or commune level allows wastewater treatment facility, although cooperative community involvement in support of sustainability agreements based on economies of scale could have to emerge. However, simply decentralizing invest- yielded significant savings. In Vietnam, in contrast, ment and management decisions does little to pro- flexibility in local management and timing of water mote sustainability unless primary users participate delivery has resulted in significant drops in water in maintenance and management decisions. use per hectare, revealing the gains from production On a larger scale, the efficient use of capital in efficiency that can result from decentralizing rural municipal utilities relates directly to performance irrigation systems (see box 10.1). measures targeted and rewarded under national accountability systems. One of the apparent para- Allocative Efficiency doxes of China's investment in municipal waste- water treatment plants is that local governments A full argument for gains in allocative efficiency are investing heavily in new plants while existing from decentralization would marshal several lines facilities operate at 50 percent or less of designed of evidence. First, central and provincial officials capacity because of lack of funds for operation fail to correctly identify the spending priorities of and maintenance. This phenomenon has been local citizens. Second, choices on spending priori- reported in Hunan province, among other loca- ties made by local officials, as well as choices made tions (Chreod Ltd. 2002). It reflects the fact that by citizens themselves through participatory proce- infrastructure targets initially included in local dures, better reflect true local preferences. For development plans and monitored by higher-level example, reliance on central quality standards, as in authorities were based on the treatment capacity of the Chinese model, may significantly reduce the 218 East Asia Decentralizes BOX 10.1 Vietnam's Red River Delta: Efficiency Gains from Decentralizing Irrigation The Red River Delta (RRD) in Vietnam has one of Cooperatives have improved allocative effi- the highest-density rural populations in the ciency by making water delivery more flexible in world. The delta depends on irrigation for crop response to farmers' demands. Empirical obser- production. Experience shows the efficiency vations over an irrigation season found that, on gains that can be obtained from decentralizing average, cooperative pumping stations provided management of an infrastructure network, as water flows within 2 days of a request, compared well as links between gains from production and with 11.5 days for centrally managed systems, allocative efficiency. which work off a predetermined rotational sched- Since Vietnam decollectivized its agricultural ule. The shorter delivery time and greater flexi- sector in the 1980s, two types of institutions bility of cooperative supply give farmers more have provided irrigation in the RRD to farmers. choice in cropping patterns, rice varieties, and One is state-owned Irrigation and Drainage growing conditions. Production efficiency can be Management Subsidiary Companies (IDMSCs), measured in terms of water use per hectare-- which centrally manage pumping stations and table 10.1B below shows observed differences in irrigation networks. The other is cooperatives, water use for comparable paddy fields. which operate on a smaller scale through joint Part of the observed differential in water use management and localized pumping stations. reflects the shorter canal networks of local coop- Cooperatives entered the irrigation business in eratives, which reduce water losses and present response to inflexibilities in water provision from fewer opportunities for diversion to illegal users. the centrally managed state companies, which Part of the savings comes from management scheduled uniform water flows in advance efficiency motivated by the desire to save on regardless of local conditions or planting pat- contracted costs for water supply. An important terns. Cooperatives obtain raw water supplies part of the savings, however, comes from through contracts with regional irrigation man- reduced corruption. Staff of the state-owned agement companies. The cooperatives are, in company reportedly made illicit agreements to effect, decentralized alternative suppliers of irri- sell water on the side, adding to the amounts gation services to farmers. In the RRD, IDMSCs pumped per eligible hectare. and cooperatives each serviced about half of the land, facilitating comparisons between the Sources: Fontenelle and Molle 2002; Fontenelle institutions. 2000. TABLE 10.1B Average Volumes of Water Pumped per Hectare, Spring Season 1996, Vietnam (cubic meters) Water used to Water used during Supplier prepare land growing season Total Local station 1,600 2,400 4,000 Centrally managed 3,900 5,900 9,800 Sources: Fontenelle and Molle 2002; Fontenelle 2000. gains in allocative efficiency from decentraliza- preferences vary substantially across locations, pre- tion if it prevents municipalities from adjusting sumably reflecting differences in both local values the quality of service to the preferences of their and conditions.Municipal officials had a statistically constituencies. significant ability to identify local preferences, while Analysts have collected evidence on parts of this provincial officials had no ability whatsoever to argument. For example, Azfar et al. (2000) found identify local preferences--with a negative correla- that Filipino households strongly favor spending tion between their predictions and actual household incremental funds on roads, but that household preferences. In particular, provincial officials vastly Decentralizing Basic Infrastructure Services 219 underestimated local demand for spending on roads has given local authorities responsibility for defining and other local infrastructure. their own procurement rules. This situation has not Household surveys in Indonesia and Cambodia only created confusion and inconsistency across the have also found strong preferences for road con- local government landscape, but has also led to pro- struction as the top investment priority. Interna- curement abuses, such as lack of competitive bid- tionally funded programs that involve community ding and technical evaluation of proposals, steering choice report a significantly higher level of invest- of contracts to particular firms by elected officials, ment in roads than among projects whose outputs and extensive price renegotiation after contracts are are negotiated at the central level. Local develop- awarded. Adoption of a uniform local procurement ment projects that incorporate citizen participa- code should be a top priority for decentralization, to tion in project selection appear to more accurately build in competition and transparency. The Gov- reflect both the general preference for roads and ernment Procurement Reform Act of 2003 in the variations across communities. For example, during Philippines offers a model of such support.An over- the first two years of the Kecamatan Development sight and regulatory board empowered to investi- Program in Indonesia, road projects were by far the gate and punish procurement abuses needs to sup- most popular local investment choice (62 percent), plement formal procedures. compared with bridges (10 percent), irrigation Best-practice local examples suggest other (8 percent), and clean water (7 percent). Follow-up steps for improving procurement. Naga City in the surveys found that even given these percentages, Philippines,for example,now publishes in the news- households felt that too much had been expended paper and on the Internet winning per unit prices on clean water. Evidence from these and other stud- for all procurement contracts. This simple action ies, in short, is that citizens have clear-cut priorities both supports transparency and encourages price for spending, and that distant representatives and competition among suppliers (World Bank and bureaucrats do not grasp these priorities or assign Asian Development Bank 2003). In administering great importance to local priorities. Moreover, the Kecamatan Development Program in Indonesia, local demand for simple infrastructure projects-- the World Bank has found that the simple expedient primarilyroads--ishighcomparedwithalternatives. of requiring that an independent (local) third party Of course,how much deference central governments as well as contracting principals sign off on all con- should pay to household preferences in allocating tracts and procurements,however small,saves funds spending across sectors is an open question, espe- and fosters a culture of transparency. ciallywhensectorslikeeducationandhealthgenerate Excess capacity in infrastructure design is a major positive externalities not fully captured by local resi- source of cost inefficiency. Paradoxically, at a time of dents. Nonetheless, for spending assigned to local large infrastructure backlogs, a number of projects investment choice,evidence supports the conclusion suffer from substantial overcapacity, raising costs that decentralization is closing the gap between local unnecessarily. In some cases, excess capacity has preferences and project selection. resulted from centralized application of standard project designs or per capita use estimates that Lessons Learned do not take local conditions into account. Some More important than the first-round impacts of Indonesian PDAMs are operating at only one-third decentralization on the efficiency of infrastructure of designed capacity because abundant groundwa- services are the lessons that can be learned to better ter sources are available and households prefer to extract potential gains. continue pumping their own water rather than pay Because the costs of corruption are high in pro- for connecting to the piped water system. viding infrastructure, countries should take practi- Many Chinese cities suffer from overestimated cal steps to reduce corruption and other inefficien- water demand because authorities failed to antici- cies under decentralized management. In most East pate the drops in demand that would occur with full- Asian countries, central authorities have tradition- cost water tariffs and the rapid decline in state- ally handled infrastructure procurement, even for owned firms--often the most inefficient water users. locally selected projects supported by significant This situation has left some local governments with local funding.Meanwhile,however,decentralization "take or pay" contracts with private companies, 220 East Asia Decentralizes which require governments to either purchase more and autonomous local governments that are water than consumers will use or displace lower-cost responsible to the citizens . . . and make deci- municipal water with higher-cost supplies. As in sions over delivery of public goods and services. all countries, grant and low-cost loan financing also · Promote the culture and practice of participatory promotes excessive scaling of infrastructure facili- development (planning, management, resource ties. When investment appears to be free to local mobilization) at local level. authorities, they tend to base estimates of future · Contribute to reduction of poverty in the demand on the most optimistic assumptions. Com- country through improvement of service condi- parable projects financed from own-source funds tions . . . and [service] improvement to poor reflect more realistic growth projections and the and deprived ones" (Royal Government of time and cost entailed in tying up funds. Again, the Cambodia 2003a, p. 108). practical lessons are straightforward. Wherever possible, major infrastructure projects should occur Countries have introduced autonomous decen- at the local level after reforms in user prices, so offi- tralization primarily through grants to local com- cials can estimate the impact on demand more accu- munities that allow citizens to directly choose small rately. Uniform per capita projections of use should investment projects and provide for community be scuttled in favor of demand studies based on local management of the projects once installed.The Seila conditions. All infrastructure projects of significant Program has brought citizen choice over small infra- scale should require substantial own-source financ- structure projects to Cambodia; the Kecamatan ing, to create incentives for realistic cost projections Development Program has done the same for sub- and savings. districts in Indonesia; as have barangay investment Predicting a country's optimal infrastructure programs in the Philippines for urban subdistricts. capacity compared with its long-term investment Several other programs follow the model of provid- needs is admittedly complex, given the "lumpiness" ing communities with investment resources that of investments and the uncertainties associated with they can allocate among eligible projects. Such long-term planning. Hence, whereas today's use of programs have typically won strong support from infrastructure facilities points to overcapacity, the the World Bank, the Asian Development Bank, the extent to which that overcapacity is likely to persist is United Nations Development Programme, and difficult to gauge. other international organizations seeking to imprint local citizen choice as the bedrock support for decentralization. Gains from Citizen Participation Experience with citizen participation in East Asia For countries pursuing autonomous decentraliza- raises two critical questions. First, does participa- tion, the hallmark of reform has been the promise of tion in fact improve sustainability and coverage of greater citizen participation in local budget deci- basic infrastructure services at the village, urban sions, investment choices, and development plan- subdistrict, and neighborhood scale? Second, can ning. Such participation is supposed to yield greater experience in direct participation be successfully citizen satisfaction with basic services and more cov- scaled up to larger political units and infrastructure erage for previously excluded groups--particularly networks? As the following evidence reveals, the the poor, ethnic minorities, and women--while lay- answer to the first question is yes, while the jury is ing the groundwork for broader democratic partici- still out on the second question.Despite some prom- pation in national government. ising local experience, scaling up has proved more Cambodia's 2003 Poverty Reduction Strategy difficult than optimistic advocates of participation illustrates the weight of expectations placed on foresaw. decentralization as a tool for achieving participation: "Decentralization has three objectives in Community-Scale Participation Cambodia: A large body of evidence supports the importance · Promote pluralist participatory democracy at of community-scale participation in infrastructure local level . . . by creation of popularly mandated choices and management. This evidence also Decentralizing Basic Infrastructure Services 221 provides guidelines on specific responsibilities that families receive latrine facilities, photos showing communities can handle to extract the maximum their convenience and the families' pride in owning value from participation. For example, in a study them are used to engage the widely dispersed rural of 88 community-managed water supply projects community in discussing the benefits of latrines worldwide, Gross et al. found that: and gain support for their adoption. The literature on decentralized fiscal choice has · Community-based water supply projects which emphasized the gains from allowing clients to were more demand-responsive--that is, those choose among a wide array of project options, con- that involved greater community choice in the strained by either a fixed budget or the requirement type of water supply, households to be covered, that users pay for the service they choose. Many and the method for paying for operations and East Asian experiments in decentralized project maintenance--were more likely to be sustained selection have followed this model, offering local and better maintained. residents a broad initial choice of investment prior- · The more broadly participatory and gender rep- ities. The Seila Program in Cambodia and the Keca- resentative the decision-making procedures, the matan Development Program in Indonesia are two higher the rate of cost recovery. examples that offer relatively open-ended choice. · Communities that set up water management However, some national programs have defined organizations (WMOs) had better project sus- community choice and participation differently, tainability records than communities that par- particularly those that retain strong roles for gov- ticipated only in initial project selection; the ernmental planners and service providers. In the more equitably WMOs represented women and poor rural province of Guizhou, China, for exam- the poor, the greater their sustainability. ple, prospective township clients can choose to · Household contributions to construction were receive private water connections at a tariff level significantly associated with better-sustained that covers operating costs plus some 75 percent of water supply services only if the community capital costs, including all debt service. Less expen- actively participated in project selection and sub- sive alternatives, such as public stand posts, are not sequent management (Gross et al. 2001). offered, nor does the community have a voice in tariff policy or technology. Community consulta- Hopkins' study of 33 sites in Flores in the Philip- tion consists of carefully explaining to residents the pines found higher rates of sustainability where kind of water service they will receive, the tariff planning included both women and men, as well costs, and the procedures for collecting the tariffs, as poor households. The World Bank has reported which entail house-by-house collection (Zhixiong higher rates of sustainability of locally selected in- 2003). The community may then embrace the pro- frastructure projects when project choice reflected gram and its rules or, in principle, express reluc- direct community participation (Hopkins 2003; tance to pay for it. Village committees identify World Bank 2003d). delivery problems and exert collective pressure on Some of the benefits to be reaped from local citizens to pay water bills. However, the plant man- participation and choice require careful listening ager remains in control of all aspects of manage- by outside experts, whether national officials or ment. A portion of his salary is deducted if tariff international advisors. In 2001 Lao PDR adopted a collection rates fall below 90 percent. new policy of allowing communities to choose their hygiene solutions and influence their design. Scaling Up Participation Follow-up work found that villagers primarily val- ued the "comfort and convenience" of latrines The difficulties of scaling up successes in community rather than the health benefits, which were fore- participation are widely recognized (World Bank most in the minds of external experts (Meadley 2003b). Ironically, scaling up community and rural 2003). This finding influenced both the design of village projects has proved easier in some respects in the latrines and the strategy for disseminating the a centralized rather than a decentralized environ- program, which relied on "champion families" ment, where many levels of government exert respected by other families. After the champion authority. Scaling up means multiplying the number 222 East Asia Decentralizes of villages and urban subdistrict communities new proposals, and then submit a priority list to the reached using the same implementation principles. next level of government (Kabupaten/Kota), which Such scaling up is constrained primarily by the adds proposals from technical officials. The project resources of a program and the central government's preferences of the Kabupaten parliament also willingness to support it through its own resources become part of the mix and may override other rec- or agreement with international donors.If the finan- ommendations. Finally, an umbrella system is sup- cial and human resources are replicable, projects can posed to coordinate local investment priorities with be reproduced throughout the country. Seila in provincial and national priorities. As part of this Cambodia is an example of a community participa- structure, the Indonesian government has promul- tion program that is rapidly scaling up to reach the gatedgeneralguidelinesforparticipatoryplanningat entire country with direct central support, without the Kabubaten and Kota level, including open meet- much interaction with other levels of government. ings that bring together representatives of communi- The Kecamatan Development Program in Indonesia ties, NGOs, and technical bureaus as well as munici- is another program that has spread quickly by repli- pal elected officials. The entire process--facilitated cating the same neighborhood approach, with cen- by a government-provided scoring sheet--should tral government and donor support. Only now, in its yield a consensus list of local priorities. third generation, is the program attempting the Case studies reveal that this process plays out more difficult task of integrating priority setting at differently in different locations. Municipal parlia- the community level with the formal planning pro- ments, technical agencies, and NGOs typically have cedures of the decentralized system. different priorities (Pratikno 2002; Indonesian Scaling up community-identified priorities Partnership on Local Governance Initiatives 2002). withinadecentralizedgovernmentstructurerequires Negotiations leading to final project prioritization finding ways to transmit community preferences on at the municipal level depend on the relative clout infrastructure investments to successively higher of these parties and the role the mayor chooses to levels of government. Community "demand" may play. Most case studies have concluded that the include both community-scale projects that need preferences of local elites, the municipal parlia- higher-level financial support and community ment, and technical agencies tend to drown out the views on the priority of village, district, and munici- preferences of community groups. The mandated pal investment projects that affect the community. participation of NGOs has not resolved this issue, The transmission of community preferences about because--instead of representing a consensus of capital projects to higher levels of government has lower-level priorities--NGOs have more often proved problematic. Part of the difficulty stems proved to be splintered advocacy groups for partic- from distrust of representative government and sus- ular priorities. A similar system of planning and picion of the willingness of municipal officials to priority setting operates in the Philippines, includ- respect community investment priorities. This dis- ing the mandated participation of community trust has been compounded by the difficult interface NGOs in local development councils. Similar diffi- between top-down national investment planning culties in sustaining grassroots participation in and bottom-up community and local planning. municipal priority setting have been reported. Both the Philippines and Indonesia have attempted A realistic reassessment of what community par- to address this challenge by directly involving com- ticipation in infrastructure decision making means munity groups in progressively higher levels of deci- at the municipal scale is in order. Intermediation is sion making. plainly required in scaling up from the community Indonesia illustrates the complexity of transmit- level to the municipal level. Elected representatives ting community preferences.5 The urban planning in the municipal parliament provide one form of process begins with village development meetings intermediation; NGOs provide another. In the attended by the village representative council,NGOs, most successful examples of public inclusiveness in and a representative of the subdistrict (Kecamatan). setting infrastructure priorities, NGOs have played A major objective is to submit project proposals to the role of intermediary between community and the Kecamatan subdistrict level. There officials government. This has required continuous involve- review and weed out community proposals and add ment by NGOs, starting with community-level Decentralizing Basic Infrastructure Services 223 meetings and extending to collaboration with the The idea that community participation in setting technical agencies of municipal government. infrastructure priorities will breed a national culture Against this preparatory background, structured of democratic decision making has seen a modest meetings on investment priorities and budget allo- amount of empirical support thus far. Just as scaling cations, such as citizen forums, can succeed. The up infrastructure priorities can be difficult, so can conditions for success, however, are demanding: scaling up expectations about democratic partici- NGOs must be willing to see themselves as partners pation in governance. The experience of allocating with local government rather than antagonists, budgets to communities undoubtedly empowers and local officials must be open to input from insti- households and raises their expectations about tutions outside the political and governmental responsive government. Whether such participa- technical sphere. Clear examples of successful tion helps consolidate national democracy remains implementation of this vision do exist. These to be seen. include the involvement of the Indonesian Partner- ship on Local Governance Initiatives with both Paying for Infrastructure Services municipal government and communities in urban forums in Indonesia, and the culture of partnership Infrastructure services must be financed at two in all decision making in Naga, the Philippines. levels. Capital resources must be mobilized to pay Efforts to include community expression under the for the initial investment in facilities. Then recur- Kecamatan Development Program and ordinary ring revenues must cover the annual cost of opera- decentralized priority-setting procedures in tions and maintenance, plus contributions to serv- Indonesia are other examples. Whether scaling up icing the debt incurred to finance the initial direct community participation in setting invest- investment. In an economically efficient world, the ment priorities is widely workable remains to be full cost of infrastructure facilities, including depre- seen, however. Many communities have reported a ciation, would be recognized and recovered through public stalemate that gives rise to the older pattern user fees and--given positive externalities--explicit of nontransparent decision making by elites. subsidies from government, in the form of capital grants or targeted subsidies. Much of the institutional unbundling of large Lessons Learned infrastructure utilities in East Asia has been moti- Community participation is essential to the success of vated by a desire to generate more reliable financing infrastructure projects at the village and subdistrict streams, and to make infrastructure services more level. Participation is required at the point of attractive candidates for commercial investment or project selection as well as in continued project lending. The latter goal has sometimes conflicted management. Meaningful participation requires with the goal of making monopoly utilities more the involvement of a cross-section of project users, directly accountable to local governments. especially women and minority groups, who are Asia faced a formidable infrastructure pricing often excluded from project management. challenge at the beginning of this decade, as it had Scaling up participation in infrastructure choice the lowest water and sanitation tariffs--both in has proved difficult. Other mechanisms, such as absolute terms and as a percentage of the costs of genuinely representative municipal governments, production--of any other region (see annex 10.1). and NGOs willing to serve as intermediaries to The median urban tariff for water supply report- both municipal government and local communi- edly covered less than 85 percent of operating and ties, are needed to make the process work. maintenance costs, with no contribution to the cost The most promising approach involves NGO par- of capital. The median tariff for sanitation covered ticipation from the start of the priority-setting process an even lower portion of operating and mainte- through the municipal meetings that establish local nance costs. If anything, the tariff ratios reported investment priorities. Only NGO involvement at the by governments are likely to underestimate the true last stage--such as through the Local Development cost gap. Councils prescribed by Filipino law--has proved In examining financing approaches, we can unsuccessful. extend the distinction introduced at the beginning 224 East Asia Decentralizes of this chapter between China and (in incipient infrastructure bank loans--typically three years, terms) Vietnam, on the one hand, and the Philip- sometimes five years. pines, Indonesia, and (in incipient and less clear- In response, the central government took three cut terms) Cambodia, on the other. Aided by strong important steps designed to ready local govern- economic growth, China has devised a broad ments to further finance local infrastructure within model for self-financed infrastructure investment a fiscally responsible framework. First, it announced and market-based capital financing. Both the that most local governments would be responsible Philippines and Indonesia have been handicapped for obtaining capital financing for infrastructure by a stronger impact from the Asian financial crisis investment from the market. State onlending and and less robust economic growth. However, they grants from the proceeds of state infrastructure have compounded the difficulty by retaining cen- bonds would be limited to the economically laggard tral control of lending to local authorities for infra- western provinces--other local governments had structure investment--an anomaly in their other- to be self-financing. Second, to support this self- wise sweeping embrace of decentralization. sufficiency, the government decreed that municipal- ities should adopt full-cost tariffs for water supply, solid waste, and wastewater treatment, on a highly China's Infrastructure Financing Strategy accelerated timetable. Full-cost pricing was defined China has pursued a clearly defined sequential strat- to include all operating and maintenance costs plus egy for financing local infrastructure. Although the debt service and a competitive return on newly model has important weaknesses, it contains les- invested capital. Better-off municipalities were sup- sons for the rest of the region. To finance the first posed to set tariffs to allow for full recovery of sys- wave of investment in local infrastructure net- temwide capital costs. This regime was designed to works, the government relied primarily on local provide adequate revenue streams to cover debt taxes and fees, supplemented by borrowing from service and attract private-sector capital into the international donor agencies and capital grants and local infrastructure sector, via either direct invest- budget assignments from central government. Start- ment or lending. ing in 1998, the government began to borrow heav- In a third--and in some respects most ily from the domestic market to finance infrastruc- interesting--initiative, the government decreed that ture. Between 1998 and 2002 it issued Ą 660 billion local government should be restructured to separate (US$79.5 billion) in infrastructure bonds--some theownershipof infrastructureandotherassetsfrom 30 percent of which was then transferred to local operating responsibilities.This last initiative had two governments, half as subloans and half as grants. goals. First, it was intended to yield more efficient Along with this use of the domestic bond market, management of municipal assets,following so-called the government communicated the priority of local New Government trends established in Australia, infrastructure lending to China's banks, all of New Zealand,andothercountries,inwhichtheasset- which are publicly owned. Short- to intermediate- owning institution levies a capital charge on users to term loans from banks have been a principal source allocate costs more efficiently. In China's case, how- of capital financing for local governments investing ever, a more important motivation was to place in infrastructure. under a single institutional umbrella assets used as One undesirable effect of the surge in local bor- collateral for municipal loans.As part of governmen- rowing to finance infrastructure investment has tal restructuring, only the asset-holding institutions, been a high and rising level of municipal indebted- known as Urban Development Investment Corpora- ness. Under existing arrangements, municipalities tions (UDICs),may now borrow on behalf of munic- had to repay outstanding debt from their general ipal government. The restructuring was supposed to budgets, placing a high degree of strain on their ensure that municipal borrowing did not exceed the finances. This was especially true in light of the collateral capacity of the municipality's asset base. inadequate structure for service fees, which did not Moreover, as only UDICs are legally authorized to recover operating and maintenance costs, much borrow, the restructuring was supposed to insulate less the costs of debt service. Further squeezing the general municipal budget from debt service municipal budgets was the short maturity of claims while providing a powerful incentive to Decentralizing Basic Infrastructure Services 225 UDICstoimplementfeesthatwouldcoverthecostof costs, and revenues that must be shared with higher capital. levels of government. Still, with 40 percent of the In some respects, China's infrastructure finance gross value representing net profit, the net value of reforms are less sweeping than they may at first Changsha's land inventory is some Ą 34 billion, appear. Although municipalities are prohibited which could finance a large part of the municipality's from direct borrowing, they continue to provide 10-year capital investment plan. comfort letters to local UDICs. These in effect com- The potential for converting Chinese land values mit a municipal government to use general budget into infrastructure assets is even greater if one con- revenues or income from municipally owned prop- siders that a large share of local infrastructure erty to help meet the UDIC's debt service obliga- investment is capitalized into the value of munici- tions, should such support become necessary. pally owned land, and that municipalities' land- Although UDICs tap the general corporate debt holdings are far from static. As population growth market, their debt instruments are a form of pushes the urban boundary outward, more and municipal borrowing. The change in the institu- more land reverts from collective rural ownership tional name of the borrower does not relieve local to municipal ownership, providing a continuing public institutions of the debt service burden cre- basis for capturing land value. Rural land at the ated by short-term borrowing to finance long-term edge of urban centers is indeed seriously under- infrastructure projects. From the banking sector's priced, as the replacement cost of rural land is perspective, the large amount of local assets held in based on its agricultural use without the premium the form of loans to local governments represents a reflecting proximity to urban centers. This gives credit risk of unknown magnitude. No defaults on municipalities strong incentives to convert land municipal borrowing from banks have been from rural to municipal ownership. And munici- reported. However, banks routinely roll over short- palities are in fact converting land values into infra- term loans as they come due. Questions remain as structure investment throughout China. In some to whether banks will continue this policy as for- municipalities studied under the City Development eign competition enters the banking sector under Strategies (Cities Alliance), land has financed as WTO rules, how much of their outstanding debt much as 70 percent of local infrastructure invest- municipalities could actually pay under existing ment, either directly through proceeds from leases schedules, and how politically feasible and eco- or indirectly by serving as collateral for infrastruc- nomically rewarding it would be for banks to fore- ture loans.6 However, concerns are growing that close on assets offered as collateral. reliance on revenues from periurban land may lead One finding of potential significance to other to unsustainable urban planning and degrade the East Asian countries has emerged from UDIC quality of urban life, calling for the development of restructuring, however, especially for periods of and integrated approach to land use and urban strong economic performance. Municipal govern- planning at the municipality level. ments in China possess undeveloped or redevel- The combination of full-cost tariffs for water opable land with great market value, which could supply, sanitation, and solid waste, coupled with finance a substantial proportion of the local infra- increments in land value created by road construc- structure investment burden. Changsha, the capital tion and expansion of urban boundaries, provides of Hunan province in China's interior, illustrates a potentially sound basis for financing municipal this situation. The municipality holds title to some infrastructure. 1.33 million hectares of land, valued at the munici- pality's minimum long-term leasing price in 2001 at Infrastructure Financing in the Philippines some Ą 105 billion. Of this total, about Ą 85 billion and Indonesia (more than US$10 billion) corresponds to land not occupied by the municipality itself that could be The structure of lending to local governments for leased. Changsha officials estimate that some 60 per- financing infrastructure has become an important cent of the gross price of leased land represents costs bottleneck to decentralization in both the Philip- for land that must be set aside as matching open pines and Indonesia. Despite plans to graduate space under planning regulations, resettlement creditworthy municipalities and local utilities to the 226 East Asia Decentralizes competitive credit market, the central government the required tariff increases? On the other hand, remains in control of credit channels in both coun- writing off these loans implies a substantial fiscal tries, acting as a monopoly intermediary between loss to the central government as well as an unde- loans provided by international financial institu- sirable precedent for future onlending. These con- tions and local governments. This position has ditions seem to lay the groundwork for loan frustrated development of sustainable sources of restructuring, but reaching comprehensive agree- domestic financing while allowing central govern- ment has proved difficult. ment institutions to restore--through loan condi- tions and discretionary loan approvals--some con- Lessons Learned trol over the local infrastructure sector formally relinquished in the decentralization process. A self-sustaining local credit market is essential for The local credit market in the Philippines illus- successfully decentralizing the infrastructure sec- trates the unequal playing field established by the tor. National government, as the original onlender government. The Development Bank of the Philip- to municipal governments and utilities, needs to pines and the Land Bank of the Philippines obtain have a strategy for developing a domestic local financing from international organizations and credit market from the outset. This strategy from the National Bank at below-market rates. requires three components: Their onlending to local governments is secured by the authority to intercept intergovernmental · Establishing a track record of timely debt repay- revenue-sharing allotments--authority that is not ment by local authorities. The injection of politi- available to private lenders. Reliable local loan cal considerations into debt repayment to gov- repayment to government financial institutions ernment financial institutions creates a credit was intended to introduce commercial banks to risk in local lending that can set back market municipal lending as a creditworthy activity. How- development for decades. The institutional ever, commercial banks cannot match the cost of weight of bad loans deters entry by private funds of the government financial institutions, as financial institutions into the subnational credit they are prohibited from serving as depository market. institutions for municipalities, which would · A policy that promotes replacement of government strengthen their ties to municipal budgets and pro- lending by private lending as quickly as feasible. vide a lower-cost source of financing. As a result, This, in turn, requires a level playing field commercial bank lending to municipalities for regarding revenue intercepts, depository func- infrastructure has yet to get off the ground, despite tions, and other regulations. In other regions, the formal policy of promoting creditworthy internationally sponsored municipal develop- municipalities to the private credit market. ment funds have proved successful by onlending Indonesia illustrates the legacy power of bad to municipalities through commercial banks, loans in thwarting development of a local credit which fully accept the credit risk. This strategy market. As of March 31, 2000, 63 percent of the introduces commercial banks to municipal borrowing accounts of water utilities through sub- lending while giving them access to longer-term, sidiary loan agreements and the Regional Develop- lower-cost funds than are available on the ment Account were reportedly in arrears (World domestic market. This approach requires a will- Bank 2003b). Although the country has launched a ingness on the part of government financial program of debt restructuring, it has made little institutions to introduce decentralization to the headway in straightening out legacy borrowing. financial sector by eliminating their monopolis- Resolution of the inherited debt runs straight to the tic role as sole municipal lenders. fundamental issues raised by political decentraliza- · A policy of substantial capital cost recovery tion. If central authorities made past investment through service tariffs. For services such as water decisions and mandated loan agreements, should a supply and wastewater removal, which cannot decentralized water utility be required to honor directly generate gains in land value, the only that debt, and, if so, how will it recover the debt if reliable recurring source of revenue is service municipal governments are unwilling to impose fees. One of the most useful standards that Decentralizing Basic Infrastructure Services 227 national government can set in the infrastructure without intervention by higher-level government,as sector is model tariff agreements between munic- local governments would ignore these impacts. For ipal authorities and water utilities. Under these, example, local decisions on regulating effluent dis- the municipality agrees to sanction cost-recovery charges into rivers have implications for users in tariffs if the water utility meets performance other jurisdictions that are part of the same catch- targets for service delivery and coverage. ment area. Similarly, spillover effects occur when municipalities are responsible for managing feeder How Far to Decentralize? roads whose benefits accrue to more than one jurisdiction. Decentralization requires unbundling the functions To correct for such spillover effects, the tier of associated with providing infrastructure services government whose jurisdiction encompasses all the and allocating them among different tiers of govern- users benefiting from such services should decide on ment. The optimal level of decentralization will vary investment priorities and allocate resources. with government's policy goals and the types of For example, managing environmental resources infrastructure. This section discusses the mandate of according to water basins is becoming more com- higher tiers of government in the context of decen- mon to correct for externalities in shared water tralized services. resources. The functions entrusted to water basin authorities include managing and conserving water- The Role of Higher Tiers of Government sheds, controlling floods, reducing pollution, and Even with aspects of infrastructure that are essen- licensing water extraction. On the other hand, over- tially local, an argument can be made for allocating use and degradation of natural resources may occur specific functions to tiers of government higher when an intermediate tier of government does not than the municipal level given one of the following take responsibility for integrated water resource conditions: management (see box 10.2). Spillover effects simi- Spillovereffects. Interjurisdictionalspillovers--or larly call for devolving management of secondary externalities--arise when the activities of one juris- road networks to intermediate tiers of government. diction affect the welfare of people in surrounding Economies of scale. A municipality responsible jurisdictions.If municipal services produce spillover for providing basic infrastructure services may be benefits or costs, service provision will be inefficient smaller than the minimum scale required to ensure BOX 10.2 Vietnam: Watershed Management In Dak Lak, Vietnam, groundwater is in high Participatory Irrigation Management Steering demand to feed the expanding cultivation of Committee to provide guidance to the Water coffee plantations. Groundwater resources in the Users Associations, along with supporting com- Ea Tul and Quang Phu catchments have so far mittees at the district level. The water associa- been freely accessible, leading to overuse and tions, which cooperate with local line agencies, degradation of the natural resource base. Grow- have encouraged farmers to view watershed ing competition in water use has led to conflicts problems more holistically and mobilized new between upstream and downstream users. forms of collective action to address overuse and Local agencies have failed to mediate these erosion. conflicts. For example, attempts by irrigation The focus on participatory irrigation manage- officers to introduce irrigation calendars failed ment is seen as a starting point for a more envi- because communes could not consolidate their ronmentally integrated approach to managing cropping calendars to fit the desired schedule. water resources. However, even in the new insti- Under the new Water Law, Water Users Associa- tutional landscape, no effective regional institu- tions have emerged to make decisions and coor- tions regulate access to and use of groundwater. dinate water resources. At the provincial level, the Province People's Committee established a Source: Dupar and Badenoch 2002. 228 East Asia Decentralizes technical efficiency, especially for services that are lagging behind in economic development. In this local in nature but require large capital invest- context, intergovernmental transfers are instru- ments, such as water supply, electricity distribu- mental in ensuring that all localities can afford to tion, and public transport.7 When excessive frag- invest in infrastructure. As an example, the govern- mentation of service provision is a concern, ment (or regulatory authority) may impose a levy clustering municipalities to provide regional on all firms operating in a market, and redistribute services can boost efficiency.8 However, regional the revenues to companies connecting new users in utilities require an institutional interface at a higher poor regions that cannot afford steep user charges. level of government in charge of setting investment Higher tiers of government may also need to retain priorities and regulating services. Spontaneous some discretion in setting investment priorities to coordination across municipalities is indeed diffi- ensure that local projects contribute to national cult to achieve and may be unsustainable when no and regional strategies for reducing poverty. higher tiers play a coordinating role. As an example, Distortion of interjurisdictional trade. Local regu- in Caracas, 23 municipalities agreed to cooperate to lation of basic infrastructure services may affect award a single water concession. However, the interjurisdictional trade, adding transaction costs resulting agreement lacked credibility to investors, for operators. For instance, local regulations gov- and the group received no responsive bids from erning transportation safety may conflict and thus private operators (Triche et al. 1993). limit or distort opportunities for trade. When local Scarcity of human resources. A scarcity of specific regulations impede trade across jurisdictions, there skills may also make multiplying the number of is an economic argument for setting homogeneous service providers undesirable. In such a context, quality standards throughout the area. As an ex- fewer larger entities may be in a better position to ample, in catchment areas cutting across several attract the minimum required skills than more municipalities, water basin authorities may play a numerous, smaller service providers.A similar argu- role in harmonizing environmental standards and ment can be made for limiting the number of regula- regulating inland waterways. Higher tiers of gov- tory entities to enhance their capacity when human ernment could similarly be entrusted with respon- resources are scarce (see, for example, Smith 2000). sibility for setting quality standards for secondary Equity considerations. Fiscal decentralization road networks (see box 10.3). may conflict with equity goals if the poorest regions Destructive competition. Decentralization may have limited leeway to mobilize financing and raise increase efficiency by promoting competition own-source revenues to meet their infrastructure among local governments. However, devolution of needs, such as through local taxation, user fees, and decision-making powers to the lowest tiers of gov- access to capital markets. This may argue for limit- ernment may turn the potential for competition into ing fiscal decentralization to allow higher tiers of a "race to the bottom," where competition among government to redistribute resources to areas local governments to attract foreign investment in BOX 10.3 Indonesia: When Transport Regulations Distort Trade One area where transport regulations may dis- decree (Keputusan) No. 10 1996, while local criminate against outsiders is the introduction of governments issue the third type of license. licenses for use of roads within a certain region. Trucks not carrying licenses are typically fined The Izin Trayek rule in South Sulawesi, Indonesia, Rp 35,000. This regulation clearly discriminates for example, requires that all transport trucks against trucks from other areas, particularly carry one of three specific licenses: for inter- because licenses are not available outside South provincial transport, for intraprovincial trans- Sulawesi. port, and for entering the regencies (Keca- matan). The first two licenses are issued at the Source: Goodpaster and Ray 2000. provincial level in accordance with gubernatorial Decentralizing Basic Infrastructure Services 229 infrastructure can prompt municipalities to bid externalities. The trend toward regional integration down taxes or other regulatory obligations (or bid is therefore an important step toward more effi- up subsidies or regulated rates of return).9 Excessive cient infrastructure service provision. competition may induce inefficient allocation of On the other hand, examples of interjurisdic- resources and overinvestment, with municipalities tional management of shared water resources, building or upgrading ports or other infrastructure where spillover effects also call for the involvement facilities in their own areas to enhance their prestige, of higher tiers of government, are still rare in China. rather than relying on facilities in adjacent regions. The main exception is the recent attempt to pro- Efficiency of revenue collection. The scope for mote shared environmental infrastructure in the decentralizing financial powers may be limited Pearl River Delta (PRD), one of the most complex when the central level can collect budgetary rev- urban systems in Asia. Many sections of the PRD enues more efficiently, and when there is little have extremely poor water quality. The municipali- opportunity for collecting cost-covering charges at ties of the Guangdong province are the highest con- the point of service. This is often the case in the road tributors to PRD pollution, and the provincial sector, for example, where financing comes largely government--through its Environmental Protec- from fuel taxes and vehicle operating fees, which are tion Bureau--has recently announced an eight-year, more efficiently collected by higher tiers of govern- US$5 billion program to invest in wastewater treat- ment. Higher levels need to redistribute those rev- ment facilities. The plan is based on the recognition enues to lower tiers where services are provided. An that investment in environmental infrastructure example is dedicated road maintenance funds should be guided by a regional development strat- financed by user charges collected at the national egy that reflects sound environmental management level (mainly through fuel taxes). Sophisticated cost- and fiscal sustainability. A pilot project will promote sharing formulas can allocate these funds among dif- development of environmental infrastructure for ferent road networks (and corresponding levels of three groups of two or more municipalities, dis- government), and robust accountability mecha- tricts, and towns. A key parallel activity is the PRD nisms can oversee use of the funds. Cleanup Campaign, which has set phased targets for meeting water quality standards. One of the goals of the campaign is to enhance intermunicipal Interjurisdictional Coordination in China collaboration. In China, sustained economic growth is spurring a In China, more effective regional coordination few major cities to develop into metropolitan areas in setting investment priorities and allocating that cut across more than one jurisdiction. These resources is also needed to help avoid excessive areas include the Pearl River Delta region (centered competition among municipalities in the provision around Guangzhou, Shenzhen, and Hong Kong), of infrastructure services. Amid economic transi- the Lower Yangzi Delta region (centered around tion and decentralization, local investment policies Shanghai), and the Beijing-Tianjin region. Coordi- are indeed driven primarily by a growth and com- nated development has started to emerge, as petitiveness agenda. As a result, municipalities tend municipalities have begun to see the benefits of to compete excessively to attract outside invest- regional integration. For example, a pilot exercise in ment in businesses and infrastructure projects. the Pearl River Delta region, led by the Ministry of Their competitive tools are mainly preferential Construction, aims to establish a metropolitan policies such as tax holidays, free land, and dis- planning model to be replicated in other regions. counted land concessions. In this context, lacking An even more ambitious project is the plan to coordination at higher tiers of government, such ad develop a Pan-Pearl River Delta Regional Coopera- hoc policies may unduly distort resource allocation tion and Development Area, which would encom- between municipalities, as well as between stake- pass almost one-fourth of China's territory, includ- holders within a municipality. ing nine provinces, Hong Kong, and Macau. The The Missing Middle: The Case of Indonesia main goal of this regional initiative is to facili- tate the management of highway and railway proj- In Indonesia, Law 22 of 1999 accords provinces ects, which are expected to generate significant two roles: as deconcentrated representatives of the 230 East Asia Decentralizes center, and as autonomous regions. While provinces in practice, these grants have tended to exacerbate can officially coordinate regional policies and rather than reduce regional inequality.11 perform joint tasks on behalf of local governments, the legislative framework provides no hierarchical The Philippines: Another Example relationship between provinces and local govern- of the Missing Middle ments. This has jeopardized the ability of provinces to facilitate cooperation among local governments In the Philippines, the national government is and establish their authority in regional functions. responsible for providing primary infrastructure, As a result, sectors with large externalities and sig- including backbone transmission grids in the power nificant economies of scale, such as watershed man- sector and primary road networks, while cities and agement, have consistently underperformed. More- municipalities are responsible for tertiary infrastruc- over, decentralization has resulted in a multiplicity ture such as roads and water. However, it is unclear of standards at the municipal level, which may dis- which tier of government is responsible for planning tort trade across jurisdictions (see box 10.3). investment and coordinating development of sec- The resulting efficiency losses from the "missing ondary networks that serve more than one local gov- middle" are compounded by the small size of some ernment and involve common resources such as local entities, which suggests diseconomies of scale river basins.While in principle provinces play a coor- and points to consolidation of regions.10 The dinating role among cities and municipalities, they narrow administrative boundaries of local govern- lack the technical and financial resources to do so. ments, combined with the limited role of pro- As a result, no intermediate tier is capable of vinces, have led to suboptimal investment decisions managing water resources shared by several local from a regional and national perspective. The argu- governments. Without a regional body to coordi- ment for strengthening the role of provinces in nate investment, local governments often argue managing road networks, whose benefits accrue to over river basin planning and management, alloca- more than one local jurisdiction, is especially com- tion of water rights, and pollution control, and pelling. Secondary road networks have been largely water-stressed municipalities have had trouble underfunded compared with the need (the country negotiating water rights outside administrative confronted an estimated 15­20 percent funding boundaries. For example, Cebu City reportedly had shortfall in 2000) (World Bank 2004). In this con- great difficulty convincing Bohol province to sup- text, giving provinces greater financial authority ply the water-starved city even though the marginal can broaden their influence on local governments value of water consumption in the city was high. and thus increase investment efficiency. One A similar situation arises in managing secondary approach is to link provincial and national road road networks.While road density in the Philippines funds, and assign the provincial government strate- is among the highest in the region,only 20 percent of gic oversight of all roads (provincial and kapub- the road network is paved. Provincial roads account aten) in the province, as well as responsibility for only 13 percent of total roads, with more than 70 for assessing investment needs and allocating percent consisting of city, municipal, or barangay resources. The provincial road funds would finance roads. Moreover, in 2000, only 21 percent of provin- maintenance and rehabilitation of local networks, cial roads were paved--a rate much lower than that provided that the kapubaten adopt sound road for national roads (62 percent), city roads (77 per- management practices. cent) and even municipal roads (34 percent) In Indonesia, the central government retains (Department of Public Work and Highways 2003). significant control in allocating resources between Under the institutional framework, volunteer coop- jurisdictions. Limited decentralization of revenue- eration among local governments is the only mecha- raising powers can help reduce regional inequality, nism for coordinating management of the secondary given redistributive mechanisms. The intergovern- road network (see box 10.4). mental transfer system includes two equalization Provinces also suffer from an acute lack of finan- grants (the DAK and the DAU, or dana alokasi cial resources. This is due largely to the fact that umum) to fund investment in infrastructure, espe- decentralization has significantly shifted own-source cially services that generate externalities. However, revenue from provinces to cities and municipalities. Decentralizing Basic Infrastructure Services 231 BOX 10.4 The Philippines: Toll Road Management The construction of a circumferential road across to raise revenues once it is operating. The coop- Cabanatuan City and adjacent municipalities is a eration reflected strong leadership from the chief good example of cooperation in planning invest- executives of the three local governments and ment and implementing a project. Cabanatuan their understanding of the benefits of a joint City signed a memorandum of agreement with approach to combating rising urban congestion. the municipalities of San Leonardo and Sta. Rosa that defined their contributions and obligations Source: Gilbert Llanto, mayor of Cabanatuan to this project. The toll road, financed by contri- City, field interview. butions from the local governments, is expected Not only are the taxing powers of provincial govern- drawing on the experiences of China, Indonesia, ments inferior to those of city and municipal govern- and the Philippines. The need to strengthen the ments, but cities do not have to share their tax rev- role of higher levels of government in providing enues with provinces. As a result, provinces depend infrastructure services appears particularly com- mostly on the Internal Revenue Allotment (IRA)--a pelling in Indonesia and the Philippines, where mechanism for transferring funds from the center. there is clear evidence of a missing middle in the Regional Development Councils (RDCs) could architecture of decentralization. The experiences of play a role in integrating regional and local infra- all three countries suggest the following lessons on structure plans with the country's overall infra- how far to decentralize infrastructure services: structure plans.12 However, the majority of RDCs Partnership between national, provincial, and are weak and ineffective in planning and coordinat- municipal governments is crucial to maximizing the ing infrastructure projects. The perception is that efficiency gains from decentralization. Infrastructure RDCs merely act as endorsers of projects initiated services entail a broad set of functions. The extent by regional offices of line ministries or local gov- to which higher levels of government perform ernments, which require international funding or a some of these functions often depends on the char- guarantee from the national government. Thus, acteristics of a particular industry. For example, rather than playing a coordinating role, RDCs are when the main economic argument for involving seen as more concerned with monitoring national provincial authorities is the presence of spillover projects implemented at the local level (Llanto and effects, higher levels of government need to retain a Lasam 2003). planning and coordinating role, while municipali- In the Philippines,as in Indonesia,the IRA is sup- ties may be better positioned to build and operate posed to correct the mismatch between revenue and facilities. When the main concern is excessive frag- expenditure assignments across different levels of mentation, decentralization of service provision government. Although local governments have be- only to the regional level can retain economies of come more and more dependent on the IRA, the scale. Clearly defining responsibilities and provid- transfer system has not contributed to increase ing mechanisms for coordinating all tiers of gov- equity. On the contrary, the IRA formula favors big- ernment are essential to maximizing the benefits of ger and richer local governments at the expense of decentralization. poorer and smaller ones, as it is based on land area A progressive approach to decentralization has and population.Hence,cities and municipalities that merit. Developing countries may need to build or are more populous and have larger land areas enjoy a strengthen institutions at intermediate and lower strong advantage, and richer local governments with levelsof governmenttocoordinateresponsibilities-- larger tax bases receive a bigger share of the IRA. institutions that are often the norm in mature infra- structure industries. This implies that countries may Lessons Learned need to phase in decentralization while building This section has discussed the role of higher tiers capacity. Another argument in favor of a progressive of government in decentralized infrastructure by approach is that once functions are decentralized to 232 East Asia Decentralizes the lowest tiers,creating a role for higher tiers of gov- control of results, and accountability for results. ernment can be very difficult, as municipalities may This lesson is elaborated in detail below. be reluctant to relinquish decision-making and Performance measurement opens the door to effi- revenue-raising powers.However,political consider- ciency gains from decentralized infrastructure ations often play a critical role in designing a decen- because it permits meaningful accountability. At the tralization strategy. For example, while a more grad- local level, standardized cost comparisons, such as ual phasing-in may have been warranted in the cost per kilometer of road construction, can imme- Philippines and Indonesia, political imperatives diately translate into savings when used as a called for a Big Bang approach, under which local guide for competitive procurement. Performance governments assumed responsibility for providing contracting--in which a municipality commits to basic infrastructure services almost overnight. authorizing tariff increases if a utility meets well- Given economies of scale, decentralizing infra- defined performance goals--can upgrade perfor- structure to the lowest tiers of government may lead mance while breaking the deadlock over setting to excessive fragmentation. The risk of excessive tariffs high enough to recover service costs, which fragmentation is particularly high when decentral- has handicapped local investment. NGOs and citi- ization is not conceived as a response to specific zens gain the power of accountable oversight only if problems but rather as a byproduct of wider they can measure performance against quantified reform. The result could be an industry structure targets. that is far from optimal from an economic point The power of upward accountability is evident of view. in the case of China, where measurable perfor- Careful design of intergovernmental transfer mance against state-determined investment targets mechanisms is needed to meet equity objectives. While has driven the infrastructure sector. On the other central intervention may be warranted to redress hand, in politically decentralized systems, perfor- regional inequalities, experiences in Indonesia and mance measures are almost totally lacking in the the Philippines show that limiting fiscal decentral- infrastructure sector. This is in striking contrast to ization has not produced the expected results in the health and education sectors, where client sur- terms of income redistribution. On the contrary, veys and output measures are far more common. intergovernmental transfers have exacerbated Simple, transparently reported measures of infra- regional difference in income, jeopardizing the abil- structure performance tracked locally and used for ity of the poorest regions to finance their infrastruc- local management would enable countries to go far ture needs. Countries need to improve the efficiency in realizing the potential of decentralization. of their equalization mechanisms to address fiscal Community participation at the project level is imbalances across regions. critical. A large body of evidence supports the importance of community-scale participation in infrastructure choices and management. Commu- Key Issues for Policy Makers nity water supply and latrine projects have proven Several key issues stand out from this review of East more sustainable--longer lasting, more fully used, Asia's experience in decentralizing infrastructure. and more financially self-sufficient--when designed Perhaps the most important is the coherence and in partnership with the community and managed alignment of administrative, financing, perfor- by community organizations. The participation of a mance measurement, and incentive policies and cross-section of users in management has been programs. Where alignment exists, anticipated found to be particularly important. Women and the results will be forthcoming, as in the case of the poor are most likely to be excluded from manage- principal-agent arrangements of China and ment, undermining project sustainability. Vietnam. Of course, this can be a doubled-edged The economic literature on decentralized fiscal benefit. If the goals make sense, performance will choice has emphasized the gains from allowing yield desirable outcomes. On the other hand, doing clients to choose among an array of project the wrong thing well is also a possibility. Thus, it is options. On the other hand, some East Asian coun- as important to ensure that coherence and align- tries offer communities a much smaller range of ment favor appropriate local decision making, choices. Rural water projects in China, for example, Decentralizing Basic Infrastructure Services 233 give communities the opportunity to sign on to for local credit, frustrating development of a self- standardized arrangements. Community consulta- sustaining domestic credit market that meets tion consists of explaining to residents the services municipal needs for financing infrastructure. they will receive and the tariff costs they will be Government financial institutions can encourage responsible for, if they participate. Such projects the emergence of a municipal credit market by have high sustainability in China despite the closed removing regulations that give them preference nature of the initial choice. over commercial banks and other lenders. Onlend- Scaling up community participation in municipal- ing of international funds to municipalities via level capital planning is difficult. Decentralization in commercial banks, rather than government finan- the infrastructure sector has been premised on the cial monopolies, would likely speed development value of community choice. Both Indonesia and the of such a market. Philippines have attempted to incorporate commu- Higher tiers of government play a critical role in nity participation in higher-level choices by includ- the architecture of decentralization. A partnership ing NGOs in efforts to set municipal investment pri- between national, provincial, and municipal gov- orities. The results of this experiment have been ernments is crucial to maximizing the benefits of mixed, at best. The presumption that NGOs repre- decentralization, even in infrastructure industries sent community consensus rather than advocate that are essentially local. For example, empower- particular points of view has often broken down in ing provincial governments to perform a planning practice.Urban forums open to all have proved valu- role is essential to correct for interjurisdictional able in stimulating public debate over investment spillovers, while limiting fiscal decentralization priorities, but a difficult vehicle for actually estab- may be warranted on equity grounds to allow for lishing capital budgets. Whether direct community cross-subsidies between geographic groups. The participation in investment decision making over- need to strengthen the role of intermediate tiers comes skepticism about local representative democ- of government is compelling in Indonesia and racy remains to be seen. Substantial differences in the Philippines, where the missing middle has the information available to stakeholders remain a resulted in poor coordination between jurisdic- significant problem, as such asymmetry contributes tions. The consequences are particularly evident to the greater influence of elites in decision making. in the transport sector, where secondary road Efficiency in capital investment is important. networks have suffered from severe maintenance Paradoxically, at a time of great pressure on infra- backlogs as a result of poor interjurisdictional structure investment budgets, large portions of coordination. existing capital remain unused. In China, waste- water treatment plants in some provinces are not functioning for lack of operating and maintenance Annex: Comparing Water Coverage in Different Countries funds while expensive new treatment plants are and Regions being built. In Indonesia, some water utilities have much excess capacity because they failed to take Truly comparable data across countries on infra- into account ample free water from household structure coverage, investment levels, and tariffs are wells. In politically decentralized systems, commu- difficult to produce and generally not available. nity participation in initial project design and Global Water Supply and Sanitation Assessment required local cost sharing can reduce excess capac- 2000, prepared by the World Health Organization ity. In upwardly accountable systems, performance and the United Nations Children's Fund, provides targets need to measure relevant outputs rather probably the most standardized reporting, but even than merely the capacity of capital facilities. the data in that volume are imperfect. Coverage A well-functioning local credit market is an rates reflect access to "improved" water and waste- essential ingredient of a decentralized infrastructure water systems. Tariff and investment ratios are sector. The supply of credit for local investment reported as averages for the decade 1990­2000 and has become a bottleneck to decentralizing infra- therefore do not take into account recent changes. structure. In both Indonesia and the Philippines, Nonetheless, the data do provide a general compar- central authorities retain control over the channels ative baseline for East Asian countries: 234 East Asia Decentralizes TABLE 10A.1 Water Supply Coverage Rates (percent of population covered) Country Year Urban coverage Rural coverage Total Cambodia 1990 -- -- -- 2000 53% 25% 30% China 1990 99% 60% 71% 2000 94% 66% 75% Indonesia 1990 90% 60% 76% 2000 91% 65% 87% Philippines 1990 94% 81% 87% 2000 92% 80% 87% Vietnam 1990 81% 40% 48% 2000 81% 50% 56% Source: WHO and UNICEF 2000. Note: Rapid rates of urban population growth mean that even where urban coverage rates declined, large numbers of households gained access to a municipal water supply. (--) not available. TABLE 10A.2 Sanitation Coverage Rates (percentage of population covered) Country Year Urban coverage Rural coverage Total Cambodia 1990 -- -- -- 2000 58% 10% 18% China 1990 57% 18% 29% 2000 68% 24% 38% Indonesia 1990 76% 44% 54% 2000 87% 52% 66% Philippines 1990 85% 64% 74% 2000 92% 71% 83% Vietnam 1990 -- -- -- 2000 86% 70% 73% Source: WHO and UNICEF 2000. Note: Given the high differentials in coverage between urban and rural areas, one of the most statistically significant ways of expanding national coverage is through rural to urban migration and other sources of urban population growth. (--) not available. Through the Millennium Development Goals, all mental investment devoted to water supply and the East Asian countries have set far more ambitious sanitation. This probably implies, as Global Water targets for coverage during the period 2000­5. Supply and Sanitation Assessment 2000 concludes, Because of decentralization, local governments will that Asian countries gave the water and sanitation be the primary instruments for implementing and sector lower priority, but the situation also reflects financing this accelerated coverage. the higher shares of public sector budgets devoted to investment in Asia. Investing in the Water Supply Investment shares in East Asia in this sector have and Sanitation Sector climbed recently as countries have focused on meet- During the decade 1990­2000, Asia lagged behind ing their coverage targets. China has also set ambi- other developing regions in the share of govern- tioustargetsfortreatingwastewaterbeforedischarge. Decentralizing Basic Infrastructure Services 235 TABLE 10A.3 Median Investment in Water a population of 125,000 or less, could reduce per customer Supply and Sanitation, 1990­2000 operating costs by increasing their scale of operation. (percentage of overall government investment) 8. However, there is often a trade-off between efficiency gains and loss of local accountability, which efforts to identify the optimal area of service provision would need to take into Region Percentage account. 9. This argument is made, for example, in Smith 2000. Africa 5.3% 10. Local governments range in population from 24,000 to 4.1 Asia 3.5% million. The per capita wage bill of local governments seems to suggest that efficiency falls sharply at the level of Latin America and Caribbean 8.3% about 500,000 people. 11. In Indonesia, revenue disparities are significant at all levels Source: WHO and UNICEF 2000. of government. These inequalities are most extreme at the local level, where the richest region accounts for 46 times the revenues of the poorest region. The richest province also TABLE 10A.4 Median Urban Tariff Rate has 32 times the per capita revenues of the poorest province. (US$ per cubic meter) 12. There are nine RDCs, including those in the Cordillera Administrative Region and the Autonomous Region of Muslim Mindanao. Region Water Sewerage Africa 0.35 0.12 Bibliography Asia 0.22 0.14 Alonzo, Ruperto P. 1998. "Local Governance and Poverty Allevi- Europe 0.67 0.59 ation." In Growth, Poverty, and Income Inequality in the Latin America 0.44 0.21 Philippines, eds. A.M. Balisacan and S. Fujisaki. Tokyo: Insti- tute of Developing Economies. North America 0.48 0.41 Azfar, Omar. 2002. "Decentralization in Indonesia: A Report from a Household Survey in West Java." Paper presented Source: WHO and UNICEF 2000. at the Regional University Consortium Conference on Regional Economic Development in a Decentralizing Indonesia. Azfar, Omar, Tugrul Gurgur, Satu Kahkonen, Anthony Lanyi, Tariff Rates and Patrick Meagher. 2000. Decentralization and Gover- nance: An Empirical Investigation of Public Service Delivery in Median urban tariffs for water and sewerage from the Philippines. College Park, MD: University of Maryland, 1990­2000 were lower in Asia than in other Department of Economics, Center for Institutional Reform regions, although Asia's lower rates partly reflect and the Informal Sector. lower production costs. Asia faces a particularly Bellier, Michel, and Yue Maggie Zhou. 2003. "Private Participa- tion in Infrastructure in China." Working Paper No. 2. challenging task in raising tariffs to commercial Washington, DC: World Bank. levels. Chanthaphone, Soutsakhone, and Santanu Lahiri. 2003. "Look- ing Backward to See Forward." In Sustainability Planning and Monitoring in Community Water Supply and Sanitation, Endnotes eds. Mukherjee and van Wijk Washington, DC: World Bank. Chreod Ltd., with the Urban Institute. 2002. China City Develop- 1. As of 2001, only 200 of China's 667 cities treated any waste- ment Strategies: Changsha, Zhuzhou, Xiangtan. Ottawa: water before discharge (Murray 2003). Chreod Ltd. 2. Local officials are evaluated based on their contribution to ------. 2003. China City Development Strategies II: Inception economic growth, which is often interpreted as reaching Missions--Lanzhou, Chengdu, Zhengzhou, Xinxiang, Erdos. investment targets. Ottawa. 3. The city of Naga in the Philippines makes similar use of Department of Public Work and Highways, Philippines. 2003. comparative disclosure at the local level. The city publishes "Roads in the Philippines." Manila: Department of Public all per unit costs from different bidders for local construc- Work and Highways. tion contracts on its website. Dupar, Mairi, and Nathan Badenoch. 2002. "Environment, 4. In the case of the Kecamatan Development Program, Livelihoods, and Local Institutions: Decentralization in 30,000 villagers were hired and trained in project develop- Mainland Southeast Asia." Washington, DC: World ment and 2,000 community facilitators were also hired and Resources Institute. trained. Fontanelle, Jean-Philippe. 2000."Water Management Decentral- 5. This discussion follows Usui and Alisjahbana 2003. ization in the Red River Delta, Vietnam: An Uncompleted 6. A city development strategy is an action plan for equitable Transition Process toward Local Governance." Paper pre- growth in cities and their surrounding regions, developed sented to the eighth biennial conference of the International and sustained through participation, to improve the quality Association for the Study of Common Property, Blooming- of life for all citizens. See Chreod Ltd. 2002. ton, Indiana. 7. For example, Tynan and Kingdom 2004 proved economet- Fontanelle, Jean-Philippe, and François Molle. 2002. "The Intri- rically that smaller water utilities, particularly those serving cacies of Water Pricing in the Red River Delta, Vietnam." 236 East Asia Decentralizes Paper presented to International Conference on Irrigation ------. 2003b. Water and Sanitation Resources: Sector Overview Water Policies, Agadir, Morocco. and Market Structure. Phnom Penh: Ministry of Commerce. Goodpaster, Gary, and David Ray. 2000. "Trade and Citizenship Smith, Warrick. 2000. "Regulating Utilities: Thinking about Barriers and Decentralization." Indonesian Quarterly 3. Location Questions." World Bank Summer Workshop on Government of China. 2000. City Dirty Water Treatment and Market Institutions, Washington, DC. Pollution Prevention and Treatment Policy. Beijing: Ministry Triche, Thelma, Abel Mejia, and Emanueal Idelovitch. 1993. of Science and Technology, Ministry of Construction, State "Arranging Concessions for Water Supply and Sewerage Environment Protection Administration. Services: Lessons from Buenos Aires and Caracas." WS-10. Gross, Bruce. 2003. "Do Project Rules Promote Sustainability Washington, DC: World Bank, Transportation, Water, and and Equity?" In Sustainability Planning and Monitoring in Urban Development Department. Community Water Supply and Sanitation, eds. Mukherjee Tynan, Nicola, and Bill Kingdom. 2004. "Optimal Size for and van Wijk. Washington, DC: World Bank, International Utilities? Returns to Scale in Water--Evidence from Bench- Water and Sanitation Centre. marking." Viewpoint note. rru.worldbank.org/documents/ Gross, Bruce, Christine van Wijk, and Nilanjana Mukherjee. publicpolicyjournal/283tynan.pdf. 2001. "Linking Sustainability with Demand, Gender, and Urban Institute. 2003. "Local Government Water Services Poverty: A Study in Community-Managed Water Supply Project Final Report." Washington, DC: Urban Institute. Projects in 15 Countries." Delft, Netherlands: Water and Usui, Norio, and Armida Alisjahbana. 2003."Local Development Sanitation Program and IRC International Water and Sani- Planning and Budgeting in Indonesia: Update." Prepared for tation Centre. the international symposium on Indonesia's Decentraliza- Hofman, Bert, and Kai Kaiser. 2002. "The Making of the Big tion Policy: Problems and Policy Directions. Jakarta. Bang and Its Aftermath: A Political Economy Perspective." World Bank. 2002a. "Selling Sanitation in Vietnam: What Paper presented at conference on Can Decentralization Help Works?" Vietnam Country Report, WSP-EAP. Washington, Rebuild Indonesia?, Atlanta, GA. DC: World Bank, Water and Sanitation Programme. Hopkins, Richard. 2003. "How Well Did Those Development ------. 2002b. "Benchmarking the Urban Water Sector Projects in Flores Work?" In Sustainability Planning and Vietnam." Hanoi. Monitoring in Community Water Supply and Sanitation, eds. ------. 2003a."China Country Assistance Strategy."Washington, Mukherjee and van Wijk. Washington, DC: World Bank, DC: World Bank. International Water and Sanitation Centre. ------. 2003b. "Cities in Transition: Urban Sector Review in an Indonesian Partnership on Local Governance Initiatives. 2002. Era of Decentralization in Indonesia." Working Paper No. 7. Various papers for the International Workshop on Participa- Washington, DC: World Bank. tory Planning Approaches for Local Governance, Bandung, ------. 2003c. "Decentralizing Indonesia: Regional Public Indonesia. Expenditure Review." Report 26191-IND. Washington, DC: Llanto, Gilberto M., and Joel Lasam. 2003. "Report on the World Bank. Forum on Planning, Programming, and Monitoring." Sub- ------. 2003d. "Indonesia: Selected Fiscal Issues in a New Era." mitted to National Economic and Development Agency and Report 25437-IND. Washington, DC: World Bank. Deutsche Gesellschaft für Technische Zusammenarbeit. ------. 2003e. "Project Appraisal Document for a Third Loehr, William, and Rosario Manasan. 1999. "Fiscal Decentral- Kecamatan Development Project." Washington, DC: World ization and Economic Efficiency: Measurement and Evalua- Bank. tion." Cambridge, MA: Harvard Institute for International ------. 2004. "Indonesia: Averting an Infrastructure Crisis--A Development. Framework for Policy and Action." Washington, DC: World Meadley, Thomas. 2003. "Community Design of Hygiene Pro- Bank, Recent Economic Development in Infrastructure. motion IEC Materials." Presentation to World Bank Water World Bank and Asian Development Bank. 2003. "Philippines: Week. World Bank, Washington, DC. Decentralization and Service Delivery--From Promise to Murray, Bruce. 2003. Presentation to Workshop on National Performance." Manila: World Bank and Asian Development Guidelines for Urban Wastewater Tariffs and Management. Bank. Manila: Asian Development Bank. WHO (World Health Organization) and UNICEF (United Orial, Lydia N. 2002. "Local Government Finance and LGU Nations Children's Fund). 2000. Global Water Supply and Bond Market Development: Philippines." Manila: Asian Sanitation Assessment 2000. www.wssinfo.org/pdf/GWSSA_ Development Bank. 2000_report.pdf. Pratikno. 2002. "Initiating Citizen Participation in Local Policy Zhixiong, Luo. 2003. "Financing Arrangements for Rural Water Making in Indonesia: The Case of Surakata Municipality in Supply and Sanitation Project in China: Experiences the Post Authoritarian Regime, 1999­2002." Gadjah Mada from Guizhou Province." Presentation to World Bank, University. Washington, DC. Royal Government of Cambodia. 2003a. National Poverty Reduction Strategy 2003­2005. Phnom Penh: Council for Social Development. 11 Governance Gone Local: Does Decentr alization Improve Accountability? Jose Edgardo Campos and Joel S. Hellman The case for decentralization is fundamentally an particularly innovative local governments, we hear argument about governance. The case is rooted in much less about the impact of decentralization on two powerful intellectual traditions: the critique of governance in the median region, beyond these economic centralism (especially central planning), shining stars. As attractive as the theoretical foun- and the perceived economic advantages of federal- dations of decentralization may be, the relationship ism. The first tradition posits that decentraliza- between decentralization and governance in prac- tion aligns government decision making more tice is still very much a matter of debate. closely with local preferences, largely because of the Moreover,the decision to pursue decentralization information advantages associated with smaller is largely political, with the underlying economic jurisdictions.1 The second tradition emphasizes the rationale secondary, if not marginal. In many cases, competition among regions sparked by decentral- the decision to decentralize is sparked by strong ization, as local governments have incentives to reactions to a prolonged period of highly authoritar- engage in a "race to the top" to attract capital and ian rule. This has certainly been the case for the two labor, or simply to build their political reputations.2 countries on which this chapter focuses--Indonesia Both strands argue that decentralization will make and the Philippines. local officials more accountable to constituents for Nevertheless, the development community has their performance. generally welcomed decentralization with some Though these arguments are deeply rooted in enthusiasm, and has responded by shifting a signif- theory, surprisingly few empirical studies have icant share of development assistance to local gov- examined the direct relationship between decen- ernments, and to support for the decentralization tralization and governance. And the majority of framework.4 As decentralization projects in East studies that have been done tend to dispute the Asia are still under way or in the early stages of expected governance gains from decentralization.3 implementation, an assessment of lessons from the Though case studies often trumpet the successes of region on the relationship between decentralization 237 238 East Asia Decentralizes and governance is not yet available. Yet there is a FIGURE 11.1 The Accountability Framework certain urgency for learning more about the impacts of decentralization. Experience so far has Policy highlighted a range of governance risks that appear makers to threaten the promised accountability gains from decentralization. This chapter describes five key governance risks that could mitigate the posited advantages of decen- Poor people Providers tralization: capture,clientelism,capacity constraints, competition over the balance of power between Source: World Bank 2004c. levels of government, and weaknesses in the interre- gional information flows that are critical for effective The second and third legs address how they exercise competition. To be sure, some of these risks can that authority, such as by formulating policies, pro- affect all levels of government. But this chapter grams,and projects (see figure 11.1),and by delegat- argues that they may be more prevalent at the ing implementation to the public bureaucracy, civil local level than at the national level. The chapter society organizations, or the private sector. also argues that decentralization itself--if poorly The key idea of the triad is to show that designed--could exacerbate these governance citizens--as clients of public services--have two distortions, undermining any positive gains in routes by which they hold providers accountable for accountability. This is not to suggest that the con- service quality. The first is a "long route," whereby ventional arguments about decentralization and citizens give feedback to policy makers about their governance are wrong. Rather, these experiences preferences, who then control the providers of pub- raise challenges that need to be addressed in design- lic services. The second is a "short route," whereby ing decentralization, assessing its appropriateness in users give direct feedback to service providers, creat- different political contexts, and assisting the process. ing pressure and providing information that helps The chapter is organized in four parts. First, to policy makers hold them accountable for their help guide the analysis, we present a conceptual performance. framework for governance building on the World Decentralization adds an alternative route to this Bank's World Development Report 2004. Second, we accountability framework by shortening the link discuss the implications of decentralization for between policy makers on the one hand and citizens governance. Third, within this framework, we ana- and public service providers on the other.Decentral- lyze the limited experience of East Asian countries ization, proponents argue, brings politicians closer with decentralization, given the above-mentioned to the people, by giving them better information risks. And fourth, we suggest possible avenues for about constituents' preferences and making it easier minimizing these risks, given the numerous con- for constituents to monitor politicians' perfor- straints that these countries face. mance. At the local level, citizens can more easily learn of the activities and programs that their local leaders have promoted and supported, discern how The Foundations of Governance much effort they have devoted to improving public The World Development Report 2004 presents a sim- services,and confirm whether they have delivered on ple triad to illustrate the multiple relationships that campaign promises. In other words, the information constitute a framework for thinking about account- that citizens need to make judgments is more readily ability in delivering public services. The triad accessible under decentralization.Hence,it strength- focuses on three basic relationships: between citi- ens political accountability (see figure 11.2).6 zens and policy makers, between policy makers and Similarly, local politicians, being much closer the bureaucracy (those responsible for providing to the action and having more direct interactions public goods and services),and between the bureau- with the local bureaucracy,are potentially better able cracy as delivery agents and the citizenry as clients.5 to monitor the performance of local agencies. Local The first leg of the triad deals primarily with how leaders can more easily find out whether doctors are policy makers acquire authority and thus power. arriving at work at local health clinics and whether Governance Gone Local: Does Decentralization Improve Accountability? 239 FIGURE 11.2 Decentralization and the Accountability Framework Central Long route government Medium Local route government Paths of accountability Citizens/clients Providers Short route Source: Author. teachers are showing up at community elementary officials need to have enough training, experience, schools. Leaders can also more quickly receive alerts and professional skills to competently make and on security problems that arise in different parts execute responsive public policies. If officials do not of the locality. In other words, decentralization have adequate capacity, then they cannot implement improves bureaucratic accountability as well. their designated tasks effectively. Capacity in this Finally, local communities are potentially better context refers to the skills public officials need to endowed with social capital, and thus face lower deliver on various mandates, the resources (capital costs in organizing into groups. Client power at the and financial) they require to support their efforts, local level is therefore likely to be much greater, and and the systems (such as budgeting systems) citizens are better able to communicate the nature that enable large numbers of bureaucrats to work and location of problems to local politicians. This together effectively. Obviously, the same considera- induces improvements in the compact between tions hold for members and staff of civil society local politicians and the local bureaucracy. organizations. However, this simple approach includes some Accountable government generally requires a important untested assumptions. First, it assumes certain degree of institutional capacity. In particu- that proximity breeds accountability, and that lar, participants need information for auditing, eval- accountability is largely a function of information. uation, reporting, investigations, and prosecution. Yet a range of other political factors also determines They also need processes, skills, and resources to accountability--factors that may not necessarily be provide the infrastructure and create the incentives more prevalent at the local level. Second, the stan- to produce the right information. Accountability is dard approach assumes a frictionless relationship thus circumscribed by both institutional and indi- between levels of government, and thus that local vidual capacity. governments can respond effectively to the con- cerns of their constituents and exercise effective The Risks of Decentralization authority over service providers. Yet competition between policy makers at the center and in the In practice, decentralization does not occur in a vac- localities could constrain the responsiveness of uum. Many factors--including historical trends, local governments to constituent pressures and institutional inertia, and class cleavages--may weaken their capacity to control service providers. undermine the simple links between decentraliza- And third, the simple model does not incorporate tion and improved governance. The next section interregional dynamics and asymmetries that could examines some of these risks. alter the responsiveness of local governments to their own constituencies and influence their con- State Capture at the Local Level trol over service providers. Even more importantly, this approach ignores At every level of government, state capture may the importance of capacity constraints. Government distort political accountability between citizens 240 East Asia Decentralizes and policy makers. State capture refers to "actions Third, countervailing powers representing a of individuals, groups, or firms either in the pub- broader range of public interests, such as the media lic and/or private sectors to influence the forma- and nongovernmental organizations (NGOs), are tion of laws, regulations, decrees and other gov- generally less developed in local jurisdictions. Even ernment policies to their advantage through the in large, geographically and linguistically diverse illicit and nontransparent provision of private countries such as Indonesia, the main media out- benefits to politicians and/or civil servants" (World lets are highly concentrated in the capital city. Bank 2000: xv).7 State capture distorts the chain At best, an emerging but still loose network of of accountability between politicians, service local newspapers and television stations resorts to providers, and constituents through asymmetries "selling news for cash" to survive, regardless of its of political influence. Analysts have only recently veracity. studied the dynamics of state capture at the local In most East Asian countries, years of authoritar- level (Bardhan and Mookherjee 2000; Zhuravskaya ian rule have systematically weakened the develop- 2000). The question is: are local governments more ment of an NGO sector. Though the number of susceptible to state capture than their national NGOs is growing, they still tend to not only concen- counterparts? The limited empirical studies sug- trate in the capital city but also depend on donor gest that they are. support. Some are even government-sponsored and First, state capture thrives in an environment thus cannot conduct their activities at arms length. where highly concentrated interest groups-- LocalNGOnetworksinmostEastAsiancountries-- especially powerful firms and families--dominate while expanding--therefore remain quite weak. the market for political influence, and where politi- With limited resources, low capacity, weak links to cal competition is weak. Local economies tend to be national networks, and significant government more homogeneous, more concentrated, and less interference in their activities, local NGOs still tend competitive than the national economy, creating to play a restricted role in holding local governments fertile ground for dominant economic actors to accountable. engage in state capture. This is particularly true in In many East Asian countries, the question of resource-rich regions in developing countries, who captures whom at the local level is less where local economies depend on a particular state straightforward than in other parts of the world, monopoly or powerful firm. In such cases, the where the private sector is more autonomous. The boundary separating the interests of the region and problem of local state actors using their power to the firm can be murky at best, and local political capture ownership of local businesses or de facto and economic elites are closely intertwined in cash flow rights is particularly serious. This has promoting state capture. been a longstanding issue in China. In Indonesia, a Second, many of the institutions normally preliminary study of regional legislation (known as expected to serve as checks on state capture are perdas) showed surprisingly frequent references to weaker at the local level than at the national level, local pemda companies--those wholly owned by especially in developing countries. Local legisla- district governments, with control exercised by tures in most East Asian countries typically do not local executive officials. Whether local governments yet serve as a significant countervailing force on the areseizingopportunitiestoexpandtheircontrolover executive. Even more than their national counter- local businesses, or businesses themselves are using parts, these legislatures lack technical and support their influence to capture local officials, the impact staff, resources, experience, and training. is generally the same: the exercise of state power For the same reasons, local judicial systems tend favors particular firms, often at the expense of the to be less effective (and more incompetent and cor- region's broader welfare. rupt) in rendering fair judgments in the face of sub- stantial asymmetries of power among contesting parties.8 As a result, the executive branch tends to Expanding Clientelist Roots dominate local polities in decentralized countries, Clientelism--in which politicians distribute with local representative and judicial institutions publicly funded goods to selected members of offering weak constraints. the electorate in return for votes and political Governance Gone Local: Does Decentralization Improve Accountability? 241 support--systematically weakens political account- result, clientelism is generally a phenomenon of ability in a variety of ways.9 It narrows the range of local politics. constituents to whom politicians are responsive. Where clientelism is prevalent, decentralization Through the distribution of "reversible" goods-- can seriously exacerbate inefficiencies and inequal- those that can be withdrawn--to exclusive net- ities in public services. Clientelist governments works, clientelism also gives politicians a mecha- tend to favor investments that generate jobs, which nism to punish voters who do not provide they can then distribute to build patronage net- continuous support. Clientelism also creates disin- works. In a common example, newly empowered centives for groups to develop collective forms of local governments may favor investing in new representation and therefore weakens interest schools as opposed to improving existing schools. group competition. Like state capture, clientelism is As a result, each school is underresourced, under- common to many different types of political sys- mining any positive impact on educational levels. tems and at all levels of government. However, it Indeed, clientelism can distort a range of decisions might be more pervasive at the local level, for about public services--from the service mix, to the several reasons. location of facilities, to standards discriminating in With smaller jurisdictions, local politicians are favor of particular constituent groups at the more likely to engage in clientelist forms of "retail" expense of more overall community welfare. politics to win elections and maintain political sup- port networks. In such jurisdictions, politicians can Amplifying Capacity Constraints more effectively identify individual voters for clien- telist networks and more easily monitor their polit- Decentralization shifts the responsibilities of local ical support, making clientelist "contracts" more government from purely implementing policy to feasible and enforceable.10 Local clientelism redi- both formulating and implementing policy. This rects the arrows linking policy makers, providers, requires a wider range of skills and experience, and clients. Rather than citizens holding local offi- which local politicians and bureaucrats may have to cials accountable, clientelist politics allows politi- develop. Hence, at least in the initial stages of cians to shape constituencies to their own advan- decentralization, local skill levels and policy- tage by selectively providing public services and making processes are likely to lag behind those at other benefits. For example, discretionary distribu- the national level.11 tion of front-line service jobs to political support- Besides skills and processes, capacity also implies ers systematically weakens the potential links appropriate management systems--accounting, between clients and service providers, as job budgeting, procurement, tax administration, audit- recruitment and promotion have little to do with ing, reporting, and personnel management. In performance. many developing countries, national governments Moreover, local elites in more homogeneous continue to struggle with reforming these systems. communities tend to be bound together by a dense Given their relative inexperience and more modest network of familial, ethnic, social, and cultural ties resources, local governments are likely to find estab- that encourage clientelist behavior and, more gen- lishing such systems and processes even more erally, corruption (Tanzi 1995; Prud'homme 1995). challenging.12 (See chapter 7 for more on the prob- The continued presence of aristocratic lineages in lems East Asian countries face in creating personnel the sugar-producing provinces in Central Luzon management systems.) and Western Visayas in the Philippines; the network Resources are often a forgotten or underesti- of tribal leaders in Papua New Guinea; and the mated aspect of capacity. To train personnel, system of ulama in East Java, Indonesia, are all develop systems and processes, strengthen account- ready-made local clientelist networks that build on ability, and ultimately deliver public goods and predemocratic legacies. These legacies have proven services, local governments need funds. Other quite adaptable to new modes of political competi- chapters discuss funding issues--in particular, tion. Such exclusive patron-client networks are far intergovernmental transfers and local taxation--in more difficult to build and maintain in a more more detail. Suffice it to say that efforts to fund diverse and competitive national playing field. As a local government present complex considerations 242 East Asia Decentralizes and challenges with which newly decentralizing prolonged period. While the initial impetus for countries--including most East Asian countries-- decentralizing power in East Asia has been broader are only starting to grapple. political transitions from highly centralized regimes Capacity problems are not limited to institu- or within such regimes, implementation has con- tions and processes within the executive branch; fronted a variety of fiscal and administrative chal- the same constraints affect other actors in the lenges. These challenges often generate conflicting accountability framework. The accountability of incentives at different levels and within different local governments--like that of the national branches of government. The result is that the government--depends largely on the strength of shape of decentralization is constantly evolving, countervailing institutions, such as the legislature reflecting changing political and economic realities, and judiciary and civil society groups, to provide conflicting interests, and shifting priorities. effective inputs and monitor policies. Here again This changing landscape has a number of poten- are reasons for concern that capacity among these tially serious implications for the impact of decen- actors is not as strong at the local level as at the tralization on local governance. First, contests over national level. As suggested, local legislatures in the extent and contours of decentralization between East Asia generally have limited access to indepen- levels of government tend to constrain local auton- dent policy expertise and analysis and therefore are omy. The shifting relationship between local politi- much more dependent on local executives. Local cians and service providers and their national courts generally remain embedded in centralized counterparts blurs lines of accountability. This can hierarchies even in the most decentralized coun- reduce the responsiveness of local governments and tries. As a result, patterns of career advancement weaken their authority over service providers, who draw court officials to the center, leaving local may be responding to incentives from national courts with less capacity. hierarchies. Comparing the capacity of civil society between Second, contests across levels of government the center and the regions is more difficult but also often lead to what might be called imbalanced essential to determine the foundation for local decentralization--especially between power and accountability. As mentioned, the local civil society financial responsibility, distorting incentives at institutions regarded as key checks to prevent state lower levels. In Indonesia, for example, local gov- capture and clientelism are not well developed in ernments have considerable control over expendi- many East Asian countries, especially those in tran- tures for basic public services but little control over sition from previously authoritarian systems, where the civil servants that provide those services. Thus, the legacy of public participation is particularly these governments cannot downsize the workforce, weak. With a few exceptions, NGO penetration at alter remuneration packages, or introduce new the local level is also lower than at the national (and perhaps more meritocratic) recruitment and level. Though no systematic quality comparisons promotion to coincide with changing expenditure exist, it is widely agreed that the quality of the priorities. media beyond the national level drops substantially In many countries local governments also face in most developing countries. In sum, it is not only strong constraints on raising revenues, limiting any local governments that face capacity constraints autonomy they receive over local services. Such relative to their national counterparts, but also the imbalances might not only constrain local govern- other institutions and actors whose participation is ments but also create incentives for them to make essential to holding local governments accountable. inefficient investment decisions, especially if they can shift costs to central budgets (see chapters 5 and 6). Creating Intergovernmental Tension Both of these factors generate substantial uncer- Regardless of the initial enthusiasm with which tainty regarding the distribution of functions and many countries have embraced decentralization, responsibilities, the extent of autonomy, and the when it comes to actual dividing power and author- balance of power between levels of government. ity across different levels of government, decentral- Such uncertainty can weaken accountability at ization generally remains contested terrain for a every leg of the triad, as it affects the expectations Governance Gone Local: Does Decentralization Improve Accountability? 243 of all actors in the framework. Such uncertainty governments can undermine national laws and reg- can also stifle investment and thus interregional ulations without likelihood that these actions will be competition. subject to central review and potential revocation. This allows regional governments to entice firms with tax evasion strategies, legal exemptions, Stifling Rather Than Promoting regional protectionism, and even shelter from court Interregional Competition actions, further undermining the center's regulatory The view that competition among jurisdictions can and enforcement capacity. Interregional competi- enhance governance has been an important justifi- tion thus becomes not a race to the top but an cation for decentralization. The basic foundation of escape route by which firms avoid the reach of this idea is simple: officials who steal or waste national authorities. In East Asia, we see this partic- resources, or fail to provide essential public goods ularly in the justice sector, where complex and high- relative to other regions, will lose residents and profile legal challenges are often deliberately shifted businesses, thus reducing the tax base. This argu- to remote regional courts, where the opportunities ment has taken on greater weight with the growing for undue influence may be greater. Such a phenom- focus on the investment climate. Given mobility of enon is also evident in regional regulations that pro- capital among countries as well as within countries, vide preferential tax and regulatory regimes for spe- businesses can seek out jurisdictions where regula- cific firms in contradiction to national legislation. tion is not overly onerous, infrastructure is sound, and trust relationships can be forged with local The Early Experience in East Asia officials. This kind of competition for investment is seen to discipline local governments, strengthening Few empirical studies of the impact of decentral- their incentives for delivering transparent and ization on governance in East Asian countries exist accountable governance. so far. This is partly because most countries in the Recent literature has stressed that competition region began the decentralization process only among regions could also have a host of negative recently: the Philippines and Vietnam in the early implications for governance. Competition for capi- 1990s, Thailand and Indonesia in the late 1990s, tal and residents could become too intense, pushing and Cambodia at the turn of the century. Of these, governments to overshoot in cutting tax rates and Indonesia and the Philippines have gone the fur- expenditures and providing public goods (Keen thest in implementing comprehensive programs, and Marchand 1997). Such competition could also and thus offer modest empirical evidence on the have negative spillover effects, such as exporting impact of decentralization on governance. While taxes or pollution to neighboring regions (Gordon the experiences of each country are undoubtedly 1983; Oates and Schwab 1988). More generally, unique, these two nations can provide a glimpse of competition for investment could lead regional the potential benefits, risks, problems, and chal- governments to distort the investment climate to lenges to governance as a country decentralizes. favor particular firms through preferential regula- Empirical evidence on these two countries tions and protectionism, or, even worse, by shelter- comes from perception-based surveys; no system- ing firms from tax policies and regulations of the atic analysis based on objective indicators is yet central government. In the latter event, competi- available, at least for East Asia. tion among regions could weaken the capacity of the central government to collect revenue, enforce Indonesia law, and regulate interregional competition to ensure welfare-enhancing outcomes. This scenario Because Indonesia implemented its Big Bang has been called "state-corroding" decentralization decentralization only in 2001, it is still quite early to (Cai and Treisman 2002). expect reliable estimates of the impact of complex The risks of state-corroding decentralization and slow-moving institutional changes on percep- are particularly high in countries where central tions of governance. Though the reform did rapidly law enforcement and respect for the constitution transfer control over a significant share of public are still quite weak. In such contexts, regional resources and direct authority over nearly 2 million 244 East Asia Decentralizes civil servants to the local level, the institutional FIGURE 11.4 Obstacles to Business, as Cited changes are still in flux and the lines of authority by Firms in Indonesia are unclear in many areas. Moreover, the country 50 has recently revised the basic laws defining decen- 45 tralization, so considerable uncertainty remains 40 about its ultimate extent and shape. To expect this (%) 35 obstacle fluid environment to exert a clear impact on 30 perceptions of governance--which generally lag citing 25 behind institutional changes--is premature. constraint 20 A number of empirical studies have attempted to firms of 15 major measure governance trends across districts in as 10 Indonesia, but few provide a benchmark for com- share 5 paring governance indicators before and after 0 decentralization. As a second best--one that local local local local focuses on how the perceptions of firms change national national national national over time--the World Bank's Productivity and customs, licenses, Investment Climate Survey (PICS) explicitly asked customs, regulations, licenses, corruption, regulations, corruption, firms to identify the direction of change on a variety labor of governance dimensions before and after decen- labor obstacles tralization. Though these data are preliminary, they provide some basis for exploring these issues. Source: PICS. The PICS survey asked firms directly about the impact of decentralization on key aspects of gov- ernance and the investment climate. Firms per- regulations worse, nearly 50 percent think policy ceived decentralization as having a negative impact uncertainty and corruption--two broad indicators in four areas: labor regulations, licenses, policy of accountability--have worsened. uncertainty, and corruption (see figure 11.3). The survey results suggest that greater proximity While just under 30 percent of the firms felt that to the client--in this case, in the business arena-- decentralization has made licensing and labor does not necessarily lead to perceptions of greater accountability. The survey also asked firms to rate the extent to which a number of standard problems FIGURE 11.3 The Negative Impact of affect their business. In customs, labor regulations, Decentralization, as Cited by Firms in licenses, and corruption, firms see little difference Indonesia between the constraints imposed by national gov- 60 ernments and those imposed by regional govern- ments (see figure 11.4). Indeed, in some areas such 50 as licenses and corruption, firms rate local govern- ments worse than the national government, despite 40 the greater proximity of the former. 30 Corruption appears to be a particular problem respondents under decentralization in Indonesia. Over one- %20 third of firms stated that informal payments had grown under decentralization, while less than 10 15 percent cited some improvement (see figure 11.5). 0 Moreover, firms pay 64 percent more in informal labor licenses policy corruption payments as a share of annual revenues to local regulations uncertainty governments than they do to national level officials, areas of impact on average (see table 11.1). Firms also spend 15 per- worse better cent more of their time dealing with local regu- Source: PICS. lations than with national regulations. Together, Governance Gone Local: Does Decentralization Improve Accountability? 245 FIGURE 11.5 Informal Payments after they reap in revenues--on small-scale economic Decentralization, as Cited by Firms in activity. Many of these levies contradict national Indonesia guidelines on the tax authority of local govern- 100 ments. Annual studies by the Jakarta-based moni- 90 toring organization Regional Autonomy Watch 80 have emphasized the negative impact of these nui- 70 sance taxes and charges on regional investment cli- 60 mates. Studies of the legislative processes under 50 which the taxes and charges have been formulated 40 respondents show little citizen consultation or participation % 30 (Lewis 2003; Regional Autonomy Watch 2004). 20 Moreover, they tend to be poorly designed, with no 10 clear links between service delivery and charges. 0 Despite a requirement that national authorities Share of revenue predictability review all regional legislation authorizing new taxes effect on informal payments and charges within 30 days, a recent estimate sug- gests that some 60 percent of these regulations have strongly increased decreased moderately increased moderately decreased strongly not been reviewed, and that regions never sent a stayed the same large share to the center for review (Lewis 2003). Though many regional governments have indeed Source: World Bank 2005. introduced innovative programs to improve public service delivery and strengthen the transparency of TABLE 11.1 Regulatory Burdens on Firms local processes, little empirical evidence so far in Indonesia reveals a race to the top to improve regional invest- ment climates. National Local government government In conjunction with World Bank projects that aim to invest in communities committed to gover- Time spent dealing with nance reform, detailed case studies have examined regulations 7.1% 8.2% governance dynamics at the district level (World (percent of Bank 2004b). These studies point to common fac- total time) tors distorting the potentially beneficial impact of Informal decentralization on governance. First, these studies payments 1.35% 2.1% emphasize weak development of local "accounta- (percent of bility infrastructure": institutions that constrain total sales) the authority of the executive branch. The heads of Source: World Bank 2005. districts in Indonesia--bupatis and walikotas--are still appointed by local legislatures, rather than these results suggest that bringing government elected, with a strong role played by national par- closer to clients--at least those in the business ties. Removing district heads requires the approval community--has not led to greater accountability of national authorities. The local legislatures them- compared with the national government. Instead, selves are extremely weak and ineffective: they have decentralization is linked with a general decline in virtually no technical expertise, and rely entirely on perceptions of government accountability at the the executive branch in key areas such as legal local level. drafting, budget analysis, and public accounting. These perceptions are consistent with analyses of Moreover, case studies of elections for local legisla- the fiscal performance of local governments under tures have shown that the chosen candidates have decentralization.Several recent studies have pointed weak accountability ties to their constituents. Can- to a widespread trend in which regional govern- didates routinely pay national party organizations ments have imposed numerous nuisance taxes and for ballot slots, and their selection is closely linked charges--levies that cost more to administer than to elite village networks. Voters are also strongly 246 East Asia Decentralizes influenced by direct payments and other transfers. decentralization in Indonesia have thus been in a As a result, the local governance environment is permanent state of renegotiation since Big Bang highly susceptible to clientelism and capture. decentralization began in 2001. Not surprisingly, Local legal and judicial institutions play only a decentralization has exerted its most negative minor role in the accountability infrastructure. A impact on policy uncertainty. detailed study of 37 cases where public officials TheWorldBank'sengagementindecentralization stole or misused local development funds con- has highlighted the lack of information on regional cluded that most communities could not overcome performance, which is essential to promoting a local power structures and elite networks to gain race to the top among regions. Decentralization-- access to justice through formal and informal dis- especially the initial shift of control over civil pute resolution mechanisms. Only in cases where servants to local governments--has undermined external actors such as national civil society groups, Indonesia's regional information systems. The outside media, and representatives of a national problem is particularly severe in key decentralized project management structure intervened were public service sectors such as health and education. corrupt local officials successfully prosecuted. Minimum service standards mandated in the Civil society groups and other external con- decentralization laws have not yet been developed. straints on government such as the media remain Only a small minority of districts have submitted weak in most districts. Efforts to promote local the required performance self-assessments based plans to address poverty have been hampered by on annual and five-year plans. Districts send only a low levels of participation, except in districts near small share of laws and decrees to the center for major universities or technical institutes. Though official review, and the Ministry of Home Affairs community radio is developing rapidly, print does not have a system for analyzing and cataloging media do not penetrate far beyond the provincial this material. While several civil society groups capital. According to a recent poll, more than have begun to develop systems for comparing per- 85 percent of Indonesians obtain their political formance across districts, especially regarding the information from television (International Foun- investment climate, these efforts do not yet serve as dation for Electoral Studies 2004). Provincial and a reliable basis for monitoring districts. As a result, national media devote minimal attention to district little information about regional performance is news. Interviews also suggest that it is standard available to stimulate competition and disseminate practice for local governments to pay provincial good and bad practices.13 and national journalists for positive stories on a per story basis. The Philippines As mentioned, a major factor weakening the impact of decentralization on governance is uncer- In contrast to Indonesia, which is still in the very tainty about the proper role of different levels of early stages of decentralization, the Philippines has government and the resulting conflicts. In Indone- several years more experience, having launched its sia, despite recent revisions to the main decentral- major reform in 1992­93 under the mandate of its ization laws (formerly Laws 22 and 25 of 1999; now post-Marcos democratic Constitution of 1987.14 Laws 32 and 33 of 2004), the roles and responsibili- Given more than a decade of experience, analyzing ties of district governments have not been fully clar- the long-term impact of decentralization on gover- ified. The role of provincial governments in decen- nance outcomes might indeed be possible. How- tralization is also poorly defined. Moreover, the ever, empirical analysis in the Philippines is nearly oversight functions of different central ministries-- as sparse as in Indonesia, and no direct survey especially the ministries of Home Affairs and measures have explicitly examined the impact on Finance--remain unclear. This has led to constant perceptions of governance or related indicators. conflict, as different levels of government and com- Existing evidence comes from an extensive study by peting agencies stake their claims over vaguely the Center for Institutional Reform and the Infor- defined roles and powers. This conflict is reaching a mal Sector (IRIS) at the University of Maryland crescendo with recent proposals to further revise the and the World Bank, based on surveys of house- main decentralization laws. The boundaries of holds and public officials at different levels of Governance Gone Local: Does Decentralization Improve Accountability? 247 TABLE 11.2 Perceptions of Corruption in the Philippines (mean statistics, scaled from 0 to 100) Type of respondent Municipal Municipal Municipal Teachers Teachers health administrative official at public at private Survey questions official official at DECS schools schools Corruption in the national government 74.00 69.23 62.77 66.35 80.85 Corruption in the provincial government 59.43 43.86 37.96 50.65 69.57 Corruption in the municipal government 43.42 29.32 24.79 36.86 62.32 Corruption in the barangay government 38.96 28.85 22.22 24.76 48.89 Source: Azfar et al. 2000. Note: DECS Department of Education, Culture, and Sports. government.15 The study presents a mixed picture 49 percent had seen or heard reports of national of the impact of decentralization on governance officials engaged in corruption, while only 27 per- and concludes that "the results do not match the cent had seen or heard reports of local officials most optimistic theoretical expectations." The au- engaged in corruption (Azfar et al. 2000). thors find that while perceptions of corruption are A deeper probe into the sources of information much lower at lower levels of government, local on politics among Philippines respondents led to an governments are no more accountable to local pref- interesting result (see table 11.3). While nearly erences than the central government, given a vari- 98 percent of respondents rely on the media-- ety of constraints on responsiveness. especially radio and TV--for their information on Perception-based measures of corruption in national politics, these sources are far less prominent the Philippines have shown significant and consis- with regard to local politics. Instead, 42 percent of tent improvement since decentralization, though the respondents used personal networks as their pri- establishing a clear causal connection is difficult. mary source of information on local politics and According to most international rankings of cor- corruption, relying on community leaders or local ruption, the Philippines began to show progress officials rather than the media. This reflects the weak in the early 1990s and maintained this progress penetration of local media and their generally weak throughout the decade. Moreover, surveys of public capacity to cover local politics. Thus, the media may officials revealed that respondents in every category play a much weaker role in monitoring local officials saw corruption as more prevalent at the national for corruption, and information on local politics level than at lower levels of government (see could be more susceptible to manipulation through table 11.2). While municipal officials have obvious clientelist networks and state capture. biases, the pattern is the same among teachers at The IRIS­World Bank survey also explored local private schools, suggesting that the responses may accountability by asking both households and offi- in fact reflect reality in the ground. cials to identify their funding priorities. Except for These differences in perceptions of corruption infrastructure (roads, canals, bridges), the professed may reflect different sources of information about preferences of municipal officials failed to match local and national levels that may be even more those of households. In the critical area of health, important in explaining the dynamics of accounta- the correlation is in fact negative (see table 11.4).16 bility. The IRIS­World Bank survey asked respon- To supplement this finding, the survey asked dents whether they had seen or heard reports of public officials how they learned of household local and national officials engaged in corruption. preferences. All the officials attested to making Among the sample of over 1,100 households, some effort to understand such preferences. About 248 East Asia Decentralizes TABLE 11.3 Sources of Information on Politics, as Cited by Households in the Philippines T-test for difference Local politics National politics between the two resultsa Television 0.221 0.635 16.12 Radio 0.332 0.307 0.91 Newspaper 0.030 0.037 0.67 Officials 0.181 0.008 10.50 Civic associations 0.014 0.003 2.04 People/neighbor/ friends/family 0.195 0.009 10.91 Inside information 0.018 0.002 2.80 Other 0.105 0 2.46 Total media 0.583 0.979 16.94 No source 0.491 0.416 3.57 Source: Azfar et al. 2000. Note: Table indicates the percentage of people using each source (given that a source is used, scale 0 to 1). a. Numbers greater than 3 indicate that the difference is statistically significant at the 1 percent level. TABLE 11.4 Correlation between Funding 30 percent consult with community leaders and Priorities of Public Officials and Households local civic organizations, while 15 percent also con- in the Philippines duct surveys and make other inquiries (Azfar et al. 2000). However, neither method significantly Municipal officials strengthened the relationship between the funding (municipal average; number of priorities of households and municipal officials. Priorities officials 78) Despite this apparent mismatch, assessments of Roads, bridges, canals 0.495 decentralization's impact on public service provi- (2.14)a sion in the Philippines are mixed. For example, Education 0.015 experts express concern about the deteriorating (0.08) quality and administration of public health pro- Health 0.703 ( 2.14)a grams, but most people express more positive views. Agriculture/irrigation 0.090 According to a June 1999 survey by Social Weather ( 0.34) Station (a nonprofit organization specializing in Garbage collection 0.099 survey-based research), 58 percent of respondents ( 0.22) said health care had improved with decentraliza- New jobs 0.030 tion, while only 8 percent said it had worsened ( 0.08) (USAID 1999). A more quantitative measure of Aid to poor 0.068 ( 0.68) service quality is administrative delays in hiring and Water, drainage 0.339 paying staff, which are significantly more severe in (1.08) education than in health (see table 11.5). This is revealing, as the education sector has seen far less Source: Azfar et al. 2000. decentralization than the health sector, even though Note: Public officials' preferences are regressed on household preferences for each priority type the Local Government Code requires devolution of where variables are differences from the basic education to local governments. Although national mean. The first number in column two performance measures before and after decentral- is the OLS regression coefficient, with t-statistics in parentheses. ization are not available, devolved services seem to a. Significant at the 1 percent level. work better on some dimensions. Governance Gone Local: Does Decentralization Improve Accountability? 249 TABLE 11.5 Responses from Facilities on Today more than 95,000 NGOs exist, with some Administrative Delays in the Philippines 7,000 operating at the grassroots level (Alonzo 2003). One study concluded that "NGOs operating Health facility Index at the local level often found themselves in direct How often has there been a delay confrontations with local elites who, understand- of more than one month in getting ably, feel threatened when NGO members build a new health worker on a payroll? 3 alliances, muster funds and other resources for use In the last year, were there any by local people outside the control of traditional nonpayments or delay of more than two weeks in the payment patrons, and gain the moral high ground in the of your salary? 17 process" (Alonzo 2003: 17).18 In the last year, were there any However, these surveys suggest that the respon- nonpayments or delay of more siveness of local governments to the new layer of than two weeks in the payment of your allowances? 7 NGOs has been limited, owing to legal and proce- dural constraints on decentralization and fears Primary school among central officials that they will lose control How often has there been a delay over the regions. While boosting local taxing of more than one month in getting authority, the Local Government Code also con- a new teacher on a payroll? 26 In the last year, were there any strains local revenue collection through rules on nonpayments or delay of more rates, assessments, appeals, and administrative than two weeks in the payment responsibilities (USAID 1999). The code also regu- of your salary? 20 lates earmarked funds such as the Special Educa- In the last year, were there any nonpayments or delay of more tion Fund, specifying the property levies through than two weeks in the payment which these funds are raised and the rules for dis- of your allowances? 17 tributing the proceeds. These requirements signifi- cantly compromise local revenue autonomy. In fact, Source: Azfar et al. 2000. Note: The higher the index, the worse the whenever fiscal (deficit) problems have become problem. The index, which is scaled from 0 to serious, the central government has conjured up 100, is constructed as follows: {(0*% responded schemes to reduce mandated, formula-based trans- never) (1*% responded sometimes) fers, further limiting the flexibility and responsive- (2*% responded most of the time) (3*% responded always)} 3. ness of local governments (Azfar et al. 2000; Alonzo 2003). Some aspects of governance and service deliv- Not surprisingly, as in Indonesia, the hypothe- ery standards have improved under decentraliza- sized race to the top among local governments has tion, although whether these changes are due to not emerged. But unlike in Indonesia, and perhaps enhanced accountability at the local level is much because of longer experience, reliable information less clear. Surveys have shown that, as in Indone- on performance is beginning to develop, spurring sia, key institutions in the local accountability some competition among local governments. The framework--such as an independent media, and power of information in fostering competition is legislative and judicial branches to check the perhaps best illustrated by the success of the Galing power of local executives--are weak. However, Pook Awards. These awards, first introduced in unlike Indonesia, the Philippines has seen civil 1993 in response to the Local Government Code, society groups mushroom at the local level. A num- were patterned after awards to U.S. local govern- ber of studies reveal that NGOs have indeed ments for notable achievements in improving become actively involved in the decision making of service delivery and public welfare. Since their the local government.17 In 1986, when the country inception, a total of 2,339 programs have competed restored democracy, 27,100 NGOs had registered for these awards, with just 175 selected and 8 con- with the Securities and Exchange Commission. sistently cited for excellence and elevated to the After passage of the Local Government Code, Hall of Fame. The awards carry no monetary com- this number quickly doubled to 50,800 in 1992. pensation; they simply provide a credible signal to 250 East Asia Decentralizes local--and potentially higher-level constituencies-- national counterparts--are subject to the risks of that a mayor or governor is doing a good job.19 capture and clientelism. Countervailing institu- tions at the local level generally lack the indepen- dence and capacity to check these risks. This does Conclusions not suggest that decentralization in such contexts Though decentralization is, fundamentally, a strat- should be avoided. Indeed, experience in the egy for improving governance, its impact on gover- Philippines shows that decentralization can encour- nance outcomes is still largely unknown. A strong age the development of a vibrant local civil society body of theory posits governance gains from network. Yet more needs to be done to support the decentralization, but recognition of how decentral- other key institutions in the accountability frame- ization can go wrong has grown. The expected work that guarantee and sustain the expected gov- gains are based on assumptions about the politics ernance gains from decentralization. of accountability at the local level and the nature of The case studies also reveal the risks associated interjurisdictional competition that need to be with sustained uncertainty about the division of examined closely, especially in developing coun- roles and responsibilities across levels of govern- tries. Using an accountability framework that links ments. In Indonesia, in particular, decentralization citizens, policy makers, and service providers, this is hotly contested terrain. The enabling legislation is chapter examined the potential risks in each link often inconsistent and leaves many issues unre- that threaten to undermine the expected gover- solved. These inconsistencies have set the stage for nance gains from decentralization. The chapter serious political conflicts across levels of govern- then applied that framework to the two East Asian ment and among agencies, further exacerbating countries with the most significant decentralization these uncertainties. Even after a decade of decentral- programs: Indonesia and the Philippines. ization, the Philippines suffers from similar prob- Not surprisingly, in neither country has decen- lems. Such an environment constrains the ability of tralization fulfilled the governance goals predicted local officials to respond to the demands of their by the most optimistic theories. In Indonesia, which constituents, and shifts their focus to bureaucratic is still in the early stages of its reform, the initial struggles to preserve their powers. Uncertainty impact on perceptions of governance and selected blunts the impact of decentralization on accounta- outcomes has not been positive. There is a widely bility at the local level. held view that decentralization has exacerbated cor- Finally, the case studies emphasize the critical ruption and significantly increased policy uncer- role of information in spurring interregional com- tainty across different levels of government. Decen- petition, yet these systems have largely been over- tralization has also led to a greater regulatory burden looked in the context of decentralization. The on firms and questionable financial management decentralization process itself tends to fragment practices. These problems have contributed to a and weaken information flows from the local level general weakening of the investment climate, which over the short term. Significant efforts are needed has harmed Indonesia's growth prospects. In the to preserve these channels and develop new stan- Philippines, which has a longer record of decentral- dards and instruments for measuring and dissemi- ization, the picture is more mixed. Overall, percep- nating information on regional performance. tions of corruption have declined, and service deliv- These flows expand the range of information avail- ery standards have improved somewhat. However, able to local voters, who surveys have shown still the link between these outcomes and improvements largely depend on social networks for information, in the accountability of local politicians is weak. which can encourage clientelism and capture. In both countries, surveys and case study These flows are also essential to enabling investors, research suggest that the most serious problem is donors, and others to compare the performance of the weakness of local institutions intended to play a different regions, and to widely disseminate good major role in the accountability framework, includ- and bad practices to maximize their impact. ing local legislatures, judicial institutions, and the The analysis in this chapter suggests a strategy media. Local officials--even more than their for enhancing the impact of decentralization on Governance Gone Local: Does Decentralization Improve Accountability? 251 governance. Such a strategy should rest on three bilityunderdecentralization.Thisapproachcomple- pillars: ments community-driven development (see chap- ter 12) by strengthening countervailing institutions · Frame decentralization within national rules that and building local demand for good governance. weaken the ability of local governments to engage in capture or clientelism. To combat local cap- ture, national authorities need to enforce strong Endnotes restrictions on the ability of local governments to offer tax and regulatory privileges to specific 1. The main proponent of this view was Freidrich Hayek firms and groups. Local legislation often codifies 1948. 2. The classic statement of this view is Tiebout 1956. such privileges, and needs to be subject to some 3. For a review of the literature, see Bardhan 2002. See also form of central administrative review. To combat Bardhan and Mookherjee (2000) for a discussion of decen- clientelism, countries also need to link decentral- tralizing infrastructure services that questions received wisdom on the impact of decentralization on governance. ization to national civil service reform, stipulat- 4. The World Bank has been particularly active in this area, ing meritocratic recruitment and advancement especially in East Asia. In Indonesia, for example, 40 per- with appropriate central oversight and appeals. cent of the Bank's lending program for 2004­7 will focus on local governments and support decentralization. The political and economic incentives for the 5. The World Development Report 2004 characterizes these center to enforce such standards in regions may bilateral relationships between agents and principals as be stronger than the incentives for the center to fostering accountability (World Bank 2004c). adhere to the same standards. 6. This argument dates from the classic works of Rousseau 1986. · Focus on strengthening local countervailing institu- 7. More specifically, the possibility of obtaining rents drives tions and collective associations, not only through influential groups and individuals to bribe politicians and bottom-up participation but by linking them high-ranking civil servants, who introduce and maintain bad laws, policies, and regulations to perpetuate their illicit to national networks, which can help these insti- earnings. Note that in this context, corruption causes bad tutions counterbalance entrenched local elites. governance. To enhance local accountability, countervailing 8. See, for example, World Bank 2004b. 9. For a review of the literature on clientelism, see Keefer and institutions and collective associations may need Khemani 2003. to marshal resources outside their localities to 10. Illustrating the contrast with the national political arena, a enhance their power and influence. Capacity- popular song on the eve of Indonesia's first democratic building programs should therefore focus on national elections exhorted voters to "take Golkar's money, but vote for someone else." See Friend 2003. forging closer links between local institutions 11. While no systematic studies have compared skills and expe- and national networks. Local legislatures may be rience across levels of government, greater prestige and more effective at checking the executive branch higher stakes and living conditions at central levels are likely to attract more competent individuals. if they are allied more closely with a national 12. John Stuart Mill raised a related argument in On Repre- legislative network. Local civil society groups sentative Government. He worried that in a decentralized may be more effective in promoting accounta- state, "the local representative bodies and their officers are almost certain to be of a much lower grade of intelli- bility if they can rely on the capacity and power gence and knowledge, than Parliament and the national deriving from their national networks. executive." · Strengthen a national framework for encouraging 13. The World Bank has introduced a local government reform competition among regions, so local capture and platform within its Country Assistance Strategy for 2004­7 to address these information gaps and help develop a system clientelismentailreputationalcostsandlowerexter- for monitoring and evaluating regional performance. nalinvestment.Providinginformationonregional 14. Ferdinand Marcos ruled the country as a virtual dictator performance measures and the quality of gover- from 1964 to 1985. 15. The study was based on nine separate surveys covering nance is particularly important, as such informa- 1,100 households; 80 municipal administrators, health offi- tion can counterbalance the powerful local pres- cials, and education officials; 20 provincial administrators, sures supporting capture and clientelism. health officials, and education officials; 160 workers at gov- ernment health facilities; and 160 school principals (Azfar Overall, this strategy entails using national struc- et al. 2000). 16. Unfortunately, the survey did not include national offi- tures and networks to enhance the ability of local cials, so we cannot determine if the correlation is stronger organizations to promote greater political accounta- than at the national level. 252 East Asia Decentralizes 17. See especially Rood 1998; Alonzo 2003; and USAID 1999. Lewis, Blane D. 2003. "Tax and Charge Creation by Regional 18. For an extended discussion, see Racelis 2000. Governments under Fiscal Decentralization: Estimates 19. Galing Pook means means "place of excellence." The Ford and Explanations." Bulletin of Indonesian Economic Foundation provided seed funding for the awards; today Studies 39 (2). the Galing Pook Foundation raises money from private Mutebi, Alex. 2003. "Challenges of Participatory Local Manage- sources, both local and international. ment under Decentralization: Lessons from Chiangrai "The Galing Pook winners have become models of good Municipality." INDES workshop. Bangkok: Asia Develop- practices in local governance. Their programs have been ment Bank. documented, published, studied and visited by countless Oates, Wallace E., and Robert M. Schwab. 1988. "Economic local and national government personnel, academics, stu- Competition among Jurisdictions: Efficiency Enhancing or dents, media practitioners, civil society leaders, and ordi- Distortion Inducing?" Journal of Public Economics 35: nary citizens from the Philippines and abroad. Local chief 333­54. executives and program managers of these programs have Porter, Doug, and Clay Wescott. 2002. "Decentralization and become sought-after resource persons in conferences and Citizen Participation in East Asia." INDES workshop. have served as mentors to other LGU's that have adopted Bangkok: Asian Development Bank. their programs. Schools of public administration and other Prud'homme, Rémy. 1995. "The Dangers of Decentralization." institutions have used the Galing Pook case studies in their World Bank Research Observer 10 (2): 201­20. academic courses and training programs. The awards have Racelis, Mary. 2000. "New Visions and Strong Actions: Civil served as a seal of good housekeeping on program excel- Society in the Philippines." In Funding Virtue: Civil Society, lence and have opened many windows of opportunity in Aid, and Democracy Promotion, eds. M. Ottaway and terms of accessing funds, particularly from donors." See T. Carothers. Washington, DC: Carnegie Endowment for www.inq7.net/globalnation/galingpook/about_gp.php. International Peace, pp. 159­87. Regional Autonomy Watch. 2004. Regional Investment Attrac- tiveness: A Survey of Business Perception, 2004. Jakarta: Asia Bibliography Foundation. Alonzo, Ruperto. 2003. "The Political Economy of Decentraliza- Rood, Steve. 1998. "Decentralization, Democracy, and Develop- tion in the Philippines." Washington, DC: World Bank, East ment." In The Philippines: New Directions in Domestic Policy Asia and Pacific Region. and Foreign Relations, ed. D. Timberman. New York: Asia Azfar, Omar, Tugrul Gurgur, Satu Kähkönen, Anthony Lanyi, Society. and Patrick Meagher. 2000. "Decentralization and Gover- Rousseau, Jean Jacques. 1986. "The Social Contract." In Political nance: An Empirical Investigation of Public Service Delivery Writings, translated and edited by Frederick Watkins. in the Philippines." College Park, MD: University of Mary- Madison, WI: University of Wisconsin Press. land, Department of Economics, Center for Institutional Tanzi, Vito. 1995. "Fiscal Federalism and Decentralization: A Reform and the Informal Sector. Review of Some Efficiency and Macroeconomic Aspects." In Bardhan, Pranab. 2002. "Decentralization of Governance and Annual World Bank Conference on Development Economics Development." Journal of Economic Perspectives 16 (4): 1995. Washington, DC: World Bank. 185­205. Tiebout, Charles. 1956. "A Pure Theory of Local Expenditures." Bardhan, Pranab, and Dilip Mookherjee. 2000. "Capture and Journal of Political Economy 64: 416­24. Governance at Local and National Levels." American Eco- USAID (U.S. Agency for International Development). 1996. nomic Review 90 (2): 135­9. "Issues in Local Government Finance: Local Taxation and Cai, Hongbin, and Daniel Treisman. 2002. "State-Corroding Other Revenue Generation." Report No. 96­02. Manila: Federalism." Los Angeles: University of California Los Ange- USAID. les (UCLA), Department of Political Science. ------. 1999. The Governance and Local Democracy Project Friend, Theodore. 2003. Indonesian Destinies. Cambridge, MA: (GOLD): Summary Assessment. Manila: USAID. Belknap Press. World Bank. 2000. "Anticorruption in Transition: A Contribu- Gordon, Roger H. 1983."An Optimal Taxation Approach to Fiscal tion to the Policy Debate." Washington, DC: World Bank, Federalism." Quarterly Journal of Economics 98 (4): 567­86. Europe and Central Asia. Hayek, Friedrich von. 1948."The Economic Conditions of Inter- ------. 2004a. "Kabupaten Reform Dynamics Study." Jakarta: state Federalism." In Individualism and Economic Order. World Bank Indonesia Resident Mission. Chicago, IL: University of Chicago Press. ------. 2004b. "Village Justice in Indonesia." Jakarta: World Hoffman, Bert, and Kai Kaiser. 2003. "Decentralization, Democ- Bank Indonesia Resident Mission. ratic Governance, and Local Governance in Indonesia." ------. 2004c. World Development Report 2004: Making Services Washington, DC: World Bank, East Asia and Pacific Region. Work for Poor People. Washington, DC: World Bank; and International Foundation for Electoral Studies. 2004. Presiden- New York: Oxford University Press. tial Election Survey. Jakarta: International Foundation for ------. 2005. "Raising Investment in Indonesia." Washington, Electoral Studies. DC: World Bank, East Asia Poverty Reduction and Economic Keefer, Philip, and Stuti Khemani. 2003. "Clientelism." Back- Management Network. ground Paper for World Development Report. Washington, Zhuravskaya, Ekaterina V. 2000. "Incentives to Provide Local DC: World Bank. Public Goods: Fiscal Federalism, Russian Style." Journal of Keen, Michael, and Maurice Marchand. 1997. "Fiscal Competi- Public Economics 76 (3): 337­68. tion and the Pattern of Public Spending." Journal of Public Economics 66: 33­53. 12 Community-Driven Development: Decentralization's Accountability Challenge Susan Wong and Scott Guggenheim Decentralization is often hailed for moving govern- often do not work well in environments where ment closer to its citizens and providing opportuni- information is scarce, open elections are unfamiliar ties for participation in decision making. Achieving or rare, and clarity about performance standards is this goal, however, depends on a variety of condi- lacking. Decentralization can also aggravate exist- tions, including establishing mechanisms to pro- ing social problems. Polarized local environments mote broad civic participation and more responsive can erupt into open conflict when decentralization and accountable local governments. These goals are leads to the exclusion of one group from govern- especially relevant for East Asian countries pursu- ment. Marginalized groups, particularly the poor, ing large-scale decentralization programs. One often fare worse under decentralization when local mechanism intendedtomakedevelopmentplanning governments do not see redistributive or highly tar- and management more responsive and account- geted social programs as priorities--particularly if able arises from what the World Bank has called the primary beneficiaries of such programs are peo- community-driven development (CDD). ple least likely to vote.A great deal of work is needed A great deal of empirical evidence reveals that to introduce transparency, consultation, participa- decentralization will not always achieve the goal of tion, and accountability mechanisms that can link making local governments more responsive and public administrations with their surrounding com- accountable. They are often susceptible to elite cap- munities and give the poor a greater voice. ture: that is, public decision making often reflects Decentralization and CDD are inherently politi- disproportionate influence by well-off and well- cal reform processes. Nevertheless, because devel- connected groups. In many countries, officials opment programs channel significant resources exploit the opportunities offered by decentraliza- into the existing administrative structure, donor tion to promote their own commercial activities. choices can help determine to what extent decen- Standard mechanisms for accountability, such as tralization promotes healthier links between com- elections, audits, and performance benchmarking, munities and governments. 253 254 East Asia Decentralizes The challenge of finding mechanisms to promote The growth of CDD programs has occurred in more responsive and accountable local governments parallel with a general trend toward decentraliza- is especially relevant in East Asian countries under- tion and democratization in countries within the going large-scale decentralization. China, for exam- region over the last decade or so, driven by varying ple, has quietly but effectively pursued a national political motivations. Given these trends, what has program for village elections, and is developing a been the relationship between the portfolio of CDD package of programs that provide new village gov- programs and decentralization reforms? How have ernments with investment resources. Indonesia, as these programs worked within evolving decentral- this chapter will show, is similarly concerned with ization frameworks? This chapter describes some of fostering more effective, adequately funded local the dynamics behind these two trends and focuses governments that can pursue a range of invest- on three main issues: ments, and with encouraging communities to feel they have a stake in local government and the means · CDD's role in improving the quality of decentral- to change it. ization by promoting greater civic participation, CDD programs are one mechanism for making voice, and accountability in local governance. development planning and management more · CDD's role in delivering cost-effective and timely accountable. Although community development is services within a decentralized context. a well-established idea in the literature, large devel- · CDD's role in informing and formulating decen- opment agencies such as the World Bank first tralization regulations. became involved with significant CDD programs in the late 1990s. Based on the above analysis, this chapter argues This chapter defines CDD as an approach that that in a decentralized environment, CDD programs gives communities or locally elected bodies control can improve the quality of the decentralization over the decision making, management, and use of framework by promoting greater civic participation, development funds.1 CDD differs significantly voice, and accountability in local governance; pro- from an earlier generation of community-based viding an effective means of delivering much-needed rural development projects in emphasizing the role services in a decentralized context in a cost-effective that communities play in driving the development and time-efficient manner; and directly informing process rather than acting as passive beneficiaries. and shaping decentralization regulations. While The long-term goal of CDD projects, especially CDD projects are too new to allow definitive state- in East Asia, has been to reduce poverty by improv- ments about their overall success or failure, prelimi- ing local governance and empowering local people. nary evaluations point to tangible contributions in While varying somewhat in design, CDD programs these areas.The chapter also discusses challenges that generally consist of a participatory planning process lie ahead for CDD in decentralized environments. at village and commune or subdistrict levels, leading to funding and implementation of priority activi- CDD as Public Sector Heresy: CDD Works in Practice but Does It ties. CDD programs emphasize giving communities Work in Theory? and locally elected bodies the power, information, and skills to determine the best use of development Specialists in public administration are often wary resources. of CDD for understandable reasons. CDD financial CDD approaches are especially relevant in coun- flows bypass formal intergovernmental transfer tries where the priorities of central and local govern- systems, and CDD sits outside integrated govern- ments with respect to poverty are weak, and where ment planning. But are there persuasive counter- governments will respond more readily to poverty if arguments as well? there is strong demand emanating from society. The One is that CDD can improve the way local gov- demand side is as important and fundamental in ernments function and thus help them work better politics as in the realm of economics. CDD hypoth- for their citizens. The decentralization literature says esizes that the community demand-side approach-- little about the role of nonstate actors in making if well designed and implemented--will exert mul- local governments work in transparent and account- tiplier effects in broader decision-making processes able ways. CDD aims to strengthen that role. By pro- at the local level. moting demand, competitive pressure, and the flow Community-Driven Development: Decentralization's Accountability Challenge 255 of information between governors and the gov- to follow cumbersome or leaky procedures for erned, CDD programs introduce a range of account- transferring funds or develop complicated mecha- ability mechanisms and participatory processes that nisms for coordinating and delivering services. improve local governance, especially in a decentral- Community groups can fulfill many of these func- ized environment. CDD encourages the creation tions instead. And, for certain types of infrastruc- and strengthening of community groups, and forges ture and services, preliminary evidence suggests new norms regarding civic behavior and expecta- that delivering funds directly to communities com- tions regarding the relationship between govern- plements investments of district governments with ment and the people. no loss of quality, and allows large-scale provision Whereas reforms of public administration tend of productive investments that would otherwise to focus on the government side of the governance not have occurred. equation, CDD emphasizes the other half of the ThischapterexamineshowCDDapproacheshave equation: strengthening citizen participation and been designed to operate in different decentralized demand. These two efforts are not mutually exclu- environments. As other chapters explain, the forms, sive, but the starting points are quite different. dimensions, and degrees of decentralization-- CDD approaches governance from the bottom political, administrative, and fiscal--vary across the up by helping poorer communities make their region. CDD programs also show wide variation in demands more audible and visible, thus influenc- design, management, and closeness to local gov- ing local planning and decision making. ernments. These variations relate to historical and CDD also responds to an efficiency problem. structural factors within specific country contexts,as East Asian subnational bureaucracies are already well as to different strategic approaches to CDD. overburdened trying to meet routine district-level A note of methodological caution is needed. requirements. All too often, faced with limited time Large-scale CDD programs at the World Bank are and resources, these governments simply delay new relatively new, with the large majority having been efforts to meet the development needs of villages, in effect for less than three years. Within such a lim- or collapse them into existing programs that may ited time frame, it is premature to draw conclu- or may not reflect local development priorities. sions, whether positive or negative, about critical Thus, for example, a 1997 review of 48 villages in aspects of CDD, such as their impacts on poverty Indonesia found that fewer than 3 percent of village and local government, evolution, and sustain- development requests proposed through the ability. Furthermore, CDD projects have proved no government's development planning system ever more immune to the general lack of rigorous evalu- received funding (Evers 1998). ation than other development projects. Such rigor- CDD programs have the potential to eliminate ous evaluation is rare in all contexts, and some some of these problems. CDD projects do not have studies indicate that the CDD community devotes "There have been spectacular successes and miser- "They [poor people] want to develop their own able failures in the efforts by developing countries organizations so they can effectively negotiate fair to make services work. The main difference partnerships with governments, with traders, and between success and failure is the degree to which with NGOs; they want direct assistance and local poor people themselves are involved in determining ownership of funds through community-driven pro- the quality and the quantity of the services which grams, with governments and NGOs accountable they receive." to them." World Development Report 2004: Making Voices of the Poor, World Bank, 2000 Services Work for Poor People (Personal accounts from over 60,000 men "The core message from poor people is a plea and women in 60 countries of the realities of for direct assistance to them, without exploitative living with poverty, and what they need to and corrupt `middlemen' and free of well-intended improve their lives) but often wasteful development programs. They call for systemic change." 256 East Asia Decentralizes substantial if not more effort than other sectors to integrated with municipal development plans. This such evaluations (Wassenich and Whiteside 2003). partly reflects the country's more evolved decen- That being said, funders are now making concerted tralization process, in which local governments play efforts to install comparable measurement systems a more significant role than in Indonesia. across the CDD portfolio. The Seila Program in Cambodia is rather differ- The lack of comparable evaluations precludes ent.2 It began as a postconflict government experi- any conclusive statements about the overall impacts ment in alleviating poverty through decentralized of CDD. Yet it is possible to draw some preliminary systems for planning, financing, and implementing conclusions about CDD performance. Projects development at the province, commune, and vil- already under way collect a great deal of useful lage levels. The program has evolved considerably, information, and several older projects are yielding and in its second phase (2001­5) has supported a growing body of evidence, including evaluations further design and implementation of deconcen- of their technical quality, audits, cost-benefit analy- tration and decentralization reforms and worked ses, case studies, and information on types and more closely with locally elected bodies, with funds levels of participation. This chapter draws on that flowing through official intergovernmental fiscal information while noting that a full assessment of transfer systems. CDD will have to wait several years. This chapter also highlights a sectoral CDD Given the diversity of CDD, this chapter project to show how they operate. Sectoral CDD addresses general approaches while honing in on projects generally work within line ministries and three countries where specific large-scale CDD pro- provide block grants to community organizations grams intertwine with significant decentralization such as school committees, health management reforms: Indonesia, Cambodia, and the Philippines. councils, and farmer or irrigation organizations, While all these CDD programs are housed within which decide how to use the grant funds. The Cam- government ministries or interministerial commit- bodia Education Quality Improvement Project tees, they differ somewhat in terms of their design, (EQIP) provides block grants directly to school links with local government entities, and historical cluster committees for use as they see fit to improve evolution. educational quality (see box 12.1). The Indonesian Kecamatan Development Pro- We chose these case studies for two reasons. gram (KDP) gives communities planning and First, they represent the latest thinking on CDD decision-makingpoweroverdevelopmentresources. design and approaches in the East Asia region KDP initially built on the government's bottom-up and are considered the most successful of the planning structure while introducing a broad range CDD portfolio. While not without significant of reforms to make that system more participatory problems and challenges, they afford a glimpse of and efficient. KDP channels funds outside the usual some of the trends and issues facing CDD. government disbursement mechanisms, allowing Second, the CDD projects in these three countries financing to flow directly from the national level to are intricately linked to their national decentral- kecamatan and village-level bank accounts con- ization reforms. Thus, they illustrate concrete trolled by communities. Since Indonesia's decentral- ways in which CDD projects can interact with ization in 2001, the second and third phases and influence reforms within a decentralized (2002­7) of KDP have emphasized greater oversight framework. from district parliaments, government monitoring, links with sectoral agencies such as education and How Does CDD Strengthen health, district matching grants, and local involve- the Decentralization Framework? ment in drafting formal decentralization regula- Promoting Civic Participation in Local Planning tions on village autonomy. and Decision Making The Kapit Bisig Laban sa Kahirapan-Compre- hensive Integrated Delivery of Social Services Intheory,decentralizationfacilitatesparticipationby (KALALHI-CIDSS) in the Philippines began in local populations. In actuality, the legacy of decades 2002. Its design is similar to that of KDP, except of authoritarian rule and embedded hierarchical that community-funded plans are more closely structures and behavior in many countries subvert Community-Driven Development: Decentralization's Accountability Challenge 257 BOX 12.1 CDD Projects in Three East Asian Countries Indonesia: The Kecamatan Development Program While the RILG Project began only recently, (1998­2007) Seila started in 1996. The first phase (1996­2000) Begun in 1998, the government's KDP aims to enabled the government to formulate and fully alleviate poverty by raising rural incomes, test in five provinces provincial and commune strengthen local government, and community systems for budgeting, planning, financing, and institutions, and improve governance. KDP implementing development. Seila initially sup- recently completed its first phase, funding more ported elected commune and village develop- than 50,000 infrastructure and economic activi- ment councils and emphasized participatory local ties and benefiting some 35 million poor people. planning and financing. The program has since Field studies and audits show that KDP projects evolved with the country's deconcentration and delivered a broad range of services at lower-than- local election process to work with the 2002 normal cost with greater community involve- elected Commune/Sangkat Councils. The second ment. KDP is now entering its second and third phase (2001­5) supports the implementation phases, which emphasize building villagers' tech- and further design of decentralization and decon- nical skills and strengthening local government centration reforms. institutions as part of the country's overall decen- Cambodia: Education Quality Improvement tralization process. The program encompasses Program (1999­2004) some 28,000 villages--almost 40 percent of the EQIP aims to develop a model for a participatory country's total. approach to improving school quality and pursu- ing performance-based management of Cambodia: Seila/Rural Investment and Local resources. The project has two main compo- Governance (RILG) Project nents. The first finances grants to provincial com- RILG Project aims to contribute to rural develop- mittees, quality improvement grants to school ment and reduce poverty by supporting the pro- clusters, and monitoring and evaluation activities. vision of priority public goods and services at the The second component supports the National commune level, and to promote good gover- Committee on Effective Schooling, policy studies, nance by enhancing participation at the com- and provincial and district education offices. mune and provincial levels. RILG works through Cambodia's Seila Program, a mechanism for Philippines: KALAHI-CIDSS Project (2002­6) mobilizing and coordinating aid that supports This project aims to strengthen participation in the government's decentralization and decon- village governance and develop communities' centration reforms. Together, RILG and Seila capacity to design, implement, and manage provide grants for rural infrastructure and development activities that reduce poverty. The related public goods identified through the project was influenced by Comprehensive Inte- planning process. The two programs also pro- grated Delivery of Social Services, the country's vide technical assistance for strategic studies to previous poverty alleviation program, as well as inform deconcentration reforms, and to review by Indonesia's KDP. The project will encompass and strengthen the regulatory framework for some 5,300 villages in the country's 40 poorest decentralization. provinces. broad-based participation and the idea that people and, in Indonesia, separate meetings for women. can hold local government officials accountable. In The local planning process varies from country to such contexts, CDD projects can provide mecha- country. In Indonesia and the Philippines, it takes nisms to accelerate participation and accountability the form of a list of development priorities. In the through broad-based planning, decision making, Philippines, local governments agree to include the and implementation. CDD can help design the plan- CDD list of projects in municipal development ning cycle to engage the poor and other marginalized plans.Activities that appear in those plans qualify for groups, thus moving toward more inclusive gover- funding through donor assistance or other sources. nance. In most cases, the CDD community planning Under CDD, decisions on allocating resources process begins with participatory appraisals that occur in a decentralized, participatory manner--in include social mapping, wealth-ranking exercises, contrast with the first generation of social funds, 258 East Asia Decentralizes which typically relied on external parties at the Local governments play an important role by provid- national level to make such decisions. Forums com- ing technical services for planning projects,to ensure posed of elected community representatives make their feasibility. Municipal development plans auto- decisions at intervillage and subdistrict levels. In matically include all the activities selected under this Indonesia, quotas ensure that women participate in process, and therefore so do the annual budgets of the decision-making process, thus broadening the local governments, safeguarding the sustainability of scope of civic participation. these investments. Cambodia's Rural Investment and Local Gover- These CDD processes are not without their nance (RILG) Project under the Seila Program challenges and limitations. First, many struggle to illustrates the link between decentralization integrate village and subdistrict planning with reforms and CDD. Begun in 1996, Seila piloted a planning at the district and provincial levels. The model for decentralized planning, financing, and subvillage or village level may not be appropriate managing development activities at the province for choosing the location of certain public facilities and commune levels. At the time, the country had or forms of service delivery, such as health clinics no formal decentralization policy. In its early and schools. "Supralocal" or interjurisdictional stages, Seila supported participatory planning and cooperation is needed to achieve economies of scale decision making through community, village, and and pool limited resources. Local planning cannot commune development councils, with local plans supplant the need for improved regional and inte- funded outside normal government channels. As grated multisector planning. Countries need to bet- decentralization reforms slowly evolved and com- ter link small, dispersed investments occurring at mune elections were held in 2002, the program was community levels with larger investment projects to integrated into local government institutions and achieve technological sophistication and economies processes. of scale. Scaled up from an initial five provinces to a As a second challenge, a community-driven nationwide program, the Seila pilot experience planning process may overlook some needs. Efforts on participatory planning and decision making fed to address violence against women and assist directly into the country's deconcentration and widows and female heads of household, respond to decentralization reforms. The Inter-Ministerial health needs, and conserve the environment often Administrative Proclamation (Prakas) on do not survive a participatory, competitive commu- Commune/Sangkat Development Planning--the nity planning process. Greater advocacy is needed to regulation governing the preparation and imple- highlight concerns such as immunization programs, mentation of a commune development plan-- accompanied by special funding windows, quotas, incorporates lessons from Seila on how to prepare a and vertical programming so the local planning and commune development plan, a commune invest- budgeting process can address them. ment plan, and an annual budget. The regulation and accompanying guideline encourage local gov- Giving Citizens a Voice ernments to open budget discussions to the public. These guidelines also highlight the need to set up One of the main apprehensions regarding decen- appropriate mechanisms to facilitate the direct tralization has been the extent to which local elites involvement of beneficiaries in the detailed design may dominate local decision making just as and supervision of all projects in the commune easily as--if not more easily than--central authori- development plans.3 ties. Particularly in highly stratified societies, local In the Philippines, the KALAHI-CIDSS Project is elites may capture the benefits arising from a designed to promote community participation and decentralized planning and financing system. development within existing government struc- Dropping funding to provincial or local levels tures.4 Communities and local governments engage where, for decades, people have not been allowed in a demand-driven, bottom-up process of planning to develop broad civic decision-making institu- and problem solving that leads to grants for commu- tions or accountability mechanisms hardly seems nity investment programs.The emphasis is on ensur- wise. ing that decision making and management of public What can be done to counter these inequalities? resources are participatory and demand-driven. Decentralization needs to be accompanied by Community-Driven Development: Decentralization's Accountability Challenge 259 reforms that increase the transparency and community-identified priority projects. Several accountability of local government. CDD programs CDD case studies, in the region as well as globally, build on the premise that gradually broadening show that one of the main reasons for the popular- participation in policy making and resource alloca- ity of CDD projects is that they do in fact disburse tion can enhance accountability and transparency funds quickly. Why did the governments of Indone- and improve equity in service delivery. sia and the Philippines initially choose a CDD CDD programs encourage citizens to exercise model when they had traditional line agency pro- voice and demand accountability and transparency grams to combat poverty at their disposal? Those through several mechanisms: programs were not disbursing funds fast enough and were not reaching the poor. · A strong emphasis on information disclosure and In Indonesia, for example, KDP's ability to pro- transparency, especially related to project bud- vide quick, high-volume disbursements of develop- gets, financing, contracting, and procurement. ment funds from the national level straight down Financial and contract information is discussed to the local level was the main reason that the gov- publicly and displayed on village information ernment chose to launch the program in 1997­98. boards. In Indonesia's KDP, village committees The traditional methods for disbursing funds must report back to the general village assembly through line ministries had failed, as shown by the at least twice during subproject implementation poor track record of the $1.2 billion national social to discuss progress and financial status. safety net program, which was canceled halfway · A grievance mechanism (Indonesia and the through implementation because of poor targeting, Philippines) that channels anonymous com- leaks, and limited effectiveness. plaints to project authorities and encourages Moreover, the direct financing mechanism efforts to address grievances. cleared up decision-making bottlenecks caused by · Strong community monitoring mechanisms. In central efforts to plan and control activities. KDP's Indonesia and the Philippines, each village disbursement system takes an average of two weeks forms an independent committee responsible between the time when a village places a request for overseeing contracts, procurement, finances, and when funds arrive in the village account. Even and implementation of development projects. during the East Asian financial crisis, KDP dis- These committees must report on financial sta- bursed funds nearly twice as fast, on average, as tus and physical progress at various stages. agriculture, health, and education projects imple- Provincial journalists and NGOs are also invited mented through line departments (see figure 12.1). to act as watchdogs over the proper use of public Bypassing intermediate levels of government development funds. Together these mechanisms enabled KDP to move quickly to respond to village- provide a system of checks and balances to help level demands. keep local governments accountable. Furthermore, as these CDD programs evolved, they opened windows of opportunity for additional Studies and preliminary evidence highlight two financing from government and private sources as other features of the CDD process. First, CDD well as communities. Under KDP, communities projects generally include a decentralized financing contribute 17 percent of project costs, on average, mechanism that operates more quickly than compa- and these contributions often equal or surpass the rable disbursements from line agencies. Second, the grant amount. Among participating districts, services delivered are cost-effective compared with 40 percent provide matching grants from their own the delivery of small-scale works by public agencies. resources to contribute to capital costs. Community-identified projects do not have to be financed by CDD: officials can establish links Decentralized Financing Mechanisms with many potential sources of financing. For The pitfall of many decentralized planning example, in the Philippines, all projects identified processes worldwide has been a lack of resources to through the CDD process are included under implement the resulting plans. CDD projects tackle municipal development funds, so the local budget this problem by providing finances directly from covers recurrent costs. This also helps attract fund- the national level to local levels to implement ing from other government and nongovernmental 260 East Asia Decentralizes BOX 12.2 Funds Arriving in Cambodian Schools The Cambodian Educational Quality Improve- There was widespread enthusiasm for EQIP ment Project (EQIP) is a CDD sectoral project among those interviewed, and a desire to build that is modeling a participatory approach to on the progress already made. Decentraliza- improving school quality and using perform- tion issues featured prominently. Stakeholders ance-based management. Cambodia's Ministry welcomed the opportunity EQIP provided for of Education, Youth, and Sports (MoEYS) has making decisions at local levels. Being able to emphasized decentralization in its Education decide what to purchase for a cluster through a Strategic Plan 2001­2005. Projects such as cooperative process at that level was greatly EQIP, as well as several other donor-funded appreciated. It was recognized that the actual school grants and cluster programs, show what process of making decisions at provincial, can happen when governments devolve respon- district and cluster levels was developing sibilities and funding to local schools and skills in prioritizing, planning, budgeting and communities. reporting . . . When asked what was the most Funded in part by a World Bank credit of important achievement of EQIP, the most pop- US$5 million, EQIP began in 1999 and supports ular answer was that it gave finances for prior- clusters of schools in three provinces. The ity needs in a timely and predictable manner. project allocates grants based on priorities iden- Enormous value was placed in this achieve- tified through school committees composed of ment. Previously there had been no resource representatives from communities and each flow from the MoEYS to schools and little if any member school. The project is highly popular to the provinces. with participating schools and communities. According to one evaluation: Source: Turner 2002. FIGURE 12.1 Project Disbursement in Cost-Effective Service Delivery Indonesia by Ratio, Fiscal Years 2000­2 Studies of Indonesia's KDP and the Urban Poverty 66.6 Project, its urban counterpart, as well as preliminary estimates from KALAHI in the Philippines, show 52.2 that infrastructure projects cost some 25 to 56 per- cent less than similar small-scale works delivered by other public agencies. These cost savings are consis- 31.8 tent with estimated savings from CDD projects in percent 28.6 29.3 24.8 other regions. For instance, a case study of CDD 21.3 19.3 projects in Brazil cites savings of 30 to 50 percent. 14.3 All these CDD programs have delivered demand- responsive public investments in infrastructure and economic and social activities. Because CDD plan- 2000 2001 2002 ning and financing is decentralized, small-scale fiscal year infrastructure projects are usually funded and imple- rural dev. (10 projects) mented within three to four months. Under the first human dev. (19 projects) stage of KDP, villages completed more than 50,000 KDP infrastructure, economic, and social subprojects. Source: World Bank Indonesia disbursement reports. Technical reviews and external evaluations show that these projects were popular with communities and government officials alike. A sample of infrastruc- sources. In Cambodia, under the RILG Project, ture projects showed that they yielded high internal district integration workshops provide a forum for rates of return ranging from 33 to 83 percent. These other government and nongovernmental actors to projects generated 25 million workdays, with fund projects identified through the local planning 2.8 million villagers gaining short-term employment process. through labor-intensive works (World Bank 1998). Community-Driven Development: Decentralization's Accountability Challenge 261 The national size and scope of these CDD programs tion Law 22. Building on the KDP platform, the have allowed governments to deliver essential public Initiatives for Local Governance Reform Project is goods and services in a demand-responsive way designed to support district policies related to while relieving overburdened national and sub- information disclosure, procurement, revenue gen- national bureaucracies of the need to manage thou- eration, budget planning, and allocation, leading to sands of small village projects and byzantine finan- higher pro-poor expenditures (World Bank 2004a). cial sign-offs at multiple levels. Similarly, the Seila Program in Cambodia has channeled five years of experience with participa- Technical Assistance in Formulating tory planning and financing into helping govern- Decentralization Procedures ment working groups complete the country's decentralized regulatory framework, draft support- Increasing Public Engagement ing guidelines, and revise and strengthen regula- with Local Governments tions. For example, technical advisors from Seila In both Indonesia and Cambodia, CDD programs have advised the government in developing the evolved as national decentralization reforms began manual of financial management and procurement to play out at local levels. In Indonesia, KDP incor- procedures for commune funds. porated greater roles for district parliaments, mon- In contrast to Indonesia and Cambodia, where itoring and supervision by local governments, and central governments are still formulating decentral- cost-sharing arrangements with district govern- ization reforms and regulations, the Philippines has ments. In Cambodia, the Seila Program was adapted had a Local Government Code for some 12 years, to work through the newly elected Commune/ and local governments are already well established. Sangkat Councils in place of the early Commune From the outset, the KALAHI-CIDSS project has Development Councils. Financial transfers now flow worked within the decentralization law and engaged to Commune/Sangkat Funds as well as provincial with formal institutions such as the Barangay investment funds. Development Council and the Municipal Develop- The CDD programs in both countries have built ment Council to make the process for planning and on lessons from participatory, decentralized plan- allocating local development resources more par- ning and financing to shape decentralization regu- ticipatory. Furthermore, to strengthen coordina- lations. The institutionalization of participatory, tion with local governments and enhance sus- transparent mechanisms in decentralization proce- tainability, the Philippines project works with dures and frameworks has been a key aim. Both municipal committees chaired by the mayor and governments had passed legislation and decrees composed of the heads of all local departments. outlining decentralization frameworks but were Local representatives of national agencies, NGOs, still formulating implementation guidelines, mak- and donor institutions also participate. These mul- ing them amenable to community contributions. tiagency committees meet every two weeks to dis- A further benefit is that national ministries cuss progress and determine needed contributions responsible for creating decentralization guidelines to KALAHI projects, including staff, salaries, and have implemented CDD programs.In Indonesia,the other recurrent costs (World Bank 2002). Ministry of Home Affairs, with guidance from Bappenas, the national planning agency, is responsi- Conclusions and the Challenges ble for issuing the implementing guidelines for Ahead for CDD decentralization. In Cambodia, the Ministry of Interior and the Ministry of Economy and Finance Community-driven development projects aim to are key members of the Seila Task Force,and the Seila influence and strengthen local governance, with the Program has provided financial and technical intention that the participatory approach will spill resources to new departments in these two ministries over into broader decision-making processes at the that regulate and support subnational authorities. community level. For CDD sectoral programs, such Drawing on lessons from the first and second as those in health, education, and water and sanita- phases, the third phase of KDP will assist district tion, the aim is to institutionalize mechanisms and village authorities in crafting 13 regulations whereby users of services can engage in decision (perdas) for village autonomy under Decentraliza- making and monitor investments and services. 262 East Asia Decentralizes TABLE 12.1 Points of Local Government Engagement in CDD Projects Indonesia Philippines Cambodia KDP KALAHI-CIDSS Seila EQIP Project Ministry of Home Department of Seila Task Force Ministry of implementing Affairs and National Social Welfare and (composed of Education, Youth agency Coordination Team. Development 5 ministries) and Sports (MoEYS). (DSWD). provides overall policy and program guidance. Implementation rests with ministries, provinces, and communes. Laws and KDP3 will help to Seila's pilot regulations draft the 13 village experience directly autonomy influenced the laws regulations. and regulations on commune/sangkat planning. Program provides technical assistance. Local planning Kecamatan and Municipal and Planning has MoEYS staff village officials join village officials join become part of and school in local planning in local planning formal government administrators join process. process. DSWD staff at the commune/ with community facilitate planning. sangkat levels. and school reps. in planning. Financing Ministry of Finance MoF transfers funds MoF transfers funds Funds flow from (MoF) transfers directly to village directly to MoEYS directly to funds directly to and intervillage commune/sangkat school clusters for village and accounts. and provincial school grants. intervillage levels. accounts. Cofinancing By 2004, under Local governments KDP3, local and communities governments will be must contribute a paying all or partial minimum of 25%. capital costs. Monitoring and District parliaments DSWD staff at Close monitoring MoEYS staff and oversight and local officials different levels, as and oversight local school cluster provide oversight well as local provided by Seila committees. and monitoring. governments. staff at all levels. Capacity Training for DSWD staff and Seila supports MoEYS line building kecamatan project local governments. training of local staff; school managers, village government staff administrators. councils, district and local commune parliaments. councils. Source: Various World Bank project appraisal documents, project reports, and discussions with task teams. CDD has improved decentralization frame- and decision making, as well for monitoring and works by: transparency, into the local governance equation. · Providing an effective means of service delivery · Promoting greater civic participation, voice, and within a decentralized context by delivering accountability in local governance. CDD intro- needed services in a cost-effective and time- duces mechanisms for participatory planning efficient manner. Community-Driven Development: Decentralization's Accountability Challenge 263 · Informing and shaping decentralization to address recurrent costs, operations and mainte- regulations. nance, and sustainability? As countries gradually integrate CDD capital and operating costs into CDD entails challenges, too. As noted, CDD intergovernmental fiscal systems, they need to projects are too new to allow definitive statements ensure that these systems and financial flows remain about their overall success or failure. Neverthe- transparent and easily tracked by the public. less, preliminary evaluations point to some challenges ahead, especially in decentralized When Is CDD Appropriate and When Is It Not? environments. First, can CDD become an overall development Community-driven development programs are not strategy, or will CDD projects provide a relatively always appropriate. Large-scale CDD operations narrow way to fulfill numerous small, discrete such as KDP, Seila, and KALAHI presume fairly development investments? Existing reviews show sophisticated management structures, conducive that communities can build village roads, water local social environments, and mechanisms that supplies, schools, and clinics. However, such allow direct fiscal transfers to reach communities reviews do not reveal how village projects will tie in and be accounted for with reasonable accuracy. with efforts to plan road networks, protect water- CDD projects in postconflict areas, where few of sheds, train teachers, and supply needed drugs. these conditions apply, seem to work reasonably Experiences in Cambodia and the Philippines sug- well within limits, but involve significant quality gest how to coordinate local government input into tradeoffs. discrete CDD investments and address recurrent Other conditions for success relate to the spe- and operational costs, but this area needs more cific design of each CDD program. KDP assumes exploration. that private markets can provide qualified technical Expansion to this higher level requires that CDD assistance. Seila assumes that local markets for programs engage more effectively with sectoral contractors are reasonably competitive and can agencies to help them become more demand- provide quality services. KALAHI relies on the driven and responsive. As CDD projects such as competence of municipal governments. Programs Seila and KDP become better established, tech- that cannot assume that such conditions exist must niques for integrating the CDD platform with adjust their design accordingly. higher-level planning and sectoral agencies become A subtle but important risk is that CDD projects clear. As the Cambodia EQIP Project has shown, may be asked to do more than their design can sus- helping the Ministry of Education, Youth, and tain. A good example is the challenge of linking Sports move responsibilities for planning and CDD operations to district planning and budget- financing down to school clusters can yield signifi- ing. The solution lies in complementing CDD cant benefits for schools and communities. The approaches with projects and strategies directed challenge will be to strengthen accountability and toward district- and province-level reform, not in participation in service delivery. transforming CDD operations into two-headed Second, can CDD improve links with private beasts that must perform both local and district sector service providers? CDD projects tap such planning and budgeting. suppliers for construction materials and some The need to supplement CDD with a broader forms of specialized technical assistance, but few reform strategy for local government is fundamen- have explored the use of private markets to procure tal. CDD can be seen solely as a low-cost way to books and teachers, qualified health services, and provide basic infrastructure and services. In certain specialized training. Local procurement of goods environments, particularly postconflict and transi- and services sacrifices economies of scale, and iden- tional countries, where virtually all standard sys- tifying qualified service providers in poorly regu- tems and normal activities have been disrupted, lated professional environments is difficult. Never- CDD projects may be the only way to operate. But theless, some scope for promoting private sector for most countries, CDD will work best if it is one service supply may exist. part of a broader reform strategy intended to Third, can governments better integrate CDD improve the quality and efficiency of government projects into their budgeting and planning process services. 264 East Asia Decentralizes Annex: CDD Portfolio Breakdown Global Trends For fiscal years 2001 to 2003, World Bank lending all Bank investments. The following tables and fig- in support of CDD has totaled nearly $2 billion ures reveal the contours of this lending worldwide annually, and averaged approximately 11 percent of and in East Asia. TABLE 12A.1 World Bank Lending in FIGURE 12A.2 Regional Trends--CDD Support of CDD, Fiscal Years 2000­3 Investments by Fiscal Year 2,000 Total Total World CDD 1,800 World Bank as % 1,600 Bank CDD of 566 1,400 No. of lendinga lendingb Bank (US$m) 192 283 CDD (US$ (US$ lending 1,200 489 projects billions) billions) (percent) 1,000 445 378 541 FY00 55 15.3 1.0 7 800 194 investment 190 FY01 78 17.3 2.3 14 600 524 601 265 FY02 78 19.5 1.9 10 400 497 yearly 368 215 FY03 83 18.5 2.0 11 200 265 43 152 174 254 47 232 Total 294 70.6 7.2 11 0 118 97 Africa and and and and Asia Source: World Bank CDD briefing note for Asia AsiaPacific East AfricaSouth Mr. Shengman Zhang, July 14, 2003, SDV. East Europe America Caribbean Central a. Includes the International Bank for Reconstruction MiddleNorth Latin the and Development and the International Development Association. FY03 FY02 FY01 FY00 b. Excludes enabling environment. Source: World Bank CDD briefing note for Mr. Shengman Zhang, July 14, 2003, SDV. FIGURE 12A.1 Bankwide CDD Lending for Fiscal Years 2000­3 2,500 2,301.3 2,000 1,866.3 1,934.4 (US$m) 1,500 1,027.4 1,000 commitment 500 yearly 0 2000 2001 2002 2003 fiscal year Source: World Bank CDD briefing note for Mr. Shengman Zhang, July 14, 2003, SDV. Community-Driven Development: Decentralization's Accountability Challenge 265 CDD Trends in the East Asia and Pacific Region FIGURE 12A.3 CDD Project Breakdown FIGURE 12A.5 Sectoral Breakdown of CDD by Country Projects by Country 1,400 12 10 CDD amount 9 1,200 10 8 1,000 7 no. of CDD projects 8 800 6 millions 6 projects projects?] 5 600 of of 4 US$ 4 no. 400 3 [no. 2 200 2 1 0 0 0 PDR Leste PDR Leste Indonesia ThailandVietnam Lao Philippines CambodiaLaoTimor Mongolia Indonesia PhilippinesCambodia Vietnam ThailandMongolia Timor Source: Authors' calculations and desk review research. agriculture/irrigation/livestock urban development social protection subnational government administration/other water and sanitation Education Environment Source: Authors' calculations. FIGURE 12A.4 CDD Project Breakdown by Sector 1,200 12 CDD amount 1,000 no. of CDD 10 projects 800 8 millions 600 6 projects of US$ 400 4 no. 200 2 0 0 protection government livestock sanitation education development environment and social agriculture/irrigation/water urban subnational administration/other sectors Source: Authors' calculations. 266 East Asia Decentralizes TABLE 12A.2 World Bank­Funded CDD Projects Open in the East Asia Pacific Region, as of June 30 2003 World Banka Number of amounts CDD amounts Country CDD projects (US$ millions) (US$ millions) 1 Indonesia 10 1,220 1,170 2 Philippines 7 341 316 3 Vietnam 5 333 217 4 Thailand 1 300 298 5 Cambodia 6 89 89 6 Lao PDR 3 38 38 7 Mongolia 1 19 17 8 Timor-Leste 2 9 9 Total 35 2,349 2,153 Source: Authors' research and calculations. a. Includes the International Bank for Reconstruction and Development and the International Development Association. Endnotes Bibliography 1. The content of the CDD portfolio has been the subject of Asian Resource Center for Decentralization. 2002. Decentraliza- wide debate within social and human development circles. tion and Power Shift: An Imperative for Good Governance. This chapter adheres to the broad definition used by the Quezon City, Philippines: University of the Philippines, World Bank's social development network: "Community- National College of Public Administration, Center for Local driven development (CDD) gives control of decisions and and Regional Governance. resources to community groups. These groups often work Evers, P. 1998. "Village Government: Does the Current System in partnership with demand-responsive private sector, Enhance Good Government?" Washington, DC: World NGOs, and central government agencies." Bank, Social Development Department. This definition raises two issues. First, who decides how Ministry of Home Affairs, Community Development Agency, to allocate resources? CDD projects are distinguished from and National Management Consultants, Indonesia. 2002. earlier Social Funds by the fact that local communities "Kecamatan Development Program: Phase One--Final or local elected bodies make such decisions. The earlier Report, 1998­2002." Jakarta: Ministry of Home Affairs. generation of Social Funds--and even many to this day-- OECD (Organisation for Economic Co-operation and Develop- involve communities in planning, preparing proposals, ment). 1997."Final Report of the Ad Hoc Working Group on and implementing projects. However, Social Fund staff Participatory Development and Good Governance." Paris: at national or regional levels actually make decisions on OECD, Development Assistance Committee. proposals. Turner, M. 2002. "Whatever Happened to Deconcentration? CDD projects, on the other hand, retain decision mak- Recent Initiatives in Cambodia." Public Administration and ing within communities through village-elected forums or Development 22: 353­64. elected local government. Thus, for the last several years, Wassenich, Paul, and Katherine Whiteside. 2003. "CDD Impact Social Funds that do not involve communities in decision Assessments Study: Optimizing Evaluation Design under making on resources have been considered separate from Constraints." World Bank, Washington, DC. the CDD portfolio. Wescott, Clay, and Doug Porter. 2002. "Fiscal Decentralization The second issue relates to the role of locally elected and Citizen Participation in East Asia." Paper for conference governments. Traditionally, CDD projects have focused on on Citizen Participation in the Context of Fiscal Decentral- community organizations and forums that are not part of ization: Best Practices in Municipal Management in Latin the formal government decision-making structure. Over America and Asia. the last two years, however, the World Bank has defined World Bank. 1998. Project Appraisal Document, Kecamatan CDD to include both locally elected community organiza- Development Project. Internal draft document. World Bank, tions and democratically elected government bodies work- Washington, DC. ing in partnership with community groups. Thus, debates ------. 1999. Project Appraisal Document, Education Quality about whether CDD is outside or inside the usual political Improvement Project. Internal draft document. World Bank, channels have blurred. Washington, DC. 2. Seila is a Khmer word meaning "foundation stone." ------. 2000. Voices of the Poor, vols. I, II, III. Washington, DC: 3. For a fuller description, see World Bank 2003b, annexes 11 World Bank. and 12. ------. 2001a. "Promoting Good Governance with Social 4. Interview with Cyprian Fisiy, former World Bank task team Funds and Decentralization." PREM Notes, Public Sector leader for the KALAHI-CIDSS Project. No. 51. Washington, DC: World Bank. Community-Driven Development: Decentralization's Accountability Challenge 267 ------. 2001b."Decentralization and Governance: Does Decen- World Bank, East Asia Poverty Reduction and Economic tralization Improve Service Delivery?" PREM Notes, Public Management Unit. Sector No. 55. Washington, DC: World Bank. ------. 2003b. Project Appraisal Document, Rural Investment ------. 2001c. Project Appraisal Document, Second Kecamatan and Local Governance Project. Internal draft document. Development Project. Internal draft document. World Bank, World Bank, Washington, DC. Washington, DC. ------. 2004a. Project Appraisal Document, Initiatives for ------. 2002. Project Appraisal Document, Kapitbisig Laban Sa Local Governance Reform Project. Internal draft document. Kahirapan/Comprehensive and Integrated Delivery of Social World Bank, Washington, DC. Services (KALAHI-CIDSS) Project. Internal draft docu- ------. 2004b. World Development Report 2004: Making Services ment. World Bank, Washington, DC. Work for Poor People. Washington, DC: World Bank; and ------. 2003a. "Decentralizing Indonesia: Regional Public New York: Oxford University Press. Expenditure Review: Overview Report." Washington, DC: T he past two decades have witnessed a fundamental transformation in the structure of government across East Asia. Prior to 1990, most East Asian countries were highly centralized. Today, decentralization is widespread throughout the region. From China to Thailand, subnational governments are now responsible for the delivery of critical services and account for a significant proportion of total public expenditure. In just two decades, local and provincial authorities have emerged as the organizational fulcrum for much of the region's development. Despite encouraging progress thus far, fundamental weaknesses in the emerging intergovernmental structures threaten local economic development, service delivery, and governance in many countries in the region. Focused action is needed to successfully address these challenges and to ensure that decentralization remains on track. East Asia Decentralizes is the first attempt to look systematically at the decentralization phenomenon that is occurring throughout East Asia. It assesses what has happened thus far, distills key messages, highlights positive experiences, and identifies areas where priority action is needed. It is broken into four main sections. a Chapters 1 and 2 provide an overview of the intergovernmental structures and frameworks that have emerged thus far, assess the status of the decentralization process, and identify key reform challenges for the future. a Chapters 3­7 examine various dimensions of local and intergovernmental finance: subnational borrowing, local revenues, public expenditure management, and the impact of the process on interregional equity and poverty reduction. The management of human resources is also covered in this section. a Chapters 8­10 focus on the impact of decentralization on investment and service delivery in three key sectors: health, education, and basic infrastructure. a Chapters 11­12 look at issues connected with local accountability and community-driven development. Although intended primarily for central and local government policy makers, East Asia Decentralizes will also be a useful resource for academics, businesspeople, and development practitioners concerned with the unfolding process of intergovernmental reform in the region. THE WORLD BANK TMxHSKIMBy360590zv":':+:&:^ 0-8213-6059-0