DISCUSSION PAPER October 2020 Governance Global Practice Deepening Decentralization within Centrally Led States: The Direction of Local Governance Reforms in Southeast Asia Zubair K. Bhatti Lachlan McDonald This series is produced by the Governance Global Practice of the World Bank. The papers in this series aim to provide a vehicle for publishing preliminary results on governance topics to encourage discussion and debate. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. Citation and the use of the material presented in this series should take into account this provisional character. For information regarding the Governance Discussion Paper Series, please contact govgplearns@worldbank.org. © 2020 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, NW Washington, DC 20433 All rights reserved. Zubair K. Bhatti is a Senior Public Sector Management Specialist (Governance Global Practice, World Bank) who works on decentralization, civil service, and e-Government reforms in the Equitable Growth, Finance, and Institutions Vice Presidency of the World Bank. He holds a master in business administration from Imperial College, London, and a master in public policy from Princeton University. Lachlan McDonald is a consulting Governance Specialist at the World Bank. He holds a doctorate in development economics from RMIT University and a master of international relations from the University of Melbourne. He has worked in a variety of low-income countries, fragile and conflict-affected situations, and middle-income countries across Asia and the Pacific, and has expertise in applied economics, macro-fiscal and public financial management, governance, health financing, and social protection. ABSTRACT In the early 1990s, a host of unitary states in Southeast Asia began pursuing, at their own pace, the process of decentralizing their systems of governance. The form and function of these reforms were documented in the World Bank's report, East Asia Decentralizes: Making Local Government Work, published in 2005. The analysis in this discussion paper updates the findings and data from this foundational work and widens the lens to new countries embarking on similar reforms. The aim is to chart the overall trajectory of reforms, examine the political dynamics behind them, and help countries locate their place in the process. Each country is taking its own path, shaped by its history, size, and political system, though there are some similarities. In general, reforms are heavily and predictably influenced by political dynamics, managed from the central government, and driven by its priorities. They result in an unsynchronized and, at times, messy transfer of funds, functions, and functionaries to lower levels. Central governments continue to wield considerable influence over subnational governments through a combination of tight political controls, concurrent functional assignments, budgetary measures, and continued bureaucratic control. Nonetheless, with all their limitations, the countries of the region are finding ways to muddle through with the general trajectory toward more, not less, decentralization. In large part, this process is happening organically and incrementally without any big bang reforms and with the direct involvement of central governments. Variously, it is helping to nudge governments toward greater political pluralism, innovation, and improved service delivery within otherwise tightly controlled states. As decentralization deepens, so too does the importance of effective coordination. Table of Contents ACKNOWLEDGMENTS 1 1. INTRODUCTION 2 2. THE CONTEXT 4 Decentralization in Southeast Asia: A Common Trait in a Diverse Region 4 A Series of Managed Transitions Driven from the Center 9 3. STRUCTURAL FEATURES OF CONTEMPORARY DECENTRALIZED GOVERNANCE 11 State Structure and the Emphasis of Decentralization 11 Functions and Finance 23 4. THE IMPLEMENTATION OF REFORM 31 An Unsynchronized Approach to Decentralization Driven by Political Factors 31 Continuing Centralized Influence Over Public Administration 33 An Evolving System of Subnational Governance 36 An Underperforming System of Conditional Transfers 40 5. KEY LESSONS AND PATHWAYS FOR REFORM 43 REFERENCES 47 ACKNOWLEDGMENTS The authors would like to acknowledge the assistance of colleagues: Deryck Brown, Maxwell Dapaah, Somneuk Davading, Jeevakumar Govindasamy, Chandana Kularatne, Nandar Soe Linn, Barjor Mehta, Deepak Mishra, Bernard Myers, Evgenij Najdov, Ildrim Valley, Fanny Weiner, Yuan Xiao, and Serdar Yilmaz. They particularly wish to thank Fily Sissoko (Practice Manager, East Asia and Pacific Governance Unit); peer reviewers Matthew Arnold (the former Country Representative of The Asia Foundation, Myanmar) and Timothy Williams Stephenson (Senior Public Sector Specialist, World Bank). The paper was prepared by the Decentralization Global Solution Group of the World Bank. Production management was provided by Ayşe Boybeyi (Senior Knowledge Management and Learning Officer, World Bank) with the editorial assistance of Barbara Rice (Consultant). 1 1. INTRODUCTION In 2005, the World Bank published its foundational report, East Asia Decentralizes: Making Local Government Work (World Bank 2005). The authors observed that a fundamental transformation was taking place in the structure of government in the region: a host of diverse unitary states were pursuing, at their own pace, a process of decentralization. The result was that subnational governments (SNGs) were being placed at the center of a range of fiscal, regulatory, and service delivery activities that was changing the fulcrum for much of the region's economic and social development. The analysis took stock of the form and function of decentralized governance across six countries and drew on lessons from around the world to help country reformers better design and implement their transformations and avoid potential pitfalls. Since then, decentralized governance has become even further entrenched in the region— broadening to two new previously centralized countries (the Lao People's Democratic Republic and Myanmar) and generally deepening within other countries. In most places, subnational governments have grown in fiscal importance and now account for a much larger share of public expenditure than they did a few decades ago. This does not, on its own, necessarily indicate a strengthened subnational level. But when combined with steps taken to empower and improve the performance of SNGs over time, it indicates that the subnational level is playing a greater role in respective systems of governance in the diverse developing economies of the region. In light of the ongoing evolution of decentralized governance, this discussion paper provides a timely update of the original World Bank (2005) analysis. It incorporates new information to consider the major structural features of contemporary decentralization in the region, including the ways that the "3Fs" of public administration (i.e., functions, funds, and functionaries) are presently organized between the different levels of government. It also focuses on the same six East Asian countries in the 2005 report but expands the analysis by including three member states of the Association of Southeast Asian Nations (ASEAN): Lao PDR, Malaysia, and Myanmar. The slight geographical expansion southward from the original six countries in the 2005 report also leads to a slight change the nomenclature for this broader region, from "East Asia" to "Southeast Asia."1 Additionally, taking decentralization as a given in the region—and a higher-level principle and a feature of any large system—this analysis focuses on identifying the underlying political drivers that have led to the decentralized systems the authors observe across countries in Southeast Asia today. To that end, it does not consider whether decentralization is "good" or "bad," per se, but rather how it works within the systems of governance. Predictably, decentralization reforms are heavily influenced by political pressures. These include 1 The six original countries in World Bank (2005) are Cambodia, China, Indonesia, the Philippines, Thailand, and Vietnam. In this discussion paper, the nomenclature could easily have remained East Asia in line with the original report. 2 geographical and cultural heterogeneity, demands for improved service delivery, burgeoning cities, and conflict dynamics, to name a few. Accordingly, reforms evolve to suit the unique characteristics of each country. Despite the unique country contexts, common themes and lessons can nonetheless be distilled. In charting the overall political trajectory of reforms, the aim of the analysis is to help countries locate their place in the process. As the world's economic center of gravity continues to drift toward Southeast Asia, the economic, social, and political implications of decentralizing reforms are also likely to radiate beyond the region, and therefore the analysis should also be of interest to a broader audience. What is clear is that the countries of Southeast Asia have not followed a technically neat trajectory of decentralization by transferring funds, functions, and functionaries to the subnational levels to align authority and accountability. Nor have they transferred authority in response to strong regional demands. Instead, transitions have been managed from the center and driven by its priorities. The result is a mix of enduring centripetal biases and a lack of synchronization among the 3Fs, which places considerable constraints on the authority of SNGs. Despite these limitations, states are muddling through with general trajectory toward more, not less, decentralization. In large part, this is happening organically and incrementally (if at times sporadically) without any big bang reforms. While the nonlinear and messy trajectory of deepening decentralization may help explain why reforms have not always met lofty expectations, this should not minimize their importance. Collectively, SNGs are playing a pivotal role in shaping the political and economic landscape of the region. However, with deepening decentralization comes more significant coordination challenges. The discussion paper is structured across five chapters. Chapter 2 sets the scene, providing important context about the Southeast Asia region and the motivations for decentralizing reforms. Chapter 3 examines the structural characteristics of contemporary decentralized governance, looking at the different ways that subnational layers are organized and the policy emphasis, as well as the legal and political frameworks that govern the assignments of functions and finance (the first two Fs) in each country. It also examines the various mechanisms that are used to overcome vertical and horizontal fiscal imbalances. Chapter 4 then examines, in more depth, some of the notable features of the reform process. In particular, it draws attention to how the central government projects its influence, in particular through its ongoing control of public administration (i.e., functionaries— the third F), how the SNGs are evolving to overcome the constraints on their authority, and how conditions can and do chart somewhat of a middle path between centralized control and autonomy. Chapter 5 concludes by looking at some of the key lessons from the Southeast Asian decentralization experience and identifying some possible areas to meet the growing challenges of coordination as decentralization gathers pace. 3 2. THE CONTEXT The context of Southeast Asia is key to understanding how and why decentralization reforms have evolved the way they have in the region. To be sure, there is no distinct Southeast Asian pathway to decentralization reform. However, several discernible traits make the countries of the region stand out. Decentralization in Southeast Asia: A Common Trait in a Diverse Region Southeast Asia includes a highly heterogeneous mix of developing countries of varying sizes, income levels, and political systems. The region comprises the demographically and economically enormous state of China, plus a combination of other countries with larger and smaller populations and economies—with Indonesia at one end of the spectrum (261 million people) and Lao PDR at the other (7 million). There is also considerable heterogeneity of political arrangements, ranging from communist one-party states (China, Lao PDR, and Vietnam) to competitive multiparty democracies (Indonesia, Malaysia, and the Philippines). There exist constitutional monarchies with weak democratic institutions (Cambodia and Thailand), and there is a democratizing republic with substantial remnants of military rule (Myanmar). Malaysia, which has been a federation for more than 60 years, is the only formally federal nation in the region. Myanmar also has some nods to federalist features, with the push to create a more federalist structure at the heart of its enduring subnational conflict (World Bank 2019a). All other countries are politically unitary states. The region is also highly dynamic with fast growth and rapid social change occurring in the past few decades (table 1). Economic growth has generally outpaced the global average, and the region is now comprised of exclusively lower-middle- and upper-middle-income countries. Growth has been fastest in China and Myanmar; the latter though coming from a particularly low base, with Myanmar's per capita income generally being among the lowest in Southeast Asia since the 1990s. Commensurate with the rapid pace of economic growth, extreme poverty in the region has fallen drastically (World Bank 2019b). Growth has also coincided with some distinct spatial shifts, with the urban share of the population rising strongly in most places. Except for Cambodia and Myanmar, the rate of growth is no longer accelerating. China, Indonesia, and Thailand now have majority urban populations. In contrast, Cambodia, Lao PDR, and Myanmar remain predominantly rural countries. 4 Table 1. Economic and Demographic Trends in Southeast Asia, 2000–2016 Location Population Gross Compound Annual Urban Urbanization (millions, National Rates of Growth Population Rate 2016) Income (current / constant (% of total per Capita, purchasing power population) Atlas parity) Method 2000– 2010– 2000– 2000– (US$, 2016) 2000 2016 2010 2016 2010 2016 World 7,442 10,308 4.9 / 2.5 4.2 / 2.4 47.1 54.3 2.3 2.1 Cambodia 16 1,390 8.4 / 6.2 7.4 / 5.5 16.9 20.9 2.3 2.6 China 1,379 9,460 12.2 / 9.9 8.8 / 6.9 35.8 56.8 3.8 2.9 Indonesia 261 3,840 6.0 / 3.8 5.9 / 4.0 42.0 54.5 3.1 2.7 Lao PDR 7 2,450 7.6 / 5.4 7.6 / 5.8 22.0 39.7 5.9 4.4 Malaysia 31 10,590 4.7 / 2.6 5.6 / 3.7 62.0 74.8 3.4 2.3 Myanmar 53 1,310 13.4 / 11.1 8.0 / 6.1 27.0 34.7 2.4 2.6 Philippines 103 3,830 5.0 / 2.8 6.4 / 4.5 58.5 44.3 1.3 1.3 Thailand 69 6,610 6.1 / 3.9 4.7 / 2.9 31.1 51.5 4.1 3.1 Vietnam 93 2,360 7.8 / 5.6 7.0 / 5.1 24.3 34.2 3.4 3.1 Source: World Bank Development Indicators Database. Note: The urbanization rate is calculated as the average annual rate of change of the urban population. Another commonality among the diverse countries of the region is the transference of resources and spending authority to lower levels of government. In the early 1990s, a host of unitary states began pursuing, at their own pace, the process of decentralizing their systems of governance. World Bank (2005) arranged countries into three broad categories: • The fast starters (Indonesia and the Philippines), which rapidly introduced major structural, institutional, and fiscal reforms—in a "big bang" approach—responding to a sudden and far- reaching political stimulus. It involved the devolution of political, fiscal, and administrative responsibilities to more autonomous subprovincial entities. • The incrementalists (China and Vietnam), which took a more piecemeal approach to decentralization, focusing on administrative and fiscal reform with modest political change and the retention of considerable central control over subnational governments. • The cautious movers (Cambodia and Thailand), which established significant elements of decentralization at the formal policy and legislative levels but made limited progress with implementation. Broadly, these changes were driven by the rapid pace of economic development and urbanization 5 underway in the region plus a variety of political imperatives, which created an environment conducive to decentralization reforms. In general, structural economic transformations and the rapid growth of cities meant that centralized models of public administration and public financial management were increasingly ill-suited for supporting the rapid development of the private sector development. While growing cultural and economic heterogeneity within countries led to ever-stronger demands for improved and responsive local governance and service delivery. In some places—particularly the large archipelagic states of Indonesia and the Philippines—increasing levels of democratization and the rejection of centralized autocratic dictatorships also led to sudden shifts in political authority down to the local level. Since the initial wave of decentralizing reforms in the six aforementioned countries, both Lao PDR and Myanmar have undertaken significant reforms of their own. Myanmar established a new level of SNG in its military-drafted 2008 constitution. With sudden devolution of political power and budgeting authority to first-tier local governments in an attempt to relieve center-periphery tensions, the impetus, if not the depth of the effort, in Myanmar mirrored that of Indonesia's “big bang” approach. In Lao PDR, the government began its decentralizing reforms with its 2012 Sam Sang directive,2 aimed at accelerating local economic development through, among other things, greater devolution and local governance. Constitutional amendments in 2015, plus updates to the Law on Local Government Administration and the State Budget Law, established the architecture for the separation of powers, including the establishment of People’s Provincial Assemblies. However, progress in implementing reform has been slow, mirroring that of its neighbor Cambodia. The design and structure of new intergovernmental fiscal relations are still being developed, based on assessments of findings from recent pilot activities and the government’s evolving policy priorities and objectives (World Bank 2019c). While accessing consistent information on subnational spending (box 1) in general is not always straightforward, the role of Southeast Asian SNGs in public finance appears to have grown over time (table 2). China clearly has the highest share of subnational spending in the region, now accounting for almost all public expenditure. Indonesia and Vietnam shift around half of the spending to their respective subnational levels, with the subnational share growing over time in both cases. In Lao PDR and Myanmar, evidence suggests that central governments are prioritizing the reassignment of expenditure to lower levels. In Lao PDR, the subnational level accounts for nearly one-third of spending. In Myanmar, the total available resources for SNGs have grown rapidly in a short time, rising threefold in nominal and per capita terms in the five years from 2012/13 to 2017/18 (World Bank 2019a). At the other end of the spectrum, public spending is much more centralized in Cambodia and Malaysia. In Cambodia, a concerted effort to shift expenditure to the SNGs has led to a doubling in the subnational share of total government spending between 2013 and 2018. Malaysia and the Philippines are notable for the retreat in the subnational share of spending in recent years, though in each case, the nominal level of expenditure has risen. 2 The Sam Sang (“Three Builds”) directive seeks to accelerate local economic development through targeted dissemi- nation of party resolutions and government policies, decentralized public administration, and development of targeted villages. It proposed villages as the development unit, districts as the integration unit, and provinces as the strategic unit. 6 Box 1. The Challenges of Accessing Subnational Spending Information The estimates of the subnational share of government spending in the World Bank’s 2005 report were drawn from country case studies prepared for the report and World Bank statistics. The World Bank had an international dataset on fiscal decentralization indicators, which has not been updated since 2001. More recently, a new resource has emerged, the Global Observatory on Local Finance (GOLF), hosted by United Cities and Local Governments (UCLG) in partnership with the Organisation for Economic Co-operation and Development. It collates public finance data from official sources, such as national and budgetary accounts, government finance statistics from the International Monetary Fund, and relevant published studies. GOLF is a useful and popular source for subnational spending. Its flagship study—Subnational Governments around the World: Structure and Finance (OECD/UCLG 2016)— is a first exploratory step. The study identifies the primary information sources at international levels as well as major methodological and information gaps that will need to be addressed for further development. GOLF provides snapshots for each country, including the subnational share of general government spending. At the time of writing, the latest available estimates were from 2016. Unless otherwise stated, the data in table 2 and figures 1 and 2 are taken from UCLG data files; specifically, the published country case studies. Table 2. Subnational Expenditure Shares in Southeast Asia (percent of total government expenditure)a Country 2001-2002 2015-2018 Cambodia NA 7 China 69 91 Indonesia 32 53 Lao PDR N/A 31 Malaysia 13 8 Myanmar N/A 12 Philippines 26 15 Thailand 10 19 Vietnam 48 54 OECD countries 32 40 Sources: Word Bank (2005); OECD/UCLG (2019); respective Treasury and Finance departments. Note: OECD = Organisation for Economic Development and Co-operation. a. Different years: for 2001–2002, data are from 2001 except for China, Indonesia, and Vietnam (each being 2002) and taken directly from World Bank (2005); for 2015–2017, data are for 2016 and sourced from OECD/UCLG (2019), except Lao PDR (2015), Indonesia, Malaysia, and Myanmar (each being 2017), and Cambodia (2018) sourced from respective Treasury and Finance departments. 7 Subnational governments have greater fiscal weight in Southeast Asia than in comparable countries globally (figure 1). Total government spending as a share of gross domestic product (GDP) among the selected developing countries of Southeast Asia (22.1 percent average in 2016) is small— generally well below the average of all other regions and below the average of middle-income countries globally (32.0 percent).3 However, the subnational share of public spending in Southeast Asia (31.4 percent) is much higher than other middle-income countries (21.8 percent) and unitary countries more generally (19.4 percent).4 The net effect is that SNG spending as a share of GDP is broadly comparable with countries at a similar level of development and state structure. China has the highest subnational share globally by some margin (figure 2), though it is not necessarily the world’s most decentralized state (see chapter 4). Vietnam is also well above the average of federal states, and Indonesia is around the average of federal states. Indeed, these three countries have the first, third, and fifth largest subnational spending shares of all unitary countries. On the flipside, Malaysia clearly has the smallest subnational spending share among federal states, while Cambodia is also among the most fiscally centralized states globally. Myanmar, the Philippines, and Thailand are around the average of the Sub-Saharan Africa and Latin America and the Caribbean Regions, where spending tends to be more centralized (OECD/UCLG 2019). Figure 1. Subnational Government Expenditure as a Percentage of Public Expenditure and GDP, 2016 (selected Southeast Asian countries compared to unitary and federal states and middle-income countries) Selected 7.4% SEA 31.4% 6.9% Unitary 19.4% 16.8% Federal 46.9% Middle 6.6% Income* 21.8%  SNG spending as a % of GDP  SNG spending as a % of public expenditure Source: OECD/UCLG (2019). Note: Middle income is a combination of lower-middle-income and upper-middle-income countries. SEA = Southeast Asian; SNG = subnational government. 3 General government expenditure as a share of GDP in the selected countries is on par with the average of low-in- come countries (22.7 percent). 4 This remains the case even after excluding China (23.9 percent). 8 Figure 2. SNG Expenditure as a Percentage of GDP and Public Expenditure, 2016 100 SNG spending as a% of total public expenditure 80 60 40 20 0 0 10 20 30 40 SNG spending as a% of GDP  Rest of World  Latin America  Africa  Selected SEA countries Source: OECD/UCLG (2019). Note: CHN = China; GDP = gross domestic product; IDN = India; KHM = Cambodia; LAO = Lao PDR; MYS = Malaysia; MMR = Myanmar; PHI = Philippines; SEA = Southeast Asian; SNG = subnational government; THA = Thailand; VNM = Vietnam. A Series of Managed Transitions Driven from the Center In general, the origins of recent decentralization reforms are notable for having been part of a managed transition, driven by ruling elites at the center against a backdrop of peace. Besides Malaysia, where decentralized governance has been a feature for many decades, for the most part, decentralization reforms since the 1990s in Southeast Asia have mostly been instigated as a deliberate strategy to improve the quality of governance, service delivery, and local participation. With some notable exceptions, conflict resolution has not been an objective. This stands in contrast to Sub-Saharan Africa and the Middle East, where decentralization has often been used as a conflict mitigation mechanism to redesign the distribution of land, power, and resources (Wolff, Ross, and Wee 2017). In the communist one-party states of China, Lao PDR, and Vietnam, decentralization was part of the development strategy of the central government. In China and Vietnam, reforms were rooted in the broader market transitions (Eaton, Kaiser, and Smoke 2011). China's sheer size and increasing complexity meant that a central government could not maintain effective oversight of state-owned enterprises (SOEs) and implement market-based reforms in the centrally planned system. In Vietnam, transferring authority to subnational administrations was seen as a way to improve the quality and 9 efficiency of government following the liberalizing Đổi Mới (Renovation) reforms. In each case, decentralization allowed the respective politburo to meet audible demands for citizen participation and good government at the local level. Lao PDR appears to be following a similar development pathway. In Indonesia and the Philippines, the region’s two large and diverse archipelagoes, decentralization was part of a centrally managed transition from authoritarian to civilian rule and used to relieve center- periphery tensions. In Indonesia, decentralization reforms followed the overthrow of the Suharto regime and occurred against the backdrop of the de facto secession of East Timor. In a country inherently difficult to manage from the center—with more than 17,500 islands spread across nearly 2 million square kilometers, isolated population centers, and considerable cultural heterogeneity— the reforms were aimed at minimizing the probability of further provincial separatism (particularly in Aceh and Papua) and arresting the declining legitimacy of the governing party (World Bank 2005). Accordingly, subnational governments—mainly at the local level—were given substantial powers to collect and expend revenues and new responsibility to provide services to their citizens.5 Somewhat similar to Indonesia, the Philippines’ original decentralization reforms were amid a central push to improve democratization and growth following the fall of the authoritarian Marcos regime. A multilayered subnational system of governance was created that diffused political and other powers across the country’s diverse geographical, cultural, and ethnological landscape. In Myanmar and Thailand, subnational governance was integrated into the enduring semi-authoritarian governance apparatus—offering increased pluralism while minimizing perceived challenges and threats to the central government. In Myanmar, following at least five decades of the subnational conflict and in the wake of a popular uprising, the ruling military junta chose to embark on a highly choreographed top-down process of decentralization. The reforms embodied in the 2008 constitution were not linked to a coherent strategy for meeting the demands of warring ethnic groups; however, nor were they linked to devolving power to state and region responsibilities over their internal affairs or for improving service provision. Instead, decentralization was primarily viewed as a tool to relieve center-periphery tensions by providing space for greater political pluralism while protecting the core security and economic interests of the old regime (World Bank 2019a). In Thailand, following widespread protests in the mid-1990s, military rulers realized that a much greater degree of pluralism needed to be tolerated to retain an important political role (Dufhues, Theesfeld, and Buchenrieder 2012). Accordingly, an initial decentralization framework was proposed by central political forces as part of the broader democratizing reforms in the 1997 constitution. After a subsequent military coup and the promulgation of the 2007 constitution, central bureaucratic and political forces have been able to protect their interests by shaping acts and regulations concerned with decentralization (Haque 2010). The latest military government, in power since 2014, has evinced few signs of being interested in local government, though the interim 2017 constitution did not fundamentally change the country’s decentralized institutional configuration (Unger and Mahakanjana 2016). 5 Concerns regarding potential secession convinced policy makers to skip over the provincial level and concentrate decen- tralization at the district level (see Ostwald, Tajima, and Samphantharak 2016). 10 In Cambodia, a post-conflict society, reforms were not in response to, or as a result of, the war, but rather occurred amid relative peace and political stability and developed at a high level in the national government (Eng and Ear 2016). The Royal Government of Cambodia’s Rectangular Strategy emphasizes improvements in the efficiency and quality of public service delivery (RGC 2013). It has been argued, however, that the initial democratization of commune (and subsequently district and provincial) governance and the transference of service delivery was instead a tool for consolidating the capacity and control of the Cambodian People’s Party in the field in a context where citizen trust of the state is relatively weak (Smoke and Morrison 2008).6 More recent reforms in the Philippines have emerged as part of a conflict settlement. Reforms to subnational governance have been a central plank in the peace settlement between the Philippines government and the Moro Islamic Liberation Front, which has fought a decades-long insurgency for self-determination. The resultant Organic Law for the Bangsamoro Autonomous Region in Muslim Mindanao, signed in 2018, created the Bangsamoro Autonomous Region in the predominantly Muslim region and devolved significant powers and fiscal capabilities to the new government. 3. STRUCTURAL FEATURES OF CONTEMPORARY DECENTRALIZED GOVERNANCE The region is now comprised of largely unitary states with distinct vertical levels of subnational government. Structurally and politically, decentralized systems of governance have evolved to fit the unique characteristics of each country (see table 3). However, several features are related to state structure, policy emphasis, functional assignment, and financing that are worth examining in more detail. State Structure and the Emphasis of Decentralization Defining subnational government as a level of administration with some degree of formal budget (expenditure) authority (see box 2), most SNGs in the region exist at the provincial and municipal levels. Six countries (Cambodia, China, Indonesia, Lao PDR, the Philippines, Thailand, and Vietnam) have at least three SNG levels, with an intermediate layer between provinces and municipalities. Malaysia has no intermediate layer, while Myanmar is unique for having only one subnational level and no formal local tier of government (though special powers are carved out for municipal governance).7 Beneath Myanmar’s SNG is a system of district, township, and village-level 6 One of the potential political drivers of decentralization is the desire of strong autocratic governments to minimize perceived threats and challenges to the center by diffuse power among the smaller and more fragmented local government level rather than the more politically powerful second (provincial) tier. Cambodia aside, where political decentralization has been most significant at the lowest commune/sangkat (CS) level, the authors do not think this has been a feature of decentralization reforms within Southeast Asia to date. 7 While it is relatively common for unitary states to only have only one level of subnational government, it tends to be at the municipal level (OECD/UCLG 2016). This makes Myanmar’s one layer of SNG at the provincial level particularly unique. 11 administration that is integrated into the centrally managed system of the General Administration Department (for more information, see World Bank 2019a). The average population of the first-tier subnational governments is between 1 and 8 percent of the national population. China aside, the largest average population for first-tier SNG is in Indonesia (7.5 million), followed by Myanmar (3.4 million). In contrast, Cambodia and Lao PDR have relatively small average provincial sizes of around half a million. Notably, most capital cities and economic centers are uniquely treated. They sit at the same level as provinces and are often directly subordinate to the national government.8 In the Philippines and Vietnam, the largest subnational jurisdictions, by some margin, are the largest cities—Metro Manila in the Philippines and Hanoi and Ho Chi Minh City in Vietnam. In Indonesia, the Javanese provinces, including Jakarta, oversee populations nearly an order of magnitude larger than the average of the remaining provinces. Malaysia’s relatively large average municipal size (about 200,000) is considerably larger than its regional peers (about 28,000 in Thailand and less than 10,000 in Cambodia, Indonesia, the Philippines, and Vietnam) (OECD/ UCLG 2019). Box 2. Defining Subnational Government in Southeast Asia There are challenges in defining precisely the characteristics of a subnational government (SNG). The Organisation for Economic Co-operation and Development (OECD) and the United Cities and Local Governments (UCLG) define SNG as “a decentralized entity elected through universal suffrage and having general responsibilities and some autonomy concerning with respect to budget, staff, and assets” (OCED/UCLG 2019, 27). They also note exceptions, for example, when subnational entities are elected through indirect suffrage or “dual” entities (i.e., having an elected council but an executive appointed by the central government). This ensures the SNGs of Southeast Asia’s communist states (China, Lao PDR, and Vietnam) satisfy the definition. The World Bank took a somewhat different approach, suggesting that level of government “refers to an organ of government with some degree of formal budget (expenditure) authority” (World Bank 2005, 11). This broader definition potentially encompasses deconcentrated districts or agencies of the central, federal, or state government. Therefore, centrally managed major cities in China and Vietnam, as well as provinces in Cambodia and Thailand, would satisfy the definition of an SNG, though possibly not the OECD/UCLG definition. As the focus of this analysis is on the practical implementation of decentralization reforms, rather than their archetypal structure, the definition used is the same as that applied by the World Bank (2005). 8 In China, Vietnam, and Thailand, such special designation is extended to four, five, and two cities, respectively. Cambodia and Lao PDR consider the national capital to be its own administrative region. The Philippines has 38 independent cities. 12 Given its enormous size, China is an outlier in the region, with provinces overseeing populations as large, or even larger, than many countries and playing a similar role in day-to-day public service delivery that central governments play in smaller countries (box 3). Officially, China has four SNG levels: 34 provincial-level governments (including 23 provinces, five autonomous regions, four centrally administered large cities, and two special regions); 334 prefecture-level cities, each with an average population of 3.8 million; 2,850 counties and county-level cities; and more than 40,000 townships (NBS 2016).9 Provinces, prefectures, and counties each have their own revenue and expenditure assignments. Villages and townships, at the sub-county level, have their own budget, but they have no own revenue source. The main focus of China’s decentralization strategy is the deconcentration of economic authority to the provincial level and the empowerment of city governments. Provincial governments implement laws, regulations, and decisions of the provincial People’s Congresses and their standing committees, to which they also report on their work. Fiscal relations between provinces and local governments (counties and townships) have been largely left to individual provinces to define, which is a feature more typical of a federal than a unitary state (UNDP 2019). There are also several special autonomous areas within countries, reflecting an explicit effort to accommodate their ethnic groups. Two provinces in Indonesia, Aceh and Papua, are granted special administrative status with additional powers. In particular, in Muslim-majority Aceh, the provincial government has powers over the judicial and education systems, with the right to apply Sharia law, even in criminal cases. Aceh also enjoys a larger share of revenues from its oil and gas resources. Papua also has devolved powers and access to additional natural resource revenues, in addition to the special Papuan People's Council to arbitrate Papuan tribal customs. In the Philippines, the recently created Bangsamoro Autonomous Region in the Muslim-majority Mindanao also has additional legislative, fiscal, and judicial powers. China and Myanmar each also have several separate autonomous regions where minority nationalities are concentrated. Unlike Indonesia and the Philippines, however, the main function of China’s five province-level autonomous regions and Myanmar’s six self- administered areas (SAAs), which are collections of individual townships, is to provide political representation to designated ethnic groups.10 In China, local congresses in autonomous regions are given partial self-governing powers through the constitution and the Law of the People's Republic of China on Regional National Autonomy (largely the preservation of local languages, folkways, and customs). They remain entirely subordinate to the central government, owing primary responsibility to the Chinese state.11 While in Myanmar, despite 9 The 23 provinces include Taiwan and two special administrative regions (Hong Kong SAR and Macau SAR), each of which have unique colonial antecedents. They are counted as subnational governments just below the central government. For more information, see the Chinese government’s website at http://www.gov.cn/test/2005-06/15/ content_18253.htm. 10 In China, autonomous regions are Guangxi, Inner Mongolia, Ningxia, Tibet, and Xinjiang. In Myanmar, the self- administered zones are Naga in the Sagaing Region and Danu, Kokang, Pa’O, Pa Laung, and Kokaung along with the Wa Self-Administered Division of in the Shan State. 11 An added benefit provided to China’s autonomous regions is that the Mineral Resources Compensation Fee is shared 60/40 with the central government, whereas in other provinces the share is 50/50 (Bauer et al. 2016). 13 their own leadership and array of legislative powers outlined in the constitution, the SAAs have limited practical autonomy. In both China and Myanmar, the autonomous regions are also subject to national laws. Box 3. Decentralization in China Chinese governance is highly decentralized, with the subnational level accounting for more than 90 percent of all government spending. Subnational governments (SNGs) at each level have wide functional mandates: being responsible for delivering essential public services, including education, health care, and social welfare; local economic development; public safety; and judicial administration within their jurisdictions (OECD/UCLG 2019). SNGs are also responsible for the bulk of public infrastructure investment and have, therefore, been key to China’s rapid urbanization (Wilkins and Zurawski 2014). However, unlike peers in the region, China’s subnational governments have no “inherent” powers. While the constitution recognizes three levels of government legally, no national law clarifies the functions of each level of local government, nor constitutional or dedicated legal basis for decentralization. Constitutionally, all powers and responsibilities of SNGs are delegated by the central government (which can also rescind powers) (OECD 2015). Accordingly, intergovernmental reforms have tended to occur as a byproduct of decentralizing economic reforms driven from the center. An early example, from 1957, involves Mao Zedong’s realization that China's sheer size and increasing complexity meant that a central government could not maintain effective oversight of state-owned enterprises (SOEs). Accordingly, authority over economic management was delegated to local authorities, which also gained authority over an array of locally derived taxes and access to a share of SOE profits (Lin, Tao, and Liu 2003). Additional waves of fiscal and administrative decentralization occurred from the late 1970s to support the implementation of (the central government’s) market-oriented reforms within the planned economy model, resulting in further expansion of the subnational share of revenue and spending (Jun 2018). Further major reforms to the intergovernmental fiscal framework system were initiated in 1994, which centralized revenue collection and ushered in a more transparent and rules-based system for determining transfers (Shen, Jin, and Zou 2012). The intention was to replace an unsystematic and pro-cyclical system of fiscal contracting between the central government and provinces with a more systematic and consistent approach to tax assignment and tax sharing between levels of government. The reforms were successful in nearly doubling revenues as a share of gross domestic product (GDP). 14 Intergovernmental transfers, in particular, from shared tax revenues, therefore, play a large role in financing subnational spending, particularly for the lower subprovincial levels. The central government acquires the majority of tax revenues, with local governments limited to a small array of property- and land-based taxes, which complement nontax revenues from SOEs, sales of land use rights, and fees and charges (Wang and Herd 2013). Of these, SNGs only have autonomy over land concession revenues. Three broad transfer types are used: general transfers used to lower disparities in expenditure; earmarked transfers used to subsidize local projects in certain areas subject to matching outlays by local government; and compensation transfers for the reductions in local revenues resulting from the 1994 reforms (Wang and Herd 2013). Each of the three groups has several subcomponents. Transfers cascade down the levels, with each level of government transferring to the level below. Large vertical and horizontal fiscal imbalances have emerged, particularly at the country level, which now has the largest fiscal mandate, about 9 percent of GDP (OECD 2019). SNGs are responsible for the largest and fastest-growing recurrent expenditure items on account of urbanizing and aging dynamics (Wingender 2018). The central government also has routinely imposed new mandates on provincial governments, generally expecting local governments to pay for them (Su, Tao, and Yang 2018). Moreover, with China’s overly pro-urban development strategy, provinces and city-level governments have been incentivized to overemphasize urban growth to win favor for promotion within the party (Lin, Tao, and Liu 2003). This led to both economic overheating as well as city governments withholding revenue sources allocated to counties and shifting expense responsibilities due to the priority of urban construction and development (Wang and Herd 2013). Reforms were implemented to try and alleviate acute shortages of resources by reorganizing the fiscal relations to shift responsibility for managing country-level fiscal resources from prefecture-level cities to higher-level provinces. However, there is little evidence that reforms have alleviated fiscal imbalances at the lower levels (Huang et al. 2017). Moreover, the reliance on shared tax revenues (shared based on where they are derived) has led to widening disparities between counties, with poorer, less urbanized areas severely disadvantaged (Wingender 2018). The upshot is that SNGs face considerable challenges funding their wide and growing mandates. Accordingly, they have turned to debt markets and other shadowy extra-budgetary funding vehicles. The reliance on land sales to finance urban development has also contributed to corruption, social unrest (protests by people evicted), and the absence of sustained sources of revenue. 15 The political emphasis of decentralization reforms varies considerably across countries. Democratic and democratizing countries tend to devolve political authority (to varying degrees), while the communist countries centralize political power. The democratic countries of Indonesia and the Philippines have the most politically devolved systems, with competitive elections held at all SNG levels. Mirroring the devolution of political authority, administrative and fiscal responsibilities are also devolved throughout the subnational level. Democratic Malaysia also devolves political authority to its first-tier SNGs, though the elected state governments appoint executives within municipalities. In Cambodia and Thailand, political decentralization has only occurred at the lower levels of government. In Cambodia, provinces and districts remain deconcentrated entitles of the central government, albeit with their own budgets and spending authority. In Thailand, the regional administrative structure of the central government is intertwined with more autonomous local governments. Democratizing Myanmar devolves political authority to elected state and region governments, albeit only partially given the military’s unelected role in politics.12 However, deconcentrated departments of central line ministries coexist with departments carrying out subnational-level functions, occasionally within the same subnational ministerial portfolio (Batcheler 2018). The communist countries of China, Lao PDR, and Vietnam devolve no formal political authority, with SNGs operating as deconcentrated political organs of the central politburo. Nonetheless, in China and Vietnam, provinces have considerably devolved fiscal and administrative roles and have de facto power to define mandates and fiscal relationships with their lower-level SNGs (UNDP 2019). 12 The constitution reserves one-quarter of seats in all legislatures for the military. A small number of seats within each state’s and region’s legislature is also reserved for designated ethnic minorities. Ethnic groups that are not included in the 135 “national races” are not represented. 16 Table 3. The Structure of Subnational Governance in Southeast Asia Country Levels of Average Average Policy Political Subnational Population of Population Orientation Representation Government First-Tier SNG of of Decentralization of Subnational (2016, millions) Municipalities Government Cambodia • Province and 0.7 9,600 Organic laws establish Higher-level PC and capital subnational tiers of DM levels controlled by • District/ administration at the the central government municipality provincial (PC) and through appointed subprovincial levels boards of governors • Commune/ (DM/CS). Involves (with the latter also sangkat deconcentration to PC having councils with and DM levels, which is indirectly elected more significant in terms councilors); only the of public expenditures, and lowest levels (CS) have devolution to lower levels. competitive multiparty CS at the lowest level elections to elect operates as a subnational representative bodies administrative level with through universal independent governance suffrage. DM level arrangements. has indirectly elected councils. China • Provinces, 41.8 3,800,000 Unique in the region for People's congresses large cities, having no constitutional exist at all levels of special or dedicated legal basis government, but only administrative for decentralization; the village level is directly regions though the constitution elected. recognizes three legal levels • Prefectures of government (provinces, • Counties prefectures, and counties), • Villages the fourth level (villages) are recognized in the budget law. The historical focus has been on political centralization alongside a high degree of fiscal and administrative devolution. Governments at all levels are general purpose with wide mandates and budgets, though provinces wield considerable control over subprovincial levels. 17 Country Levels of Average Average Policy Political Subnational Population of Population Orientation Representation Government First-Tier SNG of of Decentralization of Subnational (2016, millions) Municipalities Government Indonesia • Province / 7.7 3,167 Constitution embeds Each level of government capital regional autonomy in has its own elected • Regencies / Indonesia’s system of leadership. city government, dividing the country into provinces, • Villages which are divided into districts and cities. Fundamental laws established the basis of administrative and fiscal decentralization: 22/1999 Regional Autonomy Law and 25/1999 establishing the fiscal relationship between the central and regional governments (both revised in 2004). In 2014, the Village Law recognized villages as self-governing areas. The initial focus was on the devolution of a broad range of government functions to cities and regencies, which replaced the earlier emphasis on deconcentration to provinces. Revisions to the Local Government Law recentralized some authority back to the central and provincial governments. Lowest levels now have a limited legal role, though in fact still have considerable autonomy. Provinces do not have hierarchical authority over local villages. 18 Country Levels of Average Average Policy Political Subnational Population of Population Orientation Representation Government First-Tier SNG of of Decentralization of Subnational (2016, millions) Municipalities Government Lao PDR • Province / 0.4 0.7 Constitutional amendment Provincial governors capital (2015) allows for the are appointed by the • District establishment of the president. District-level People’s Provincial chiefs are appointed • Village Assemblies for the first by the prime minister. time. This is set to function Provincial governors like the national assembly exercise considerable at the provincial level. authority over their Local government is budget allocation set up at the province, and operations. district, and village level. Administration at the The ministries at the village level is conducted central government level by local elected village constitute the policy tier, chiefs and village while the provinces and the councils. districts are engaged in the delivery of public services and implementation of development programs. Malaysia • State 2.4 205,350 Constitution (1957) Members of the • Local establishes multiethnic unicameral State federation of states Assembly are elected in and allocates executive competitive multiparty and fiscal powers elections every five years between federal and from single-member state legislatures. Local constituencies. Mayors Government Act (1976, and municipal councilors amended in 2016) sets are appointed by the out responsibilities state governments. financial provision of Local elections were local government (city suspended in 1965, councils, municipal councils, after which the local and district councils). government tier was Governance is centrally filled with state- dominated, with the bulk appointed officials and of legislative and fiscal brought under the direct powers allocated to the control of federal and center. State governments state governments. have a limited role, though they are responsible for oversight of local governments. 19 Country Levels of Average Average Policy Political Subnational Population of Population Orientation Representation Government First-Tier SNG of of Decentralization of Subnational (2016, millions) Municipalities Government Myanmar • States / 3.6 N/A Constitution (2008) Semi-democratic regions establishes a formal unicameral legislature subnational tier of (Hluttaw) located in government, involving 14 each state and region constitutionally equivalent elected through states and regions, competitive multiparty with partially elected elections. However, parliaments and a limited range of finance and one-quarter of seats administrative functions. in all legislatures are Most formal power reserved for the military, remains concentrated which also automatically at the union through receives the critical an interlocking web of Ministries of Defense, political, administrative, Home Affairs, and and financial provisions. Border Affairs. All chief Municipal governance is the ministers and mayors only area over which state of the two largest cities and region governments are appointed by the have sole accountability. All other powers shared with president. the union government. Philippines • Provinces / 1.2 2,398 Constitution (1987) includes Directly elected bodies autonomous decentralization and local exist at all subnational regions / government autonomy levels of government. independent among the fundamental cities principles, though, in reality, the Philippines is a unitary • Cities / state, so local autonomy is municipalities subservient to the national • Barangay congress and the president. The Local Government Code makes provisions for the basic structure, roles, and responsibilities of local government units.a The 2018 Bangsamoro Or- ganic Law created a new autonomous government in the Bangsamoro Region in Mindanao. A new draft constitution was presented to Congress in 2018 to establish a federal state, though progress has stalled. Involves a high degree of fiscal and administrative devolution across the three tiers of local government units. Each unit has its own tax administration, its own budgets, and bureaucrats that it can hire, fire, and promote. Barangays, the smallest unit, have a wide array of formal powers for relatively small populations. 20 Country Levels of Average Average Policy Political Subnational Population of Population Orientation Representation Government First-Tier SNG of of Decentralization of Subnational (2016, millions) Municipalities Government Thailand • Province / 0.9 28,282 Constitution (2017) Provincial governors major city enshrines the principles are appointed by the • Municipalities of local autonomy and central government, elected local government except in the Bangkok • Subdistricts that were first included Metropolitan (tambon) in the 1997 constitution. Administration, where State Administration Act the governor is popularly (1991) designates three tiers elected. PAOs headed of government (national, by a chairperson and a provincial, and local). At the provincial committee. subnational level, the de- concentrated apparatus of Chairman of the PAO the central state, through is head of the PAO regional administration executive branch and is (provinces and districts), directly elected in local intertwines with the largely elections constituents. autonomous system of The provincial legislative local government. Pro- assembly is directly vincial levels include 75 elected by local citizens provinces (changwats) that for a period of four are headed by a central- years. Each municipality ly appointed provincial or tambon is led by an governor and the munici- elected local council and pality of Bangkok. Within mayor. each province, a Provincial Administrative Organization (PAO) serves as the de facto government, with separate executive and legislative branches and functional control over municipalities and subdistricts. The historical focus has been on deconcentration to provinces and districts, but since 1997, framework shifted toward devolution to municipalities, districts, and subdistricts. Now involves a dual system of (deconcentrated) admin- istration in provinces and self-government in munic- ipalities.b Implementation has lagged, however, with the central government retaining a high degree of control; fragmentation within institutions leading to coordination problems; and a lack of effective local accountability mechanisms. 21 Country Levels of Average Average Policy Political Subnational Population of Population Orientation Representation Government First-Tier SNG of of Decentralization of Subnational (2016, millions) Municipalities Government Vietnam • Province / 1.5 8,511 No constitutional basis for A dual system of major city decentralization, with the legislative bodies at • District 2013 constitution noting all levels, including that for administrative the People's Councils, • Commune convenience, the country the political organ will be divided into accountable to the provinces and five cities National Assembly; under direct central rule. and the People’s The Law on Organization Committees of local (1994) and State Budget government, the local Law (2002), assign executive. Committee functions and resources to chairs participate in the subnational governments corresponding People’s (provinces, districts, and Council. People’s communes). The State Councils elect the Budget Law was revised People’s Committee. in 2015 to clarify rules for assignment of powers. Focus is on political centralization and deconcentration with a strong role for provincial councils and committees. Provincial governments have considerable administrative, functional, and fiscal responsibilities approaching devolution. Provinces also wield substantial regulatory control over subprovincial levels and are responsible for assigning functions and defining fiscal relations with subprovincial entities. Sources: World Bank (2005, 2019c); Rab et al. (2015); OECD/UCLG (2019); Berman (2011). Note: Population is based on the respective country population census or intercensal population census (e.g., Government Gazette in Thailand). "Level" refers to an organ of government with some degree of formal budget (expenditure) authority. In some cases (e.g., Indonesia), this can be highly circumscribed, particularly at the lowest levels. Myanmar includes the population of Naypyidaw, which remains a union territory. a. In the context of the Philippines, subnational government and local government unit are terms that are used interchangeably. b. This arrangement is seen in Napoleonic administration systems. 22 Functions and Finance The formal legal basis for defining the parameters of decentralized governance in Southeast Asia is generally strong. The functional responsibilities of SNGs tend to be enshrined in foundational laws (i.e., constitutions in Indonesia, Lao PDR, Malaysia, Myanmar, the Philippines, and Thailand; and Organic Laws in Cambodia and Vietnam). China is the notable exception, with SNGs having no inherent powers (see box 3). Apart from the more significant constitutional change that presaged decentralization reforms in Lao PDR and Myanmar, other notable changes include a new Organic Law that created the Bangsamoro Autonomous Region in the Philippines (see box 4) and Organic Laws in Cambodia that establish the three subnational tiers of administration. Thailand’s 2017 interim constitution reaffirmed the role of local government and provided a framework for their autonomy. Elsewhere, for the most part, these laws have remained relatively static, with minor legislative changes to clarify functional and expenditure assignments between levels of government. Box 4. The Organic Law for the Bangsamoro Autonomous Region in Muslim Mindanao The Organic Law for the Bangsamoro Autonomous Region in Muslim Mindanao (BARMM) creates a parliamentary-democratic system of local government in Bangsamoro. An independent 80-member parliament will be created, with the first elections expected in 2022. From this legislature, an executive will be elected with powers to enact its laws across 55 areas (ranging from health to operating public utilities and from quarantine regulations and food security to the administration of justice) within its territory. The BARMM will have an automatic allocation of the annual block grant, equivalent to 5 percent of the net national internal revenue of the Bureau of Internal Revenue and the Bureau of Customs, while the region will receive a 75 percent share of locally. Sharia courts will also have jurisdiction over cases exclusively involving Muslims in the region. Source: Cook (2018). For the most part, SNGs are either wholly or partly responsible for delivering social services (table 4). This is a common feature in decentralized systems, reflecting the fact that localized decision- making in these areas can help to improve the efficiency and effectiveness of service delivery and ensure infrastructure is responsive to local needs. In China and Vietnam, SNGs account for almost all recurrent spending on education and health, though there is considerable overlap between the different levels of SNGs (Rab et al. 2015; Wingender 2018). In Indonesia and the Philippines, social services are a devolved responsibility, and SNGs, therefore, exercise considerable local political, financial, and administrative autonomy. Cambodia is in the process of transferring health and education functions that were previously managed by deconcentrated ministerial offices to the subnational authorities. In Thailand, health and education are shared functions, though, in 23 reality, deconcentrated spending overshadows devolved spending (UNDP 2019). Both Malaysia and Myanmar stand apart. Constitutions in both countries centralize responsibility for health and education. In Myanmar, social services are also centralized; in Malaysia, it is a shared function. Table 4. Subnational Allocations of Social Services in Southeast Asia Functional Allocations Country Education Health Social welfare Cambodia Provincial Provincial Provincial China Provincial, local Provincial, local Provincial, local Local (primary and junior high Local, provincial (some Local, provincial (some Indonesia by district, high school by hospitals are managed by institutionalized care is province) provinces) managed by provinces) Lao PDR Provincial Provincial N/A Malaysia Central Central Central, provincial Myanmar Central Central Central Philippines Central, provincial, local Central, provincial, local Central, provincial, local Thailand Central, provincial Central, provincial Central, provincial Vietnam Provincial, local Provincial, local Provincial, local Source: World Bank data. Note: N/A = Not applicable; "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). In Myanmar, the equivalent is what would be the states and regions. Centralization of major revenue handles leads to large vertical fiscal imbalances and dependence on fiscal transfers. In all of the countries examined, major sources of taxation, such as personal and corporate income taxes, tend to be assigned to the central level of government, while land- and property-related taxes, excise taxes, and smaller fees and charges are assigned to lower levels.13 To a large degree, this aligns with the broad principles of effective revenue assignment: aligning revenues with the benefits of public spending and avoiding economic distortions or negative externalities (Bird 2006). It implies, however, that the own-source revenues (OSR) of SNGs are generally insufficient to cover the cost of delivering their functional mandates as well as their operating costs (table 5). The upshot is that SNGs are dependent on intergovernmental transfers to finance their expenditures. An extreme case is Cambodia, where almost all tax and most nontax revenues are currently being centrally collected and transfers fund almost all SNG expenditure outside of the capital (ADB 2018). 13 Some taxes may be collected by SNGs, however, the power to set rates and bases is generally assigned to the central government, as well as the bulk of the revenue. 24 Even in the Philippines, where despite the relatively broad powers assigned to all SNGs to levy taxes, fees, and other charges, wide gaps in taxing and collection capacity mean that OSR remains small. Of course, SNGs are not homogenous, and differences in fiscal capacity mean that there is also variation in OSR between SNGs within countries. Myanmar is a notable example. There is a considerable discrepancy between the commercial centers and growth poles of Mandalay and Yangon and the rest of the country. While OSR accounts for a majority of total financing in Mandalay and Yangon, elsewhere it is closer to 10 percent, comparable with regional peers (World Bank 2019a). Table 5. Estimated Own-Source Revenue of Subnational Governments (percentage of total subnational government revenue) Country Percentage Cambodia 1 China, 2016 a 45 Indonesia, 2016 14 Lao PDR N/A Malaysia, 2016 13 Myanmar, 2017/18 21 Philippines, 2016 32 Thailand, 2018 16 Vietnam, 2013 9 Sources: Wang and Herd (2013), Morgan and Trinh (2016), OECD/UCLG (2019), UNDP (2019), Chachavalpongpun (2019). Note: Figures include only official, legal revenues. a. Own-source revenue is buttressed by earnings of SOEs, which can account for up to 30 percent of revenue. Resource transfers commonly include grant and shared tax revenues, with the specific mix unique in each country (table 6). Unconditional block transfers play a particularly important role in Cambodia, Indonesia, Lao PDR, and Myanmar. Myanmar’s general grant transfer accounted for 87 percent of all funds transferred to SNGs in 2017/18 and play a dominant role in financing expenditure in a majority of states and regions outside of the growth poles (World Bank 2019a). Similarly, in Indonesia the main unconditional grant, Dana Alokasi Umum (DAU), plus a host of smaller unconditional grants funded nearly two-thirds of district-level spending in 2016 (World Bank 2017). In the Philippines, the Internal Revenue Allotment (IRA) accounted for 62 percent of total local government revenues in 2013, and even more in smaller and poorer cities (Canare 2016). While in Cambodia, two targeted unconditional grants from the central government budget, known as the District/Municipality (DM) Fund and the Commune/Sangkat (CS) Fund, account for the vast majority of SNG spending, save for in the capital 25 city.14 In Vietnam, unconditional balancing transfers interplay with the shared tax system, being used to narrow vertical fiscal imbalances after revenue-sharing arrangements have been applied. Earmarked, or conditional, grants are also used in most countries to finance specific programs, address emergencies, or stimulate local development.15 They are most prominent in China, Indonesia, Thailand, and Vietnam. Lao PDR and Myanmar are unique for not including some form of conditionality in the transfer system. Thailand makes considerable use of what it deems “specific” grant transfers, which account for more than half of central transfers and more than one-quarter of total local revenues (Varanyuwatana 2017). General purpose grants are then used as a top-up of local resources to bring SNGs’ revenue share of net government revenue to 25 percent as required by law. China combines its system of general purpose transfers with a widespread and highly granular system of earmarked transfers of equivalent size that are used to subsidize a host of local programs. At one point in 2013, there were some 200 narrowly earmarked transfers across some 20 categories.16 However, the trend is toward fewer earmarked grants and greater use of general purpose transfers to support basic service delivery. The number of earmarked grants was also cut to 76 in 2017 (Wingender 2018). In Indonesia, Dana Alokasi Kabupaten (DAK)—a special purpose grant that is earmarked for capital and recurrent expenditures in 19 priority sectors—has become an increasingly important transfer mechanism, rising sixfold in the 10 years to 2017, peaking at 27 percent of the total value of transfers in 2016. In Vietnam, conditional transfers that focus on both specific National Target Programs (NTPs)—urgent objectives of national significance—and regional economic development, in general, account for around one-quarter of local spending, on average. The central government sets targets and objectives for these funds, though local authorities still have responsibility for implementation (and can reallocate funds within each NTP). There are also several non-NTP target transfer schemes. Elsewhere, there appears to be a gentle shift toward conditionality. Conditional transfers are among the six fiscal transfer instruments available to Cambodia’s Ministry of Economy and Finance.17 While not widely used at this stage, it is envisaged that conditional transfers will be used to support the transfer of obligatory and permissive functions previously undertaken by the central government and 14 In Cambodia, both the intermediate district/khan (DK) and lower commune/sangkat (CS) levels directly receive unconditional grants from the central government, plus a share of tax revenues collected at the upper commune/ province (CP) level. From the perspective of the district/ municipality (DM) and CS administrations, these latter flows are also an unconditional subnational fiscal transfer. While the CP relies more on own-source revenues, there is also a small unconditional transfer from the national government, which is an essential source of financing for Provinces outside of the capital city. However, this mechanism is not created in law and there is no official guidance on the formula used to allocate funds between provinces. 15 Though conditional transfers have been largely in a fragmented and haphazard way that focuses on control rather than policy alignment (discussed further in chapter 4). 16 Mainly targeting transportation, social housing, agriculture, forestry, and water as well as energy saving and pollution abatement (Wang and Herd 2013). 17 The others are the District/Municipality Fund; formula-based provincial transfer mechanism; Commune/Sangkat (CS) Fund; Subnational Investment Fund; and own-source revenue. 26 for other obligatory functions (World Bank 2018). In the Philippines, the IRA has also become mildly conditional (Canare 2016). Beginning as an unconditional lump sum fiscal transfer, the Department of Interior and Local Government, which manages the IRA, has issued occasional guidelines on how the IRA should be spent. A 2007 memorandum stipulated that 20 percent of the IRA should be spent on development projects. Further, a 2012 order required 1 percent of IRA funds received by local governments to be used for programs for children.18 Revenues from centrally assigned tax handles are also shared with SNGs, usually based on where taxes are collected (the “derivation principle”). Commonly shared taxes include value-added taxes, corporate income taxes, and personal income taxes. Myanmar also shares stamp duty taxes with local government, though these are more often assigned to local government on account of their immobile base (World Bank 2019a). In Indonesia and the Philippines, tax sharing occurs through the transfer system, with both the DAU and IRA, respectively, being block grants based on fixed shares of total government revenues. The Philippines also shares tobacco excise taxes alongside shared revenues from a host of other tax handles. Cambodia’s DM and CS Funds are also explicitly linked to the revenue performance of the central government. In addition, revenue from eight shared tax handles also accrues to the upper capital/provincial (CP) level of subnational administration, which then transfers a fixed share on as an unconditional transfer to constituent authorities at the lower DM and CS levels. Lao PDR has also piloted revenue sharing in its Sam Sang initiative (LADLF 2015). For most countries, shared taxes account for a relatively small share of overall SNG financing. The notable exceptions are China and Vietnam. In China, shared taxes account for more than half of SNG revenues (OECD/UCLG 2019). In Vietnam, shared taxes account for around one-fifth of overall provincial revenues, though the share varies widely across provinces (Morgan and Trinh 2016). SNGs tend to have no practical authority over shared taxes. Even though SNGs may collect some of these taxes, the fact that they usually have little policy autonomy over setting rates and the base means there is little incentive for them to embark on reforms to improve the base, rate, and administration. Revenues from natural resources are also generally shared between levels of government. Indonesia allocates natural resource revenues to SNGs through its Dana Bagi Hasil transfer instrument, which includes a mix of “revenue sharing by origin” and “equal grant allocations” (UNDP 2019). The two special administrative provinces of Aceh and West Papua, with large natural resource wealth and a history of conflict, are assigned 70 percent of all oil revenues in their regions until 2027 (after which they will receive 50 percent). The remaining resource-rich local governments receive a considerably smaller share (NRGI 2018). In the Philippines, around two-fifths of resource revenues are shared at the subnational levels, split across the different levels of government. In Malaysia, a major producer of oil and natural gas, most revenues are centralized and codified in the constitution. A fixed royalty of 5 percent, however, is transferred to producing states according to an agreement with Petronas, 18 Other minor categorical, but not highly conditional, grants in the Philippines include the Municipal Development Fund, the Local Government Empowerment Fund, and the Calamity Fund. 27 the national oil company (Bauer et al. 2016).19 In Vietnam, the natural resource tax (except petroleum) is 100 percent locally retained revenue. In China, royalties on oil and gas production and corporate income tax from mining are shared between the central government and the provinces. Resource-rich Myanmar, where natural resource revenue is highly centralized, is an important counterexample despite the issue being intertwined in its enduring civil conflict. A disproportionately large share of Myanmar’s resource revenues is collected in conflict-affected areas, and equitable distribution of the country’s mineral wealth is a key demand of ethnic stakeholders in the peace process. Nonetheless, the vast majority of the most lucrative natural resource revenues (on oil, gas, and minerals) accrues to the central government.20 With the general trend toward natural resource revenue sharing in the region, it seems possible that Myanmar may ultimately introduce some sort for revenue sharing agreement. However, given the multitude of objectives (including benefit equalization, horizontal equity, and conflict prevention), the design of any such system will need to be carefully considered (Bauer et al. 2016). Table 6. Subnational Fiscal Structure of Selected Southeast Asian Countries Fiscal Transfers Own-Source Revenue Country Debt Revenue Unconditional Conditional Shared taxes from natural resources Low (except Cambodia Very low High Low None None Phnom Penh) China Moderate Moderate Moderate High Moderate High Moderate / Indonesia Low High Moderate Low Low high Low to Lao PDR High n.a. Low None None moderate Low (except Myanmar Low High n.a. None None Yangon) Malaysia Moderate Moderate Low Low Low Low Philippines Moderate High Low Low Moderate Low Thailand Low Low High Low Low Vietnam Low Moderate Moderate Moderate High Low Sources: UNDP (2019); Bauer et al. (2016); table adapted from World Bank (2005). Note: "Low," "moderate," and "high" refer to the rough proportions of total subnational revenues attributable to each revenue source relative to international experience. 19 When a new government was elected in 2018, the federal government floated the idea of reviewing and increasing the amount of the oil royalty up to 20 percent; however, that did not come through. The oil producing states of Sabah and Sarawak have since introduced petroleum sales taxes and imposed them on companies such as Petronas. 20 SNGs are assigned taxes on less lucrative natural resources, such as gravel and salt. 28 Rules-based approaches are generally used to predictably and transparently determine the overall size of the transfer pool and its allocation between jurisdictions. Variously, formulas consider criteria related to local development conditions, local administrative capacities, and fiscal gaps. In the Philippines, for example, the IRA is allocated to provinces, cities, municipalities, and barangays according to legislated fixed shares. Allocations among each of the four levels are based on various weighted criteria: 50 percent for population, 25 percent for land area, and 25 percent for equal sharing. In Cambodia, transfers are guided by formulas specific to each level of subnational administration and are divided into an administrative component and a development component, with the latter based on population and poverty rates. Central discretionary balancing transfers are used to address inequalities in China, Malaysia, and Vietnam, and are typically targeted at the poorest locations. However, as noted by the World Bank (2005), transfers continue to be only mildly equalizing, which is reflected in large and persistent inequalities in the available resources between regions. While there have been improvements over time, a common criticism is that the intergovernmental transfer systems continue to pay insufficient attention to the regional disparities in fiscal capacities. In Thailand and Myanmar, the equalization pool only constitutes 2 percent and 6 percent of the total fiscal transfers, respectively (UNDP 2019; World Bank 2019a). In Indonesia, DAU and other allocations are skewed toward districts with smaller populations. Urban areas, with larger populations and concentrations of poor populations, are not getting the financing they need to meet local development challenges and infrastructure needs (UNDP 2019). In both the Philippines and Thailand, grant transfers do not take into account the fiscal capacity of local government, leading to inequitable distribution of resources. China’s general transfers have become more redistributive over time, though they still do not fully compensate for regional inequality. The disparity of government spending per capita is increasing, with richer counties seeing the fastest rises (Wingender 2018). Moreover, sharing taxes based on where they are collected, not incurred, also exacerbates regional inequalities, inherently favoring more economically developed and populous locations (and locations where corporate headquarters are domiciled). Consequently, in places with where economic development is highly spatially unequal, such as Cambodia and Myanmar, shared taxes are an important mechanism for reinvesting in growth poles, but are negligible elsewhere. Fixed shares also mean that inequality of tax per capita will mirror disparities in GDP per capita (Wang and Herd 2013). In contrast, in Vietnam, the rules-based system of intergovernmental fiscal transfers, combining a redistributive mechanism within its shared tax system and balancing transfers (see box 5), has played a significant role in narrowing vertical and horizontal fiscal imbalances respectively and providing SNGs with highly predictable sources of revenue (Rab et al. 2015). 29 Box 5. Tax Sharing and Balancing Transfers in Vietnam In Vietnam, the shared tax system is geared toward redistributing locally collected revenues, which are predominantly collected in a small number of provinces. Sharing rates are defined as the proportion of locally collected revenues that are retained within a province and are based on the estimated fiscal gap and negotiations between provincial and central authorities. Sharing rates are revised every five years (known as a “stability period”). Poorer provinces with a positive fiscal gap are able to retain 100 percent of locally collected shared taxes, while richer provinces retain only a share, with the rest transferred to the central government to be redistributed through balancing transfers. In the 2017–2020 Stability Period, 50 out of 63 provinces retained 100 percent of locally collected revenues. In general, the retained shares in major cities have been declining across stability periods. Unconditional formula-based balancing transfers are then applied to reduce fiscal imbalances between provinces and across districts. Formulas consider total spending needs and equity objectives. They are geared toward narrowing regional gaps in economic development and providing capital services where they are most needed. Fiscal incentives for greater revenue collection are built into the criteria. Because such formulas are determined before each stability period, both the sharing rates and balancing transfers are fixed in nominal terms over each stability period, providing SNGs with highly predictable sources of revenues. Sources: Rab et al. (2015); OECD/UCLG (2019); Morgan and Trinh (2016). Lao PDR stands alone in its continued reliance on deficit financing, with no formula for equalization in place.21 Until recently, Myanmar also relied on deficit financing. This created perverse incentives for subnational spending control as well as for OSR collection in states and regions. It also introduced additional unpredictability and volatility in the volume of resources available every fiscal year (World Bank 2019a). However, as part of the medium-term fiscal framework (MTFF) reforms introduced in 2014, the grant transfer amounts each year became part of the annual budget preparation process, with the initial ceilings on the transfer determined when the MTFF is being prepared. Local government debt is mainly a feature in China and to a somewhat lesser extent Vietnam. SNGs in China have increasingly turned to debt to bridge the large gap between expenditure and revenues and to finance investment. Prior to 2015, due to formal restrictions on local borrowing, most funding originated outside the regular banking system using opaque off-balance-sheet entities known as local government financing vehicles (LGFVs). However, concerns that LGFVs may threaten financial stability led to reforms that made it possible for local governments to issue bonds to finance investment. Growth has been fast, with China’s local bond market now the world’s largest (Holmes 21 A formula-based allocation for fiscal transfers to provinces is being developed with assistance from development partners, which would bring Lao PDR in line with its regional peers (World Bank 2019c). 30 and Lancaster 2019) and local government debt now accounting for nearly half of all government debt (OECD/UCLG 2019). In Vietnam, provinces and centrally managed cities can borrow, though within strict limits and solely for capital investment projects that can generate returns to service debt. Generally, local borrowing remains a small share of total public debt with much of this concentrated within the five largest cities (OECD/UCLG 2019). Elsewhere, local borrowing is generally permitted, though subject to central government approval or strict prudential regulations, and generally small. 4. THE IMPLEMENTATION OF REFORM Equally as important as the structure of intergovernmental fiscal and other relations is how decentralization reforms function in practice and whether they lead to improved performance and outcomes for citizens. This is variously shaped by a range of factors, including the political dynamics driving reform, norms of intergovernmental cooperation, and the configuration of the civil service, which ultimately implements core governance functions. Chapter 4 examines some of these factors and how they influence the trajectory and coherence of the reform process. An Unsynchronized Approach to Decentralization Driven by Political Factors In the communist one-party states of China, Lao PDR, and Vietnam, the deconcentration of spending coincides with tight political control from the central government. In each country, SNGs politically are upwardly accountable to the center. At each level of government, local organs of central state power (known as People’s Congresses in China, Party Committees in Lao PDR, and People’s Committees in Vietnam) sit alongside SNGs. Subnational executives in provinces and subprovincial levels are selected by, and report to, these subnational tiers of the central political apparatus, and they are judged by criteria set out by the central government. Accordingly, while subnational shares of spending are generally high in these countries—reflecting a process of deconcentration based on the principle of subsidiarity—this does not necessarily translate into practical decentralized authority. Indeed, in the case of China, local governments have been characterized as “agents” working on behalf of the central politburo (OECD/UCLD 2016). Similar characterizations can be made for Lao PDR and Vietnam. Specific mechanisms ensure that SNGs in Cambodia, Myanmar, and Thailand remain politically accountable to central governments. In both Cambodia and Thailand, the central government appoints the executive in the first-tier provinces, though the nominally autonomous lower levels remain under the control and supervision of the center. In Cambodia, this control is exercised through the one-party system. In Thailand, central political power flows through the appointment of provincial governors, which have the power to approve annual budget plans and local regulations and dissolve local councils (Unger and Mahakanjana 2016). In Myanmar, the central government exerts considerable political authority over subnational governments through the central appointment of chief ministers and the military’s reserved parliamentary seats. Moreover, the president appoints 31 the mayors in the country’s two largest cities, Mandalay and Yangon, thus diluting the sole devolved responsibility of subnational governments (World Bank 2019a). Concurrent functions across levels of government also create ambiguities over spending assignments and undermine the autonomy of SNGs. In Myanmar, the narrow window of authority given to subnational governments is largely shared with, and in effect subordinate to, the union. Indeed, SNGs are often viewed by union government officials as being located within a hierarchy, with the president at its apex, rather than being autonomous political entities responsible for developing and implementing policies in the interests of their constituents. In politically federal Malaysia, which on the face of it is the most decentralized country in the region (Shair-Rosenfield, Marks, and Hooghe 2014),22 intergovernmental relations are highly centralized and politicized, with the most power over policy and planning residing within the Prime Minister’s Department.23 This compounds the narrow functional mandate of the state and explains why Malaysia has been described as a ”federation without decentralization” (Ostwald 2017). Functional overlaps between multiple tiers of subnational government, and horizontally within the same level, are evident in China, Indonesia, the Philippines, and Vietnam, and lead to losses in efficiency due to misalignments and overlaps and task shifting from higher to lower levels with less capability (OECD/ UCLG 2019). In Lao PDR, provinces are presently viewed under the law as executing agencies of the centralized state budget government budget. The Law on Local Administration (2015) does not clearly articulate the assignment of functions. There is no clear separation of delegated or devolved functions and little consideration so far of scale economies and capacity. Even when functions are clearly assigned, SNGs face considerable difficulties exercising their authority. An enduring challenge is the lack of capacity at lower levels of government, which makes implementation a perennial challenge. Besides, SNGs face unfunded mandates for a variety of reasons. Indonesia is a case in point. Rapidly urbanizing suburban districts face acute challenges resourcing necessary infrastructure due to distortions in the allocation of general and specific transfers (World Bank 2017). In Cambodia and Myanmar, labor costs consume a substantial share of transferred resources and crimp the available fiscal space for development spending at the subnational level (World Bank 2018, 2019a). China’s SNGs also face challenges in funding wide and growing mandates (see box 3). More intentionally, Malaysia has seen a hollowing out of subnational responsibilities and autonomy through a process of organizational duplication. The central government has progressively assumed responsibility for development policy, in particular, through the creation of federally led Malaysia Economic Corridors, which gives the central government 22 This finding is based on the Regional Authority Index, a composite index that estimates the authority of regional governments (see Shair-Rosenfield, Marks, and Hooghe 2014). 23 To put the extent of Malaysia’s concentration into perspective, Ostwald (2019) observes that, in the 2012 budget, the Prime Minister’s Department had a budget 10 times larger than the state budget of economically powerful Selangor and 15 times larger than Penang’s. This has grown proportionally larger in the 10th and 11th Malaysia Plans (2011 to 2020 inclusive). 32 extensive leverage on subnational administration (Ostwald 2017). Some central governments also wield considerable direct financial control over the subnational level. In China, Lao PDR, and Vietnam, subnational planning and budgeting exist within a nested hierarchy. Each level of local government is delegated responsibility to allocate resources within their jurisdiction, though must have budgets approved by the immediately superior level of the party apparatus (World Bank 2019c; OECD/UCLG 2016; Rab et al. 2015). The national legislature ultimately adopts a state budget for the entire country, being the consolidation of the central and local budgets. Myanmar has a similar nested budgetary system, with the central government formally approving subnational budgets, including the budgets through the Financial Commission (Shotton, Yee, and Oo 2016). This vetting allows the center to exert considerable influence over subnational finances and effectively hands it a de facto fiscal veto over subnational expenditures. Continuing Centralized Influence Over Public Administration Most places, central governments also retain a high degree of control over the civil service working at the subnational level. Although subnational levels tend to account for a sizable share of civil servants and total public staff expenditure bill in most countries (table 7), it is common for the central governments to retain a significant degree of control over public administration (table 8). Conventionally, this is in financially sensitive areas such as the size, compensation structure, and wage levels of the civil service. The discretion, whether delegated or devolved, of the subnational governments over the appointment, posting, transfer, promotion, evaluation and training of working at lower levels also varies substantially. This, in turn, substantially shapes the accountability of civil servants and, in turn, the authority and autonomy of local governments. Table 7. Subnational Share of Civil Service Spending Country Percentage Cambodia 0.7 China N/A Indonesia 53.4 Lao PDR 95.1 Malaysia N/A Myanmar ~8.0 Philippines 18.0 Thailand 16.7 Vietnam 89.8 Sources: OECD/UCLG (2019); Lao PDR Ministry of Finance State Budget Implementation Fiscal Year 2015–2016; Myanmar Ministries of Planning and Finance and Labor, Immigration, and Population and Central Statistical Organization. 33 Table 8. SNG Discretion over Civil Service Employment and Management Philippines Cambodia Indonesia Schedule 2a Myanmar Malaysia Thailand Vietnam Lao PDR Type China Cities Establishment control Control over overall staff N S N N N N S S N S numbers Control over staff numbers N S S S N N Y S N S in individual local offices Recruitment Recognized as a formal Y Y N N Y N S S S Y employer Authority to hire S Y S N Y N S Y N Y Independent merit-based N S S N Y N S S N N recruitment mechanism Career management Mandatory training N/A N/A N/A S Y N N Y N/A N/A Promotion N Y S S Y S S Y N N Horizontal inter-SNG N N N S N N Y S N N transfers Intra-SNG transfers, N Y Y S Y N Y S N Y postings Performance management Direct and supervise Y Y S S Y S Y S Y Y activities Conduct performance N Y S Y Y S Y S N N evaluations Offer financial rewards N N S S Y N S S N N Discipline staff N N S S Y N Y S N S Set pay policy N N N N N N N S N N Set local financial incentives, S Y S N S N S S N S salary top-ups Provide nonmonetary perks N/A Y N/A N Y N Y S N/A N/A Note: N = total control of central government; N/A = data not available; S = partial control of central government; SNG = subnational government; Y = complete control of the SNG. a. Limited to subnational departments carrying out subnational-level functions, pursuant to Schedule 2 “Region or State Legislative List” of the Myanmar constitution. 34 Myanmar is an example of extremely centralized control over civil servants. Myanmar’s SNGs have no dedicated permanent bureaucracy and instead rely on the centralized General Administration Department (GAD) for administrative and coordination functions (Nixon et al. 2015). Moreover, despite paying the salaries and other benefits of permanent employees working within subnational departments, subnational governments have no legal authority over civil service management functions (appointment, posting, transfer, promotion, evaluation, and training), which are highly centralized. Performance management is also organized through the verticalized line ministry structures managed at the union level. Most staff are therefore accountable, in one way or another, to their union minister.24 Staff working in city administrations are an exception, being solely accountable to the SNG, which makes city administrations a de facto ministry of SNGs and a useful precedent for entirely handing over civil services to Myanmar's SNGs. At the other end of the spectrum, Indonesia and the Philippines devolve considerable authority over civil servants to subnational governments (Shair-Rosenfield 2016). Each tier of local government has discretion over many of its own human resource management functions, subject to national civil service regulations. This includes the ability to hire, fire, and promote staff. Local governments are also involved in the performance management of staff. The central government retains control over local establishment size, while salary scales and benefit entitlements are also centrally determined. Other countries fall somewhere in between these extremes: • In China and Vietnam, there is some decentralization in administrative authority, though appointment and promotion decisions are still politically influenced (e.g., the People’s Congress or the People’s Committee at the next-highest level of government usually decides to hire or fire local civil servants, which can be based on seniority and political factors). The blurred political and administrative structure of government means that local staff is beholden to both the directives of the central government and the demands of local assemblies. • In Thailand, the central government exercises general authority over personnel decisions at each level of subnational government through the Civil Service Commission. The central government rigidly prescribes the number and grade level of authorized staff positions at each level of local government. • In Cambodia, where local administrative capacity was hollowed out following an extended period of conflict, many civil service functions are controlled from the central government. However, the stated aim of the government is to accompany the transfer of functions to lower levels with increased subnational autonomy over the management of human and financial resources; in particular, taking more of the lead in recruiting, appointing, and disciplining some 24 Strictly speaking, staff working in subnational departments performing functions under Schedule 2 of the constitution are accountable to both a union and subnational minster, while staff working in deconcentrated departments of ministries performing functions under Schedule 1 of the constitution are solely accountable to their respective union minister, though informal consultation with the subnational level is encouraged. See World Bank (2019a) for more information. 35 cadres of administrative staff, according to national guidelines (Cambodia NCDDS 2014). An improved regulatory environment and guidelines for decentralized subnational human resources management have been developed.25 However, implementation has been slow, and as of early 2018, only one of the 17 required guidelines has been completed (World Bank 2018). Malaysia is unique for its cadre of generalist civil servants in the Administrative and Diplomatic Service who are spread throughout government and play a vital policy-making and coordination role.26 With a more rigorous selection process, a focus on continuous learning, and links throughout the strategic and key decision-making positions in the Malaysian civil service, the Administrative and Diplomatic Service (ADS) is considered the premier track for public servants. With a managed rotation system, ADS officers may be transferred from one ministry or agency to another, either horizontally or vertically in cases of promotion. While recruited by the national government, they are also shared between the central government and states. This breadth of experience facilitates coordination (World Bank n.d.). With similar colonial antecedents, the supra-governmental nature of the ADS, staffed with a cadre of generalist administrators, is similar to the Indian Administrative Service (IAS) (Potter 1986). In Myanmar, the GAD—also an artifact of British colonialism—serves a somewhat similar function as a permanent senior bureaucracy as well as being central to grassroots governance and service delivery. A key difference is that subnational ADS and IAS staff are accountable to the state government, whereas in Myanmar, all GAD are ultimately accountable to the central government. An Evolving System of Subnational Governance A feature of decentralized governance in the region is that it is continually evolving within the space provided. Despite the constraints on their authority, SNGs are nonetheless playing key roles in influencing the local political and economic landscape. In Myanmar, for instance, the new SNGs are providing a critical platform for new forms of political contest and helping to promote a more sustainable, inclusive, and accountable system of local governance that reflects Myanmar’s considerable social and economic diversity. This is particularly relevant in ethnic states, given the long-standing challenges of civil wars waged by several minority ethnic groups (World Bank 2019a). Subnational legislatures have also generally grown more assertive and autonomous under National League for Democracy (NLD) leadership at the national level. A signal achievement of the Yangon Region Government was the 2018 Yangon City Development Law, which has profound implications for the way Yangon City Development Committee is elected and managed. In addition to their legislative outputs, subnational parliaments are increasingly holding the executive to account (Batcheler 2018). 25 There are 17 guidelines: (i) human resource planning and personnel projection; (ii) recruitment; (iii) study visit; (iv) official’s appointment into cadre; (v) promotion; (vi) appointment; (vii) training and capacity development; (viii) local and overseas on the-job training; (ix) performance-based and incentives management; (x) shifting workplace (transfer); (x) staffing in a free status without pay; (xi) staffing out of earlier cadre stationed or to be out of position; (xii) leave management; (xiii) imposing punishment; (xiv) legal proceeding through court system; (xv) conflict resolution concerning civil service functions; (xvi) leaving civil service due to the disabilities; and (xvii) official cadre officially removed from the list, death, and retirement. 26 Sometimes better known as Pegawai Tadbir dan Diplomatik or PTD. 36 The subnational level is also proving to be fertile ground for experimentation and innovation. World Bank (2005) observed numerous examples of local innovation occurring at the subnational level, focusing in particular on health innovations in both the Philippines and Vietnam. In Myanmar, in an otherwise staid and static bureaucracy, SNGs are demonstrating a spirit for innovation and experimentation, which is leading to positive deviance. The Mandalay City Development Committee, for example, is making early forays into “Smart City” technology, with several digital innovations, including the adoption of closed-circuit television networks to address public security functions, real-time tracking of garbage trucks, and the use of satellite imagery to improve land use mapping and tax collections. Individual states and regions are implementing proactive feedback mechanisms with users of administrative services to curb petty corruption. Administrations in smaller cities are also taking great strides toward increased use of digital infrastructure to improve the way they deliver services in smaller urban areas, raise revenues, and engage with the public. Further, and more generally, in China, decentralization has helped stimulate local policy experiments and has been central to the Chinese development model in the one-party state for some time, though not without its challenges (box 6). In Indonesia, devolution has given rise to a new cadre of powerful and assertive regional leaders. President Joko Widodo ascended from being a city mayor and then governor of Jakarta. Box 6. “Crossing the River by Feeling the Stones”: Experimentation and Innovation at the Subnational Level in China The quote, attributed to Deng Xiaoping, involved initially piloting important market-oriented reforms at the local level to test their success before scaling up successful reforms nationwide. The approach has been accredited with China’s extraordinary economic performance and continued authoritarian resilience (The Economist 2018). Fundamentally, the approach relies on China’s subnational system of government and, in particular, the considerable latitude given to Chinese cities and special economic zones to govern themselves. Faced with both an opportunity and strong career incentives for policy innovation, individual local officials looking to solve local problems have given rise to many salient examples of “positive deviance” that have led to replicable and scalable policy successes (Teets and Hasmath 2020). Successes have come in a wide variety of areas, including municipal services, rural land reform, health care policy reform, green finance zones, and even experimentation with transparency systems and anti-corruption agencies, to name a few (The Economist 2018; He 2018; Florini, Lai, and Tan 2012). Central governments have aided and abetted experimentation at the local level, comfortable in the knowledge that it absolves central policy makers of possible failures (Heilmann 2008). Empowering local authority also led to problems. Policy competition has variously led to market fragmentation, duplication of production, and local protectionism as regions compete for investment and profits (Li, Qiu, and Qunyun 2003). Moreover, the label of “policy 37 experimentation” has, at times, provided a fig leaf for local officials to pursue personal enrichment or “face innovation” and other image-building projects (Teets and Hurst 2014). Experiments are also becoming rarer under the current leadership, which has reemphasized a system of top-down policy making. The result has been a sharp decline in pilot projects at the subnational level, with the number of provincial-level policy pilots declining from around 500 in 2010 to about 70 in 2016 (Heilmann 2016). On the flip side, highly devolved functional responsibilities can also heighten the costs of fragmentation. The Philippines, which is easily the most decentralized system in the region, provides a salutary example. Considerable political, functional, and fiscal responsibilities are assigned to the lowest level barangays; potentially beyond their competence in some instances. This has led to highly fragmented governance and costly duplication of common functions (Punongbayan 2018).27 More problematically, the immense resources invested at local levels have also created new opportunities for local elite capture and patronage systems to become entrenched, which have constrained improvements in service delivery and growth (Shair-Rosenfield 2016). Fragmentation can also erode efficiencies. China and Vietnam both go so far as to decentralize their safety nets—pension and unemployment insurance in China and social health insurance in Vietnam. With pools being managed, and effectively separate at the local level, many of the advantages of scale economies and risk pooling are weakened (Escolano et al. 2015; Somanathan et al. 2014). Ongoing recalibrations and improvements and are also being made to intergovernmental fiscal relations in countries with more mature systems of subnational governance. • In Vietnam, the publication of updated stability period rules and regulations (2016–2020), which determine fiscal relationships between provinces and the districts and communes within their jurisdiction, have provided greater clarity over expenditure assignments in health and education between the province and between subprovincial authorities (Rab et al. 2015). • In China, the State Council has initiated further intergovernmental fiscal reform. The reforms, which are expected to culminate in a new Intergovernmental Fiscal Relations Act in 2020, aim to improve the clarity, coherence, and predictability of expenditure assignments (particularly focusing on services where there is presently a joint mandate); increase the fiscal resources of less- developed regions; and improve the supervision of local government (Wingender 2018). • Indonesia’s first wave of decentralization devolved considerable political, administrative, and financial authority to lower levels, bypassing the provinces. The latest wave of reforms in 2014 recentralized powers from districts to provinces and sought to contain subnational administrative expenditures through earmarked allocations for wages in DAU grants. These reforms are perceived 27 A salient example is that each barangay procures and installs its own unique street light design. 38 as correcting excesses from the initial decentralization reforms while preserving their overall intent (Ostwald, Tajima, and Samphantharak 2016). Two new village-level transfers were implemented (Dana Desa and Alokasi Dana Desa, channeled from the central and district governments, respectively) aiming to reduce poverty and improve living standards. This represents a considerable increase in resources for the lowest level. However, inequalities persist given the large weight placed on equal allocations between villages. Progress is not consistent, however, with administrative weaknesses thwarting the desired pace of reforms in Cambodia and Lao PDR. In Cambodia, a strategic goal of the government’s National Program for Subnational Democratic Development is decentralizing the delivery of core services from line ministries and provinces to districts and municipalities. Subnational authorities have been assigned a general mandate for the welfare of citizens in their constituencies though progress in transferring statutory functions to SNGs has been slow. A strong top-down push from the highest levels of government has helped accelerate the reform process in recent years. In response, 21 functions across five line ministries were transferred to the provincial level in 2019. Additional steps are also being taken to integrate all the 13 separate line offices, along with their 55 functions and nearly 20,000 staff, into the district level. However, implementation continues to be impeded by a lack of clear and consistent strategic guidance to key stakeholders regarding decentralization reforms. In Lao PDR, 15 line ministries were involved in a pilot to delegate responsibilities from the central to provincial- and district-level agencies.28 Some improvements were reported in routine procedures and service delivery and increased revenue collection due to the clearer delegation of authority. However, the overall implementation of decentralizing reforms has been slow, owing to a lack of subnational public administration capacity and wide-ranging complementary reforms to public administration and delays in reassigning budgetary authority to match changed functional assignment (LALDF 2015). In the Philippines, the formal legal framework allowed local governments to challenge the central government over transfers. In the 2019 “Mandanas ruling,” the Philippines Supreme Court determined that the calculation of IRAs should include customs revenues, which have previously been excluded. The result, which amounts to a court-driven acceleration in fiscal decentralization, is estimated to increase transfers from the central government to local government units (LGUs) by around 30 percent. Malaysia and Thailand provide exceptions to the advancement of decentralization in the region. Against a backdrop of general opposition to decentralization from the powerful Ministry of the Interior, the previous Thaksin government of Thailand enhanced the powers of the unelected provincial governors and placed considerable political, financial, and administrative authority in the prime minister’s office. The more recent military coup has since further altered the character of local governance by recentralizing political power (Lowatcharin 2014). The military government has also attempted to implement nationwide compulsory municipal consolidation. 28 The pilot devolution strategy was implemented and tested nationwide between 2012 and 2015 in 109 targeted villages (out of 8,470) located in 51 targeted districts (out of 145) (World Bank 2019c). 39 In Malaysia, the dominant feature of center-state relations—at least since the 1980s under Barisan Nasional (BN)—has been a concentration of power, both the personalization of power in the Prime Minister’s Department as well as the hollowing out of state government autonomy (Ostwald 2017). SNG discretion over human resource management and efforts made to open up negotiations around revenue-sharing arrangements between the federal and state governments indicate that Malaysia’s federalist system continues to function. In addition, the incoming Pakatan Harapan government in 2018 pledged in its manifesto to “revive the true spirit of federalism” including the transference of more functions and finance to the states and elections at the local level. While some small changes were announced, little progress was made on deep institutional restructuring, especially those involving the concentration of power (Ostwald 2019).29 Moreover, with political power shifting back to a coalition involving the BN, it is not clear whether the momentum for change will continue. A further notable feature of the recent evolution of reforms is that attempts at significant change have generally not been successful. In the Philippines, a campaign priority of President Duterte was to replace the country’s unitary model of governance with a federal one. In part, this centers on providing individual states with more autonomy and weakening economic and political dynasties. It is also linked to efforts to bring peace to Mindanao. A new proposed constitution has been drafted by a Consultative Committee that divides the country into 18 regions with their own regional assemblies and executive agencies (Heydarian 2018). However, the constitutional reform process has stalled. In Myanmar, an emerging dialogue on constitutional change, driven by the ruling NLD, ultimately came to naught. Fundamental reforms were vetoed by the military. The suggestion by the opposition to eliminate the discretion of the central government in appointing state and region chief ministers—a change likely to both improve representative governance at the local level and favor the smaller, regional parties—was not agreed to by the ruling NLD. Why the NLD opposed this change is unclear. It could be partly due to their lukewarm interest in devolving power, the absence of quid pro quo concessions by the opposition and the military, or simply bad timing. It nonetheless places into stark relief the fact that “big bang” constitutional reform is off the table, and future reform opportunities are likely to occur within the constitution itself (World Bank 2019a). An Underperforming System of Conditional Transfers In a region where there is limited formal expression of political autonomy at the subnational level, conditional grants potentially offer central governments a practical and politically feasible mechanism for transferring responsibilities. While the stated preference of central governments across the regions is the transference of administrative and fiscal responsibilities to lower levels, delays in formalizing transfers and enduring overlaps also reveal concerns about how money is used at lower levels. Sitting somewhere between outright devolution of functions and centralized control, national governments can use conditional allocations for one (or both) of the following purposes: (i) to fund subnational governments (in order to implement their functions/national policy objectives and/or 29 For example, water management has seen significant decentralization to Malaysia’s states, and half of the tourism tax collected is to be distributed back to the states. 40 compensate for asymmetric costs and externalities; and (ii) to incentivize subnational governments to improve certain aspects of their performance (Broadway and Shah 2007; Spahn 2012). The conditional nature of the transfer ensures that the central level retains some degree of influence. This then places emphasis on ensuring that they are well designed and fit for purpose. As noted in chapter 3, conditional transfers are a feature of most intergovernmental fiscal frameworks in Southeast Asia. To a large extent though, conditionality has been used as a tool of input control—influencing what SNGs spend money on—rather than improving performance. China’s widespread use of conditional transfers has been partly a function of weaknesses in monitoring at the local levels, which has undermined the national government’s trust in SNGs, and in turn, led to the national government channeling transfers in the form of narrowly earmarked programs that limit the discretion of subnational governments (World Bank and DRC 2014). In Indonesia, the scale and the scope of the conditional DAK have considerably expanded since it was initially introduced to fund reforestation. It now covers 19 priority sectors. Given that any local government activity can be classified as a potential priority, it has been an approach favored by central ministries reluctant to accept the devolution of power and authority (Fadliya and McLeod 2010). Thailand’s system of earmarked “specific” grants is attractive to the central administration because control remains firmly entrenched at the central level, allowing for central prioritization of different policy objectives (Varanyuwatana 2017). Moreover, in several cases, the implementation of conditional transfers has been unsystematic, discretionary, and opaque, which has weakened their overall effectiveness in controlling inputs and has impeded their effectiveness in encouraging performance and accountability. For example: • China’s conditional grants are allocated on an ad hoc or negotiated basis, determined by individual ministries and not the finance ministry. Moreover, the system is likely to have imposed considerable efficiency costs, being highly administratively burdensome on the central government, which often lacks the monitoring mechanisms to ensure the effective use of these earmarked resources (Wang and Herd 2013). • In Thailand, rules determining the allocation of grants have varied from year to year and from one regime to another, thus making it difficult to describe or predict a pattern of specific grant allocation (Chachavalpongpun 2019). • In Indonesia, a recent World Bank review suggests that the DAK funding flow has become increasingly fragmented across sectors, reducing its focus and prioritization and undermining its responsiveness to needs. The introduction of a proposal-based approach, while on the face of it gives districts a larger say role in planning, also seems to have increased administrative costs without improving implementation (World Bank 2017). Moreover, the allocation of DAK is yet to be linked to performance, meaning that the results orientation remains weak. • While in Vietnam, a general lack of predictability in targeted transfers plus complex, input-based 41 guidance on implementation reduce their effectiveness and increase the administrative burden on SNGs (Rab et al. 2015). As a counterpoint, there are examples of transfers becoming more focused on improving SNG performance; though the primary focus is on improving institutional processes rather than outputs or outcomes (UNDP 2019). The Philippines has strengthened the performance framework ensuring responsive governance among SNGs. Through its Seal of Good Local Governance (SGLG), introduced in 2014, the Philippines aims to improve the transparency and accountability in the use of public funds at local levels. The Department of Interior and Local Government, which has oversight over all LGUs, manages the performance-based system. All LGUs are assessed annually against six sets of criteria (i.e., disaster preparedness, social protection, business friendliness, peace and order, environmental management, and local revenue-raising efforts). Top-ranked LGUs are awarded a financial prize (UNDP 2019). Cambodia already has some experience with conditionality, particularly within the health sector, originally involving performance-based financing within the deconcentrated tiers of the centrally managed Ministry of Health and more recently with conditional transfers between levels of government following the assignment of key health service delivery functions to the subnational level (box 7). Box 7. Conditional Financing in Cambodia’s Health System Performance-based financing has been an important part of financing the supply side of Cambodia’s health system for some time. Originating under the Second Health Sector Support Program 2008–15 (HSSP-2) in 2009, it involves Service Delivery Grants (SDGs), which are an internal contracting between different tiers in the health system to define performance indicators and establish monitoring and accountability. The Ministry of Health signs performance contracts with the Provincial Health Department, which in turn signs a service delivery agreement with the respective Special Operating Agencies (SOAs)—either the Operational District or the Provincial Referral Hospital. Health facilities, in some cases, also sign performance agreements with the health staff. SDGs represent additional revenues to SOAs and can account for up to 40 percent of the total SOA funding. Contracted districts and hospitals are incentivized to improve performance by linking payments to outcomes. They are paid 80 percent upfront, of which 60 percent can be used for staff bonuses, with the remaining 20 percent paid on demonstrated performance of services delivered. Owing to the success of the SDG mechanism, it has been expanded nationally as part of the donor-supported Health Equity and Quality Improvement Project. In 2019, the Royal Government of Cambodia also transferred the Provincial Health Department functions to the capital/provincial (CP) subnational administrations accompanied by an earmarked fiscal transfer. However, the transfer of functions and finance are both somewhat hybrid arrangements. CP administrations are accountable to the Minister of Health for 42 following national health policies and guidelines; while the budgetary processes for the transferred funds must follow the budget circular for the national level, not the one designated for the subnational level. In other words, even after the transfer to the CP level, the funds are still considered part of the national budget. Matching is also a useful way to yield the benefits of improved accountability without overly burdening administrative capacity. Requiring subnational governments to match (or cost share) the central government spending on policy priorities can help ameliorate the possible displacement of subnational funding (Spahn 2012). It also requires a relatively smaller administrative investment in monitoring. Essential to the success of matching are conditions that the matching grants (i) do not include expenditures that subnational governments would intend to finance anyway; and (ii) do not encourage subnational governments to draw budgetary resources away from other priority areas. Also important is that matching programs account for the possible inequities in the greater capacity of wealthier jurisdictions to raise matching funds. As part of the SGLG In the Philippines, a Performance Challenge Fund provides matching funds for “high quality and priority” projects submitted by LGUs to encourage more strategic investments. Many of China’s conditional grants are provided on a matching basis. Matching conditions are also included in Indonesia’s DAK, which provides grants to selected SNGs to promote capital investment spending in several sectors on a matching ratio of 10:1; though it has exceeded expectations, mobilizing even an additional 12 rupiahs of SNG revenues for every 10 rupiah of transfers (Lewis and Smoke 2017). 5. KEY LESSONS AND PATHWAYS FOR REFORM A key lesson from the Southeast Asian experience is that decentralization evolves in different ways: it does not necessarily require federalism. In a highly heterogeneous region, with a wide range of political and economic systems and a preponderance of unitary states, a common element is the transference of political, fiscal, and administrative authority to lower levels of government. To be sure, there are considerable variations between countries, with each taking a path, shaped by its own history, size, and political system, which is reflected in the various differences in the extent of political decentralization and decentralization across the 3Fs (table 9). A common theme, however, is that reforms have been driven by the allure of decentralization’s purported benefits—increased political pluralism, responsiveness to local preferences, as well as an improved capacity for experimentation and innovation in governance. 43 Table 9. Summary of Decentralization in Southeast Asia across Politics and the 3Fs Political Decentralization Decentralization Decentralization Country Decentralization of Functions of Funds of Functionaries Cambodia Low Moderate Very low Low China None High Very high Moderate Indonesia Moderate High High High Lao PDR None Moderate Moderate Moderate Malaysia High Low Very low High Myanmar Low/ moderate Very Low Low Very low Philippines Moderate High Low/moderate High Thailand Low Moderate Moderate Low Vietnam None High High Moderate Source: Drawn from the information in Tables 2, 3, 4, and 8. Reforms have been a largely political process, driven from the top and largely against a backdrop of peace. There are examples (for instance, in Indonesia and the Philippines) of central governments using decentralization as a tool to relieve emerging tensions between ethnic and cultural groups within heterogeneous societies. Decentralization has also been a key part of transitions from authoritarian rule and embedded within it. In the communist countries of China, Lao PDR, and Vietnam, central governments have viewed decentralization as essential to their development strategy. In the main, however, central governments have carefully choreographed reforms, ensuring that social stability (and often the power of the central government) was not threatened. The strong political calculus behind decentralization is likely a key reason why reforms have not had the impact they might otherwise have had if efficiency and equity were given greater weightage. Decentralizing reforms in the region have also not always met lofty expectations, with persistent inequality between regions, strong centripetal biases in governance, and a range of political, administrative, and financial mechanisms constraining the autonomy of subnational governments (Malesky and Hutchinson 2017). Indeed, studies in other contexts have found that when decentralization is driven from above, it leads to lower economic growth than when reforms are driven from bottom—that is, driven by regional demands (Rodríguez-Pose, Tijmstra, and Bwire 2009). This potentially reflects differences in the perceived legitimacy of SNGs and thus the likelihood of unfunded mandates and continued central control over subnational expenditure capacity. Moreover, the political nature of reforms has meant that the transference of responsibilities between levels of government is rarely a neat process. This is reflected in the general 44 lack of synchronization between the allocation of functions, finances, and functionaries between different layers of government. It does not necessarily follow, though, that the SNGs in Southeast Asia are not important. SNGs are playing a key role in shaping the political and economic landscape of Southeast Asia. Decentralization has increased political pluralism in a culturally and ethnically heterogeneous region where power tends to be otherwise centralized and tightly held. In some places (though notably not all), subnational governments are providing explicit space for local political representation through the ballot box. By way of their growing expenditures, SNGs are increasing in importance in service delivery and public investment. Despite the general messiness of the reform process, and the constraints on SNGs, the general trajectory appears to be toward more, not less, decentralization in Southeast Asia. Decentralization has broadened to new countries in the region and is generally deepening within countries. This finding is evident in both the shifting weight of public spending toward the subnational level and the steps that have been taken to empower and improve the performance of local governments. In large part, this is happening organically and with the direct involvement of central governments, or at least their acquiescence. The overriding challenge is to make decentralization work better in the region to promote a more sustainable, inclusive, and accountable system of governance and fiscal management that is appropriate for each country’s level of social and economic heterogeneity. As systems of decentralized governance mature further, several important policy questions will need to be settled. In Indonesia and the Philippines, for instance, these may center on what level of government authority should lie. Cambodia and Thailand will continue to wrestle with what decentralization can achieve under an authoritarian central government and in a single- party state, respectively. While doubtless in China, Lao PDR, and Vietnam, the intergovernmental fiscal and administrative relationship will continue to evolve, albeit from within the confines of a politically socialist system. In Myanmar, as the nascent subnational level finds its voice, it could help in achieving the elusive dream of a more federal and democratic system of government—all from within the boundaries of the current constitution. If history is any guide, reforms are not likely to be the result of “big bang” approaches, but rather time-consuming processes of incremental, and sometimes hesitant, change. Moreover, as decentralization gathers pace, coordination challenges will continue to grow. Effective coordination and oversight are enduring challenges in decentralized systems. The further decentralization reforms go, the bigger this challenge is likely to be. This is likely to be most acute in the democratic or democratizing countries, such as Malaysia, Myanmar, and the Philippines, where some political authority is also devolved. Indeed, the Philippines suffers from substantial coordination challenges. Communist countries have the party hierarchy at national, provincial, county, and unit levels to keep everyone moving in the same direction. In a somewhat similar vein, Cambodia mostly functions as a one-party state, and Thailand has the central oversight of the military. 45 Effective coordination requires clear lines of sight between planning, resourcing, service delivery, and performance. Information flows are essential to this process; being critical for identifying local service delivery needs, for appropriately calibrating planning and budgeting, and for ensuring that the quality and performance of the last mile of service delivery is in line with expectations. Without timely, actionable feedback, the incentive to perform erodes, blame-shifting prospers, and available resources are wasted. A cohesive civil service also has an important role to play in facilitating effective coordination. In Malaysia, the federal level has an elite national cadre of Administrative and Diplomatic Service to coordinate across the country and ministries. Myanmar has a proto version of a separate senior public service in the GAD. The bureaucratized structure of public service and the aggregation of particular technical skills within specific agencies means that narrowly defined shared identities within the government are hard to avoid. Nonetheless, governments can draw on some of the salient features of these elite cadres; for example, taking steps to foster a greater esprit de corps within the civil service by socializing a common ethic and a cohesive culture across the whole of government. Also, shared training of lower-level cadres and even job rotations can support knowledge sharing and foster a common sense of identity throughout the civil service. Increased conditionality on intergovernmental transfers may also be an important way to balance autonomy and accountability, but there are pre-conditions for their success. The World Bank (2005) suggests that transfers may be necessary to make fiscal decentralization politically viable. While conditional transfers undermine local discretion, and thus a local government’s capacity to respond to local needs, they offer a mechanism for central governments to channel funding to lower levels in a way that ensures funds are spent in accordance with sectoral parameters. They offer a way to allay the concerns of central governments while still capturing the benefits of decentralization in improving service delivery and responding to local needs. Paradoxically, increased conditionality in the intergovernmental transfer system may help unlock efficiencies by paving a path for the clearer transfer of functions and better alignment between the 3Fs. Performance-based grants and matching grants have the potential to make the fiscal transfer system more accountable and more efficient. At present, conditional grants are fragmented, impose high bureaucratic and other costs, and may well, on balance, be inhibiting subnational performance. With careful design and a stronger focus on national goals and performance instead of process, such grants could leverage off stronger oversight and coordination capacity. They could be used to strengthen the institutional capacity of SNGs and their performance, while at the same time limiting bureaucratic burdens and better aligning local authority and accountability. Ultimately, the experience of countries in Southeast Asia shows that the path to decentralized governance is not a linear and technical but is rather incremental and sometimes hesitant in nature. The primacy of politics driving reform means that countries should not expect to follow a technical roadmap for decentralizing, nor are there likely to be international best practices that can be easily emulated. Instead, experience shows that the reform process is 46 likely to be messy in practice, with unexpected roadblocks, detours, and even backsliding along the way. China, for instance, has variously embraced centrifugal and centripetal forces as the need arose. Indonesia and the Philippines, the region’s “fast starter” countries, found that the outcomes were not sustainable and made further refinements over time to arrive at a more stable political equilibrium. To that end, they have become became more “incrementalist” in nature over time. Cambodia has required a top-down push from the highest levels of government to overcome institutional inertia and has driven noticeable recent change, though overall the reform process remains inherently incremental. Despite their various challenges, the mostly unitary states of Southeast Asia are finding ways to muddle on through and are generally deepening decentralization. They are taking meaningful steps to improve the quality and coherence of subnational governance and now lead the world in terms of the share of public resources that are expended at lower levels. Variously, the results are being manifest in greater political pluralism, increased innovation, and improved quality and equity of service delivery. REFERENCES ADB (Asian Development Bank). 2018. Fiscal Decentralization Reform in Cambodia: Progress over the Past Decade and Opportunities. Manila: ADB. https://www.adb.org/sites/default/files/ publication/479961/fiscal-decentralization-reform-cambodia.pdf. Batcheler, Richard. 2018. State and Region Governments in Myanmar. Yangon: The Asia Foundation. Bauer, Andrew, Uyanga Gankhuyag, David Manley, Sofi Halling, and Varsha Venugopal. 2016. Natural Resource Revenue Sharing. New York: Natural Resource and Governance Institute. http://www.resourcegovernance.org/analysis-tools/publications/natural-resource-revenue- sharing. Berman, Evan M., ed. 2011. Public Administration in Southeast Asia: Thailand, Philippines, Malaysia, Hong Kong, and Macau. New York and Oxford: Routledge. Bird, R.M. 2006. “Local and Regional Revenues: Realities and Prospects.” In Perspectives on Fiscal Federalism, edited by R.M. Bird and F. Vaillancourt, 177–196. Washington, DC: World Bank. Broadway, Robin, and Anwar Shah. 2007. Intergovernmental Fiscal Transfers: Principles and Practice. Public Sector Governance and Accountability Series. Washington, DC: World Bank. Cambodia NCDDS (National Committee for Sub-National Democratic Development). 2014. 3-Year Implementation Plan, Phase II (2015–17) of the National Program for Sub-National Democratic Development (IP3-II). Phnom Penh: Royal Government of Cambodia. 47 Canare, T. 2016. “The Relationship between IRA and Local Government Expenditures: Evidence from a Cross-Section of Philippine Cities.” Philippine Political Science Journal 37 (3): 167–189. Chachavalpongpun, P. 2019. Routledge Handbook of Contemporary Thailand. New York: Routledge. Cook, Malcolm. 2018. “Three Challenges Facing the Bangsamoro Organic Law.” ISEAS Perspective 2018 (82). https://www.iseas.eduCoz.sg/images/pdf/ISEAS_Perspective_2018_82@50.pdf. Dufhues, T., I. Theesfeld, and G. Buchenrieder. 2015. “The Political Economy of Decentralization in Thailand: How Past and Present Decentralization Affects Rural Actors’ Participation.” European Journal of Development Research 27 (5): 793–810. Eaton, Kent, Kai Kaiser, and Paul J. Smoke. 2011. The Political Economy of Decentralization Reforms: Implications for Aid Effectiveness. Washington, DC: World Bank. http://documents.worldbank. org/curated/en/232401468155133502/The-political-economy-of-decentralization-reforms- implications-for-aid-effectiveness. The Economist. 2018. “Local Experiments with Reform Are Becoming Rarer under Xi Jinping.” The Economist, August 18, 2018. https://www.economist.com/china/2018/08/18/local-experiments- with-reform-are-becoming-rarer-under-xi-jinping. Eng, N., and S. Ear. 2016. “Decentralization Reforms in Cambodia.” Journal of Southeast Asian Economies 33 (2): 209–223. Escolano, J., D. Benedek, H. Jin, C. Mulas Granados, M. Nozaki, J. Pereira, G. Rota-Graziosi, L. Sinn, and J. Torres. 2015. “Distribution of Fiscal Responsibilities in Federations.” In Designing a European Fiscal Union: Lessons from the Experience of Fiscal Federations, edited by C. Cottarelli and M. Guerguil, 13–59. New York and Oxford: Routledge. Fadliya, and Ross H. McLeod. 2010. “Fiscal Transfers to Regional Governments in Indonesia.” Working Papers in Trade and Development No. 2010/14, Australian National University, Canberra, Australia. Florini, A., H. Lai, and Y. Tan. 2012. China Experiments: From Local Innovations to National Reform. Washington, DC: Brookings Institution Press. Haque, M.S. 2010. “Decentralizing Local Governance in Thailand: Contemporary Trends and Challenges.” International Journal of Public Administration 33 (12–13): 673–688. He, Alex Jingwei. 2018. “Manoeuvring within a Fragmented Bureaucracy: Policy Entrepreneurship in China’s Local Healthcare Reform.” The China Quarterly 236: 1088–1110. Heilmann, S. 2008. “Policy Experimentation in China’s Economic Rise.” Studies in Comparative International Development 43: 1–26. 48 Heilmann, S. 2016. “Leninism Upgraded: Xi Jinping’s Authoritarian Innovations.” China Economic Quarterly 20 (4): 5–22. Heydarian, Richard. 2018. "The Implications of Duterte’s Proposed Constitutional Changes.” Asia Unbound (blog), July 24, 2018. https://www.cfr.org/blog/implications-dutertes-proposed- constitutional-changes. Holmes, Alex, and David Lancaster. 2019. “China's Local Government Bond Market.” RBA Bulletin, June. https://www.rba.gov.au/publications/bulletin/2019/jun/chinas-local-government-bond- market.html. Huang, Bin, Mengmeng Gao, Caiqun Xu, and Yu Zhu. 2017. “The Impact of Province-Managing- County Fiscal Reform on Primary Education in China.” China Economic Review 45 (C): 45–61. Jun, Zhang. 2018. “China’s Governance Tightrope.” Project Syndicate, October 1, 2018. https://www. project-syndicate.org/commentary/china-governance-reform-growth-stability-by-zhang-jun- 2018-10?barrier=accesspaylog. LADLF (Laos-Australia Development Learning Facility). 2015. Sam Sang in Practice: Early Lessons from Pilot Implementation. Canberra, Australia: Australian Aid. Lewis, B. D., and P. Smoke. 2017. “Intergovernmental Fiscal Transfers and Local Incentives and Responses: The Case of Indonesia.” Fiscal Studies 38 (1): 111–139. Li, Jie, Larry D. Qiu, and S. Qunyan. 2003. “Interregional Protection: Implications of Fiscal Decentralization and Trade Liberalization.” China Economic Review 14 (3): 227–245. Lin, Justin Yifu, Ran Tao, and Mingxing Liu. 2003. “Decentralization, Deregulation and Economic Transition in China.” Paper prepared for the conference, “The Rise of Local Governments in Developing Countries,” London School of Economics, May. Lowatcharin, Grichawat. 2014. “Along Came the Junta: The Evolution and Stagnation of Thailand’s Local Governance.” Kyoto Review of Southeast Asia 16. https://kyotoreview.org/yav/along- came-the-junta-the-evolution-and-stagnation-of-thailands-local-governance/. Malesky, E.J., and F.E. Hutchinson. 2017. “Varieties of Disappointment: Why Has Decentralization Not Delivered on Its Promises in Southeast Asia?” Journal of Southeast Asian Economies 33 (2): 125–38. Martinez-Vazquez, Jorge. 2011 Fiscal Decentralization in Asia Challenges and Opportunities. Mandaluyong City, Philippines: Asian Development Bank. https://www.adb.org/sites/default/ files/publication/28805/fiscal-decentralization.pdf. 49 Morgan, P.J., and L.Q. Trinh. 2016. ”Fiscal Decentralization and Local Budget Deficits in Viet Nam: An Empirical Analysis.” ADBI Working Paper 613, Asian Development Bank Institute, Tokyo. https:// www.adb.org/publications/fiscal-decentralization-local-budgetdeficits-viet-nam. NBS (National Bureau of Statistics of China). 2016. China Statistical Yearbook 2016. Beijing: China Statistics Press. Nixon, Harnish, Cindy Joelene, Kyi Pyar Chit Saw, Thet Aung Lynn, and Matthew Arnold. 2015. “State and Region Governments in Myanmar.” Yangon: The Asia Foundation. https:// asiafoundation.org/publication/state-and-region-governments-in-myanmar-report/. NRGI (Natural Resource Governance Institute). 2018. “Subnational Revenue Distribution: Natural Resource Revenues in a Decentralized Context.” NRGI Primer, May 2018, NRGI, New York, NY. OECD (Organisation for Economic Co-operation and Development). 2015. OECD Urban Policy Reviews: China 2015. Paris: OECD Publishing. http://dx.doi.org/10.1787/9789264230040-en. OECD (Organisation for Economic Co-operation and Development). 2019. OECD Economic Surveys: China. Paris: OECD Publishing. https://www.oecd.org/economy/surveys/china-2019-OECD- economic-survey-overview.pdf. OECD/UCLG (Organisation for Economic Co-operation and Development/United Cities and Local Governments). 2016. Subnational Governments Around the World: Structure and Finance. Paris: OECD Publishing. https://www.uclg.org/sites/default/files/global_observatory_on_local_ finance_0.pdf. OECD/UCLG (Organisation for Economic Co-operation and Development/United Cities and Local Governments). 2019. 2019 Report of the World Observatory on Subnational Government: Finance and Investment–Country Profiles. Paris: OECD Publishing. Ostwald, K. 2017. “Federalism without Decentralization: Power Consolidation in Malaysia.” Journal of Southeast Asian Economies 34 (3): 488–506. Ostwald, K. 2019. “Power Distribution and Decentralisation in New Malaysia.” ISEAS Perspective (2019): 66. https://static1.squarespace.com/ static/51d99eede4b0a0494bc4a96b/t/5d6aa27b0e4ae800018d790f/1567269499714/ Ostwald+%282019%29+Decentralisation+in+Malaysia+%28ISEAS%29.pdf. Ostwald, K., Y. Tajima, and K. Samphantharak. 2016. “Indonesia's Decentralization Experiment Motivations, Successes, and Unintended Consequences” Journal of Southeast Asian Economies 33 (2): 139–56. Potter, David C. 1986. India's Political Administrators 1919–1983. Oxford: Clarendon Press. 50 Punongbayan, B.R. 2018. “The Barangay System.” Business World, August 16, 2018. https://www. bworldonline.com/the-barangay-system/. Rab, H., J. Martinez-Vasquez, A. M. Shah, Q.H. Vu, M.V. Nguyen, K.F.N. Monkam, A. Prasad, Q.H. Doan, and I. Iyer. 2015. Making the Whole Greater than the Sum of the Parts: Fiscal Decentralization Review in Vietnam. Washington, DC: World Bank. RGC (Royal Government of Cambodia). 2013. “Rectangular Strategy for Growth, Employment, Equity and Efficiency, Phase II,” Royal Government of Cambodia of the Fifth Legislature of the National Assembly, Phnom Penh, September 2013. https://www.ilo.org/wcmsp5/groups/public/---asia/--- ro-bangkok/---sro-bangkok/documents/genericdocument/wcms_112939.pdf. Rodríguez -Pose, A., S.A.R. Tijmstra, and A. Bwire. 2009. “Fiscal Decentralisation, Efficiency, and Growth.” Environment and Planning A: Economy and Space 41 (9): 2041–2062. Shair-Rosenfield, S. 2016. “The Causes and Effects of the Local Government Code in the Philippines: Locked in a Status Quo of Weakly Decentralized Authority?” Journal of Southeast Asian Economies 33 (2): 157–71. Shair-Rosenfield, S. Marks, G. and Hooghe, L. 2014. “A Comparative Measure of Decentralization for Southeast Asia.” Journal of East Asian Studies 14: 85–107. Shen, C., J. Jin, and H-F Zou. 2012. “Fiscal Decentralization in China: History, Impact, Challenges and Next Steps.” Annals of Economics and Finance 13 (1): 1–51. Shotton, Roger, Zin Wint Yee, and Khin Pwint Oo. 2016. State and Region Financing, Planning and Budgeting in Myanmar: What Are the Procedures and What Are the Outcomes? Yangon: Renaissance Institute and the Asia Foundation. Smoke, P., and J. Morrison. 2008. “Decentralization in Cambodia: Consolidating Central Power or Building Accountability from Below?” GSU Paper 0836, International Center for Public Policy Working Paper Series, Andrew Young School of Policy Studies, Georgia State University, Atlanta, Georgia. Somanathan, Aparnaa, Ajay Tandon, Huong Lan Dao, Kari L. Hurt, and Hernan L. Fuenzalida-Puelma. 2014. Moving toward Universal Coverage of Social Health Insurance in Vietnam: Assessment and Options. Directions in Development. Washington, DC: World Bank. http://documents. worldbank.org/curated/en/383151468138892428/Moving-toward-universal-coverage-of-social- health-insurance-in-Vietnam-assessment-and-options. Spahn, P.B. 2012. “Conditioning Intergovernmental Transfers and Modes of Interagency Cooperation for Greater Effectiveness of Multilevel Government in OECD Countries.” Paper presented at the Organisation for Economic Co-operation and Development (OECD) workshop on “Effective Public Investment at Sub-National Level in Times of Fiscal Constraints: Meeting the Co- ordination and Capacity Challenges,” Paris, June 21. 51 Su, F., R. Tao, and D.L. Yang. 2018. “Rethinking the Institutional Foundations of China’s Hyper Growth: Official Incentives, Institutional Constraints, and Local Developmentalism.” In The Oxford Handbook on the Politics of Development, edited by C. Lancaster and N. V. D. Walle, 626–651. Oxford: Oxford University Press. https://www.oxfordhandbooks.com/view/10.1093/ oxfordhb/9780199845156.001.0001/oxfordhb-9780199845156-e-8. Teets, J.C., and R. Hasmath. 2020. “The Evolution of Policy Experimentation in China.” Journal of Asian Public Policy 13 (1): 49–59. Teets, J.C., and W. Hurst, eds. 2014. Local Governance Innovation In China: Experimentation, Diffusion, and Defiance. New York and Oxford: Routledge. UNDP (United Nations Development Programme). 2019. Fiscal Transfers in Asia: Challenges and Opportunities for Financing Sustainable Development at the Local Level. Bangkok: UNDP. https://www.undp.org/content/dam/undp/library/prosperity/inclusive-growth/RBAP-2019-Fiscal- Transfer-in-Asia.pdf. Unger, D., and C. Mahakanjana. 2016. “Decentralization in Thailand.” Journal of Southeast Asian Economies 33 (2): 172–187. Varanyuwatana, S. 2017. “Fiscal Decentralization and Intergovernmental Transfer in Thailand.” Thammasat Review of Economic and Social Policy 3 (2). Wang, X., and R. Herd. 2013. “The System of Revenue Sharing and Fiscal Transfers in China.” OECD Economics Department Working Papers No. 1030, OECD Publishing, Paris. https://doi. org/10.1787/5k4bwnwtmx0r-en. Wilkins, K., and A. Zurawski. 2014. “Infrastructure Investment in China.” RBA Bulletin, June Quarter 2014: 27–35. https://www.rba.gov.au/publications/bulletin/2014/jun/4.html. Wingender, P. 2018. “Intergovernmental Fiscal Reform in China.” IMF Working Paper WP/18/88, International Monetary Fund, Washington, DC. Wolff, S., S. Ross, and A. Wee. 2017. “Subnational Governance and Conflict.” Background Paper for UN-WB Flagship Study, “Pathways for Peace: Inclusive Approaches to Preventing Violent Conflict,” World Bank, Washington, DC. World Bank. 2005. East Asia Decentralizes: Making Local Government Work. Washington, DC: World Bank. http://documents1.worldbank.org/curated/en/305911468032971991/ pdf/344010PAPER0Ea101official0use0only1.pdf. World Bank 2017. “Improving Quality of Spending In Indonesia—2018 Budget and Beyond.” PER Update Phase 2, May, World Bank, Washington, DC. 52 World Bank. 2018. Cambodia’s Cross-Cutting Reforms—Public Financial Management, Decentralization, and Public Administration Reforms: Achievements, Coordination, Challenges, and Next Steps. Washington, DC: World Bank. World Bank. 2019a. Subnational Public Expenditure Review 2019: Fostering Decentralization in Myanmar. Washington, DC: World Bank Group. http://documents.worldbank.org/curated/ en/400111577821333322/Subnational-Public-Expenditure-Review-2019-Fostering- Decentralization-in-Myanmar. World Bank. 2019b. World Bank East Asia and Pacific Economic Update, April 2019: Managing Headwinds. Washington, DC: World Bank. https://openknowledge.worldbank.org/ handle/10986/31500. World Bank. 2019c. Lao People's Democratic Republic—Public Expenditure and Financial Accountability (PEFA) Assessment. Washington, DC: World Bank. World Bank. n.d. “Senior Public Service: High Performing Managers of Government.” Note, World Bank, Washington, DC. http://www1.worldbank.org/publicsector/civilservice/epublishdocs/ SPS%20note%201216.pdf. World Bank and DRC (Development Research Center of the State Council, P. R. China). 2014. Urban China: Toward Efficient, Inclusive, and Sustainable Urbanization. Washington, DC: World Bank. https://openknowledge.worldbank.org/handle/10986/18865. 53