103669 v2 Making The Whole Greater Than The Sum Of The Parts: A Review of Fiscal Decentralization in Vietnam 2015 Making The Whole Greater Than The Sum Of The Parts: A Review of Fiscal Decentralization in Vietnam 2015 Abbreviation BVT Business Value Tax CBFA Committee on Budgetary and Finance Affairs CBRC China Banking Regulatory Commission CBIS Community Boards for Investment Supervision CIT Corporate Income Taxes DOF Department of Finance DP Development Partner DSA Debt Sustainability Analysis GDP Gross Domestic Product GOV Government of Vietnam HIFU HCMC Investment Fund for Urban Development LDIF Local Development Investment Funds LG Local Government LGFV Local-Government Financing Vehicles MOET Ministry of Education and Training MOF Ministry of Finance MOH Ministry of Health MOHA Ministry of Home Affairs MPI Ministry of Planning and Investment NA National Assembly NTP National Target Program ODA Official Development Assistance OECD Organization for Economic Cooperation and Development PAPI Provincial Governance and Public Administration Index PEFA Public Expenditure and Financial Accountability PER Public Expenditure Review PFF Public Financial Funds PIBs People’s Inspection Boards PIT Personal Income Tax PM Prime Minister PPC Provincial People’s Committee PSE Public Sector Efficiency PSP Public Sector Performance SBL State Budget Law SEDP Socio-Economic Development Plan SOE State Owned Enterprise UNDP United Nation Development Program VAT Valued Added Tax VBSP Vietnam Bank for Social Policies VDB Vietnam Development Bank VHLSS Vietnam Household Living Standard Survey WB World Bank Foreword This report provides a review of fiscal decentralization policies in Vietnam and their impact on the Government’s development objectives. It aims to inform reform of central-local fiscal relations in Vietnam to further promote growth and poverty reduction. The report was prepared under the overall guidance of Victoria Kwakwa (Country Director, WB Vietnam), Sudhir Shetty (Chief Economist, WB East Asia and Pacific Region), Rob Taliercio (Lead Economist, WB), Sandeep Mahajan (Lead Economist, WB), James A. Brumby (Practice Manager, WB), and Deepak Mishra (Lead Economist, WB) by a core team including: Quang Hong Doan (Senior Economist, WB); Indira Iyer (Senior Fellow, National Council of Applied Economic Research, India); Jorge Martinez-Vazquez (Regents Professor and Director, Department of Economics, Georgia State University); Nara Monkam (Research Director, Africa Tax Administration Forum, South Africa); Minh Van Nguyen (Senior Economist, WB); Cuong Dinh Pham (Former General Director at the Vietnam Ministry of Finance); Abha Prasad (Senior Debt Specialist, WB); Habib Rab (Senior Economist – Task Team Leader, WB); Anwar Shah (Director of the Centre for Public Economics, Chengdu/Wenjiang, China); and Quyen Hoang Vu (Economist, WB). Research assistance and technical inputs were provided by: Huong Dang (Consultant, WB); Hai Anh La (Economist, Vietnam Academy of Social Sciences); Phuong Anh Nguyen (Research Analyst, WB); Lucy Pan (Economist, WB); Hien Thu (Economist, Mekong Economics); Le Quang Thuan (National Institute of Finance). Linh Anh Thi Vu (Program Assistant, WB) provided administrative support. The State Budget Department and the State Treasury Department at the Vietnam Ministry of Finance; and the Committee for Budgetary and Financial Affairs (CFBA) at the Vietnam National Assembly provided very helpful guidance and advice throughout the study. The team is very grateful to the UNDP Vietnam and CFBA project team on “Strengthening the Capacities of Budget Oversight for People’s Elected Bodies in Vietnam” for support in facilitating dialogue with local authorities in Vietnam. The team is very thankful to local authorities from the following provinces for extensive discussions and feedback throughout the study: Lam Dong; Da Nang; Cao Bang; Bac Kan; Khanh Hoa; Quang Nam; Tay Ninh; Hau Giang; Can Tho; Dong Nai; Son La; Nghe An; Yen Bai; Lao Cai; Ca Mau; Kien Giang; Dien Bien; Quang Ngai; Ha Noi; Ho Chi Minh City; Hai Phong. The team thanks the following colleagues for advice and inputs across different parts of the report: Jairo Acuna-Alfaro (Policy Advisor, UNDP); Cut Dian R.D. Agustina (Economist, WB); James Anderson (Senior Governance Specialist, WB); Cuong Duc Dang (Senior Urban Specialist, WB); Gabriel Demombynes (Senior Poverty Economist, WB); Kari Hurt (Health Cluster Leader, WB); Chris Jackson (Lead Economist, WB); Sang Minh Le (Consultant, WB); Iain Menzies (Senior Water and Sanitation Specialist, WB); Lan Thi Thu Nguyen (Natural Resources Economist, WB); Quang Vinh Nguyen (Senior Water and Sanitation Specialist); Ngoc Thi Nguyen (Consultant, WB); Suhas Parandekar (Senior Economist, WB); Madhu Raghunath (Senior Urban Specialist, WB); Paul Vallely (Senior Transport Specialist, WB); Lorena Vinuela (Public Sector Specialist, WB); Linh Hoang Vu (Poverty Economist, WB); Lan Anh Vu (Human Development Specialist, WB); Ha Thi Tu Vu (Consultant, WB). Review comments at concept and decision stages were gratefully received from: Do Viet Duc (Ministry of Finance, Vietnam); Jonas Frank (Senior Public Sector Specialist, WB); Robert Gilfoyle (Senior Financial Management Specialist, WB); Nguyen Van Hao (Ministry of Finance, Vietnam); Tran Kim Hien (Ministry of Finance, Vietnam); Kai Kaiser (Senior Economist, WB); Lili Liu (Lead Economist, WB); Estelle Liu (Senior Economist, IMF); Vu Nhu Thang (President of National Institute of Finance, Vietnam Ministry of Finance); Theo Thomas (Lead Economist, WB); Vu Thanh Tu Anh (Lecturer of Economics and Director Research, Fulbright Economics Teaching Program); Michel Welmond (Program Leader, WB). The report findings and recommendations were presented at several workshops with the National Assembly; the Ministry of Finance; scholars and researchers at the Vietnam Academy of Social Sciences (VASS), Ho Chi Minh City University of Economics (HEU), and the Harvard University Fulbright Economics Teaching Program; and members of the media Support for the preparation of this report is gratefully acknowledged from the Australian Department for Foreign Affairs and Trade and the United Kingdom Department for International Development. Table of contents Executive Summary 11 CHAPTER 1/24 Introduction CHAPTER 2/34 Expenditure decentralization trends in vietnam Institutional arrangements for expenditure assignments 35 46 Expenditure Decentralization and Institutional arrangements for budgeting and accountability at local level 53 Accountability Conclusions and recommendations 61 CHAPTER 3/64 Credibility of spending plans and local budgetary discipline Relative spending on and efficiency of service delivery across provinces 65 74 Expenditure Performance of Local Conclusions and recommendations 90 Authorities CHAPTER 4/92 Local revenue sources and trends Revenue performance across local authorities 94 103 Local Revenue Arrangements Possible reforms to revenue sharing arrangements 112 Potential for greater revenue autonomy 115 Conclusions and recommendations 122 CHAPTER 5/124 Addressing fiscal imbalances across and within provinces Equity across and within provinces 126 138 Intergovernmental Fiscal Transfers Targeted transfer programs 145 Conclusions and recommendations 153 CHAPTER 6/156 Institutional framework and trends in local borrowing Local debt sustainability analysis: hcmc case study 157 164 Local Borrowing Managing fiscal risks from local borrowing 171 Institutional reforms for local borrowing 176 Conclusions and recommendations 182 List of Tables Table 2.1: Explaining local per capita spending levels 38 Table 2.2: Explaining decentralization of spending to district authorities 45 Table 2.3: Division of responsibilities across government levels in the education sector 51 Table 2.4: Responsibility for education service delivery across provinces 51 Table 2.5: Division of responsibilities across government levels in the education sector 52 Table 2.6: Information presented in provincial budget tables 57 Table 2.7: Explaining citizens’ satisfaction with local public service delivery using PAPI results 60 Table 3.1: Explaining lack of local budget credibility 69 Table 3.2: Determinants of provincial efficiency performance in Vietnam 85 Table 4.1: Taxonomy of Subnational Revenue Arrangements 95 Table 4.2: Shared revenue sharing rates (%) across stability periods 95 Table 4.3: Stochastic frontier estimation of potential revenue, Vietnam, 2006-2011 105 Table 4.4: Summary statistics for tax effort (from models in table 4.3) 106 Table 4.5: Explaining differences in tax effort at the provincial level, Vietnam 2006-2011 107 Table 4.6: Direct to indirect tax ratio, 1995-2008 115 Table 4.7: Assignments of taxes to subnational governments 117 Table 5.1: Formula to estimate capital investment needs in provinces 131 Table 5.2: NTP spending 2012-2013 (vnd million) 149 Table 5.3: Features of traditional and output-based conditional grants 151 Table 6.1: Summary of different sources and types of local borrowing 164 Table 6.2: International Sub-National Borrowing Practices 170 Table 6.3: Sources of Fiscal Risks 171 Table 6.4: Coverage of public debt (Vietnam and IMF definitions) 178 List of Figures Figure 1.1 State Administrative Structure in Vietnam 26 Figure 1.2 Nested Budget System in Vietnam 27 Figure 2.1 Local government spending in the local economy 35 Figure 2.2 Annual local government spending in the local economy 35 Figure 2.3 Local government spending and per capita gdp 36 Figure 2.4 Local government spending and population density 36 Figure 2.5 Local government per capita spending (06-11) 36 Figure 2.6 Annual local government per capita spending (06-11) 36 Figure 2.7 Per capita spending pre central transfers 37 Figure 2.8 Per capita spending post central transfer 37 Figure 2.9: Expenditure by local authorities as a share of total expenditure 2006-2011 (%) 39 Figure 2.10: Central share of social security spending 39 Figure 2.11: Share of local recurrent spending for selected government 39 Figure 2.12: Funding source for local spending (%) 40 Figure 2.13: Expenditure and revenue decentralization 41 Figure 2.14: Expenditure decentralization and dependence on transfers 41 Figure 2.15: Expenditure decentralization and local education share 42 Figure 2.16: Expenditure decentralization and local health share 42 Figure 2.17: District expenditure/local expenditure (06-11) 42 Figure 2.18: Annual District expenditure/local expenditure (06-11) 42 Figure 2.19: District capital/local capital (06-11) 43 Figure 2.20: Annual District capital/local capital (06-11) 43 Figure 2.21: District education recurrent/local recurrent 44 Figure 2.22: District health recurrent/local recurrent 44 Figure 2.23: Annual District education recurrent/local recurrent 44 Figure 2.24: Annual District health recurrent/local recurrent 44 Figure 2.25: Decentralized revenue for provincial spending 46 Figure 2.26: Decentralized revenue and district spending 46 Figure 3.1: Actual/budgeted local expenditure (%) 66 Figure 3.2: Actual/budgeted expenditure of provincial authorities (%) 66 Figure 3.3: Actual/budgeted capital by local authorities 67 Figure 3.4: Actual/budgeted capital by provincial authorities 67 Figure 3.5: Average divergence between actual and budgeted capital spending 06-11 by region (%) 67 Figure 3.6: Average divergence between actual and budgeted local revenue 06-11 by region (%) 67 Figure 3.7: Off-budget spending/total local spending 68 Figure 3.8: Off budget spending and per capita GDP 70 Figure 3.9: Carry over spending/local budget (%) 71 Figure 3.10: Factors that may impact carry over spending 71 Figure 3.11: Average carry over spending/local budget 2006-2011 by region (%) 72 Figure 3.12: Lower than average PSP and higher than average PSE 72 Figure 3.13: Higher than average PSP and higher than average PSE 72 Figure 3.14: Lower than average PSP and lower than average PSE 77 Figure 3.15: Higher than average PSP and lower than average PSE 77 Figure 3.16: Provinces in top quarter on DEA score 77 Figure 3.17: Provinces in second quarter on DEA scores 77 Figure 3.18: Provinces in third quarter on DEA scores 79 Figure 3.19: Provinces in fourth quarter on DEA scores 79 Figure 3.20: Efficiency frontier 79 Figure 3.21: Central Highlands Provinces against efficiency frontier 79 Figure 3.22: Mekong Delta Provinces against efficiency frontier 80 Figure 3.23: NCCC Provinces against efficiency frontier 82 Figure 3.24: Northern Mountain Provinces against efficiency frontier 82 Figure 3.25: Red River Provinces against efficiency frontier 82 Figure 3.26: South East Provinces against efficiency frontier 83 Figure 3.27: CentralHighlands 83 Figure 3.28: Mekong Delta 83 Figure 3.29: Northern Central and Central Coast 89 Figure 3.30: Northern Moutains 89 Figure 3.31: Red River Delta 89 Figure 3.32: South East 89 Figure 3.33: Red River Delta 89 Figure 3.34: South East 89 Figure 4.1: Overview of types of revenue retained by different tiers of government 94 Figure 4.2: Decentralized revenue/local GDP (%), 2006-2011 96 Figure 4.3: Annual decentralized revenue/local GDP (%), 06-11 96 Figure 4.4: Revenue composition (% GDP) 96 Figure 4.5: Share of Tax Burden and Output, by Companies ranked by Tax Payable 96 Figure 4.6: Central Highlands 98 Figure 4.7: Mekong Delta 98 Figure 4.8: NCCC 98 Figure 4.9: Northern Mountains 98 Figure 4.10: Red River Delta 98 Figure 4.11: South East 98 Figure 4.12: Central and local revenue (% of GDP), 2006-2011 period average 99 Figure 4.13: Central and local revenue as a share of total revenue (%) 99 Figure 4.14: Local revenues (% of total government revenues) 100 Figure 4.15: Dec rev/local receipts 100 Figure 4.16: Decentralized revenue/local core receipts at province and district levels (2006-2011 period 101 average) Figure 4.17: Revenue retained 100%/local receipts (06-11) 101 Figure 4.18: Annual revenue retained 100%/local receipts (06-11) 101 Figure 4.19: Own source revenue taxes/GDP (%) 102 Figure 4.20: Shared revenue/local revenue 06-11 (%) 102 Figure 4.21: Annual shared revenue/local revenue 06-11 (%) 102 Figure 4.22: Local revenues/1000 population (avg, vnd mil) 103 Figure 4.23: Annual off-budget revenue/local revenue (%) 103 Figure 4.24: Stochastic Frontier and Tax Effort 104 Figure 4.25: Central Highlands 108 Figure 4.26: Mekong Delta 108 Figure 4.27: NCCC 108 Figure 4.28: Northern Mountains 108 Figure 4.29: Red River Delta 109 Figure 4.30: South East 109 Figure 4.31: Average tax effort, Vietnam 2006-2011 110 Figure 4.32: Actual/local budgeted revenue (06-11) 111 Figure 4.33: Annual actual/local budgeted revenue (06-11) 111 Figure 4.34: local shared/local budgeted shared revenue (06-11) 111 Figure 4.35: Local 100%/local budgeted 100% revenue (06-11) 111 Figure 4.36: Local shared/local budgeted shared (period avg 06-11) 112 Figure 4.37: Local 100%/local budgeted 100% (period avg 06-11) 112 Figure 5.1: Vertical imbalance across provinces (06-11) 127 Figure 5.2: Annual vertical imbalance across provinces (06-11) 127 Figure 5.3: Decentralized revenue/local spending (06-11) 127 Figure 5.4: Annual decentralized revenue/local spending (06-11) 127 Figure 5.5: Vertical imbalance below province (2006-2011) 128 Figure 5.6: Annual vertical imbalance below province (2006-201 128 Figure 5.7: Decentralized revenue/district spending (06-11) 129 Figure 5.8: Annual decentralized revenue/district spending (06-11) 129 Figure 8.9: Pedictability of balancing transfers to provinces (06-11) 132 Figure 5.10: Predictability of balancing transfers to districts (06-11) 132 Figure 5.11: Decentralized revenue as % of local receipts 135 Figure 5.12: Balancing transfers as % of local core spending 135 Figure 5.13: Coverage of provincial horizontal imbalance (06-11) 136 Figure 5.14: Annual coverage of horizontal imbalances (06-11) 136 Figure 5.15: Coverage of district horizontal imbalance (06-11) 136 Figure 5.16: Annual coverage of horizontal imbalances (06-11) 136 Figure 5.17: Central balancing transfers/local core spending (06-11) 137 Figure 5.18: Annual central balancing transfers/core spending (06-11) 137 Figure 5.19: Province balancing transfers/district spending (06-11) 137 Figure 5.20: Annual prov. Balancing transfers/district spending (06-11) 137 Figure 5.21: Average nominal per capita budget revenues of provinces, Vietnam 2006-2011 139 Figure 5.22: Average real per capita budget revenues of provinces, Vietnam 2006-2011 140 Figure 5.23: Average per capita nominal and real expenditures of provinces, Vietnam 2006-2011 140 Figure 5.24: Average per capita cumulative budget revenues and average per capita expenditure 141 of provinces, Vietnam 2006-2011 Figure 5.25: Coefficient of variation of per capita cumulative budget revenues and per capita 142 xpenditure of provinces, Vietnam 2006-2011 Figure 5.26: Ratio of maximum to minimum of per capita cumulative budget revenues and per 143 capita expenditure of provinces, Vietnam 2006-2011 Figure 5.27: Coefficient of variation of per capita cumulative budget revenues and per capita 143 expenditure of districts, Vietnam 2006-2011 Figure 5.28: Coefficients of variation of per capita resourcing and expenditures across districts 144 Figure 5.29: District per capita spending and per capita GDP 144 Figure 5.30: District per capita spending and population density 144 Figure 5.31: Central target transfers/local core spending (06-11) 146 Figure 5.32: Annual central target transfers/local core spending (06-11) 146 Figure 5.33: Province target transfers/district spending (06-11) 146 Figure 5.34: Annual province target transfers/district spending (06-11) 146 Figure 5.35: Actual vs. Budget central target transfers (06-11) 147 Figure 5.36: Actual vs. Budget central target transfers (06-11) 147 Figure 5.37: Per capita spending on NTPs by region (2006-2011) 148 Figure 6.1: Local government debt 2006-2012 158 Figure 6.2: Borrowing in the five large cities 159 Figure 6.3: Average Local debt (VND million) 159 Figure 6.4: Local government debt (VND million) 159 Figure 6.5: Borrowing/local receipts and development 160 Figure 6.6: Borrowing/local receipts and fiscal autonomy 160 Figure 6.7: Borrowing/local receipts and fiscal sustainability 160 Figure 6.8: Borrowing/local receipts and share of recurrent spending 160 Figure 6.9: Borrowing and urbanization rate 160 Figure 6.10: Borrowing/local receipts and budget credibility 160 Figure 6.11: Composition of local debt, 2006-2012 161 Figure 6.12: HCMC Local Financing (% of local GDP) 165 Figure 6.13: HCMC Local Spending (% of local GDP) 165 Figure 6.14: HCMC Local Revenue and Spending Aggregates (% change real) 165 Figure 6.15: HCMC Local Fiscal Outturns 165 Figure 6.16: HCMC Local Revenue (% of GDP) 167 Figure 6.17: HCMC Local Expenditure (% of GDP) 167 Figure 6.18: HCMC Debt-to GDP ratio 168 Figure 6.19: HCMC Debt service-to-revenue ratio 168 Figure 6.20: HCMC Interest payment/Recurrent Expenditure 168 Figure 6.21: HCMC Overall and Primary Balance (% of GDP) 168 Figure 6.22: HCMC Revenue/Recurrent Expenditure 169 Figure 6.23: HCMC Debt-to-capital expenditure (%) 169 Figure 6.24: Largest Local Public Financial Funds 173 List of Boxes Box 4.1: Country experience in revenue sharing 114 Box 6.1: Key Macroeconomic Assumptions for Baseline Scenario (2013-2025) 166 Box 6.2: Contingent liabilities of local authorities in China 172 Box 6.3: Current debt management strategy in Vietnam 177 EXECUTIVE SUMMARY 11 Local authorities in Vietnam are responsible for over half of total government spending, thanks to fiscal decentralization policies implemented over the past eighteen years. Total government spending in Vietnam is close to 30 percent of Gross Domestic Product; local authorities’ spending is close to 17 percent of GDP. Study of fiscal decentralization therefore is central to understanding government spending in Vietnam and its significant impact on the country’s successful record on economic growth and poverty reduction. To date however there has been little analysis of how fiscal decentralization has enabled local authorities to effectively, efficiently and accountably spend money on public services for development. Spending decentralization has grown more quickly than information and knowledge on local fiscal policies in Vietnam. This has prompted a number of questions from the National Assembly, the Central Government, and other stakeholders including: How are local spending choices aligning with national level objectives? How much spending responsibility do local authorities actually have? Are current policies and institutions sufficient to ensure that the money is spent well and in line with citizens’ needs? How much do provinces spend on delivering a given level of public services and what might explain the differences? What results are being achieved? How effectively are local authorities able to raise revenues to finance local spending? Are fiscal transfers sufficient to effectively cover all local spending needs? Are transfers being equitably distributed? Are there more opportunities for provinces to borrow to meet investment needs? These are some of the questions that this report tries to answer to help inform future changes to central- local fiscal relations. The report finds that fiscal decentralization policies have helped to channel more spending to the poorest parts of the country where development needs and costs of service delivery are higher. There is fairly wide variation in spending efficiency across provinces. Some of it is due to factors outside of local authorities’ control (e.g. population density, geographic location). But the report also finds that money could be better spent through improvements in budget management and accountability. The level of local revenue has increased over time, and the evidence suggests that the current policies do not impact negatively on revenue collection efforts. Another positive feature is that revenue and fiscal transfers have on the whole been quite equitable. However, local authorities currently have little to no autonomy over revenue policy and administration. Some autonomy over revenue can increase accountability between state and citizen and promote better use of the resources. It can also enable certain provinces to better meet their investment needs. The study finds that the latter can be complemented with some increase in local borrowing levels. A combination of these measures should help to maintain the government’s successful redistributive policies to meet social objectives, and enable high economic potential provinces to invest in growth enhancing initiatives, whilst enhancing transparency to ensure accountability. Lack of transparency and coordination in intergovernmental fiscal relations can lead to fragmentation and unequal development across the country. Intergovernmental fiscal relations should ensure that development efforts of all provinces exceed the sum of their individual parts so that they can help eliminate extreme poverty and promote shared prosperity in Vietnam. 12 INTRODUCTION Local fiscal policies Local authorities have been granted increasing levels of fiscal responsibilities have a significant since the adoption of the State Budget Law (2002). Local fiscal policies impact on Vietnam’s in Vietnam have a significant impact on the country’s overall development development. This trajectory. They determine decisions on public service delivery, and report looks at the therefore the overall development trajectory of the country. Poor spending extent to which fiscal or revenue decisions at local level, coupled with lack of transparency can decentralization lead to inefficient spending, unequal development across the country and policies have achieved their development reversal of the country’s successful record of growth and poverty reduction. objectives. Despite the importance of fiscal decentralization in Vietnam’s development, there has been little study on the successes and failure of policies and institutions in this area. This report responds to demands for more analysis of fiscal decentralization policies in Vietnam and the extent to which these have delivered on their intended objectives of equity and efficiency to promote economic growth and poverty reduction. It aims to inform future changes to the system of intergovernmental fiscal relations through revisions to the SBL 2002 and adoption of 2016-2020 Stability Period regulations. The report aims to The SBL 2002 establishes the key principles for spending assignments, inform changes to revenue arrangements, and intergovernmental fiscal transfers. It also the SBL 2002 and grants provincial authorities a fair degree of autonomy to determine fiscal 2016-2020 Stability relationships with districts and communes within their jurisdiction. There Period regulations, is general acknowledgement that the SBL 2002 has provided a solid which provide the framework for public finance management including intergovernmental frameworks for fiscal relationships. intergovernmental fiscal relations in After 10 years of implementation of the SBL 2002, a number of issues arise Vietnam. in relation to the system of intergovernmental fiscal relations. These include: clarity of spending responsibilities and local level accountability including for national priorities and objectives; spending performance of local authorities; the effectiveness of local revenue arrangements in meeting spending needs and the potential for increased revenue autonomy for selected provinces; the extent to which provincial authorities are promoting or impeding central government’s redistribution efforts; and the potential for increased debt financing for local authorities within prudent limits. This report builds on existing research and aims to provide new analysis and perspectives on the above areas through extensive study of available data, consultations with central and local authorities, and reviews of laws, regulations and policies at central and local level. It analyzes five pillars of fiscal decentralization to assess the extent to which these are promoting or hindering the government’s development objectives. These five pillars are: (i) expenditure decentralization and accountability; (ii) expenditure performance of local authorities; (iii) local revenue arrangements; (iv) intergovernmental fiscal transfers; and (v) local borrowing. EXPENDITURE DECENTRALIZATION 13 Local government Local government spending constitutes an important share of provincial spending has played economies in Vietnam. It has played a particularly important role in poorer an important role in provinces that are sparsely populated. This reflects efforts to expand delivering services to service delivery into poorer areas, where costs of service delivery are the poorest parts of also higher. Central transfers to local authorities have helped in this the country. redistribution process. Local authorities Local authorities are now responsible for just over half of total government are responsible for spending. Local authorities’ share of total government recurrent spending over half of total in important areas like education, health, economic services, and public government spending. administration has been high and increased further between 2006 and 2011. For education, local authorities account for 80-90 percent of total recurrent spending and for health the ratio is around 75-80 percent. Decentralized revenue and unconditional balancing transfers help to cover over three quarters of local spending, suggesting high level of discretionary resources available to local authorities. There is significant Vietnam exhibits a high degree of spending decentralization within decentralization within provinces, which has increased over the last two Stability Periods. Half of all provinces, bringing observations in 2011 showed that districts were spending at least 45 percent resource allocation of total local recurrent spending. District authorities have been responsible decisions closer to for most of the recurrent spending in both education and health within the people. most provinces. Districts’ share of local recurrent spending in education has remained relatively high and constant in the last two Stability Periods, but in health there has been a marked shift in spending decentralization to districts. In contrast, most of the capital spending (around 70 percent on average) is carried out by provincial authorities. Higher capacity In Vietnam, the evidence shows that on average higher capacity local local authorities authorities have higher levels of spending responsibilities. Experience have higher levels in other countries has shown that decentralizing too quickly can have of spending serious adverse impacts when local administrative capacities are low. The responsibilities. evidence in Vietnam also shows that provinces that have more people in rural areas have also decentralized more responsibility to the districts. This is a good sign since more rural provinces tend to have lower population density, therefore decentralization can help to better take account of local preferences. There is scope for The SBL 2002 currently assigns the same expenditure responsibilities to clarifying spending both central and provincial authorities i.e. most spending assignments are responsibilities shared/concurrent. Sector legislation provides some clarity in terms of between center and assigning exclusive responsibility for regulation to central government. With province. the exception of transportation, health and education however, because so many functions are shared between center and province, there is ambiguity over spending assignments. Some functions Some functions should rest exclusively with central authorities because their 14 should rest costs and benefits are national in scope e.g. national defense and security, exclusively with foreign policies). In Vietnam, however, local authorities are also expected central government. to participate in funding of these services. Several local authorities have indicated that this creates ambiguities and pressures on local budgets, which should be addressed through clearer assignment of exclusive responsibilities to central government. In the case of non-exclusive/shared functions, the explicit responsibilities of center and provincial authorities should be clearly set out based on agreed criteria. Within a jurisdiction, Provincial level resolutions on revenue and expenditure assignments service delivery have provided significant clarity on spending and revenue assignments for responsibility should provincial, district and commune authorities. However, in many cases, both rest with only one province and district authorities are also assigned the same service delivery level of government. responsibility. Therefore unlike center-province division of responsibilities, there are explicit assignments for authorities within provinces; but in some cases, all tiers within the province are explicitly responsible for the same services. Whilst different orders of government can be responsible for different tasks within a jurisdiction, it is better to avoid assigning different orders of government to perform the same task within a jurisdiction. This leads to overlap and makes it difficult to hold each level of government accountable. The nested budget Although the SBL 2002 has introduced significant clarity on the roles and system complicates responsibilities of different actors, the nested budgeting system complicates budget preparation budget preparation and monitoring. Vietnam is one of few countries that still and monitoring. operates such a nested system, which creates several challenges. Firstly, Some SBL 2002 the multi-layered budget approval process significantly compresses the provisions also dilute budget calendar. This gives Provincial Councils little time to review the draft accountability to budget. Secondly, the authority of, and therefore accountability to, local People’s Councils. councils is diluted by the following provisions in the SBL 2002: “veto rights” of higher levels of government over budgets approved by lower tiers (Article 47) – which albeit is applied only in cases where lower tiers have made incorrect allocations; and the lack of formal requirement to seek legislative approvals before the executive makes changes to budget appropriations. Central fiscal rules Despite increased responsibility over spending, central fiscal rules and and norms on norms affect local autonomy over budget decisions in selected areas and minimum allocations distort resource allocation. There are minimum allocations set for education impact adversely on and science and technology without due consideration of actual needs or local autonomy. the level of service provision by central authorities within a province. A minimum allocation to an area such as science and technology without link to capacity can lead to waste. Rules on the need to spend 50 percent of over-realized revenue on wages in the next budget year and 50 percent on important capital projects fail to take advantage of recognized budget priorities and leads to the inefficient use of resources. Central fiscal rules Great efforts have been made to ensure publication of budgetary information, and norms on but current practices on budget disclosure do not encourage citizen 15 minimum allocations participation. The general public is not able to contribute to the budget impact adversely on preparation process as the budget is only published after it is approved local autonomy. by the legislature. The current budget classification structure, budget table templates, and coverage of fiscal activities in budget documents do not facilitate active citizen participation in scrutinizing public finances and providing meaningful feedback to local authorities. The highly technical presentation of budget documents prevents the public from understanding let alone analyzing budget policies. There is much Budget disclosure should be complemented with efforts to encourage citizen budget information participation as disclosure by itself will have limited impact on accountability in the public domain for service delivery. At a minimum this would involve publication of the but limited public draft budget submitted to the National Assembly and People’s Councils. participation. But in addition, the government should look at further developing its existing Citizens’ Budgets to communicate the State Budget in a simple way so that it is understood by as many people as possible. Evidence from the PAPI survey show a very clear link between participation in public finance and project design and satisfaction with service delivery at the local level. Based on these findings, the study recommends: (i) more explicit, and when possible exclusive, assignments on service delivery for different tiers of government within a jurisdiction; (ii) decompression of the budget preparation calendar to allow greater scrutiny of local budgets by People’s Councils; (iii) greater autonomy over budget approval authority of People’s Councils and clearer provisions on budget appropriations; (iv) review of minimum allocation requirements for sectors and minimum allocation by types of expenditure, which hinder local autonomy over resource allocation; and (v) increased transparency and citizen participation in the budget process through further development of existing Citizens’ Budgets and other instruments to improve communication. EXPENDITURE 16 PERFORMANCE Local government With increased spending responsibility delegated to local authorities, it is spending has played important to look at expenditure performance across provinces. This has an important role in been a topic of high interest, particularly for the National Assembly and delivering services to central authorities because reporting on service delivery performance has the poorest parts of not kept pace with the increased spending decentralization. In addition to the country. this, tightening fiscal conditions in Vietnam calls for a better understanding of how much different provinces are spending to deliver the same services, which should help identify areas of potential efficiency gains. Local spending plans The study finds that local spending plans lack credibility, which reduces lack credibility, which transparency and impacts on efficiency. This is particularly the case for reduces transparency capital spending, which is often more than 50 percent of what was budgeted, and impacts on well above good practice guidelines of maintaining spending within 5 efficiency. percent of budget. The lack of budget credibility is partly due to the flexibility accorded to the executive to change budget appropriations approved by the legislature. As recommended in chapter 2, this warrants review of Article 49 in the SBL 2002 to clarify rules on changes to appropriations. The higher than budgeted spending is also due to spending from contingency reserves, and extra budgetary funding sources for capital investment projects. Policies on over There are several institutions and policies embedded in the SBL 2002 realized revenue, off- that enable the executive to take advantage of flexibility over budget budget financing, and appropriations, and which this study recommends be reviewed. The first carry overs reduce is around use of over-realized revenue, for which, according to regulations, budget credibility. half has to be channeled to salaries and the other half to capital regardless of actual spending needs. The second policy area is on the use of off- budget financing mostly for capital investments, which ranges from 5 to 20 percent of the budget. These extra-budgetary sources lack transparency and create a distorted picture of budget allocations. The third policy area contributing to loss of budget credibility relates to the practice of carry overs. Unlike other countries, there are currently no limits on the level of carry overs, which can go up to 50 percent for some provinces. Lack of spending plan The above institutions and policies on budget management can impact credibility dilutes link adversely on the efficiency of spending. Maintaining a credible spending between budgets and plan is a necessary, albeit not sufficient, condition to promoting more plans. effective and efficient government spending. In addition, poor credibility of spending plans dilutes the link between local development plans and local budgets. Despite the strong role of local spending in provincial economies in Vietnam, the role of public finances is insufficiently covered in local development plans reviewed during this study. Differences in The study looks at various measures of spending efficiency across provinces spending efficiency and the extent to which current budget management practices affect those. 17 across provinces are Using available data on three sectors (education, health and transportation), due to factors outside the study computes Public Sector Performance indices, reflecting the of province’s control outputs and outcomes achieved in the three sectors. It uses this to derive but also due to lack of Public Sector Efficiency scores. The analysis finds that although there are credibility in spending some clear patterns in terms of service delivery performance and spending plans. efficiency across regions, within regions there are disparities between provinces. Some of these disparities are associated with differences in population density. But poorer performers also tend to have less credible spending plans and higher levels of carry overs compared to their peers in the same region. There are also cases where provinces are very efficient but are performing poorly on critical service delivery outcomes, meaning that these provinces could consider increasing local spending in those particular sectors. On average provinces From the above, a productive efficiency frontier is estimated based on in poorer regions the maximum level of output attained by local authorities in Vietnam for spend 35-40 percent a given set of inputs. The purpose is to review the efficiency of provinces to deliver the same relative to such an efficiency frontier. The frontier is derived based on PSP level of services as scores against per capita spending of the most efficient provinces. This the most efficient analysis provides an approximate idea of the difference in spending across provinces. provinces for achievement of a given level of outputs and outcomes. It finds that on average, provinces in the Northern Mountains region spend 42 percent more and those in the Central Highlands 35 percent more to achieve the same outputs/outcomes as the most efficient provinces. In the South East and the Mekong River Delta on the hand, the average is closer to 7 percent and 9 percent respectively. As noted above, some of these differences are associated with geographical characteristics and population density, but also partly due to budget management practices. Within regions, Within the regions, the analysis finds a number of outliers for which provinces that are productive efficiency is considerably lower than for other provinces in the less efficient also same region. For the poorer regions (Northern Mountains and Central have less credible Highlands), the outliers also happen to have the least credible spending spending plans. plans and highest levels of carry over spending. In the Northern Central and Central Coast region, the province of Danang performs very well in terms of service outputs and outcomes, but it also spends much more despite high population density and urbanization. Determinants of Econometric analysis of possible determinants of the relative efficiency relative efficiency of of provinces in Vietnam confirms some of the hypotheses in the fiscal provinces in Vietnam decentralization literature. Findings show that dependence on transfers are in line with the can impact negatively on efficiency. There may be other factors also (not fiscal decentralization covered in the analysis) such as capacity to develop sound budgets, timely literature. budget allocations and capacity to implement projects. Findings also show that population density, urbanization rates, and budget transparency have a positive impact. Surprisingly, higher per capita GDP was found to reduce 18 efficiency, though this may point to residents in poorer localities being more active in monitoring the quality of service delivery. The credibility of spending plans has a slight negative impact on efficiency, but the share of recurrent spending in overall spending seems to have a strong negative impact on efficiency. Richer provinces tend On the latter, the study looked specifically at how the composition of to have a lower share recurrent spending, in particular the share of salaries and wages in overall of recurrent spending recurrent spending, interacts with productive efficiency across regions and going to wages and provinces. Overall, provinces in wealthier regions not only have lower salaries. local spending as a share of the local economy, but also a lower share of recurrent spending going on wages and salaries – this may be because poorer provinces have to offer higher wages and salaries to attract staff, and not necessarily because the civil service is leaner in richer provinces. Within poorer regions, there are outliers in terms of high share of salaries and wages and low productive efficiency. It is important to review those cases in more detail to ensure that the wage bill is not crowding out critical operations and maintenance or goods and services spending, which would further exacerbate efficiency concerns. To further strengthen expenditure performance of local authorities in Vietnam, the study recommends: (i) spending of over-realized revenue should not be tied to specific categories and should be approved by legislature; (ii) extra budgetary activities should be integrated into the overall budget; (iii) limit the level of carry over spending as a share of budget and to capital spending only; (iv) introduce population density criteria in allocation norms to estimate spending needs of local authorities; (v) review wage bill and application of salary reform policies. LOCAL REVENUE ARRANGEMENTS 19 Levels of local Local authorities in Vietnam have seen increased levels of local revenue revenue have in the 2006-2011 period to match spending decentralization, but continue increased but revenue to have little to no autonomy over revenue policy and administration. autonomy has not. Local authorities in Vietnam have two main sources of revenue outside of transfers from upper tiers of government: (i) revenue retained 100 percent by local authorities, which is the closest that local authorities come to “own source revenue;” and (ii) revenue that is shared with upper tiers of government, which are pooled and redistributed across the country. Decentralized revenue Compared to other countries, decentralized revenues in Vietnam constitute as a share of GDP relatively large shares of national GDP (9-10 percent. General government are relatively large in revenue to GDP is around 20 percent of GDP) and general government Vietnam. revenue (33 percent excluding extra budgetary sources). Decentralized revenue as a share of local GDP on the whole is relatively small – less than 7 percent for half of all observations from 2006 to 2011. The reason is that central revenues constitute the bulk of general government receipts. It therefore does not necessarily reflect lower revenue effort on the part of local authorities. The share of The share of decentralized revenue over total local revenue however has decentralized revenue declined over time. This is partly because richer provinces have been however has fallen. making higher contributions of shared revenue to central government for the purposes of redistribution through transfers. Transfers in turn have constituted a growing share of local financing. There is however One concern about this trend is that provinces where sharing rates declined good evidence of in subsequent Stability Periods may have disincentives to maximize strong revenue effort revenue effort. The marginal benefit of raising more revenue for these by provinces. provinces gets diminished with the decreasing tax sharing rates, which get adjusted at the beginning of each stability period with the objective of channeling those funds to the central government for redistribution and other budget objectives. An analysis of revenue effort by local authorities does not suggest that On average, provinces this has had a significant negative impact on revenue effort by the net in Vietnam have contributing provinces. Performance across provinces however varies relatively strong a lot due to factors such as local capacity and the level of revenue revenue effort. decentralization as measured by share of 100 percent locally retained revenue. This, together with the needs of richer, high growth potential provinces, highlights the need to consider opportunities for more revenue autonomy. The revenue sharing Whilst the revenue sharing arrangement does not necessarily point to 20 arrangement warrants lower incentives to collect revenue on the part of net contributing provinces, reform including there are a number of elements that warrant review. Shared revenues in revamping the system Vietnam are split based on where revenues are actually collected rather of assigning revenue than where the tax is incurred (the so-called “derivation principle”). This to where they are raises questions concerning the fairness of the system, especially for the collected rather than Value-Added Tax (VAT) and the Corporate Income Tax (CIT). For example where the tax is incurred. if an enterprise is operating in Son La province, but is headquartered in Hanoi, the enterprise’s VAT and CIT liabilities will be owed to the province of Hanoi (unless it is a firm with unified accounting, in which case CIT will be paid to the central government). The major taxes therefore end up credited primarily to the few jurisdictions where the headquarters of the enterprise are located or the place of business registration, and not necessarily where the outputs are produced or sold. There are several options to address this including sharing VAT on a formula basis, or centralizing CIT and natural resource taxes. There is currently little interest to move to such a model because of a perception that VAT is a tax in expansion and CIT is one in contraction in their relative shares in total revenue. Increasing revenue Aside from this, however, increasing revenue autonomy is one of the most autonomy is one of significant issues with the current local revenue arrangements in Vietnam. the most significant As noted in chapter 3, this is important for efficiency of spending, but also policy issues with the to enable richer provinces to fill their infrastructure financing deficit, as current local revenue discussed in chapters 5 and 6. There are several options that Vietnam arrangements. can consider to enhance local revenue autonomy. This would entail Several options are providing local governments a closed list of selected taxes with discretion proposed in the study. to set rates within a band. It might also involve increasing autonomy over determination of user fees at subnational level, and over the longer-term considering the full introduction of modern property taxation. Based on these findings, the study recommends: On revenue sharing arrangements: (i) consider moving to sharing of VAT on a formula basis rather than on a derivation basis; (ii) centralize natural resource taxes, with offsetting transfers to poorer provinces that rely on these taxes; (iii) share CIT on a formula basis or centralize CIT (if allowing surtax on CIT); or to simplify, considering the option to establish a general pool of shared taxes (including VAT, CIT, and PIT) at the central level and apply the same sharing principles for the pool as a whole. On revenue assignments: (iv) enable provinces to impose surtaxes on PIT, local businesses and some excises; (v) increase autonomy over determination of user fees at local level; (vi) consider the introduction of the property tax in the long run. These reforms should improve accountability, spending efficiency, transparency and equity of the revenue arrangements, while promoting fiscal responsibility, spending efficiency, and eventually local economic performance and revenue efficiency. INTERGOVERNMENTAL FISCAL TRANSFERS 21 Vietnam has a Vietnam has developed a relatively transparent, rules-based system of relatively transparent, intergovernmental fiscal transfers. Revenue sharing arrangements and rules-based system of the system of balancing transfers in Vietnam have played a significant fiscal transfers. role in narrowing vertical and horizontal fiscal imbalances respectively. A general trend both across and within provinces, however, is growing vertical imbalances due to local authorities’ increased spending responsibilities. In other words, 100 percent locally retained revenues are not keeping pace with increased spending decentralization. Revenue sharing arrangements are helping to cover the fiscal gap, but overall spending responsibilities are growing faster. As a result, a selected number of richer provinces are contributing a growing share of their shared revenues for redistribution through balancing transfers. Richer provinces are The study argues that trading off richer provinces’ right to retain a higher share transferring higher of shared revenue for growth enhancing spending to allow more redistribution revenue to allow more for greater equity seems like the correct policy choice in Vietnam. The redistribution. evidence shows that increased spending responsibility of both provincial and district authorities are being covered by increased levels of balancing transfers. The latter at the center to provincial level are underpinned by a transparent, norms-based system. At province to district level, the publication of Stability Period rules and regulations by the provincial governments have also helped to improve transparency. The current system The current institutional arrangements for fiscal transfers have helped to has promoted more promote greater equity in resource distribution both across but also within equity across and provinces. The effectiveness of central government’s redistribution policies is within provinces, but a matter of concern in countries where intermediate provincial authorities are there is wide variation granted responsibility for determining transfers to lower orders of government. in per capita spending However, the evidence shows that balancing transfers are helping to equalize across districts. levels of resourcing across provinces and across districts. However, there is evidence a fairly wide variation in per capita spending across districts. This may be reflecting the diversity in conditions of different districts but there is no conclusive evidence on this. Targeted transfers Targeted transfers continue to play an important role in local spending even should become though its relative share has fallen over time. The lack of predictability in more predictable targeted transfers poses serious challenges for both provincial and district and shift towards a authorities. It impedes on planning, budgeting, and delivery of NTP targets. performance-based The level of resourcing however is not closely aligned with the targets and framework. objectives, which are quite ambitious. Complex, input-based guidance on NTP implementation reduces flexibility and increases burden of reporting. There is a real opportunity to review the institutional arrangements for NTPs for the coming Stability Period. In light of the above findings, the study recommends to: (i) Link Stability Period balancing transfers to inflation so do they do not decline in real terms over the period; (ii) make some adjustments to allocation norms to determine spending needs; (iii) ensure greater predictability of targeted transfers to provinces and districts; (iv) simplify institutional structure for NTPs and link resourcing to performance. 22 LOCAL BORROWING Some provinces are The system of intergovernmental fiscal relations in Vietnam, as illustrated not able to satisfy in the previous chapters 4 and 5, is geared to redistributing locally collected capital spending revenues, the bulk of which are contributed by a handful of provinces, which needs through at the same time increasingly point to a growing infrastructure financing existing revenue and deficit. Local borrowing has accordingly emerged as an important topic in transfers. Vietnam particularly for those provinces that are not able to satisfy their capital spending needs through the existing local revenue and transfer arrangements. Local borrowing has Borrowing by local authorities over the 2006-2011 period has remained very remained low in the low (less than 3 percent of GDP). The five largest cities account for just over 2006-2011 period. 40 percent of total local debt. Local authorities have several sources of debt financing including development banks (38 percent of total in 2011), State Treasury (29 percent), local bonds (22 percent), and central government on-lending (8 percent). An illustrative The SBL 2002 requires local authorities to maintain total outstanding debt debt sustainability from these sources at below 30 percent of annual capital budgets, except analysis of HCMC for HCMC and Hanoi for whom the ceiling is 100 percent of annual capital shows that under budget. Currently 13 provinces have already breached this ceiling, which specific assumption, in the first place is not a good indicator of local authorities’ actual borrowing the province could capacity. potentially borrow more than the current An illustrative debt sustainability analysis of HCMC shows that under limits in the SBL 2002. specific assumptions, the province could borrow more than the current This warrants review of current fiscal rules limits prescribed in the SBL 2002 and maintain sustainable levels of debt. However, without greater access to decentralized revenue, the province would face liquidity pressures, thereby reinforcing recommendations on revenue autonomy in chapter 5. Aside from direct debt liabilities, local authorities also need to look more closely at indirect liabilities and contingent liabilities, potential sources of which include: Public Financial Funds; local SOEs; and risks to the banking sector stress from local authorities’ payment arrears to local contractors. Aside from direct In light of the above findings, the study recommends to: (i) restate fiscal liabilities, local rules related to local debt in the SBL 2002 and adopt more standard authorities also need qualitative fiscal rules along international standards in the revised law, to watch indirect and retaining the current provisions on borrowing only for capital investment; contingent liabilities. (ii) adopt specific thresholds on local debt stock and debt servicing in line with debt sustainability targets; (iii) improve transparency and reporting on public debt, including through statistical debt bulletins in more advanced provinces; (iv) transition to a two-track local debt system with central government facilitating more market access for advanced provinces and greater ODA on-lending for less advanced provinces. 23 INTRODUCTION 01 The State Budget Law 2002 (SBL 2002) has enabled decentralization of important fiscal responsibilities to local authorities over the past ten years. This report responds to demands for more analysis of fiscal decentralization policies in Vietnam and the extent to which these have delivered on their stability, equity and efficiency objectives. It aims to inform future changes to the system of intergovernmental fiscal relations through revisions to the SBL 2002 and adoption of 2016-2020 Stability Period regulations. The SBL 2002 establishes the key principles for spending assignments, revenue arrangements, and intergovernmental fiscal transfers. It also grants provincial authorities a fair degree of autonomy to determine fiscal relationships with districts and communes within their jurisdiction. The current system allows a reasonable level of differential treatment across provinces to take account of their specific circumstances. There is general acknowledgement that the SBL 2002 has provided a solid framework for public finance management including intergovernmental fiscal relations. At the same time, after 10 years of implementation a number of issues arise such as: clarity of spending responsibilities and local level accountability including for national priorities and objectives; spending performance of local authorities; the effectiveness of local revenue arrangements in meeting spending needs and the potential for increased revenue autonomy for provinces; the extent to which provincial authorities are promoting or impeding central government’s redistribution efforts; and the potential for increased debt financing for local authorities. This report builds on existing research and aims to provide new analysis and perspectives on the above areas through extensive study of available data, consultations with central and local authorities, and reviews of laws, regulations and policies at central and local levels. It covers the following five pillars of fiscal decentralization: (i) expenditure decentralization and accountability; (ii) expenditure performance of local authorities; (iii) local revenue arrangements; (iv) intergovernmental fiscal transfers; and (v) local borrowing. CHAPTER 1: INTRODUCTION 25 1. Background and motivation for report on fiscal decentralization in Vietnam: Vietnam over the past 18 years has gradually decentralized more fiscal responsibilities to local authorities. Local spending today accounts for just over half of general government spending. Local revenue accounts for over a third of general government revenue, and just over half when extra budgetary sources are included. These are significant shares when compared to other countries, particularly those at a similar level of development to Vietnam. Fiscal decentralization was mandated in large part by Vietnam’s first State Budget Law in 1996 and its subsequent revision in 2002. The objectives were to promote sustainable development underpinned by local preferences and economic stability; equity across provinces; efficient service delivery; and enhanced transparency and accountability in public finances. 2. As the National Assembly and the Government of Vietnam prepare to further update the State Budget Law 2002 (SBL 2002), and embark on a new Stability Period from 2016 to 2020, there is significant demand for analysis on the effectiveness of fiscal decentralization policies to date. Several studies in the past have made important contributions to informing the current institutional framework for intergovernmental fiscal relations. Stakeholders in the National Assembly, the government, local authorities, and Development Partners would now like to see more evidence of how this framework has delivered in terms of stability, equity and efficiency. 3. In this regard, a particular gap in the current literature relates to fiscal relations between provincial and sub-provincial authorities. With some limited exceptions, most of the past studies on fiscal decentralization in Vietnam have focused mainly on the relationship between central and provincial authorities. However, since the adoption of the SBL 2002, provincial authorities have been granted important authority for determining the institutional framework for fiscal relationships across different orders of government within their jurisdiction. Analysis of developments and institutional arrangements at this level is therefore critical to understanding fiscal decentralization in Vietnam. 4. Objectives of Vietnam Fiscal Decentralization Review: The original study concept note had set out the following three intentions, which were followed through over the course of research process : (i) take stock of the current institutional framework for intergovernmental fiscal relations; (ii) empirically assess the extent to which this framework has delivered on its intended objectives; and (iii) use the latter to provide technical inputs and inform policy and institutional reforms for fiscal decentralization in Vietnam. 5. Based on the above, the study’s specific objectives are to: (i) inform discussions on the revision to the SBL 2002 with specific recommendations on the Law’s provisions relating intergovernmental fiscal relations; (ii) recommend sound principles for fiscal relations within provinces to help guide the development of local fiscal decentralization rules for the 2016-2020 Stability Period; (iii) enable stakeholders to better understand and monitor intergovernmental fiscal relations through the development of a standard of indicators; (iv) illustrate how different analytical tools and approaches can be used to assess issues such as relative efficiency of local resource mobilization and use, equity of resource distribution across and within provinces, and debt sustainability of local debt; and (v) use the aforementioned to inform prioritization of resources, including donor projects. 6. State administrative structure in Vietnam: To set the context, it is helpful to have an overview of the State 26 administrative structure. Vietnam is a unitary state with four tiers of government: central; 63 provinces; around 680 districts; and around 11,000 communes. Each tier of government has both legislative and executive authorities. At central level legislative authority rests with the National Assembly, and executive authority rests with line ministries and agencies. At local level, each tier of government has People’s Councils to exercise legislative authority; and People’s Committees and line departments to exercise executive authority. 7. The Constitution of the Socialist Republic of Vietnam, adopted by the National Assembly in November 2013, recognizes the Central Government as “the organ of the National Assembly, the highest organ of State administration of the Socialist Republic of Vietnam” (Article 109) and provides a sweeping mandate for it to carry out any tasks that it deems necessary “for the fulfillment of the political, economic, cultural, social, national defense, security and external duties of the State” (Article 109). Article 112.1 mandates the establishment of a unified government under central control as an important and overriding objective. 8. The Constitution maintains in Article 118 that for administrative convenience, the country will be divided into provinces and cities under direct central rule. Provinces will be divided into districts, provincial cities, and towns. Cities under direct central control will be subdivided into urban and rural districts and towns. Districts are divided into communes and townlets; provincial cities and towns are divided into wards and communes and urban districts will be divided into wards. Figure 1.1 State Administrative Structure in Vietnam Central Government Provincial authorities City authorities Districts Cities Towns Urban Rural Communes Townlets Wards Communes Communes Communes 9. Overview of intergovernmental fiscal relations in Vietnam: The State Budget Law 2002 (SBL 2002) provides the overall institutional framework for intergovernmental fiscal relations. In many countries, the Organic Budget Law usually only covers the central government. Intergovernmental fiscal relations, and sub-national budget management arrangements, are usually treated separately in Decentralization Laws and Local Government Laws. Budget Laws in China, Lao PDR and Vietnam cover all levels of government. This reflects the unique nested budget systems in these countries. Budgets of local authorities are prepared and submitted through a bottom up process, in which local legislatures review and appropriate the local budgets before submitting these to the upper tier of government. The national legislature ultimately adopts a State Budget for the entire country, being the consolidation of the central and local budgets.1 1 WB, “Vietnam: Review of the State Budget Law, 2002” (April 2013), Section 1, paragraph 6 10. Within this nested budget system, provinces, districts and communes, collectively referred to as local authorities, have significant delegated responsibility to allocate resources within their jurisdiction. Administrative 27 departments at each level of government operate on the basis of dual subordination. This means that they are horizontally accountable to the People’s Council within their tier, but also vertically accountable to their functional line department at the immediate higher level of government, and ultimately central line ministries. The latter are usually responsible for setting national policies and standards. The budget at each level has to be approved by both the People’s Council at that level and the upper level of government. The latter look primarily at budget aggregates rather than detailed allocations but they do have the right to veto or suspend decisions by People’s Councils at the lower tier of government. The National Assembly has ultimate legislative authority over budgetary decisions at all levels of government. Figure 1.2 Nested Budget System in Vietnam Approve State = Prepare National Assembly Central + Local Central authorities National policies Local = Approve Prepare Province Provincial People’s Councils + District + Provincial authorities Commune Local policies Approve District Prepare District People’s Councils = District + District authorities Commune Local policies Approve Prepare District authorities Commune Commune authorities budget 11. The SBL 2002 sets out the budget preparation calendar for the entire State Budget. The annual process begins in May with a Prime Ministerial Directive issued to central and local authorities, providing high level guidance and principle for budget formulation. Vietnam has a dual budgeting system, whereby the Ministry of Finance leads the preparation of the recurrent budget, and the Ministry of Planning leads the preparation of the capital budget. MOF and MPI then issue Guiding Circulars before June 10 with indicative budget estimates for recurrent and capital spending respectively across central and local budgetary units. The latter have to submit their budget proposals for the next year by July 20 for negotiations with MOF and MPI. The negotiated budgets are then reviewed by the government before being submitted to the National Assembly through the Committee on Budgetary and Finance Affairs. The National Assembly in turn agrees the budgets for central authorities and budget aggregates for local authorities by November 15. Local authorities then make their detailed allocations, and report these to the MOF before the end of the fiscal year on December 31. 12. The SBL 2002 establishes the key principles for spending assignments, revenue arrangements, and 28 intergovernmental fiscal transfers. As discussed in chapter 2, most if not all spending responsibilities are shared by all tiers of government as set out in the SBL 2002; de facto detailed assignments are agreed separately through a combination of regulation and traditional practice. Local authorities in Vietnam do not have any policy autonomy over local revenues. Tax bases and rates are determined centrally. The SBL specifies three types of revenue: (i) 100 percent central; (ii) 100 percent local; and (iii) shared. Shared revenue help to close the fiscal gap between 100 percent local revenue and local spending needs. The horizontal fiscal gap, which remains after shared revenue are added to 100 percent local revenue, are filled by a system of balancing fiscal transfers to provincial authorities, which are calculated on the basis of a formula. On top of this there are targeted (or conditional transfers), and local authorities also have access to extra-budgetary financing including limited borrowing. 13. The above provides the broad framework for fiscal relations between different tiers of government. Central authorities use the above framework as a basis to determine fiscal relations with individual provinces. These central-province fiscal relations are asymmetric across provinces in terms of division of spending responsibility; the revenue sharing rates; and the level of fiscal transfers. Since the adoption of the SBL 2002, provincial authorities have been granted a fair degree of autonomy to determine fiscal relationships with districts and communes within their jurisdiction. These in turn can also be asymmetric to reflect the particular circumstances and needs of a local jurisdiction. The details of the fiscal relations within a province are set out in resolutions of Provincial People’s Councils. These resolutions are adopted for five-year Stability Periods, during which key fiscal parameters remain constant. This study overlaps, at least partially, with three Stability Periods: 2004-2006; 2007-2010; and 2011-2015. Though the data only covers 2006-2011. 14. The asymmetry in fiscal relations between central and provincial authorities, and between provincial authorities and lower tiers, means that the system allows for differential treatment of local authorities to take account of their specific circumstances. For example revenue sharing rates are based on local revenue raising capacity; fiscal transfers are based on formulae that account for fiscal gaps and local development conditions; local administrative capacities help to inform decisions on spending assignments; and more recently, the government has introduced higher borrowing thresholds for selected provinces that have more borrowing capacity. Whilst some of these arrangements do involve negotiations, on the whole the system is based on transparent rules and regulations geared at promoting stability, equity and efficiency. 15. There is general acknowledgement that the SBL 2002 has provided a solid framework for public finance management including intergovernmental fiscal relationships. At the same time, the above structure makes for quite a complex picture of fiscal decentralization in Vietnam and has raised several issues in the context of discussions around the revision of the SBL 2002: roles and responsibilities of different tiers of government including budget approval authority; clarity of spending responsibilities and institutional arrangements to ensure accountability of local authorities, including for delivery of national priorities and objectives; actual performance of local authorities both in terms of budget management and service delivery; the effectiveness of local revenue arrangements in helping to meet spending needs, and the potential for increased autonomy for selected provinces; the extent to which provincial authorities are promoting or impeding central government’s redistribution efforts through the transfer system; and the potential for increased debt financing for local authorities. These issues are discussed further in the report outline below, before which it is useful to take stock of existing studies on fiscal decentralization in Vietnam. 16. Existing studies on fiscal decentralization in Vietnam: One of the early investigations into fiscal decentralization issues was conducted by the World Bank (1996) shortly before the adoption of the State Budget Law 1996. The study looked specifically at the quality of service delivery in rural areas, focusing in particular on primary education, basic health care, infrastructure and social relief. At that time, the study found large variations in service quality both across and within what at that time were the 53 provinces in 29 Vietnam. The main reasons for this were lack of funding at local levels, poor administrative capacity and weak targeting of scarce resources. 17. At the time expenditures were centrally mandated with little provincial discretion. Local authorities tried to find flexibilities, but the process lacked transparency, which created mistrust. Based on local level surveys, the analysis showed that provincial and district spending policies were not redistributive vis-à-vis lower orders of government, despite higher levels of central transfers to poorer provinces. In other words, central transfers did help to redistribute resources across provinces, but once it got to the provinces, poorer districts and communes did not necessarily benefit. This phenomenon, as discussed later, is not uncommon in other countries also, and is an important part of the analysis in this report. The same problem arose with centrally targeted transfers designed to address development deficits in the poorest provinces. Despite central efforts to target needy areas, the objectives were often thwarted by decisions of local authorities. 18. The report at the time recommended increased fiscal decentralization and strengthened capacity of the central government to ensure adequate quality and targeting. This would involve clearer delineation of financial and administrative responsibilities of different tiers of government; simplification of revenue sharing and transfer arrangements; greater responsibility in allocation of most public expenditures, except basic services of national importance, which would be guaranteed by central authorities. 19. The SBL 1996 provided the first step to addressing some of the issues highlighted by the World Bank (1996). The share of sub-national spending in general government spending went from 26 percent in 1992 to 43 percent in 1998 (World Bank 2000). The joint Government of Vietnam-World Bank Public Expenditure Review in 2000 reported that the Budget Law specified more clearly the taxes assigned to each level of government, even though rates and bases were central responsiblities. Local authorities were given some power to raise revenue from fees, charges and tolls for provision of selected services. The PER however also found overall per capita spending seemed to be positively correlated to per capita GDP. This trend, as will be discussed in chapter 2, has been reversed in that poorer provinces are now seeing higher levels of per capita spending by local authorities. The PER 2000 also found that capital spending had been more oriented to growth (e.g. large infrastructure) than equity (e.g. schools and hospitals). In light of the above trends, the PER 2000 recommended clearer criteria for determining central transfers, in particular to promote greater equity and redistribution. 20. The SBL 2002 further enhanced the system of intergovernmental fiscal relations. The joint Government of Vietnam-World Bank PER in 2004 and Martinez-Vazquez (2005) provide the most comprehensive reviews of fiscal decentralization reforms adopted in the SBL 2002. These studies indicated that one of the major changes in the SBL 2002, as noted in paragraph 13 above, was the autonomy accorded to provincial authorities to determine fiscal relations within their jurisdictions. Expenditure assignments between central and provincial authorities were not significantly altered, but provinces were now allowed to determine expenditure assignments for district and commune authorities. It also prohibited unfunded mandates, whereby upper tiers of government assign new responsibilities to lower tiers without associated increase in funding. 21. On revenue arrangements, the SBL 2002 lists the types of revenue that fall under each of the revenue categories noted in paragraph 12 above (i.e. central, local and shared). Although revenue sharing rates are not specified in the SBL 2002, the process for determining these (i.e. through estimation of fiscal gap) is provided for. The sharing rates remain constant over the Stability Period, which is also a new concept that was introduced during this period. Another new provision in SBL 2002 is one that allows provinces to revenue that were realized over and above what was budgeted. 22. On fiscal transfers, an important innovation was the adoption of a more transparent formula-based 30 system. The formula uses criteria that targets more explicitly equity objectives. In 2007, the government also introduced a formula to estimate capital spending needs, which were matched against estimated revenues to determine total balancing transfers needs. The revenue sharing arrangement essentially complemented this transfer system by aiming to redistribute revenue collected in richer provinces to poorer ones. Nguyen Hoang and Schroeder (2010) looked at the effectiveness of the shared revenue and balancing transfers system in generating across-province fiscal equity and found that on the whole these were more equalizing as they reduced variation in revenue per capita across provinces. 23. A number of issues however were highlighted in Martinez-Vazquez (2005), which are investigated further in this report. The first is the lack of clarity over spending assignments. The SBL 2002 de jure assigns the same functions to all tiers of government (i.e. all functions are shared), which can create overlap and loss of accountability. On revenue arrangements, it was suggested that the institutional arrangements at the time could lead to several perverse incentives. For example, the ability to retain over collected revenues for own spending can lead to persistent underestimation of revenue in budget, diluting the credibility of the budget and its links to local development plans. A subsequent review by Martinez-Vazquez and Nguyen (2011) looked more specifically at the revenue arrangements and presented options for reform, including possibly more revenue autonomy for selected provinces. However there has not been much development on this front. 24. Another issue raised by Martinez-Vazquez (2005) was how the vertical structure of government and the provincial authorities’ independence over arrangement of sub-provincial fiscal relations may make it more difficult to implement central redistributive or other development objectives. The experience in a number of other countries has shown that intermediate layers of government (i.e. provincial authorities) can capture resources and not delegate to lower tiers of government. This can lead to inequitable outcomes in resourcing across lower tiers of government, and a mismatch with spending responsibilities. Uchimura and Kono (2012) for the first time studied developments below the provincial level and find that in the 2004-2006 period, the revenue sharing arrangements and balancing transfers system have had a positive impact in bridging respectively the vertical and horizontal imbalances across districts. 25. Aside from institutional reviews of fiscal decentralization in Vietnam, some studies have also looked more closely at possible linkages between fiscal decentralization and local development outcomes. A paper by Nguyen (2008) looked in particular at the links between fiscal decentralization and poverty. It found that despite decentralization of spending responsibilities to provinces, poverty levels had increased. This was linked in part to the fact that provinces had not adopted pro poor allocation norms vis-à-vis its districts. This is directly related to the point noted above by Martinez-Vazquez (2005) and Uchimura and Kono (2012) on whether or not provincial autonomy makes it difficult to implement central redistributive policies. 26. Another study by Bjornestad (2009) looks at fiscal decentralization in Vietnam and pro poor outcomes, by analyzing linkages between local spending and levels of poverty. It concludes that the system of fiscal transfers has contributed to more public resources in poorer provinces, but that for the outcomes to be more pro poor, there should be greater participation and accountability in local development. Malesky, Nguyen and Tran (2013) compared service delivery outcomes between provinces where the government started piloting the abolishment of District People’s Councils in 2009 versus other provinces. They find that the ‘recentralization’ of resource allocation decisions to Provincial People’s Councils in the pilot provinces has contributed significantly to improved public service delivery in areas important to central policy-makers, especially in transportation, healthcare, and communications. Kono and Uchimura (2012) on the other hand find positive impact of fiscal decentralization on income inequality and selected health outcomes. 27. Fiscal decentralization and the macroeconomic context in Vietnam: The global economic crisis in 2008 had an important impact on both central and local government finances in Vietnam. As illustrated in chapter 31 4, growth in central government receipts and decentralized revenue (i.e. shared revenue and 100 percent local revenue) declined quite significantly in 2008-2009 due to the impacts of the crisis on economic growth. To offset this, the government implemented a fiscal stimulus package, which increased both recurrent outlays on social protection programs and debt financed public investment projects. The overall budget deficit went from under 2 percent of GDP in 2008 to over 6 percent of GDP in 2009. Although the economy began to recover as a result, a delayed withdrawal of the fiscal stimulus led to economic overheating. By the Summer of 2011, inflation had reached over 20 percent; there was extreme volatility in the foreign exchange market; external reserves were rapidly dwindling; and vulnerabilities from state enterprises and the banking sector began to rise. 28. Despite the relatively large share of spending managed by local authorities, central government has been able to implement counter-cyclical policies for macroeconomic stabilization. Part of this is due to the fact that most of general government revenue accrues to the center, which in turn is redistributed through transfers to local authorities. This enables the central government to maintain a fair degree of control over aggregate fiscal policies. Other features of the intergovernmental fiscal system in Vietnam, which have helped maintain aggregate fiscal discipline are limits on local revenue autonomy (chapter 4) and limits on local borrowing (chapter 6). Although there is some local borrowing and establishment of extra budgetary funds (chapter 6), they are still not a major source of general or local government financing. 29. The central government in recent years has also demonstrated its ability to establish stricter control over local spending growth to maintain stability. When the economy started overheating in 2011, policies were adopted to establish greater control over the capital investments, more than 70 percent of which are managed by local authorities. Oversight from the central Ministries of Finance and Planning and Investment on local capital spending budgets was strengthened through the issuance of Directive 1792 in October 2011. This led to stricter criteria for approving capital investment projects, which led to a fairly sharp drop in capital spending as a share of GDP over the next two years. Lack of control over capital spending decisions in previous years led to a build-up of arrears to construction companies (chapter 6). Again the central government imposed requirements on local authorities to clear these arrears as a matter of priority before any new capital spending could be approved. 30. Scope of study and report outline: This report builds on the above body of research and more recent issues highlighted by stakeholders during discussions on the revision of the SBL 2002. It covers the following five pillars of fiscal decentralization: (i) expenditure decentralization and accountability; (ii) expenditure performance; (iii) revenue arrangements; (iv) intergovernmental fiscal transfers; and (v) local borrowing. 31. Expenditure decentralization and accountability: The second chapter looks at trends in the decentralization of spending responsibilities; the institutional arrangements on allocation of those responsibilities across different tiers of government; and budget management practices linked to local level accountability for expenditure decisions. The objectives are to assess the extent to which public spending decisions have been brought closer to the people, and whether the institutional arrangements around spending assignments and accountability are conducive to promoting allocative and productive efficiency. It uses available data to take stock of the importance of local authorities’ spending in the local economy and how much of this spending has been delegated down to below provincial level. It then looks more closely at the clarity of expenditure assignments at different tiers of government and the basis on which these are determined, highlighted in earlier studies as a critical area of weakness (Martinez-Vazquez 2005), which in turn can negatively affect efficiency. The final part looks at how budget management practices impact on local accountability through legislative oversight, links between planning and budgeting, fiscal rules, budget 32 transparency and participation. 32. Expenditure performance of local authorities: The third chapter builds on the above institutional review of expenditure decentralization and accountability to look more directly at selected dimensions of expenditure performance across provinces. The objectives are to determine if specific spending management practices are impacting on fiscal discipline and spending efficiency to identify reforms that would help institute harder budget constraints. The first part looks at the spending performance in terms of quality of budget outcomes, in particular the maintenance of fiscal discipline through hard budget constraints, highlighted in previous studies as an issue (GOV 2013, WB 2013). The latter is a necessary, though not sufficient, condition for credible and transparent public spending plans, linking those spending plans with local Socio-Economic Development Plans, and ultimately the efficiency of spending. The second part looks more closely at productive spending efficiency across provinces. Using available sector performance and spending data across all provinces, the chapter derives Public Sector Performance and Public Sector Efficiency Scores to look at how provinces perform against each other on these dimensions. The final part looks at what might explain the differences in performance and implications for policy reforms, in particular to promote harder budget constraints. 33. Local revenue arrangements: The fourth chapter looks at the extent to which the above trends in spending decentralization have been complemented by matching levels of and responsibility over local revenues. The objectives are to assess whether the current revenue arrangements are impacting on the revenue performance of local authorities, and how these could be addressed through reforms of the revenue sharing arrangement and the potential for more revenue autonomy in selected provinces. It takes stock of current institutional arrangements for local revenues and its different components. It analyzes recent trends and developments in local revenues and how these compare to trends in other countries. The second part of the chapter takes a similar approach to the spending efficiency analysis to assess the relative revenue performance across provinces. It tries to estimate how much revenue a province could potentially raise and compares this to actual collections to derive an indicator of revenue effort. Based on this, it explores possible reforms to the current revenue sharing arrangement to tackle any potential disincentives to revenue collection. It also explores the possibility of increasing revenue autonomy for selected richer provinces that have high economic growth potential. 34. Intergovernmental fiscal transfers: The fifth chapter looks at how effectively the fiscal gap created by the spending needs of local authorities on the one hand, and the local revenues available to them on the other, is being filled by the current system of intergovernmental fiscal transfers. The objectives are to assess the extent to which balancing transfers are achieving their aim to equalize resource availability not just across provinces but also across districts, and recommend opportunities for reform. Another objective of this chapter is to identify ways to reform the system of targeted transfers to more effectively deliver on their intended targets. The chapter looks at how effectively the revenue sharing arrangements at both provincial and district level have helped to close vertical imbalances, the institutional arrangements for determining fiscal transfers, and how effective the latter have been at addressing horizontal imbalances. It then focuses on the extent to which the system has promoted, or hindered, equity both across but also within provinces. This is an issue that in the past was highlighted as a potential concern, but with limited analysis in the past. The final section looks at the institutional arrangements for National Target Programs and draws on international experience with sub-national conditional grant programs to provide ideas for reform. 35. Local borrowing: The sixth and final chapter investigates local borrowing, which has emerged as an important topic particularly for provinces that are not able to satisfy their capital spending needs through the existing local revenue and transfer arrangements. As discussed in chapters 4 and 5, the current system is geared to redistributing locally collected revenues, the bulk of which are contributed by those provinces 33 that point to infrastructure financing deficits. The objectives are to assess whether selected provinces could potentially borrow more to address this deficit, and highlight reforms to the existing institutional framework to enable this. The chapter analyzes the existing levels and types of local borrowing, and the legal and institutional framework underpinning these. It undertakes an illustrative debt sustainability analysis of Ho Chi Minh City to assess borrowing capacity under different scenarios. It uses this to discuss possible institutional reforms relating to fiscal rules, transparency in local debt monitoring and reporting, and management of sub-national fiscal risks. 36. Study approach and methodology: This report aims to provide new analysis and perspectives on the above areas through extensive study of available data, consultations with central and local authorities, and reviews of laws, regulations and policies at central and local level. On data, the study undertook a major effort to compile publicly available fiscal information reported by local authorities to the Ministry of Finance between 2006 and 2011. Between November 2013 and February 2014, local spending, revenue, and fiscal transfers’ information was manually entered into a single dataset by teams of data entry specialists. The information was collected from over 5,300 local budget and expenditure tables covering the period 2006- 2011 recorded in State Budget Reports and Final Account Reports. This process highlighted large gaps in reporting by local authorities and inconsistencies in classification of fiscal information. Between March and April 2014, the dataset went through extensive review and quality check before compilation of key indicators and data analysis. 37. In addition to the publicly available information, the Ministry of Finance also shared selected sector level data across provinces over the 2006-2011 period. This helped in the review of sector decentralization and spending trends, and in the compilation of Public Sector Performance and Efficiency scores across provinces. The report accessed a wide range of data on national accounts, social sector, poverty, population, public administration and infrastructure reported by local authorities through the General Statistics Office of Vietnam to analyze the linkages between spending and service delivery performance. This was complemented by valuable information from the Provincial Competitiveness Index and the Public Administration Performance Index surveys. 38. The study team has conducted extensive consultations with central and local authorities. At central level, the team has held several rounds of discussions with the Ministry of Finance and members of the National Assembly. It participated in several workshops on the revision of the State Budget Law, organized by the UNDP Project with the National Assembly Committee on Financial and Budgetary Affairs, the World Bank and others. At local level, questionnaires were sent out to local authorities soliciting their feedback on the current system of intergovernmental fiscal transfers, and their recommendations for reform. This was followed by interviews with over 20 provincial authorities (People’s Councils, People’s Committees, and line departments at both provincial and district level) in the period between December 2013 and February 2014. Questionnaire and interview responses in addition to a wealth of information collected by the UNDP-CFBA project during consultations with local authorities on the amendment of the SBL 2002, have played a central role in informing this report. 39. Finally, the study team has also conducted detailed reviews of different laws, regulations, and policy papers at both central and local level. A detailed review of the SBL 2002, including provisions on intergovernmental fiscal transfers, was presented to the Ministry of Finance, the Ministry of Planning and Investment and the CFBA in February 2014. Central policy documents on the process and system of balancing and targeted transfers have been reviewed in detail. Local level resolutions on revenue and spending assignments, and intergovernmental fiscal transfers, have been collected from and reviewed across nearly 30 provinces. EXPENDITURE DECENTRALIZATION AND 34 ACCOUNTABILITY 02 Key issues: local authorities in Vietnam have historically accounted for a significant share of total public sector spending, which has increased further since the adoption of the SBL 2002. Feedback from implementation of the SBL 2002 and earlier studies have highlighted that transparency and accountability of local budgets have not kept pace with the rising delegation of spending responsibility. Objectives: the objectives of this chapter are to assess the extent to which public spending decisions have been decentralized in Vietnam, and options for strengthening the institutional arrangements around spending assignments and accountability to promote allocative and productive efficiency. Key findings: local authorities in Vietnam represent a large share of the local economy, particularly in poorer provinces. The latter have also seen higher levels of per capita spending compared to richer provinces, thanks to the redistributive nature of fiscal transfers in Vietnam. Local authorities in Vietnam are responsible for around 55 percent of general government spending. They account for over 75 percent of total capital spending, and the majority of spending in key social service delivery areas such as education (90 percent), economic services (80 percent) and a growing share of health (from 72 percent of recurrent spending in 2006 to 88 percent in 2011). Decentralization within provinces, down to district authorities has also increased in the 2006-2011 period. In half of all observations, district spending represents more than 45 percent of total local spending. District authorities account for the majority of recurrent spending in both health and education within provinces, though decentralization of capital spending has been less pronounced. An institutional review of expenditure assignments across different tiers of government suggests that the current arrangements would benefit from some specificity, including on: exclusive spending mandates of central government, exclusive responsibilities of different tiers of government within one jurisdiction; and exclusive responsibilities of local authorities. On budgeting and accountability, the chapter finds that the compressed budget calendar and current appropriation structure warrant review to strengthen legislative oversight. At the same time, there is considerable scope for improving budget transparency and participation through clearer communication of budget policies, and publication of the draft budget proposals. There is currently limited participation in budget preparation in Vietnam, when evidence from elsewhere, and partly also from Vietnam shows better outcomes and satisfaction with service delivery when participation is promoted. Recommendations: (i) more explicit assignments on service delivery for different tiers of government within a jurisdiction; (ii) decompression of the budget preparation calendar; (iii) greater autonomy over budget approval authority of people’s councils and clearer provisions on budget appropriations; (iv) elimination of minimum allocation requirements for sectors; and (v) increased transparency and participation in the budget process. CHAPTER 2: EXPENDITURE DECENTRALIZATION AND ACCOUNTABILITY 35 1. Background: Local authorities in Vietnam have historically accounted for a significant share of total public sector spending. This has increased further since the 2002 revision of the State Budget Law (SBL 2002). The latter enhanced local executive and legislative authorities’ decision making powers over resource allocation within their provinces. The literature on fiscal decentralization argues that assigning greater spending and policy responsibilities to local authorities should improve public sector efficiency and development outcomes. Local authorities have better information on local conditions and preferences, which in theory makes them better placed to allocate resources to their highest valued use. 2. This chapter looks at the level of spending decentralization in Vietnam and institutional factors that might influence how increased local spending responsibility can affect development outcomes. The first part analyzes spending decentralization trends in Vietnam based on available data from 2006 to 2011. The second part reviews the existing institutional arrangements in law and in practice for deciding on expenditure assignments. Based on a review of existing legislation and regulations, and interviews with local authorities, this part makes recommendations on ways to further clarify spending assignments. The third part briefly looks at institutional arrangements for budgeting and accountability of local expenditure. It reviews local authorities’ autonomy over resource allocation decisions. It also looks at transparency of and participation in local budgets to promote allocative efficiency. EXPENDITURE DECENTRALIZATION TRENDS IN VIETNAM 3. Local government spending in the local economy: Local government spending constitutes an important share of provincial economies in Vietnam. As illustrated in figure 2.1, between 2006 and 2011, in half of all the observations, local expenditure as a share of local GDP exceeded 30 percent, which is above the national average over the same period. This excludes spending by central authorities in the provinces. The overall distribution in the share of local spending in the local economy has remained fairly constant in the 2006-2011 period (figure 2.2). Figure 2.1 Local government spending Figure 2.2 Annual local government spending in the local economy in the local economy -1SD Median +1SD .005 .01 .015 .02 .025 .03 .035 .04 2006 2007 2008 2009 2010 2011 .03 Density Density .02 .01 0 0 0 20 40 60 80 100 120 140 160 0 20 40 60 80 100 120 140 160 Loca expenditures as % of local GDP Loca spending/local GDP Source: Staff estimates based on published state budget data 4. Local governments in Vietnam in 2006-2011 have played a bigger role in poorer provinces that are not 36 densely populated. This is illustrated in figure 2.3, which shows a strong negative correlation between local spending to local GDP on the one hand and per capita GDP on the other. This is to be expected because local governments would play a bigger role in providing public services in poorer provinces. Moreover, the share of non-state activities in wealthier provinces is also expected to be higher. Figure 2.4 shows that there is also a strong negative relationship between the size of local government (local spending/local GDP) on the one hand and population density on the other. This is also expected due to the high costs of providing public services in sparsely populated areas (which is reflected in spending estimation norms as discussed in chapter 5); and the potential for economies of scale in more densely populated areas. Figure 2.3 Local government spending and per Figure 2.4 Local government spending and capita GDP population density 40 60 80 100 120 140 160 80 100 120 140 160 Local spending/local GDP Local spending/local GDP Fitted values Fitted values Corr=-0.624 (p<0.01) 60 Corr=-0.476 (p<0.01) 40 20 20 0 -20 0 8 8.5 9 9.5 10 10.5 11 11.5 12 11.5 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 Logrithm of per capita GDP Logrithm of population density Source: Staff estimates based on published state budget data 5. Distribution of spending per capita across provinces: Figure 2.5 shows that in at least half of all the observations between 2006 and 2011, local governments spent more than VND 5 million on an annual basis, with some spending over VND 10 million per capita per year. Every year the distribution of per capita spending has gradually become more widely distributed (figure 2.6). The median level of per capita spending has of course increased between 2006 and 2011 from VND 2 million to VND 6 million. But what is interesting is that some local authorities over time have started to spend much more on a per capita basis than the median value. Figure 2.5 Local government per capita Figure 2.6 Annual local government per capita spending (06-11) spending (06-11) -1SD Median +1SD 1.5e-04 2006 2007 2008 .0002 .0003 .0004 2009 2010 2011 1.0e-04 Density Density 5.0e-05 .0001 0 0 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 Local per capita spending at 2011 price Local per capita spending Source: Staff estimates based on published state budget data 6. The gradually widening dispersion in per capita spending seems to be driven by higher per capita spending over time by local authorities in poorer provinces. Figure 2.7 shows that local authorities’ per 37 capita spending pre central transfers (2006-2011 average) is relatively low in the poorest provinces located in the Northern Mountainous Region and Central Highlands, and also in the Northern Central Coastal Region and the Mekong River Delta. But once central transfers are added, it is clear that the poorest provinces are spending more on a per capita basis than wealthier provinces. In fact there is a strong positive association between per capita spending and size of transfers, suggesting successful redistribution of resources by central government. The differential in per capita spending between poorer and wealthier provinces will be even greater when central per capita spending in provinces is added because overall central spending in poorer provinces will likely be higher than in richer provinces. Figure 2.7 Per capita spending pre central transfers Figure 2.8 Per capita spending post central transfer 7. Further analysis of factors that might explain the level of local per capita spending reaffirms the importance 38 of central government’s redistributive role. Table 2.1 below summarizes the estimation results of local per capita spending regressed against population density and the share of central balancing transfer over local core expenditure. It shows that central balancing transfers as a share local core spending has a positive and statistically significant impact on per capita spending. It also shows a strong negative relationship between per capita spending and population density. This is likely because of high incidence of poverty in lower density, mountainous areas of Vietnam, where service delivery costs are higher. This will be an important issue to take into account when looking at efficiency of spending in the next chapter. Table 2.1: Explaining local per capita spending levels Local per capita spending (1) (2) Population density (persons/sq km) -1.591*** -1.683*** (0.577) (0.501) Central balancing transfers/local core expenditure 32.81*** 24.49** (12.32) (11.34) Year 2007 375.5 (565.8) Year 2008 838.6 (551.7) Year 2009 1775.6*** (574.9) Year 2010 2761.1*** (575.0) Year 2011 4215.8*** (557.2) Logarithm of population density Constant 3840.0*** 2435.8*** (533.2) (589.4) Observations 244 244 R2 0.080 0.329 Standard errors in parentheses; * p<0.1, ** p<0.05, *** p<0.01 8. Decentralization of spending from central to local authorities: Local authorities in Vietnam are responsible for over half of total government spending. This is expenditure that is executed directly by local authorities (provincial, district and commune) out of revenues raised and retained within their jurisdiction and out of transfers from upper orders of government. Local authorities have therefore played a significant role in service delivery for poverty reduction in the country. Figure 2.9 below shows that spending by local authorities as a share of all government spending, excluding central transfers to provinces, went from 47 percent to 54 percent over the 2007- 2010 Stability Period. This share had increased from around 26 percent in 1992 and 43 percent in 1998 following the 1996 State Budget Law and subsequent 2002 revision. 9. By 2011, local authorities were responsible for executing nearly 77 percent of overall capital spending (67 percent on average between 2006-2011), with central government’s share declining to less than a quarter (figure 2.9). 10. The decentralization of capital spending started in 2004, and accelerated over the 2006-2010 Stability Period. There was a spike in local capital spending in the 2009-2011 period, as local authorities implemented fiscal 39 stimulus programs in response to the effects of the global financial crisis. Nonetheless the rapid rise in local capital spending has raised concerns over efficiency of public investment due to lack of capacity, coordination and planning. This in turn has led to significant capital spending consolidation over the past three years. Figure 2.9: Expenditure by local authorities as a share of total expenditure 2006-2011 (%) 100% 56% 54% 54% 80% 53% 64% 63% 62% 65% 73% 77% 52% 51% 60% 50% 40% 48% 48% 47% 47% 46% 20% 51% 50% 51% 53% 53% 55% 44% 0% 42% 2006 2007 2008 2009 2010 2011 Local recurrent/Total recurrent Local capital/Total capital Local/Total Source: Staff estimates based on published state budget data 11. Share of functional spending by central and local authorities: Local authorities’ share of overall recurrent spending has also increased, even if a bit more slowly than the share of capital spending. The central government in 2011 was responsible for around 45 percent of recurrent expenditure (figure 2.9). Around 27 percent of this on average between 2006 and 2010 was for social security (figure 2.10). Although the central government’s share of social security has declined from 87 percent in 2006, it still remains high at around 71 percent in 2011. This is not unusual compared to other developing countries where funding of social assistance remains primarily a central government responsibility because of national social development objectives, economies of scale inherent in social security programs, and their positive externalities across jurisdictions. Local authorities are usually responsible for providing beneficiary information to enable better targeting of these social assistance programs. Figure 2.10: Central share of social security Figure 2.11: Local recurrent spending in selected spending functions (% of total) 35% 100% 95 30% 91 90 90 90 80% 88 88 87% 86% 88% 86% 87 86 25% 85 79% 20% 71% 60% 80 80 80 79 78 79 15% 30% 29% 40% 75 27% 28% 25% 10% 23% 72 72 71 70 70 70 69 20% 68 5% 66 66 65 64 64 0% 0% 62 60 2006 2007 2008 2009 2010 2011 2006 2007 2008 2009 2010 2011 Social security/Central Recurrent (%) Administration Economic Central government share of social security (%) Education Health Source: Staff estimates based on published state budget data 12. Other than social security, local authorities’ share of total recurrent spending in selected government 40 functions has been high and increased further between 2006 and 2011 (figure 2.11). In education, for example, local authorities are responsible for 91 percent of total recurrent spending in education in the country. In health, the share has increased from 72 percent to 88 percent in 2006-2011. Central ministries and agencies play an important role in setting and monitoring national standards. But local authorities are responsible for allocation of the above resources across and within sectors in their respective jurisdictions. 13. Discretionary resources available to local authorities: Trends in funding sources can provide a rough idea over the level of autonomy that local authorities have over allocation of the above resources. Figure 2.12 shows that around 60 percent of local spending is funded out of decentralized revenues.2 As shown later, there is significant variation in these ratios across provinces given different fiscal capacities. But in general, the chart below illustrates that on average, if taken with balancing transfers and other resources (mostly off- budget), local authorities have a fair amount of discretionary resources at their disposal. It is important to note, however, that Vietnam does not have fiscal autonomy in terms of setting of tax rates and bases. 14. Balancing transfers (also referred to as unconditional transfers) are not earmarked and therefore provide autonomy over resource allocation. Target (or conditional) transfers are not insignificant, averaging 25 percent of local spending. Although the central government sets targets and objectives for these funds, local authorities still have responsibility over implementation. In reality of course, the true level of autonomy in policy-making, designing of programs, and allocation of resources across and within sectors will depend on institutional rules and practices reviewed in the next section. Figure 2.12: Funding source for local spending (%) 15. International comparisons in spending decentralization: Compared to other countries, local authorities’ share of government spending in Vietnam is high. International comparisons are difficult and should be treated with caution because of the problems in getting fully consistent and recent sets of comparable indicators.3 Figure 2.13 shows that sub-national expenditure as a share of total expenditure in 2 Decentralized revenue as discussed further in chapter 4 include two components: (i) revenue types that are 100 percent retained by local authorities; and (ii) revenue types that are collected at the local level but need to be shared with the central authorities. 3 This report uses the same definitions as in the World Bank Fiscal Decentralization Indicators (http://www1.worldbank.org/publicsector/ decentralization/fiscalindicators.htm), which also provide international comparators (1972-2000) based on data drawn from the IMF Government Finance Statistics database. Vietnam (averaged over 2006-2011) is significantly higher than the average for other countries (dotted line, showing values averaged over 1990-2000). The chart shows a direct relationship between expenditure and 41 revenue decentralization across all countries. However as noted above and discussed in chapter 4, there is very limited revenue decentralization in Vietnam in terms of policy autonomy even though higher levels of local spending in Vietnam is matched by higher levels of local revenue. Figure 2.13: Expenditure and revenue Figure 2.14: Expenditure decentralization and decentralization dependence on transfers Sub-national Expenditures (% of total expenditures) Sub-national Expenditures (% of total expenditures) 70 80 70 80 60 60 Vietnam Vietnam 50 50 40 40 30 30 10 20 10 20 0 0 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 90 100 Sub-national Revenues (% of total revenues) Transfers to sub-national(% of total sub-nation revenues & grants) Source: WB fiscal decentralization indicators and staff estimates 16. Compared to other countries, local authorities in Vietnam are less dependent on central transfers. This is illustrated in figure 2.14, where the dotted lines show the average level of spending decentralization and the average level of dependence on central transfers as a share of local receipts across all countries in the sample. Countries in the top left hand quadrant like Vietnam exhibit higher than average spending decentralization but lower than average dependence on transfers. As noted above, in Vietnam unconditional transfers do not impose restrictions on resource allocations. But overall, with higher of local spending, Vietnam has higher discretionary resources compared to other countries. These trends point to the need for strong local accountability discussed in the next section. 17. Data on sector composition of local expenditure shows that local authorities in Vietnam allocate a larger than average share of their resources on education and health compared to other countries. Vietnam in figures 2.15 and 2.16 falls in the top right quadrant, with high spending decentralization and high share of local spending on education and health. One point to note is that the data for Vietnam shows local recurrent spending in education as a share of total local recurrent spending rather than total spending on education and health. Data on total spending within a function is difficult to obtain publicly because capital spending broken down by government functions is not consistently reported. The main point, however, is that with greater spending decentralization, local authorities spend a large share of their budgets on social sectors. This further highlights the significant role played by local authorities in poverty and social development outcomes. Figure 2.15: Expenditure decentralization and local Figure 2.16: Expenditure decentralization and local 42 education share health share Sub-national Expenditures (% of total expenditures) Sub-national Expenditures (% of total expenditures) 70 80 70 80 60 60 Vietnam Vietnam 50 50 40 40 30 30 10 20 10 20 0 0 0 10 20 30 40 50 60 0 5 10 15 20 25 30 35 40 45 Education (% of total sub-national expenditures) Health (% of total sub-national expenditures) Source: WB fiscal decentralization indicators and staff estimates 18. Spending decentralization across provinces: Assessing the level of spending decentralization across provinces would require data on total spending in the province, which unfortunately was not available for this study i.e. spending by local authorities divided by spending by both the central and local authorities within the province. Although this information is accounted for and recorded in the government financial management information systems, it is not reported in the annual State Budget report or Final Accounts. This makes it difficult to have a comprehensive picture of spending in any one jurisdiction (see part 3 of this chapter). 19. Spending decentralization within provinces: Vietnam exhibits fairly high degree of spending decentralization within provinces, which seems to have increased over subsequent Stability Periods. Figure 2.17 shows that for half of all the observations between 2006 and 2011, district spending constituted more than 45 percent of total local spending. Figure 2.18 also shows that the distribution of district spending as a share of local spending has consistently shifted right from 2006 (median value of around 35 percent) to between 2007 and 2010 (median value of around 43 percent) and then to the current stability period starting in 2011 (median value of around 45 percent). Figure 2.17: District expenditure/local Figure 2.18: Annual District expenditure/local expenditure (06-11) expenditure (06-11) -1SD Median +1SD .005 .01 .015 .02 .025 .03 .035 .04 .03 Density Density .02 .01 0 0 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 District expenditure/local expenditure District expenditure/local expenditure Source: WB fiscal decentralization indicators and staff estimates 20. There is less capital spending decentralization within provinces than between center and province. For a sample of district level capital spending data, figure 2.19 shows that for half of all observations 43 between 2006 and 2011, district capital spending was around 30 percent of local capital spending. Figure 2.20 shows that although district capital spending in local capital spending had increased over the 2007- 2010 Stability Period, this seems to have receded again in 2011. This is most likely related to the capital consolidation measures that were introduced in 2011 whereby central authorities instituted more discipline on capital spending because of efficiency concerns. Figure 2.19: District capital/local capital (06-11) Figure 2.20: Annual District capital/local capital (06-11) -1SD Median +1SD .025 2006 2007 2008 2009 2010 2011 .025 .02 .02 .015 .015 Density Density .01 .01 .005 .005 0 0 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 District capital spending sa percent of local capital spending District capital spending sa percent of local capital spending Source: Staff estimates based on published State Budget data 21. District authorities have been responsible for most of the recurrent spending in both education and health within most provinces in the 2006-2011 period. Figure 2.21 shows that in half of all observations, district authorities were responsible for more than 75 percent of local recurrent spending on education. Similarly figure 2.22 shows that in half of all observations, district authorities were responsible for more than 60 percent of local recurrent spending on health. Districts’ share of local recurrent spending in education has remained relatively constant across Stability Periods. But in the health sector, figure 2.24 shows that there has been a clear shift between 2006 and 2011 towards greater decentralization of recurrent spending down to the district level. This is in line with the growing share of local authorities’ health spending in overall health spending illustrated in figure 2.16 above. By the end of 2011, half of all observations show that districts were responsible for more than 75 percent local recurrent spending in health. Figure 2.21: District education recurrent/local recurrent Figure 2.22: District health recurrent/local recurrent 44 -1SD Median +1SD -1SD Median +1SD .025 .04 .02 .03 .015 Density Density .02 .01 .01 .005 0 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 District education expenditure/local education expenditure District health expenditure/local health expenditure Figure 2.23: Annual District education recurrent/local Figure 2.24: Annual District health recurrent/local recurrent recurrent 2006 2007 2008 2006 2007 2008 2009 2010 2011 2009 2010 2011 .04 .03 .03 .02 Density Density .02 .01 .01 0 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 District education expenditure/local education expenditure District health expenditure/local health expenditure Source: Staff estimates based on published State Budget data 22. There are several factors that might explain the level of decentralization within a province. Provincial authorities in interviews have indicated that decisions on delegation of spending responsibilities to district authorities depended a lot on capacity. Analysis on this was conducted by Uchimura and Kono (2012) who found that the shares of rural population and the level of local capacity in provinces are important factors in explaining the level of district expenditure in local expenditure. A similar relationship was estimated for the available data from 2006-2011, eYit= α +βXit +γTdt+ μit Where Y is the share of district spending in province i at time t; and X are control variables including per capita GDP, share of rural population, and the share of employees in administrative and non-profit organizations that have Bachelors’ or Engineering Degrees as a proxy for capacity in the province. The new estimation using more recent data, which is summarized in Table 2.2 below, also shows that the share of rural population and local capacity are positive and statistically significant determinants of spending decentralization down to districts. This highlights efforts to bring spending closer to people, and reaffirms provincial authorities’ feedback that the degree of decentralization is also a function of local capacity. Table 2.2: Explaining decentralization of spending to district authorities 45 District expenditure/ local (1) (2) (3) (4) (5) expenditure -0.0005*** Per capita gdp (0.0001) -4.602*** Log of per capita gdp (1.462) % of bachelor, engineer in 0.460*** 0.226* administrative and non-profit (0.130) (0.130) organizations Rural population share 0.289*** 0.263*** (0.0449) (0.0475) Year 2007 2.360 2.812 1.869 1.733 1.828 (2.588) (2.602) (2.663) (2.502) (2.530) Year 2008 3.023 2.974 1.132 1.074 1.206 (2.648) (2.685) (2.690) (2.526) (2.556) Year 2009 2.698 2.136 -0.212 -0.0486 0.120 (2.798) (2.845) (2.784) (2.629) (2.645) Year 2010 5.833** 4.795 1.415 1.321 1.663 (2.893) (2.927) (2.752) (2.598) (2.615) Year 2011 10.45*** 8.144*** 3.285 3.840 3.963 (2.971) (2.918) (2.573) (2.431) (2.448) Constant 45.76*** 83.18*** 12.01 19.41*** 6.788 (2.092) (13.45) (8.486) (3.883) (8.117) Observations 284 290 281 284 281 R2 0.079 0.044 0.052 0.137 0.148 Standard errors in parentheses; * p<0.1, ** p<0.05, *** p<0.01 23. Spending decentralization to districts is driven more by the level of provincial transfers than by decentralized revenue collected at district level. Figure 2.25 shows positive and significant correlation between provincial share of local spending and decentralized receipts at provincial level. Figure 2.26 on the hand shows a slight negative and significant relationship between district share of local spending and decentralized receipts at district level. This suggests that district authorities rely more on transfers to meet their expenditure responsibilities because of relatively lower levels of decentralized revenue at their disposal. 24. This is likely due to higher tax administration capacity and collection responsibility at provincial level compared to district level. A number of districts also indicated that tax payments of large enterprises accrue to provincial rather than district authorities. Whilst transfers do not necessarily mean less discretionary resources (and therefore less autonomy over spending decisions), the fiscal decentralization literature also argues that efficiency gains from spending decentralization are higher when it is matched with higher levels of decentralized revenue (Oates, 1972). On the other hand, reliance on mostly local financing would lead to differential quality and access as well as fiscally induced migration of the needy and the poor to richer local jurisdictions. Greater local tax burden could lead to business moving out at the same time leading to an erosion of tax base and the capacity to provide better social services. Fiscal transfers combined with local provision therefore could overcome such externalities. Figure 2.25: Decentralized revenue for provincial Figure 2.26: Decentralized revenue and district 46 spending spending 70 80 90 100 70 80 Corr = 0.137 (p<0.05) Corr = 0.240 (p<0.01) 40 50 60 40 50 60 10 20 30 20 30 0 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 Province decentralized revenue as % of provincial receipts District decentralized revenue as % of district receipts Province expenditures as % of local expenditures Province expenditures as % of local expenditures Fitted values Fitted values Source: Staff estimates based on published State Budget data 25. An important point from this analysis is that provincial authorities in Vietnam are enabling decentralization of spending responsibilities through the transfer system. Studies have shown that the existence of intermediate provincial authorities can lead to less decentralization within a province. In China, for example, Wu and Wang (2013) find that transfer dependency of provinces negatively impacts on decentralization within the province. Provincial authorities retain central transfers at their level rather than channeling it down for spending by lower tiers of government. Similar results are found by Freinkman and Plekhanov (2009) in Russia, where higher transfer dependence leads to lower decentralization within regions. In Vietnam, however, it is clear that provincial and district authorities have gradually accounted for a growing share of total general and local government spending respectively. This trend points to an equalizing effect of the transfer system, which is discussed further in chapter 5. INSTITUTIONAL ARRANGEMENTS FOR EXPENDITURE ASSIGNMENTS 26. Trends in spending and discretionary resources across different tiers of government show that Vietnam is gradually bringing decisions on public spending closer to people. But local authorities’ actual autonomy over spending and the impact this has on service delivery will depend on a range of institutional factors. This includes the clarity of expenditure assignments to different tiers of government. Clarity on expenditure assignments, and the basis on which they are determined, are critical to avoid overlaps, which lead to inefficiency. It is also important for ensuring accountability of governments for performance in assigned functions. 27. Principles for determining spending assignments: The fiscal decentralization literature outlines some broad principles for determining spending assignments. The literature notes that central government should have exclusive responsibility in selected areas, including: (i) providing public services that are national in scope (e.g. defense, monetary policy); (ii) correcting inequity across provinces arising out of differential fiscal capacities of provinces; (iii) ensuring overall macroeconomic stability, including through oversight of sub-national fiscal practices that might adversely affect stability; and (iv) regulating service delivery to ensure adequate quality and standards across the country. 28. Beyond these responsibilities, the general principle for expenditure assignments is that public services should be provided “by the jurisdiction having control over the minimum geographic area that would 47 internalize benefits and costs of such provision” (Oates 1972, p.55). The benefits of defense, for example, accrue nationally, and therefore should be provided by central government. Primary education or urban transport on the other hand should be decentralized as benefits are mostly local. When local services benefit citizens from outside the jurisdiction (e.g. due to temporary migration, tourism) higher tiers may need to assume responsibility or provide compensation. The institutional framework should be clear about any exclusive assignments to a particular tier (e.g. defense to central government). In the case of exclusive responsibilities, one order of government is responsible for regulation, financing and service delivery. If on the other hand, there is shared responsibility for a function, it should be clear which tier is responsible for regulation, financing and service delivery. 29. Different tiers can have different responsibilities for a function within a jurisdiction, but they should avoid performing the same task for that function within a jurisdiction. For example, central, provincial and district authorities should not all be involved in delivering primary schooling in a particular district. This can increase coordination costs and lower efficiency. Instead, for primary schools in a district for example: central government could regulate and oversee standards for primary education, provincial authorities could provide financing, whilst districts are responsible for service delivery. The division of responsibility could differ across different provinces depending on local circumstances. But only one order of government should be explicitly assigned regulation, financing and service delivery responsibility for a function within a jurisdiction. 30. Spending assignments in the State Budget Law 2002 (SBL 2002): Articles 31 and 33 in the SBL 2002 enumerates central and local government spending functions respectively, but the two lists are almost identical, which means that all functions are shared. The few exceptions include “additions to the State reserve fund” and “support for the social insurance fund”, which are exclusive responsibilities of the central budget. On local level assignments the SBL 2002 also states that “there must be tasks of investment in construction of public schools, lighting, water supply and sewage, urban traffic, and other public welfare construction works.”(Article 34.1 (d)).4 The Government’s Decree 60/2003 supporting SBL 2002 provides a longer list of expenditures assigned to central (article 21) and local government (article 24), but no function is exclusively assigned to any particular tier. 31. The two articles on spending assignments in the SBL 2002 could be simplified by stating that all tiers may spend on any functional area but that some functional areas are the exclusive responsibility of central government (e.g. national defense, foreign affairs). For these areas, the central government would be exclusively responsible for regulation, funding and service delivery. Although there are spending areas that are also typically exclusive to lower tiers of government (e.g., refuse collection, urban street lighting), these are usually very detailed and best identified in regulations. If however, the government later decided to assign exclusive responsibilities to local budgets, a revised SBL would need to reflect this e.g. some countries assign different levels of education across government levels: tertiary education with central government, primary/secondary education with lower levels. But this may be too rigid. It is often better to have shared responsibility for certain functions and different expenditure assignments to different provinces depending on local circumstances and capacity. These assignments can evolve over time without requiring budget law or other legal/constitutional amendments. 4 These tasks are assigned exclusively for towns and cities under provincial control in correspondence with regulations in Article 34.1.c where those towns and cities enjoy at least 50 percent of the registration fees, except for land and housing, collected in their jurisdiction 32. Assignment of responsibilities in shared functions: In the case of shared responsibilities for specific 48 functions, as noted above it is important to have clarity over which order of government is responsible for what task within a jurisdiction. Sector legislation helps to provide some clarification on explicit responsibilities of different tiers. The Education Law (2009) Article 51 clarifies selected decision powers for central, provincial and district authorities. According to the Law, the Prime Minister has to approve the establishment of a university, the Minister of Education and Training decides on the establishment of colleges, and the heads of central government agencies can decide on vocational schools. MOET can also authorize creation of technical secondary schools. The Heads of provincial people’s committees have the authority to establish general secondary schools and the heads of the district people’s committee are authorized to approve pre- schools, primary schools, lower secondary schools and semi-boarding private schools. 33. Policy making and regulatory responsibilities in education rests with the central government and is exercised through the Ministry of Education and Training (MOET). All educational institutions require accreditation by MOET. Further MOET is responsible for teacher certification, staffing controls, hiring and firing of teachers and establishing curriculum and prescribing textbooks. The Ministry of Home Affairs decides on teachers’ salaries and supplementary allowances and benefits. In addition to the Education Law (2009), the sector also has Decree No. 115/2010/ND-CP, which provides further clarity on the assignments to Provincial Departments of Education (mainly upper secondary schools and above, vocational training centers) and District Departments of Education (mainly lower secondary schools and below), and also the Commune People’s Committees (mainly community education centers with some sharing of responsibility with districts on lower secondary schools and below).5 34. The Law on the Protection of Public Health (1989) Article 3 states that the central government has the overarching responsibility for setting policies and priorities for health care. To this end, the Ministry of Health is responsible for managing, improving and developing health care systems. The Peoples Councils at province, district and commune levels are responsible for ensuring adequate financing of health care at own levels. In addition, there is inter-ministerial (Ministry of Health and Ministry of Home Affairs) Circular No. 03/2008/TLLT-BYT-BNV (April 25, 2008), which assigns financing and service delivery responsibilities to different tiers of government according to ownership of health facilities. 35. There are however several issues with Circular No.03. Firstly, the inter-ministerial Circular does not have the same legal status as Prime Minister’s Decree 115 in the education sector, which impacts adversely on compliance with the former. In discussions with local authorities, the study as a result found inconsistent implementation of the Circular; a number of Northern provinces reported that district hospitals were funded out of district budgets, whereas Southern provinces indicated that district hospitals were being financed out of provincial budgets. 5 Other agencies such as the General Department of Vocational Training are responsible for establishing vocational colleges; and Provincial People's Committees are responsible for establishing vocational schools within their jurisdiction. The authority for hiring and firing of teachers is decentralized for ministries, local authorities and autonomous universities. The authority to make salary and supplementary allowances and benefits belongs to the Prime Minister with recommendations from Ministry of Home Affairs, Ministry of Education and Training and Ministry of Finance. 36. Explicit assignments between center and province: There is therefore some clarity on responsibility for regulation in health and education, which rests explicitly with the center. But there is scope to clarify 49 the sharing of responsibilities between central and provincial authorities for funding and service delivery within a province. At the moment, because all functions are shared, central and provincial authorities can both be involved in funding and service delivery of a particular function within a province e.g. both central and provincial authorities can fund and provide hospital services in a province. The decision on who funds and provides those services can be based on who owns the facility (i.e. centrally owned hospital will be funded and managed by central government). It is also a function of the decentralization of socio-economic management responsibilities, which will be discussed as part of new legislation that the Ministry of Home Affairs on Local Governments. Being more explicit about central and provincial assignments in shared functions within a province would help reduce ambiguity. 37. Regulations on assignment of responsibilities within provinces: Explicit spending assignments for local authorities in each province are set out in provincial regulations. These have contributed clarity in spending assignments of provincial, district and commune authorities. The empowerment of provincial authorities to determine spending assignments is an important feature of the SBL 2002. Each Provincial People’s Council has issued resolutions accompanied by regulations issued by the Provincial People’s Committees on decentralization of revenues and expenditures for the stability period from 2011-2015. This allows provinces to assign explicit spending responsibilities asymmetrically across districts depending on capacity and other factors. However, as discussed further below, a number of these resolutions also allow all levels of local authorities to provide the same service in a particular jurisdiction. Therefore although there is more clarity on explicit responsibilities between province and district compared to those between center and province, there is still an issue of different orders of government being involved in the same task within a jurisdiction. 38. From legal provisions to actual practice in expenditure assignments: The above points to three issues in the current legal framework on spending assignments. Firstly the lack of enunciation of exclusive responsibilities of central government could (and does as discussed below) unduly shift the burden of some of those responsibilities to lower tiers. Secondly, in the case of shared functions, there is scope to more explicitly assign responsibilities between the center and provinces rather than relying on facility of ownership to determine assignments, which can create ambiguity. Thirdly the current legal framework enables different orders of government to provide a particular service within the same jurisdiction, which can lead to duplication and inefficiency. 39. The study tried to take stock of whether these three issues, and any other factors affecting the clarity of spending assignments, act as institutional constraints to efficient service delivery. To do so requires a review of actual practice in the assignment of responsibilities in different functions. This is difficult because of the large number of local authorities, and because much of the practice is based on tradition and historical norms. An attempt was made through questionnaires sent to provincial authorities followed up by extensive interviews with officials from over twenty local authorities. 40. Sharing responsibilities in areas that should be exclusively assigned to the center: In Vietnam the central government practices exclusive responsibility for external relations, foreign trade and foreign assistance, food safety and drug regulation. Responsibility for all other public services is shared with provincial and local governments. Two interesting cases are national defense and social insurance and protection. As the benefits and costs of these services are national in scope, both in principle and in practice, they should typically be the exclusive responsibilities of central government. 41. In Vietnam, however, provincial and local governments play a role in delivery of these services and also 50 provision of supplementary finances. This is despite the fact that the responsibility for the social insurance system has been assigned exclusively to the central budget (Article 31.2.g of SBL 2002). Provinces and local governments are expected to contribute to the operations of armed and security forces in their jurisdictions. While these contributions are typically very small, they disproportionately affect poorer border provinces (e.g. Tay Ninh and Quang Tri provinces) and sometimes these expenses cannot be anticipated and therefore require diverting local resources from other services. 42. This feedback therefore reaffirms the earlier need to formalize exclusive responsibility of central government in the SBL 2002. Central and local authorities already have a clear idea of what those responsibilities are, but formalizing would avoid some of the above issues faced by local authorities. The guiding principles are that the central government should retain dominant role in policy making, legislation and oversight and financing to meet national objectives and direct delivery of exclusive central functions such as defense and national security, currency and macro stability, and social insurance. Provinces should have a coordinating role for implementation of central policies and inter local functions and below province local governments should be responsible designing and delivering public services to meet national objectives as well as to satisfy local preferences. 43. Determining assignments between the center and province within a province: Despite a lack of clarity on explicit assignments between center and province for selected tasks within a province, the current system has some desirable features. The central assignment to provinces is asymmetric and it recognizes the special character of provinces and provincial cities. Provincial cities and fiscally advantaged provinces enjoy greater fiscal and administrative autonomy. Central line agencies have a more expansive role in fiscally disadvantaged provinces. Within provinces, greater clarity on explicit assignments has not resulted in wide variations in the structure and roles of local governments across provinces. But it has also enabled provinces on average to delineate expenditure responsibilities based upon the fiscal capacity and rural and urban characteristics of local governments as was presented in the analysis earlier in the chapter. 44. Beyond this however, a number of provincial authorities did indicate that more explicit assignments between center and province would help to reduce ambiguity and transaction costs. Aside from fiscal capacity and rural/urban characteristics of local government, as noted above, shared responsibilities are also decided by ownership of public assets. Officials noted that traditions and historical norms within jurisdictions offer additional clarity to officials on who is responsible for what. Despite this, provincial authorities indicated that the current arrangements often require endless meetings and communications to clarify respective roles in areas where there are significant overlap of responsibilities. For example, provinces are often required to maintain central roads and other facilities and receive reimbursements for tasks performed on behalf of the center. But reimbursement is often delayed due to conflicting interpretations of central and provincial shares of costs. These issues can be overcome if, in addition to the explicit assignments of local authorities, there could also be regulations setting out the division of responsibility between center and province. 45. In the education sector, a review of provincial resolutions shows that de jure, different levels of local authorities can be involved in delivering the same services within the same jurisdiction. Table 2.3 below does a comparison of practice in Vietnam versus stylized recommendations in the fiscal decentralization literature. In Vietnam, pre-kindergarten and kindergarten schooling is financed, delivered and managed by provincial, district and local governments. Primary schooling is financed by provincial and district governments but delivered and managed by provincial, district and commune governments. Secondary education is financed, delivered and managed by provincial and district governments. University and college education is financed, delivered and managed by national and provincial governments and by the private 51 sector. Vocational training is financed, delivered and managed by provincial and district governments. Table 2.3: Division of responsibilities across government levels in the education sector Regulation Financing Provision/ownership Management Stylized Vietnam Stylized Vietnam Stylized Vietnam Stylized Vietnam University/college N, p N N,p N,p P,d ,ps N,p, ps B N,p Secondary D N P,d,c P,d D,c P,d, ps B P,d Primary D,c N D,c P,d D,c P,d,c, ps B P,d,c Pre-k and kg C N Ps P,d,c Ps P,d.c, ps Ps D,c Vocational P,d N P, ps P,d Ps P,d, ps B P,d N= national/center, p = provincial, d = district, c = commune, b= board/institutional, ps=private , ngo= civil society/professional association 46. This is further illustrated by a comparative perspective of responsibility for service delivery in education across 18 provinces. Table 2.4 below shows that at the pre-school level all local authorities are involved. The districts play an important role in the provision of primary and secondary education in most provinces whereas provincial and central institutions dominate the university and college education. Vocational and short term training is mostly the responsibility of upper tiers though some provinces did report that there were overlapping responsibilities between provinces and districts. Table 2.4: Responsibility for education service delivery across provinces EDUCATION LEVEL PROVINCES Pre-school గ ᆩ గᆩ గᆩ గᆬᆩ ᆬగᆩ గ గ గᆩ గ ᆬగᆩ గᆩ ᆬగᆩ గᆩ గᆩ గᆩ గᆩ గ గᆩ Primary గ ᆬగ గᆩ గᆬᆩ గ గ గ గ గ గ గ ᆬగ గ గ గ గ గ గᆩ Lower secondary గ ᆬగ గᆩ గᆬᆩ గ గ గ గ ᆬగ గᆩ గ ᆬగ ᆬ గ ᆬగ గ ᆬ గ ᆬగ Higher secondary ᆬ ᆬగ ᆬ ᆬ ᆬగᆩ ᆬ ᆬ ᆬ ᆬగ ᆬ ᆬ ᆬగ ᆩᆬ ᆬ ᆬ ᆬగ ᆬ ᆬ Vocational training ᆬ ᆬగ ᆬగ ᆬగ ᆬ ᆬ ᆬగ ᆬ ᆬ ᆬ ᆬ ᆬగ ᆬ ᆬగ ᆬ ᆬ ᆬ ᆬ University, colleges ᆬ प࣫झ ᆬ ᆬగ ᆬ ᆬ ᆬ ᆬ ᆬ ᆬ ᆬ ᆬ ᆬ ᆬ ᆬ ᆬ ᆬ Short term training ᆬ గ ᆬగ ᆬ ᆬ ᆬగᆩ ᆬగ ᆬ ᆬగ ᆬ గ ᆬగᆩ ᆬ ᆬగ ᆬగ ᆬగ ᆬ గ ᆬగ Commune learning centers ᆩ ᆩ ᆩ గ ᆩ ᆩ Central level ᆬ Provincial level District level ᆩ Commune level గ 47. Earlier analysis in the chapter showed that much of the local service delivery spending in education is in fact delegated to district authorities. Therefore the scope for overlaps in service provision by different tiers within the same jurisdiction may not be that big a problem. If it is indeed the case that sub-provincial authorities have most of the service delivery responsibility for certain types of education, the provincial resolutions should more clearly assign exclusive responsibilities for those types of education. For example, they should state clearly that for at least pre-school and primary education, service delivery responsibility will rest with districts and communes. Higher capacity provinces could do the same for lower secondary. 48. In reality, a number of district authorities did point to a lack of coordination in provision of primary schooling 52 by provincial, district and commune authorities. As noted earlier in the chapter, districts do not have a full picture of spending by provincial authorities within their jurisdictions. This lack of a comprehensive planning view, including objective criteria regarding school access, means that some areas remain underserved. It also means that some schools are not properly maintained as the government that built the facility may fail to provide for adequate funds for proper operation and maintenance. These problems could be avoided if exclusive responsibility for a given task is assigned within a jurisdiction. 49. In the health sector, central government through its Ministry of Health defines health policy, oversees all health and pharmaceutical institutions and dispensaries, and licenses doctors and pharmacists. It approves drugs and sets staffing standards, controls and salary ranges for medical professionals. Financing, provision and management of health research is a central government responsibility. Central and provincial governments finance, provide and manage teaching hospitals, specialized care institutions. Public health and preventive care is financed by central and provincial governments but delivered and managed by ownership by national, provincial and district governments. Table 2.5: Division of responsibilities across government levels in the education sector Regulation Financing Provision/ownership Management Stylized Vietnam Stylized Vietnam Stylized Vietnam Stylized Vietnam Health research N, p, ps N N,p,ps N N,p,ps N B N Teaching hospitals P N N,p N,p N,p, ps N,p B N,p Hospitals P N N,p,d N,p,d N,p,d,ps N,p,d, ps B N,p,d,ps Clinics P N Ps P,d,c Ps P,d,c Ps P,d.c Health centers P N D, ps P,d D,ps P,d,c B P,d,c Specialized care P N N,p,ps N,p P, d N,p B N,p Public health/pc N,p N N,p N,p P,d N,p,d B N,p,d Drugs approval N N Drug procurement N,p,d N,p Licensing pharmacies P N Doctor registration P N Pharmacist reg. P N Doctor salaries B N Staffing B N N= national/center, p = provincial, d = district, c = commune, b= board/institutional 50. Functional responsibilities in the health sector therefore seem clearly outlined, but linking the investment in and ownership of health facilities to responsibility for service delivery raises some issues. Generally, upper tiers of government take responsibility for larger investments. They then also own the facility and are responsible for its financing and service delivery. Therefore, financing and management of hospitals depends on whether they are owned by national, provincial and district authorities. Similarly financing and management of clinics depends on whether they are owned by provincial, district and commune governments. The practice of ownership and responsibility by size creates a bias for larger projects because these are more likely to be built and maintained by higher orders of government. Whilst the upper tier can provide capital investment and operational financing, there is no reason for not delegating management responsibility for that facility to lower tiers. This could promote coordination across orders of government and improve investment efficiency. 51. Summary of recommendations on expenditure assignments: The current legal framework would benefit from: (i) Clarifying the exclusive expenditure mandates of the central government. This could 53 be done in the amendment of SBL 2002. Local authorities should not be burdened with responsibilities that belong to central government; (ii) Defining the explicit responsibilities of center and province within a province. This could be done through provincial level regulations, which would supplement existing Stability Period Resolutions; (iii) Clarifying exclusive responsibilities for local authorities. Local authorities should avoid having multiple layers of government providing the same service within a jurisdiction, even if those different authorities could play a different (e.g. regulation, financing, service delivery) within the function. This could be reflected in the provincial resolutions. INSTITUTIONAL ARRANGEMENTS FOR BUDGETING AND ACCOUNTABILITY AT LOCAL LEVEL 52. Aside from clarity over spending assignments, growing spending responsibility and discretionary resources for lower tiers of government call for strong accountability of local authorities. A wide range of political, administrative, social and budget management factors can impact on local level accountability. This part briefly touches on how budget management practices can impact on local accountability and possible ways to address these in the amendment of the SBL 2002 or through other policy changes. It looks at how the current nested budgeting and planning system affects local accountability by introducing multiple layers of budget approval authority. It then discusses budget transparency issues at local level, before concluding with a brief review of citizen participation. 53. Nested budgeting system and accountability to legislative bodies: The structure of SBL 2002 reflects the nested planning and budgeting system in Vietnam. By law, each level of government has its own budget, which is prepared by the local executive (People’s Committee) and approved by the local legislature (People’s Council) at each tier of government. Each level’s budget is then integrated into the upper tier’s budget; the entire State Budget is consolidated and approved by the National Assembly. 54. The multi-tiered and consolidated budgets approved by the National Assembly and provincial/sub- provincial Peoples’ Councils complicate the preparation calendar in Vietnam, which is a critical element of budget accountability. The start of the annual planning and budgeting preparation cycle starts in May every year. It allows one month and a half for central and local authorities to submit their plans and budgets to the Ministry of Finance and the Ministry of Planning and Investment before the start of budget negotiations in the second half of July. After Government and National Assembly deliberations between mid-September and mid-November, there is only one month left for local authorities to finalize their detailed allocations. 55. This is very compressed given that there are three more levels of administration for which annual budgets need to be approved before December 31st. The budget approval time at Provincial Peoples’ Councils and sub-provincial levels is very short. In fact, Provincial Peoples’ Councils have only until December 10th (SBL 2002 Art. 45 (3)) – about 3 weeks -- to consider and adopt their annual budgets. District People’s Councils and below then have till the end of December to have their budgets approved. This means that all local legislative authorities at have around one and a half months (or three weeks for legislative authorities at each local tier) to approve a cumulative 50 percent of general government budget. This does not even take account of citizens’ participation, which is restricted due to lack of publication of budget proposals submitted to legislative authorities. Whilst there is no international best practice on the time needed to consider annual budgets, good practice is to allow around two to three months for legislative authorities to scrutinize draft budgets. 56. A related issue is that despite the relatively decentralized planning and budgeting framework, higher 54 levels of government have certain “veto rights” over budgets adopted by provincial and sub-provincial authorities (SBL 2002, Article 47). Furthermore the budget law authorizes higher level governments to modify local budgets. It states that the “Chairman of a People’s Committee may request a lower level People’s Council to readjust the draft budget if the disposition of local budget does not conform to the decision of the upper-level People’s Council.” If the provincial administrations had earlier and/or more detailed notification of their spending allocations approved by the NA, there would be little need to maintain these “veto rights”. In reality, it is reported the veto power is only applied in cases where there is a violation of the SBL 2002. Though legal violation can be treated separately, and since the provincial and sub- provincial are best placed to know their spending priorities, it may be useful to review Article 47 with a view to providing local authorities more autonomy in the allocation of spending, subject to broad parameters decided at central level. 57. A final issue on the nested budgeting system and accountability to legislative authorities are the current provisions on budget appropriation. The budget appropriation is the allocation that is agreed in law by the legislature. The State Budget Law 2002 (SBL) has provisions on the appropriation structure for the State Budget. Articles 15 and 25 of the SBL specify the authority of the NA and Provincial Councils respectively in relation to expenditure allocations. Articles 31 and 33 specify the various “functions/spending tasks” for appropriating the central and local budgets respectively. The “functions/spending tasks” in the appropriation structure include: capital development, recurrent (by 11 functions, as defined internationally), debt servicing, aid provision, replenishment of financial reserve funds and others. 58. The SBL 2002 provisions on budget appropriation however could be strengthened to promote more budget discipline. The structure of appropriations in Organic Budget Laws differs from country to country depending on national accountability arrangements for budget management. The Budget Law however should state that any changes to budget appropriations will require parliamentary approval. There may be some flexibility in terms of delegated authority to the executive e.g. for changes to appropriations up to a certain limit. 59. In Vietnam, Article 49 of the SBL indicates that in case of big changes to appropriations, the government should submit to the NA (and Provincial Councils) for approval. But in case of emergency or smaller changes, the revised budget is submitted to the NA Standing Committee, and the NA (or Provincial Councils) is (are) informed about budget adjustments. This warrants review to promote greater clarity on circumstances in which appropriations can be changed to ensure that the approval authority of and therefore the accountability to legislative authorities are not diluted. Any adjustments to appropriations need to be transparent, adequately justified, and usually mandated by the legislature. 60. Links between local budgeting and planning: The nested budget system is linked in parallel to a nested planning system, where national level strategies and plans cascade down to local authorities to inform budget plans and priorities. A ten-year Socio-Economic Development Strategy (SEDS, 2011-2020) sets the long-term development priorities and vision for the country. This is formulated based on sector and regional SEDS’. The SEDS is in turn supported by a five-year Socio-Economic Development Plan (2011- 2015), setting out specific objectives and how these will be met. The SEDP is disaggregated into local level and sector master plans. Therefore each local authority has its own five-year SEDP, which is meant to align with the national SEDP. The five-year SEDPs, in turn, are supported by annual development plans, which are meant to substantively inform annual budget allocations. 61. These annual development plans layered by local SEDPs , the national SEDP, sector and regional Master Plans leads to fragmentation with overlapping and at times inconsistent targets that lack prioritization and that cannot be meaningfully linked to resources. Multiple plans prepared at different levels can also confuse accountabilities, and thereby make it difficult to implement national priorities at the local level. It is 55 therefore worth looking at ways to streamline the planning process, which otherwise dilutes the quality of SEDPs at different levels, and thereby their relevance for budget allocations and accountability. Each local authority’s SEDP should identify a limited set of priority objectives and highlight how the local budget will be used as an instrument to achieve those objectives. 62. One way to do this is to shift the emphasis in local SEDPs from inputs and set out more clearly how local authorities’ budgets aim to achieve service delivery performance or national minimum standards of access and service quality. The focus on inputs is often derived from allocation norms set by provincial authorities for district authorities. This level of control however undermines local agility, autonomy and flexibility to find innovative cost-effective solutions to local concerns and needs. Central/provincial allocation norms used are based upon previous year spending rather than service requirements e.g., road conditions. This leads to inequity in allocation as well creates disincentive for innovative management responses. Similarly capital allocation norms use a Christmas tree approach to allocate small funds to all jurisdictions rather than directing funds where they may be needed to overcome identified infrastructure deficiencies. 63. Coordination of budget policies at a regional level: The lack of regional coordination on budget allocation decisions has exacerbated the inefficiencies created by the nested budgeting and planning systems. As noted in the 2012 Vietnam Development Report (2012 VDR), whilst provincial authorities have been able to mobilize resources for public investment, this has led to levels of investments that go beyond market demand. Central government has had to impose greater discipline on local capital investment decisions. Directive 1792 for example, adopted in October 2011, has strengthened oversight responsibilities of the Ministry of Finance and the Ministry of Planning and Investment. New projects are now only approved on an exceptional basis. Priority for capital budget allocations include completion of existing projects, clearance of arrears, or counterpart funding for externally financed projects. 64. Strengthening regional coordination would help to considerably strengthen the quality and efficiency of large infrastructure projects and other public projects that have big externalities. As illustrated in chapter 3, economies of scale play an important role in determining relative efficiency of local authorities. Such scale can be achieved through improved coordination at regional level. As recommended the VDR 2012, this could be pursued through various channels including more funding for regional projects, and stronger independent regulation of major infrastructure decisions. 65. Fiscal rules and norms and local autonomy: One issue that most local authorities raised as constraining autonomy over their expenditure assignments are central fiscal rules and norms for particular sectors. Central government imposes uniform minimum expenditure requirements as a share of local budget for education (20 percent of local budget), science and technology (2 percent of local budget), and contingencies (2 percent – 5 percent of local budget). The health budget needs to increase by at least the same growth rate as the overall budget. In addition any increase in realized revenues over projected revenues that is retained by the province 6 by local government must be apportioned 50 percent for wages in the next budget year and 50 percent for important capital projects. 6 The share of over-realized revenue retained by provinces is the same as the sharing rate for shared taxes. 66. In some instances, these fiscal rules help to ensure that critical social expenditure does not fall 56 below minimum levels. But they do also undermine local priorities and allocative efficiency as they do not recognize fiscal and economic climate of local jurisdictions. For example, infrastructure needs in fast growing provinces like Da Nang require higher priorities for capital spending and the centrally determined fiscal rule constrains financing available from own reserves. Further, not all local governments can and must strive to establish science and technology corridors and parks. For rural and remote locations, basic health and education would be a priority rather than science and technology. For major urban centers, the private sector may fill the gap for education, health and science and technology. 67. Prescribed fiscal rules fail to take into account private provision of public services. For example, there is significant involvement of private sector in providing schooling in Hanoi. A public sector mandate for minimum educational expenditure must therefore take into account resources available from all sources for schooling. Fiscal rules in the absence of objective standards of access and quality of service are in any case not very useful; there is no one to one correspondence between level of spending and quality and quantity of service provision. Properly designed fiscal rules, can be helpful in ensuring fiscal discipline and are discussed further in chapter 6. 68. Transparency of budget documents: A recent review of fiscal transparency in Vietnam by the World Bank found that despite extensive requirements in law on the publication of budget information, there is much scope to improve communication of budget policies.7 This is confirmed by the most recent results of the Vietnam Provincial Governance and Public Administration Index (PAPI) survey results in 2013.8 The PAPI survey looks specifically at transparency in commune budgets. Although the PAPI results show some improvement over time on transparency of commune budgets, overall awareness understanding of local citizens is low. Nationwide, around 37 percent of survey respondents noted that they were aware of the commune budget. Out of this, one third actually read the commune budget, and around three quarters of those that read the documents were satisfied with the quality. This means that less than 10 percent of respondents were aware, understood and were satisfied with commune level budgets. 69. Since the Fiscal Transparency Review has already covered in detail issues of budget coverage and comprehensiveness, we focus here on one particular issue which is that of budget classification and reporting templates of local budgets. As noted in the first part of this chapter, the current classification structure and reporting templates in published State Budget reports severely limit the analysis of overall spending within a particular province. This hampers budget transparency and accountability to the National Assembly, but also to citizens within a province because spending by central authorities is not reported, or to citizens within a district because provincial authorities’ spending is not reported. Knowing how much is spent by what tier of government within a jurisdiction is also important in the context of transparency over spending assignments within a province (ref above discussions on explicit assignments between multiple orders of government, and multiple orders of government providing the same service). 7 World Bank, “Vietnam Fiscal Transparency Review: Analysis and Stakeholder Feedback on State Budget Transparency” (2013) 8 CECODES, VFF-CRT, UNDP, “The Vietnam Provincial Governance and Public Administration Performance Index PAPI 2013: Measuring Citizens’ Experiences,” (2013) 70. The standard tables for reporting on the local budget therefore warrant review. The study reviewed in detail the ten standard tables on local level budgets presented to the National Assembly as part of the 57 State Budget. The templates for these tables are set out in Resolution 387 of the National Assembly (2003). These tables currently provide: (i) aggregate spending and revenue of all local authorities, with a breakdown of totals for provincial authorities on the one hand and sub-provincial (districts plus communes) on the other; (ii) gross receipts broken down by types of revenue collected by all local authorities; (iii) recurrent functional and total capital spending by provincial authorities; (iv) breakdown of spending by provincial administrative units; (v) breakdown of capital projects of provincial authorities; and (vi) aggregate receipts of and spending by districts. 71. The current tables therefore provide a fair amount of detailed budgetary information. In particular, they provide information on the overall size of local spending with capital and recurrent breakdown; details on sources of funding (decentralized revenue, intergovernmental fiscal transfers, off-budget receipts); the relative share of provincial versus sub-provincial spending in local expenditure; and spending across government functions by provincial authorities. It is therefore possible to analyze from this information the degree of fiscal autonomy within the province, level of vertical fiscal imbalance, dependence on transfers, and the extent of spending decentralization down to below provincial authorities. 72. However, there are also important issues that limit the transparency and analytical content of local budget reports. The tables do not provide a consolidated picture of spending across different functions by all local authorities. They only report on recurrent spending across functions by provincial authorities. But as noted above, in key areas like education and health, this may only represent 10 percent of local spending within the province, and even less as a proportion of total spending within the province if central authorities’ spending is also taken into account. Additionally, with the exception of education and science and technology, the budget tables do not provide the capital breakdown of functional spending by provincial authorities. This means that it is not possible to have a complete picture of functional spending even just by provincial authorities. 73. The budget tables should provide not only a complete picture of functional spending by each tier of government, but also a consolidated view of spending within the province by both central and local authorities. It would be too much to do this in the National Assembly reports for spending within districts. But at district level, the authorities should have a consolidated picture of spending by provincial authorities and their own spending within the district. For reports to the National Assembly, table 2.6 below illustrates the information already available in provincial budget tables (√) and what should be available in provincial budget tables to ensure adequate budget coverage but currently is not (χ). Table 2.6: Information presented in provincial budget tables Central Provincial Totals Total education χ √ Total spending for a government function within recurrent spending χ √ a province not available. capital spending χ √ Total in other functions χ χ recurrent spending χ √ capital spending χ χ Totals Total spending by different tiers within a province not available 74. Aside from purely financial information, however, the government should consider further developing its 58 existing citizens’ budgets to summarize local budgets in a format that is easily accessible to citizens.9 This is vital for public participation in the budget process. The government can only capitalize on the benefits of publishing budget information if people also participate in reviewing and questioning that information. But the information has to be accessible and easily understood. In addition to this, the government should also publish the draft budget when it is submitted to the legislature. Even if figures are preliminary, there are great benefits to publishing at this stage, and promoting participation. This is a further reason for decompressing the budget calendar recommended earlier in the chapter. 75. Participation in the budget process: Budget transparency is a necessary but not a sufficient condition to ensuring greater accountability of local authorities. Translating budget transparency into accountability requires participation in the budget process. This includes participation in highlighting budget priorities to ensure it reflects local needs and preferences, but also participation in monitoring implementation. Several studies have sh own that disclosure of financial information in itself will not lead to more accountability if there are impediments to citizen participation in the budget (Fukuda-Parr et al 2011, Khagram et al 2013). 76. In Vietnam, as discussed above, the lack of transparency in the budget itself is an impediment to participation. Citizens cannot participate if budget information is not communicated in an accessible manner. In addition, there are several provisions in the current institutional arrangements, which limit participation such as the compressed budget calendar and the non-publication of the draft budget proposal when it is submitted to the legislature. 77. The 2013 Constitution has introduced several provisions to enable citizens to participate in government affairs and hold the government to account. Article 6 of the constitution states that “The people practice state power under the forms of direct democracy and indirect democracy through the National Assembly, the People’s Councils and other state agencies”. This article seems to empower people by having the right to referenda on major projects and right to petition for continuation or termination of specific programs. Further the Constitution establishes people’s right to access to information on government operations (Article 25); requires GOV to create conditions for people’s participation in administration (Article 28) and for the People’s Councils to regularly report to people (Article 115). 78. In addition to this the Grassroots Democracy Ordinance issued in 2007 sets out detailed provisions on the People’s Councils’ and Committee’s responsibilities in terms of consultation with citizens.10 It sets out in particular what information needs to be disclosed to the public and requirements for consultation before specific laws and regulations are formally adopted. This includes local development plans, details on investment projects, levels and purpose of new debt financing, and results of inspection processes. It also mentions the annual budget estimates, but it is those that are already approved by the local legislature (Article 5). This is confirmed by Article 19, which lists a number of policy areas where draft documents will be shared with citizens before they are formally adopted, and the State Budget does not appear on this list. There are however quite significant provisions requiring citizens to participate in the prioritization and monitoring of investment projects (Article 10). This is done through People’s Inspection Boards and Community Boards for Investment Supervision, which are required to be established in every commune. 9  For examples on Citizens’ Budgets see “International Budget Partnership”: http://internationalbudget.org/what-we-do/open-budget- survey/research-resources/citizens-budget/ 10 Standing Committee of National Assembly, “Ordinance on Exercise of Democracy in Communes, Wards, and Townships,” No 34/2007/ PL-UBTVQH11 (April 20, 2007) 79. Despite these extensive provisions, evidence suggests that actual participation on budget issues at local level is quite low. This is what is expected from the summary above on factors that impede active 59 citizen engagement on the budget. The annual Vietnam PAPI survey provides a wealth of evidence on the level of budget transparency and participation, particularly at the level of communes based on citizens’ feedback.11 Results of the 2013 survey suggest that only 14 percent of respondents were aware of the exact roles of the People’s Inspection Boards and the Community Boards for Investment Supervision. On a more positive note, the share of respondents that participated in the design of investment projects increased from 34 percent to 45 percent between 2011 and 2013; and the share of respondents that participated in monitoring of project implementation also increased from 22 percent to 28 percent in 2013. 80. Looking at the results of PAPI surveys in the last three years could help identify potential links between citizens’ satisfaction with public service delivery on the one hand, and budget transparency and participation on the other. The analysis looks at what potentially explains citizens’ satisfaction with public service delivery as measured by the overall public service delivery dimension in PAPI. This dimension includes education, health and infrastructure services. The dimensions on citizens’ satisfaction with public service delivery are a reflection of citizens’ perception of the accessibility, quality and availability of basic public services in their locality. It does not reflect the relative performance of service across localities, but rather citizens’ views about public service performance in their locality. This can therefore be a good proxy for allocative efficiency i.e. the extent to which resource allocation is responding to local preferences. 81. The PAPI survey includes several dimensions on the level of citizen participation in the design and monitoring and public projects. These include citizens’ views on the overall level of vertical accountability, which incorporates interactions with local authorities, and the coverage and effectiveness of People’s Inspection Boards (PIBs) and the Community Boards for Investment Supervision (CBIS). Vertical accountability and the coverage and effectiveness of PIBs and CBIS’ are expected to impact positively on satisfaction with public services. The sub-dimension on transparency of commune budgets (see paragraph 67) will likely have little if any impact on satisfaction with service delivery. This is because current levels of budget transparency do not induce citizens’ participation in public resource allocation and monitoring. 82. The analysis of potential determinants of satisfaction with public service delivery confirms a strong, positive impact of citizens’ participation. The table below summarizes the results of the model estimation. The latter controls for per capita GDP growth, which exhibits a strong relationship with satisfaction with service delivery, and rural population share, which exhibits a slight negative relationship. As expected, budget transparency has little impact on satisfaction with public services. But if more participation in the budget were facilitated, then this would likely also have a strong positive link. Limits on corruption have a positive relationship, even though it is not statistically significant. The biggest impact is through the dimension of coverage and effectiveness of People’s Inspection Boards. 12  ECODES, VFF-CRT, UNDP, “The Vietnam Provincial Governance and Public Administration Performance Index PAPI 2013: Measuring C Citizens’ Experiences,” (2013) Table 2.7: Explaining citizens’ satisfaction with local public service delivery using PAPI results 60 Public services (1) (3) (4) (5) (6) (7) (8) (9) Logarithm of /capita 0.091 0.283*** 0.283*** 0.280*** gdp (0.053) (0.052) (0.053) Rural population -0.008*** -0.008*** -0.009*** -0.010*** -0.009*** share (0.001) (0.001) (0.001) (0.001) (0.001) -0.042 0.170 Communal budgets transparency (0.129) (0.107) Vertical accountability -0.162 0.092* 0.125** (0.119) (0.049) (0.048) 0.505*** 0.228** 0.366*** People’s inspection boards (0.185) (0.099) (0.098) Community 0.284 0.225* 0.324*** investment boards (0.194) (0.120) (0.117) Limits on corruption in 0.161 0.195 0.211 0.186 0.243* 0.179 0.187 0.160 service delivery (0.140) (0.145) (0.143) (0.146) (0.140) (0.140) (0.136) (0.141) Year 2012 0.145** 0.159*** 0.156*** 0.155** 0.158*** 0.156*** 0.150*** 0.149** (0.056) (0.059) (0.059) (0.059) (0.058) (0.057) (0.056) (0.057) 0.148** 0.191*** 0.179*** 0.179*** 0.185*** 0.189*** 0.168*** 0.170*** Year 2013 (0.060) (0.060) (0.060) (0.061) (0.059) (0.058) (0.058) (0.059) 5.633*** 2.936*** 3.000*** 3.096*** 6.563*** 6.337*** 6.395*** 6.484*** Constant (0.808) (0.635) (0.600) (0.600) (0.340) (0.345) (0.299) (0.308) Observations 189 189 189 189 189 189 189 189 R2 0.315 0.216 0.224 0.217 0.251 0.267 0.294 0.271 Standard errors in parentheses; p<0.1, ** p<0.05, *** p<0.01 83. The above results could be broken down further to help take a first step in identifying possible policy priorities for selected provinces. To do this, the scatter plots below show the correlation between satisfaction with public services and: communal budget transparency and PIBs. The adjoining charts split the correlation charts to show which provinces are performing above the average and which below in each dimension. 84. For example, Hau Giang scores lower than average on budget transparency, but higher than average on budget participation and satisfaction with service delivery. This might suggest to prioritize efforts to improve budget transparency. Bac Giang on the other hand displays higher than average satisfaction with budget transparency and participation levels (chart X), but lower than average satisfaction with public service delivery levels. Therefore, it may point to other issues that have a relatively bigger impact on satisfaction with service delivery. Budget transparency and public service (papi 2011) Provincial performance against average for each dimension 61 correlation = 0.152 p-value = 0.037 Year 2011 7.2 7.4 8 An Giang Hai Duong Ba Ria-Vung Tau Kien Giang Ben Tre Kon Tum Bac Lieu Hau Giang Binh Duong Lang Son 7.5 Bac Ninh Ninh Thuan Dong Thap Long An Ha Noi Nam Dinh Binh Dinh Thanh Hoa 7 Ha Tinh Quang Binh Da Nang Thua Thien- Hue Hai Phong Quang Nam Ho Chi Minh City Quang Ninh 6.2 6.4 6.6 6.8 Dong Nai Vinh Long 7 Khanh Hoa Quang Tri Public Services Vinh Phuc Son La Binh THuan Ninh Binh Bac Giang Ha Nam 6.5 Ca Mau Phu Tho Bac Kan Hoa Binh Cao Bang Phu Yen Binh Phuoc Lao Cai Dien Bien Soc Trang Can Tho Nghe An 6 Ha Giang Tay Ninh Dak Lak Quang Ngai Hung Yen 6 Tra Vinh Dak Nong Thai Binh Lai Chau Tuyen Quang Gia Lai Thai Nguyen 5.5 5.6 5.8 Lam Dong Yen Bai Tien Giang 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 Communnal Budgets Transparency 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 Public Services Fitted values Communnal Budgets Transparency Participation and public service (papi 2011) Provincial performance against average for each dimension correlation = 0.167 p-value = 0.021 Year 2011 7.2 7.4 8 An Giang Ho Chi Minh City Ba Ria-Vung Tau Lang Son Bac Lieu Khanh Hoa Bac Ninh Long An Binh Duong Nam Dinh Ben Tre Kien Giang 7.5 Dong Thap Quang Binh Binh Duong Kon Tum Ha Noi Quang Ninh 7 Da Nang Ninh Thuan Ha Tinh Quang Tri Dong Nai Quang Nam Hai Duong Son La 6.2 6.4 6.6 6.8 Hai Phong Thua Thien- Hue Hau Giang Thanh Hoa 7 Public Services Vinh Long Vinh Phuc Binh Thuan Lai Chau Bac Giang Lao Cai Ca Mau 6.5 Lam Dong Can Tho Ninh Binh Bac Can Nghe An Cao Bang Phu Yen Binh Phuoc Phu Tho Dack Lak Quang Ngai 6 Dak Nong Soc Trang Ha Nam Thai Binh Dien Bien Tay Ninh 6 Gia Lai Tra Vinh Hoa Binh Thai Nguyen Tuyen Quang 5.5 Ha Giang 5.6 5.8 Hung Yen Tien Giang Yen Bai 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 Public Services Fitted values People’s Inspection Boards Source: Staff estimates based on published State Budget data 85. The above should of course be treated only as a very first step in using available cross-province evidence to identify possible policy priorities for improving citizen satisfaction with service delivery through improved budget transparency and participation. In reality, despite some improvements between 2011 and 2013, the PAPI results, complemented by the earlier analysis on institutional factors inhibiting transparency and participation, suggest that significant efforts are needed to improve local level budget transparency and accountability. CONCLUSIONS AND RECOMMENDATIONS 86. Higher levels of local spending in Vietnam have been targeted to the poorest parts of the country. Vietnam exhibits fairly high levels of spending responsibility for local authorities, both provincial and district. This seems to have moved in tandem with local capacity and been accompanied by more discretionary resources. With spending moving closer to people, there are some key institutional reforms that would be expected to improved allocative and productive efficiency. The chapter recommends in particular: (i) more explicit assignments on service delivery for different tiers of government within a jurisdiction; (ii) decompression of the budget preparation calendar; (iii) greater autonomy over budget approval authority of People’s Councils and clearer provisions on budget appropriations; (iv) elimination of minimum allocation requirements for sectors; and (v) increased transparency and participation in the budget process. Recommendations on expenditure decentralization 62 Issues Recommendations Simplify Articles 31 and 33 in the SBL 2002 stating Lack of exclusive spending assignments for that all tiers may spend on any functional area central government in SBL 2002 leads to local but that some functional areas are the exclusive authorities taking on the responsibilities of responsibility of central government. This may also central government. need to be addressed in the upcoming Law on Local Government. The current budget preparation calendar is Review Article 45 of SBL 2002 to decompress the too compressed to enable local councils to budget calendar, including earlier start to budget scrutinize the draft budget. preparation process. State Budget Law 2002 SBL provisions that give upper tier authorities Review Article 47 of SBL 2002 to assign more “veto rights” over budgets approved by lower clearly authority to local councils for budgets at tier governments dilute local councils’ authority. their own level. SBL provisions on changes to budget Review Article 49 of SBL 2002 to indicate that appropriations, undermines accountability to any adjustments to appropriations need to be local councils as the latter are not required to mandates by the legislature. approve re-appropriations. The general public is not able to contribute to the budget preparation process as the budget Require publication of budget proposal when it is is only published after it is approved by the submitted to the local legislature. legislature. The lack of explicit assignments between Each provincial resolution should set out the center and province for service delivery can Stability Period regulations explicit responsibilities of center versus provincial lead to ambiguity and undue pressures for local authorities within that jurisdiction. budgets. For shared functions, enabling multiple orders of government to provide the same service/ Provincial resolutions on revenue and spending perform the same task within a jurisdiction assignments should set explicit responsibility for can lead to overlaps and poor planning and specific tasks within a jurisdiction. budgeting. Central fiscal rules and norms affect local Reconsider minimum allocation norms for sectors autonomy over budget decisions and distort or types of expenditure in the next Stability Period. resource allocation. Budget Law regulations The current budget classification structure, Further develop the existing Citizens’ Budget to State budget table templates, and coverage of fiscal communicate in a more accessible manner budget activities in budget documents. policies of local authorities. 63 EXPENDITURE PERFORMANCE OF LOCAL 64 AUTHORITIES 03 Key issues: Earlier studies have indicated a lack of credibility in public spending plans, due to current policies on use of over realized revenue, extra budgetary financing and carry over practices. This leads to a loss of budget transparency, which can impact negatively on productive efficiency. This is a matter of prime concern to policy makers now as Vietnam faces tightening fiscal conditions with slowing revenue mobilization. Objectives: The objectives of this chapter are to determine if the above spending management practices are impacting on spending efficiency, and the implications for reform to promote greater budget credibility and transparency. Key findings: Local spending plans lack credibility when compared to actual outturns. This is particularly the case for capital spending, which is often more than 50 percent of what was budgeted, well above good practice guidelines of maintaining spending within 5 percent of budget. The lack of credibility arises partly due to the flexibility accorded to the executive to change budget appropriations approved by the legislature, which was discussed in chapter 2. There are however several institutions and policies embedded in the SBL 2002 that enable the executive to take advantage of the flexibility over budget appropriations, including: use of over-realized revenue; access to extra budgetary financing; and carry over practices. These can impact on productive efficiency. Analysis of productive efficiency finds that there are some clear differences across provinces. On average, provinces in the Northern Mountains spend 42 percent more and those in the Central Highlands 35 percent more to achieve the same outputs/ outcomes as the most efficient provinces. In the South East and the Mekong River Delta on the hand, the average is closer to 7 percent and 9 percent respectively. Some of these disparities are associated with differences in population density, and therefore conditions that local authorities cannot control. But poorer performers within a region also tend to have less credible spending plans and higher levels of carry overs compared to their peers in the same region. The share of wages and salaries in recurrent spending also seems to play a role, including in terms of crowding out other critical spending. Recommendations: (i) spending of over-realized revenue should not be tied to specific categories and should be approved by legislature; (ii) extra budgetary activities should be integrated into the overall budget; (iii) limit the level of carry over spending as a share of budget and to capital spending only; (iv) introduce population density criteria in allocation norms to estimate spending needs of local authorities; (v) review wage bill and application of salary reform policies. CHAPTER 3: EXPENDITURE PERFORMANCE OF LOCAL AUTHORITIES 65 1. Background: Building on the spending trend analysis and institutional review of expenditure decentralization in chapter 2, this chapter looks more directly at selected dimensions of expenditure performance across provinces. Tightening fiscal conditions in Vietnam calls for an increased understanding of the relative spending on public services across provinces. The first part looks at local level spending performance in terms of quality of budget outcomes, in particular the maintenance of fiscal discipline through hard budget constraints, which has been highlighted in previous studies as an area of weakness. The latter is a necessary, though not sufficient, condition for credible and transparent public spending plans, linking those spending plans with local Socio-Economic Development Plans, and ultimately the efficiency of spending. The second part of this chapter looks more closely at productive spending efficiency across provinces. The final part looks at what might explain the differences in performance and implications for policy reforms, in particular to promote harder budget constraints 2. Reviewing spending performance across local authorities is a challenging task for a number of reasons. Firstly, data limitations impact on accuracy and the availability of full sets of fiscal information across all local authorities. This can make it difficult to compare outcomes across local authorities. Secondly, even after correcting for accuracy and completeness issues, any comparisons should be treated with caution. Differences in outcomes across provinces can be due to a variety of different factors, which may be very context specific and difficult to control for. Thirdly, the purpose of the analysis is to identify reforms in overall fiscal management based on how current practices may impact on spending performance, and not to identify province or sector specific recommendations. CREDIBILITY OF SPENDING PLANS AND LOCAL BUDGETARY DISCIPLINE 3. To help review the credibility of spending plans and local budgetary discipline, this section looks at the extent to which actual spending (spending outturns) diverge from expenditure plans (budgeted spending). Spending outturns that consistently exceed budgeted spending reflect lack of credibility in budgeting and poor budget discipline. Lower outturns can also point to potential problems of implementation capacity or optimistic financing projections. Other factors that impact on budget credibility include budget coverage and comprehensiveness. For example, if there are significant off-budget expenditures or if service fees and charges are accounted for net, official numbers will only provide a partial and distorted picture of the budget. The budget is the key instrument for implementation of local authorities’ plans and priorities. Therefore a lack of budget credibility also impacts negatively on the meaningfulness of local development plans. The lack of budget credibility may be related to local capacity issues. But there are also a number of institutional factors that contribute to low credibility of spending plans, which are discussed further below. 4. Credibility of aggregate, capital and recurrent spending plans: To review the credibility of aggregate spending plans, the study looks at the 2006-2011 period average of local spending outturn to budgeted spending. Local spending is the total expenditure of all local authorities (provinces, districts and communes) net of transfers to lower tiers. In addition to core recurrent and capital spending of local authorities, local spending also includes other expenditure from off-budget sources (e.g. bond financed infrastructure, service delivery units spending directly out of collected fees), carry over spending, spending out of contingencies, and transfers to reserves. The reason for including the non-core items is to assess the extent to which they impact on budget credibility. This could point to possible reforms to existing institutional norms and practices that relate to these non-core items. 5. Available data highlights poor credibility of aggregate spending plans at the local level. Figure 3.1 66 shows that for half of all local authorities, total actual spending per year was at least 150 percent of planned spending for the year. This is considerably higher than good practice benchmarks in the Public Expenditure and Financial Accountability (PEFA) framework of actual spending remaining plus/minus within 5 percent of the budget in at least one out of the last three years.12 The threshold for the lowest PEFA score in this dimension is outturn exceeding budget by 15 percent or more in at two of the last three years. In the case of Vietnam, as illustrated in figure 3.1, very few local authorities are able to meet even this threshold when looking at period average of outturn to budget. Figure 3.1: Actual/budgeted local expenditure (%) Figure 3.2: Actual/budgeted expenditure of provincial authorities (%) -1SD Median +1SD -1SD Median +1SD 16 18 14 16 14 12 12 10 Frequency Frequency 10 8 8 6 6 4 4 2 2 0 0 110 120 130 140 150 160 170 180 190 200 210 100 120 140 160 180 200 220 240 260 280 300 Actual vs. budgeted local expenditure (%) Actual vs. budgeted provincial expenditure (%) Source: Staff estimates based on published state budget data 6. Provincial authorities’ budgets display similar lack of credibility. As illustrated by figure 3.2 above, for half of all observations, actual spending by provincial authorities (net of transfers to districts) exceeded budgeted spending by at least 50 percent. No provincial authority achieves good practice of maintaining spending within 5 percent of the budget. The distribution of actual to budgeted spending ratio is slightly more skewed for provincial authorities towards even higher ratios of outturn over budget. 7. Recurrent expenditure outturns are more closely aligned to budgets than capital spending outturn. For recurrent spending, in half the observations, actuals are within 20 percent of planned expenditure (figure 3.3). Half of observations at the level of provincial authorities exceeded budget by less than 10 percent (figure 3.4) – though in some cases spending was less than budget, which as noted above may also signal some difficulties. A large share of recurrent spending are fixed costs in terms of wages and salaries, which means less room for divergence. Despite this, however, for half of all observations, divergence between actuals and budgets is greater than 20 percent. This is due to policies on the use of over-realized revenues as discussed below. 9 PEFA Secretariat, “Public Financial Management Performance Measurement Framework,” Performance Indicator 1 on aggregate expenditure outturn compared to original approved budget (January 2011) Figure 3.3: Actual/budgeted recurrent by local Figure 3.4: Actual/budgeted recurrent by provincial authorities authorities 67 -1SD Median +1SD -1SD Median +1SD 16 16 18 14 14 12 8 10 12 10 Frequency Frequency 86 6 4 4 2 2 0 0 100 110 120 130 140 150 160 60 70 80 90 100 110 120 130 140 150 160 Actual vs. budgeted local recurrent expenditure (%) Actual vs. budgeted provincial recurrent expenditure (%) Source: Staff estimates based on published state budget data 8. Actual capital spending exceeded plan by more than 50 percent for half of observations in figure 3.5 below. For around 13 local authorities, capital spending exceeded plan by more than double. There is not much difference between performance for local authorities as a whole and provincial authorities (figure 3.6), because the latter make up most of local capital spending (70 percent on average). These points to important issues on the quality of public investment management. In a few cases there is also under-execution, which may be linked to lower implementation capacity or unrealistic budgets in selected provinces. This would need to be reviewed more carefully through more detailed analysis of those particular cases. Figure 3.5: Actual/budgeted capital by local Figure 3.6: Actual/budgeted capital by provincial authorities authorities -1SD Median +1SD -1SD Median +1SD 25 25 20 20 15 Frequency Frequency 15 10 10 5 5 0 0 50 100 150 200 250 300 350 400 450 500 550 50 100 150 200 250 300 350 400 450 500 550 Actual vs. budgeted local capital expenditure (%) Actual vs. budgeted provincial capital expenditure (%) Source: Staff estimates based on published state budget data 9. Capital budgets in richer provinces on average are more credible than those in poorer provinces. This is illustrated in figure 3.7 below, which shows that the average divergence between actual and budgeted capital spending between 2006 and 2011 is greater for poorer regions (i.e. Central Highlands, Northern Mountains, and Northern Central and Central Coast). This may be linked to better investment management capacity, or more debt financed infrastructure in richer provinces as discussed below. Figure 3.7: Average divergence between actual and budgeted capital spending 2006-2011 by region (%) 68 190 2006 2007 2008 2009 2010 2011 140 90 40 -10 Central Highlands Northern Mountain Northern Central and Mekong River Delta Red River Delta South East Central Coast Source: Staff estimates based on published state budget data 10. Institutional factors that contribute to lack of credibility of spending plans: Lack of credibility in spending plans arise because local authorities are able to relax spending limits set in approved budgets thanks to institutional practices that allow them to finance higher spending without cost or consequences. Part of this is driven by the relatively relaxed provisions on budget appropriations discussed in chapter 2. This allows the executive fairly significant flexibility to alter overall allocations over the course of the fiscal year. 11. The appropriations set by the National Assembly and People’s Councils are at quite a high level – they appropriate by 11 broad functions for recurrent spending and for capital as a whole (aside from debt servicing, and some other non-discretionary items). Appropriations are generally to administrative units at the same level of government (i.e. National Assembly appropriates for central government, Provincial People’s Councils for provincial departments and so forth). It is therefore possible to shift resources within these categories without going back to the National Assembly or relevant People’s Councils. However, a number of current policies undermine the ceilings in approved budgets. 12. The first are policies on the use of over-realized revenue, which from the available data, are the biggest contributors to the lack of local budget credibility. Table 3.1 below summarizes the estimation results from regressing local budget incredibility (i.e. the higher the value, the lower the credibility) with a set of possible explanatory variables. This is based on available data for the 2006-2011 period. 13. The results show that: (i) the lack of credibility in revenue projections have a direct and statistically significant impact on the lack of credibility in spending projections; (ii) per capita GDP, as a measure of local development and capacity, although expected to be positively associated with budget credibility exhibit a slight negative relationship, which is discussed further below; (iii) the level of off-budget expenditure as a share of the local budget, contributes to lower divergence between spending plans and outturns. Table 3.1: Explaining lack of local budget credibility 69 Local expenditure/local budget (1) (3) (4) (5) (5) 0.318*** 0.323*** 0.305*** 0.274*** Actual vs. Budgeted local revenue (0.073) (0.068) (0.068) (0.066) -0.0002 0.001*** Per capita gdp (0.001) (0.0001) 14.37*** Logarithm of per capita gdp (3.322) Actual local off-budget spending/ -0.565 -0.534* budgeted local expenditure (0.385) (0.292) Targeted transfers/local core -0.0847 expenditure (0.182) Local decentralized revenue/ local -0.375** core expenditure (0.148) Central balancing transfers/ local -0.699*** core expenditure (0.178) Year 2008 13.85** 14.83*** (5.854) (4.826) Year 2009 11.52 16.29*** (6.972) (5.131) Year 2010 19.71** 23.57*** (8.589) (4.996) Year 2011 35.90*** 30.37*** (11.34) (4.976) Constant 150.3*** 104.1*** -20.35 110.3*** (15.68) (8.633) (31.77) (8.454) Observations 141 167 167 167 R2 0.398 0.199 0.218 0.323 Standard errors in parentheses; * p<0.1, ** p<0.05, *** p<0. 14. Local authorities are allowed to spend over-realized revenue beyond the ceilings appropriated in the approved budget without having to revert back to Peoples’ Councils for approval. National Assembly regulations allow for spending of half of over-realized revenue on salary reforms and the other half on capital projects. Local authorities are therefore within their legal mandate to spend above the approved budget. But this weakens the credibility of spending plans and dilutes accountability to legislative bodies. There are several factors that contribute to poor revenue forecasts, which are discussed further in chapter 5. But the above also suggests incentives to underestimate revenue. 15. There is no clear pattern on the level of divergence between actual and budgeted revenue across regions. The above estimation suggested that there may be a slight negative relationship between per capita income and credibility of spending plans. In other words, more developed regions are likely to have less credible spending plans, which could be due to poor revenue projections. Figure 3.8 below shows that on average, the wealthier regions of Red River Delta and the South East have the second and third worst records in terms of credibility of revenue projections. The Northern Mountainous region has the least credible revenue projections. 16. One possible reason for this is that wealthier provinces may have a slightly bigger incentive to 70 underestimate revenue. It may be more worthwhile for richer provinces because decentralized revenues are more substantial both as a share of total local receipts and in absolute terms compared to poorer provinces. Therefore at the margins, higher collections have a bigger impact. Of course the margin of error in forecasting revenue may be naturally bigger in richer provinces because of more complex revenue structure. But as illustrated in figure 3.8, the difference across different regions is not very significant. In later years, poorer provinces also seem to be underestimating revenue by a large margin. Figure 3.8: Average divergence between actual and budgeted local revenue 2006-2011 by region (%) 50 40 2006 2007 2008 2009 2010 2011 30 20 10 0 Northern Source: Mountain Staff estimates Red River based Delta on published South state East budget dataNorthern Central and Central Highlands Mekong River Delta Central Coast 17. A second policy area that contributes to lack of spending plan credibility is the use of off-budget financing. Most of the off-budget category is borrowing through bond issuances by local authorities, development banks, and local investment funds. Under the SBL 2002, local borrowing is treated outside the balance sheet because provinces are required to maintain a balanced budget. It also only allows borrowing for capital investments. The level of borrowing is reported in budget documents and the government sets aggregate ceilings for local borrowing, but it is not part of the formal appropriation approved by local Peoples’ Councils. 18. The estimation in table 3.1 however seemed to suggest that off-budget financing contributes to lower divergence between spending plans and spending outturn. Figure 3.9 shows that for half of all observations, the 2006-2011 period average for off-budget spending as a share of local spending was between 5 and 20 percent. Figure 3.10 shows as expected a direct relationship between share of off- budget spending in local spending and per capita GDP. Figure 3.9: Off-budget spending/total local spending Figure 3.10: Off budget spending and per capita GDP 71 -1SD Median +1SD 0 20 40 60 20 16 18 Actual local o -budget 14 expenditure as % 10 actual local 8 10 12 expenditure Frequency 0 60 Per 6 40 capita GDP(- 4 million 20 2 VND 0 0 5 10 15 20 0 10 20 Actual local o -budget expenditure as % actual local expenditure Source: Staff estimates based on published state budget data 19. The positive link between off-budget financing and budget credibility might be because local authorities exercise more discipline in planning and budgeting for debt financed infrastructure compared to those funded out of regular receipts. This might be an added explanation as to why richer provinces that are able to borrow for investments exhibit more credible capital budgets as discussed above. But this would need to be looked at separately and in detail using project level data. What is important, however, is that even if debt financing is a relatively small share of overall local financing at the moment, this may increase over time. This would in turn exacerbate the lack of spending plan credibility, and therefore as argued in chapter 6, it is important to include all borrowing on-budget and report on it transparently in budget reports. 20. A third policy area that contributes to loss of spending plan credibility is the practice of carry overs. Local authorities carry over spending (i.e. because of under-spending in the budget) at the same time as spending more than what they have budgeted. The reason is because the over-spending may come from spending of over-realized revenues on items that were not initially appropriated (e.g. salary reform and capital projects). The under-spending, which is linked to carry overs, could be because local authorities are not able to spend what is officially appropriated. 21. There are three broad elements to carry over practices in Vietnam. Firstly, part of over-realized revenue, which is not spent in the fiscal year, can be carried over to the following year for capital spending and salary reform. Although the level of carry over amount is appropriated by the NA, the detailed allocations are not, which means that these can be spent across different appropriation categories at the executive’s discretion. Secondly, administrative units have discretion to spend savings from blocks grants in the next fiscal year. The amounts are not large, but they are also not appropriated in the new budget. Thirdly, unfinished capital works can also be carried over, but expenditure is attributed to the year that it was appropriated rather than the year it was spent. For expenditure that is carried over, there is no economic or functional breakdown in budget reports. 22. Figure 3.11 below shows that for half of all observations, the 2006-2011 period average for carry over spending over budgeted local spending ranged between 25 and 50 percent. This is significant and highlights real issues in terms of the transparency and credibility of the budget, particularly given the above practices of not reporting on the details of carry overs. 23. Figure 3.12 shows that the level of carry over spending is positively related to the share of capital spending 72 in the budget, and negatively related to the share of recurrent spending in the budget. This suggests that most of the carry over may be due to capital spending. Local authorities indicated that this reflects delayed start in construction works, underestimating project costs, or cost adjustments due to inflation. Another interesting trend is that carry over spending has increased over the 2006-2011 period across all regions, and levels of carry over as a share of budget are particularly high in the Red River Delta provinces. Figure 3.11: Carry over spending/local budget (%) Figure 3.12: Factors that may impact carry over spending -1SD Median +1SD 0 20 40 0 500 Actual local 60 18 carry-over spendings as % 40 budgeted local 20 16 expenditure 0 40 8 10 12 14 Local capital spending as 20 % local spending Frequency 0 60 Local recurrent spending as % local 40 spending 20 500 Actual vs. 6 budgeted targeted transfers from center to 4 province 0 100 2 Decentralized revenue as % 50 0 local revenue 0 10 20 30 30 50 60 70 0 0 20 40 60 20 40 60 0 50 100 Actual local carry-over spendings as % budgeted local expenditure Figure 3.13: Average carry over spending/local budget 2006-2011 by region (%) 60 50 2006 2007 2008 2009 2010 2011 40 30 20 10 0 Red River Delta Northern Central and Central Highlands Northern Mountain Mekong River Delta South East Central Coast Source: Staff estimates based on published state budget data 24. In many countries, unspent appropriations are usually re-appropriated in the following year’s budget. Some countries have started to allow unspent budget spending authority to be carried over into the next fiscal year. For recurrent spending, if allowed at all, a firm quantitative limit is set on carry-over of recurrent spending authority – and on the type of recurrent spending. Carryover of investment spending is more frequent, with restrictions.13 In some countries, the Government or the MOF approves carryover requests. In Japan and Korea, end-year carryover of unspent appropriations is allowed following approval from the National Assembly. This section draws on WB, “Revising Vietnam’s State Budget Law 2002: Proposals Drawing on International Experience,” (2014), 13   paragraphs 101, 103-105 25. In Japan, the Ministry of Finance approves ministries’ individual carryover requests on a case by case basis. In Korea, carryover is authorized for contracts made in the fiscal year, but where payment 73 is not possible in the same year due to unavoidable reasons. In other countries, instead of discretionary control by the Ministry of Finance, quantitative limits are imposed on carryover. Sweden, for example, has a carryover limit of up to 10 percent provided for in its 1996 State Budget Act. In practice, this upper limit is administered more restrictively by the government, which operates with a guideline of a 3 percent carryover limit for agencies’ cash appropriations. Norway specifies the appropriations that are eligible for carry-over. Concerning investment, expenditures for building, construction and materials can be carried over, if approved by Parliament, and for up to two years. 26. While there is no specific norm for the extent to which carryover of investment could be approved in Vietnam, it may be appropriate to begin cautiously, by allowing carryover for investment spending only in the SBL, without specifying a quantitative limit in the law. The Ministry of Finance and the Ministry of Planning and Investment could issue a joint regulation limiting end-year carryover of investment project spending to 5-10% for example, perhaps with differing percentages for different types of investment spending. 27. As an alternative to formalizing carryover for investment spending, the SBL could allow multi-year appropriations for investment spending. This would need to be aligned with the Medium-Term Investment Plan, developed recently by the Ministry of Planning and Investment, and formalized under the recently approved Public Investment Law. The addition to the annual budget’s cash-limit appropriations for 12 months, investment spending could be provided to allow multi-year appropriations for spending commitments (when the government enters into a contract that will involve payments at a future time period). An upper limit would be set for spending to be committed to investment projects, for which payments take place over time. This option would require government agencies and the MOF to be able to account for spending at the commitment stage, as well as at the payment stage. Its suitability for Vietnam needs careful investigation – it may be premature. 28. Implications of current policies on planning and budgeting outcomes: The above policies and practices contribute to a loss in spending plan predictability, which in turn dilutes the links between local development plans and the budget. Budget appropriations are approved by the legislature on the basis of agreed development plans. If the executive has the discretion to change appropriations, then the legislature can only verify ex post whether the revised budget is aligned with the original plan. Budget adjustments can of course be a good thing, and can also lead to a better outcome in terms of implementing development plans. Additionally, not every adjustment to appropriations needs to be approved by the legislature. But these adjustments need to be transparent, and adequately justified to the legislature, which approves the original appropriations. 29. Local development plans should also explicitly set out the role of local budgets in helping to meet socio- economic goals and targets. Despite the significant role of local expenditure in provincial economies as highlighted in chapter 2, there is little coverage of public finances in selected provincial-level Socio-Economic Development Plans (2011-2015) reviewed during this study. All these SEDPs set out growth and poverty reduction targets, including service delivery objectives. But there is no discussion of potential costs of these targets, how they will be financed, and the role of the public sector relative to other actors in service delivery. A stronger link between local SEDPs and public finances is essential and is a criteria that the government may wish to include in the preparation of SEDPs for the 2016-2020 period. 30. Although spending over budget has not led to debt problems like in other countries, local authorities still need to be mindful of its impact on medium to long-term fiscal sustainability. Despite regularly overshooting planned spending, consistent underestimation of local receipts has helped to keep in check the fiscal balances of local authorities. It is important to note however that the policy of spending 50 percent of 74 over-realized revenue on salary in the next budget year and the other 50 percent on capital projects create significant future commitments to the budget. There are no medium-term projections of these commitments in local budgets. However, salary reform expenditure creates non-discretionary compensation entitlements for civil servants. Capital expenditure on the other hand creates commitments in terms of operational and maintenance needs. In addition, the situation could be compounded by pro-cyclical fiscal policies as local authorities spend with limited control at times when revenues are buoyant, further exacerbating sustainability issues. 31. These policies can also impact negatively on efficiency of spending. As discussed in chapter 2, these fiscal rules and norms not only restrict local autonomy but also earmark spending on areas that may be sub-optimal in relation to actual needs of the provinces. The ideal of course would be to either have more realistic revenue forecasts or institute tighter controls over spending of over realized revenues. Either way, addressing this should have positive knock effects on the credibility of spending plans, levels of carry over and general transparency of the budget. The expectation is that this would be positive for efficiency of spending overall, which is discussed in more detail in the next section. RELATIVE SPENDING ON AND EFFICIENCY OF SERVICE DELIVERY ACROSS PROVINCES 32. Tightening fiscal conditions in Vietnam calls for an increased understanding of the relative spending on and efficiency of public service provision across provinces. A slowing economy and increasing counter- cyclical and investment-related tax incentives have in recent years led to slowing revenue collections. This has had a very direct impact on inter-governmental fiscal relations in the last two to three years. In particular, the central government has introduced stricter limits on spending of over-realized revenue, requiring these to be channeled to existing appropriations rather than salary reforms or new investments. If provincial authorities raise below what is budgeted, central authorities have stipulated that they have to first find financing from existing sources (e.g. Financial Reserve Fund) before requesting transfers from the center. 33. A key consideration for central government therefore is the level of output and outcome achieved for a given level of input across local authorities, and ways to improve this. This section attempts to look at this issue in three steps. The first is by comparing the different levels of output achieved by provinces for a given level of inputs. This is done by computing Public Sector Performance indicators (PSP) using a range of sector output and outcome data, and Public Sector Efficiency Indicators by relating PSPs to spending by local authorities. The second is to analyze the relative efficiency of provinces against a production possibility frontier that is determined by the most efficient provinces. This is based on Data Envelopment Analysis (DEA), which helps to assess how much less resources a province could theoretically use (all other things constant) to produce the same level of output it is currently producing. The third part tries to analyze what factors might explain the differences in relative efficiency across provinces. It tries to relate this analysis to discussions in chapter 2 on spending decentralization and accountability, and the first part of this chapter on the quality of spending plans. 34. This approach is based on methodology used in the existing literature on productive efficiency of sub- national authorities. In summary, Afonso and Fernandes (2007) note two strands of empirical research on efficiency and decentralization: overall efficiency of local authorities, and efficiency in provision of specific services. Afonso et al (2013) highlight that there is only a small strand of literature looking at efficiency of sub-national authorities. This may be due to data limitations that prevent having complete input and output information across all local authorities, particularly in developing countries. The World Bank has also conducted studies of local government efficiency in Colombia (WB 2009) and Peru (WB 2010). These looked at relative efficiency of providing specific services including health, education, and transportation. 75 35. In Vietnam, the study attempts to assess the relative spending overall public service delivery across different local authorities. It uses output/outcome and public spending data in three sectors (education, health and road transportation) to assess overall public sector efficiency of a local authority. Local authorities have most of the responsibility over these services, particularly education and health as seen in chapter 2. Therefore using these services could ensure a better link between local spending and observed outputs, than compared to if central authorities had significant responsibility for providing these services at the local level. 36. In reality of course, it is not possible to fully attribute performance in these sectors to public spending because results may also be due to private provision of services or out-of-pocket expenses (e.g. school or hospital fees). In addition, outputs and especially outcomes may depend on conditions that are outside the influence of local authorities. For example, education outcomes can be highly determined by parents’ education levels, income per capital levels, and other initial endowments. Similarly in health, a large share of financing comes out of health insurance, whereas the analysis below only takes direct spending out of local budgets. 37. The results of the efficiency analysis therefore need to be carefully interpreted. There are clearly agency and measurement challenges around the proposed results chain from inputs to outputs. Agency challenges as noted above come from the fact that multiple levels of government (or even private and other non-government sector) are either spending or engaging on an output or outcome. Measurement challenges arise as it is difficult to measure the extent to which inputs are confined to a particular jurisdiction. On top of this, the quality of data across the chosen sectors may vary, and as noted above attribution is an issue. 38. Nonetheless, it is important to start investigating sub-national spending efficiency in Vietnam. The approach in this study is to identify emerging issues based on cross-sectional analysis. It focuses in particular on whether there are any links between intergovernmental fiscal policies discussed in this report and the efficiency of local authorities. It is therefore important to note that the purpose of the analysis is not to make recommendations on the appropriate spending mix to achieve better results within a particular sector. The analysis should provide the central fiscal authorities (Ministry of Finance and Departments of Finance) a first review of relative productive efficiency across provinces, and make recommendations on how these could be improved through reform of central-local fiscal policies. 39. Measuring inputs using public expenditure data: The input data for the efficiency analysis is per capita recurrent expenditure in education, health and population, road transportation of all local authorities within a province (provincial, district and commune authorities). This is expenditure that is fully assigned to local authorities. The input data therefore does not include central government expenditure in provinces. As noted earlier, this could potentially overestimate the impact of local spending on outputs/outcomes in education, health or transportation, particularly in provinces where central provision is high or where private financing is also high. 40. The input data does not include a breakdown of spending by each tier (i.e. spending by provincial authorities vs. spending by district authorities vs. spending by commune authorities). This means that it will not be possible to look at the productive efficiency of province authorities vs. that of district authorities for example. However, the analysis is still valid and important because it will allow a review of productive efficiency of all local authorities within a province. Moreover, the output/outcome data is for provinces as a whole, and more difficult to obtain at district level across so many sectors. 41. Measuring Public Sector Performance (PSP) using output and outcome data: The study took 76 local level output and outcome indicators in the education, health and transportation sectors to derive a PSP score for each province. The full list of output and outcome variables for each sector is provided in annex A. The output/outcome data was converted to a composite index for each of the sectors in each of the provinces. This was done by: (i) computing the overall average for an output/outcome variable across all provinces; (ii) normalizing the output/outcome value in each province by dividing it by this overall average; (iii) averaging the normalized values of all output/outcome indicators for a sector in a province to derive the sector performance index in that province. The three sector performance indices are then averaged to derive the overall PSP score for a given province. PSP scores range from 0.77 to 2.28, with an average of score of 1. 42. Public Sector Efficiency (PSE): PSE scores are then computed for each province by dividing the overall PSP score by the average per capita spending across all three sectors. PSE scores range from 0.62 to 2.82 with an average of score of 1.4. The results comparing performance versus efficiency are summarized in figures 3.14 to 3.17 below. The four charts group provinces into four quadrants according to their relative performance on PSP and PSE scores: (i) the bottom left quadrant are provinces that have below average scores on PSP and PSE (i.e. poor sector outputs and outcomes, and high spending service delivery); (ii) the top right quadrant are provinces that have higher than average scores on PSP and PSE (i.e. good sectors outputs and outcomes and low spending on service delivery); (iii) the bottom right quadrant are provinces that have higher than average scores on PSP and lower than average scores on PSE (i.e. good outputs and outcomes but high levels of spending); and (iv) the top left quadrant are provinces that have below average scores on PSP but higher than average scores on PSE (i.e. poor performance but efficient). 43. Possible implications of PSP and PSE scores: Patterns emerge when looking at performance on service delivery outputs and outcomes versus efficiency of service delivery at local level. It is clear that the poorest provinces face the most challenges in terms of service delivery performance and efficiency of spending. All of the provinces in the Northern Mountainous Region as well as most of the provinces in the Central Highlands Region are in the bottom left quadrant. For a given amount of public spending, it is difficult to achieve the same outputs and outcomes in these regions compared to others. These are also provinces where, as discussed in chapter 2, local spending constitutes a large share of the local economy. On the other hand most of the provinces in the wealthier South East, and half of the provinces from the wealthier Red River Delta, perform above average on service delivery and efficiency. Public sector performance and efficiency indicators across provinces 77 Figure 3.14: Lower than average PSP and higher Figure 3.15: Higher than average PSP and higher than average PSE than average PSE 1.2 Public Sector Performance Public Sector E ciency 2.5 Public Sector Performance Public Sector E ciency 2.5 3.0 1.0 2.0 2.0 2.5 0.8 1.5 2.0 1.5 0.6 1.5 1.0 1.0 0.4 1.0 0.2 0.5 0.5 0.5 0.0 0.0 0.0 0.0 Lam Dong Binh Duong Hai Duong Hai Phong Dong Nai Tien Giang Binh Dinh Nam Dinh Khanh Hoa Can Tho Ho Chi Minh Tay Ninh Thai Binh Ha Noi Ben Tre BRVT Hau Long Kien Vinh An Dong Bac Phu Binh Ninh Bac Ha Hung Binh Giang An Giang Long Giang Thap Lieu Yen ThuanThuan Ninh Nam Yen Phuoc Mekong River Delta Northern Central Red River Delta South and Central Coast East CH Mekong River NCCC Red River Delta South East Delta Figure 3.16: Lower than average PSP and lower than Figure 3.17: Higher than average PSP and lower than average PSE average PSE 1.2 Public Sector Performance Public Sector E ciency 1.6 Public Sector Performance Public Sector E ciency 1.0 1.4 1.4 1.4 1.2 0.8 1.0 1.2 1.2 0.6 0.8 1.0 1.0 0.4 0.6 0.4 0.8 0.8 0.2 0.2 0.0 0.0 0.6 0.6 Lang Son Tuyen Quang Thai Nguyen Quang Binh Dak Nong Quang Nam Soc Trang Cao Bang Bac Giang Ha Giang Lai Chau Son La Vinh Phuc Ninh Binh Quang Tri Kon Tum Dak Lak Phu Tho Lao Cai Hoa Binh Dien Bien Ha Tinh Yen Bai Gia Lai Bac Kan Tra Vinh TTH Ca Mau 0.4 0.4 0.2 0.2 0.0 0.0 Central Mekong Northern Northern Mountain Red Da Nang Nghe An Thanh Hoa Quang Ngai Quang Ninh Highlands River Central and River Delta Central Coast Delta Northern Central and Central Coast Red River Delta Performance Public Sector 74 44. There are some outliers both in the top and bottom performers’ quadrants. Wealthier Vin Phuc and middle income Ninh Binh in the Red River Delta both perform relatively poorly on service delivery and efficiency. One of the reasons for this might be because both of these provinces have lower population density relative to other provinces in the same region, which increases spending on service delivery. It is also interesting to note that Ninh Binh has spent 90 percent over budget on average between 2009 and 2011, and Vinh Phuc around 70 percent, compared to an average of 50 percent across all the provinces. Therefore both provinces perform below average on credibility of spending plans, which itself is high in the first place. 45. On the other hand, Lam Dong in the Central Highlands, and Tien Giang, Can Tho and Ben Tre, all in the Mekong River Delta, perform strongly on service delivery and efficiency compared to their peers in the same region. Lam Dong has higher level of urbanization and population density, lower levels of poverty, and the most credible expenditure plans compared to other provinces in the same region (Dak Nong, Dak Lak, Gia Lai, and Kon Tum). Similarly, Tien Giang, Can Tho and Ben Tre are among the more densely populated in the Mekong River Delta. However, aside from Can Tho, the other two provinces perform below average in terms of budget credibility. The next section further below explores in more detail the relative efficiency between provinces and possible explanatory variables for these. 46. A small group of provinces in the Northern Central and Central Coast, and Quang Ninh in the Red River 78 Delta, perform above average on service delivery but with relatively high spending. Interestingly all of these provinces also perform below average in terms of budget credibility, spending nearly 60 percent above budget on average between 2009 and 2011. Though it is also important to note that with the exception of Danang, these provinces are also among the poorest in the region and have the lowest population density. Therefore it is worth exploring further some of possible causes for the relatively high spending on service provision in Danang. 47. Despite higher population density and urbanization and lower poverty, some provinces in NCCC perform below average on service delivery but above average on efficiency. There are also a number of provinces in the Mekong and Red River Deltas, which fall in the same category. One interesting pattern is that in most of these cases, local spending as a share of local GDP is relatively low at below 21 percent. The exceptions are Ninh Thuan, Hau Giang, Ha Nam and Phu Yen where local spending is above 26 percent of local GDP – though these provinces have relatively higher poverty rates compared to others in this group and are more dependent on balancing transfers. Therefore, for those provinces that are achieving higher efficiency levels, but lower than average service delivery, there may be room to increase local public spending on service delivery. The fiscal space would need to be through higher balancing transfers. The possibility of increasing service fees could also potentially be considered in provinces where poverty rates are relatively lower (e.g. Binh Thuan, Vinh Long, An Giang, Kien Giang, Long An, Bac Ninh). But this would first require a review of fee setting practices in those provinces. 48. How efficient provinces are relative to each other: Building on the above analysis, this section looks more closely at how each province performs against an “efficiency frontier”. The idea is to calculate or estimate an efficient frontier function, which reflects the maximum output (as measured by the PSP discussed above) attained by local authorities in Vietnam, using a given set of inputs (as measured by per capita spending) and existing production technologies. The performance of each province can then be measured against this efficient frontier function. This can give an estimate of the amount by which each province is spending over and above the minimum needed to achieve the same level of output (input method).14 A province may of course need to spend more than the minimum defined by the efficient frontier function because of factors outside its control that lead to high spending on service delivery (e.g. difficult access in mountainous regions, low population density). These issues are considered in the next section of this chapter.15 49. To estimate efficiency scores of provinces relative to the best performing provinces, the study uses the Data Envelopment Analysis technique.16 In the DEA, also referred to as the non-parametric approach, the functional relationship between inputs and outputs (or the production function) is calculated via a mathematical programming model or an econometric technique applied to a sample of observed data. From this sample of observed data, a frontier envelopment surface is defined and the provinces that lie on that frontier are termed productively efficient and are assigned a DEA score of one. The provinces that lie inside the frontier are considered productively less efficient relative to those on the frontier (Murillo- Zamorano 2004, Ramanathan 2003). The percentage difference between the score for a province and the score of 1 is an estimate of the spending differential between that province and the most efficient province in delivering a certain level of output. 14  With the input-based approach, an inefficient province is one that produces the same level of output with more inputs compared to an efficient province. With the output approach, an inefficient province is one that produces less output with the same level of inputs compared to an efficient province. 15 As noted above, some factors are not possible to control for in terms of their impact on outputs and outcomes e.g. the impact of central government programs, private sector, initial endowments. 16 Further detail on the empirical methodology is provided in annex B. 50. The results of the DEA for each province are summarized in figures 3.18 to 3.21 below. The provinces are grouped in each chart according top, second, third and fourth quarter in terms of their relative productive 79 efficiency. Figure 3.68 therefore includes the top 16 provinces in terms of delivering a given set of public services or outputs at the lowest possible spending. Provinces such as An Giang, Hau Giang, Tien Giang and others with a score of 1 are those that come out as the most productively efficient. This as discussed further below is not necessarily an indication of whether a province is delivering an adequate level of public services. It is an indication that the province is delivering a certain level of services with lower spending than other provinces that are delivering the same level of services. The average DEA score across all provinces is 0.75, which suggests that on average, local authorities in Vietnam could have theoretically achieved the same level of service delivery with 25 percent fewer inputs. Provinces in the fourth quarter have the highest spending for delivering a given set of services. On average they spend 50 percent more to provide the same level of services as the most efficient province. Figure 3.18: Provinces in top quarter on DEA score Figure 3.19: Provinces in second quarter on DEA scores 1.01 DEA PSE 3.0 DEA PSE 1.0 0.940.90 0.88 0.85 2.5 1 1 1 1 1 1 1 1 1 1 1 1 1 0.89 0.86 0.83 0.81 1 0.9 0.83 0.82 0.79 0.79 1.00 2.5 0.80 0.79 0.77 0.8 0.74 2.0 0.99 0.7 2.0 0.98 0.6 1.5 1.5 0.5 0.97 0.96 0.96 0.4 1.0 1.0 0.96 0.3 0.2 0.5 0.95 0.5 0.1 0.94 0.0 0.0 0.0 Ho Chi… Hung Yen Dong Thap Soc Trang Da Nang Ninh Thuan Binh Duong Phu Yen Vinh Long Binh Thuan Can Tho Hai Duong Hoa Binh Long An Bac Lieu Tay Ninh Bac Ninh Dong Nai Ha Nam TT Hue Hau Giang An Giang Tien Giang Kien Giang Binh Dinh Binh Phuoc Nam Dinh Ca Mau Ben Tre Thai Binh Ha Noi Tra Vinh Figure 3.20: Provinces in third quarter on DEA Figure 3.21: Provinces in fourth quarter on DEA scores scores DEA PSE 0.7 0.59 0.58 DEA PSE 1.6 0.73 0.71 0.56 0.8 0.73 0.68 1.8 0.59 0.58 0.52 0.71 0.69 0.65 0.62 0.61 0.57 0.50 0.49 1.4 0.66 0.6 0.55 0.7 0.65 0.60 0.59 1.6 0.45 0.62 0.61 0.51 0.50 0.60 1.2 1.4 0.5 0.47 0.6 0.5 1.2 0.38 1.0 0.4 1.0 0.8 0.4 0.8 0.3 0.3 0.6 0.6 0.2 0.2 0.4 0.4 0.1 0.2 0.1 0.2 0.0 0.0 Tuyen… 0.0 0.0 Quang… Quang… Quang… Quang… Lam Dong Lang Son Dak Nong Thai Nguyen Khanh Hoa Thanh Hoa Bac Giang Quang Tri Lao Cai Phu Tho Dak Lak Ninh Binh Yen Bai Gia Lai BRVT Hai Phong Nghe An Cao Bang Ha Giang Lai Chau Son La Kon Tum Vinh Phuc Dien Bien Ha Tinh Bac Kan Source: Staff estimates 51. Another way of illustrating the above results is to look at where the different provinces lie relative to 80 the productive efficiency frontier. This can be done by plotting for each province its per capita spending on education, health, and transportation against its PSP scores. The efficiency frontier in the plot will be defined by selected provinces that have achieved a DEA score of one. The horizontal distance between the efficiency frontier and where a particular province lies is approximately that province’s input-based DEA score as derived above. 52. To estimate the efficiency frontier, the figure below plots the log of per capita spending against PSP scores for the fourteen provinces that have a DEA score of one. The efficiency frontier is defined around two provinces: Hau Giang and Hanoi, which is illustrated in figure 3.20 below. It would in theory be possible to draw the frontier even further out by including: Tien Giang; or Binh Duong and HCMC; or An Giang, Dong Nai and HCMC. The reason for not doing so is because these provinces are outliers in terms of having high per capita income on the one hand and a low share of local spending in the local economy on the other. The probability of private sector provision of education, health and transportation services is likely to be higher in these outlying provinces. This means as discussed earlier that there may be an upward bias in the PSP score when comparing it to per capita public spending. Figure 3.22: Efficiency frontier 2.5 Ho Chi Minh Hanoi 2.0 1.5 PSP scores Binh Duong Dong Nai Tien Giang Thai Binh Binh Dinh Nam Dinh An Giang 1.0 Vinh Long Hau Giang Binh Phuoc Tra Vinh 0.5 0.0 12.9 13.0 13.1 13.2 13.3 13.4 13.5 13.6 13.7 13.8 Log of per capita spending Source: Staff estimates 53. The same could of course apply to Hanoi, where the share of private sector provision may be relatively high compared to poorer provinces. At the same time, however, local government spending as a share of the local economy in Hanoi (23 percent of local GDP) is considerably higher than the average for the outliers (14 percent of local GDP). This may of course be due to spending that is not related to education, health or transportation. However, average per capita spending in these three areas is not only higher in Hanoi than in the outliers but also in the top quarter when compared to all provinces. Part of this may be related to the merger of Hanoi with Ha Tay in August 2008, when levels of public service delivery by Hanoi authorities had to increase to meet demands in Ha Tay. 54. Based on the above, figures 3.23 to 3.28 below plot provinces in each region against the production efficiency frontier. A similar pattern emerges to the earlier analysis of the Public Sector Performance versus Public Sector Efficiency. Spending on service provision in poorer, mountainous provinces in the Central Highlands (figure 3.24) and the Northern Mountains (figure 3.26) is clearly very high. Provinces in the Northern Mountains region on average spend 42 percent more to deliver the same level of services as the most efficient 81 provinces; provinces in the Central Highlands spend on average 35 percent more. This is followed by the Northern Central and Central Coast where spending is on average 29 percent higher, and the Red River Delta where on average spending is 21 percent higher. Provinces that deliver services at least cost relative to the efficiency frontier are located in the South East (figure 3.28), where spending is on average 7 percent higher compared to the most efficient; and the Mekong River Delta (figure 3.24) where spending is on average 9 percent higher. 55. Within regions, it is interesting to look at the relative performance of some of the outlying provinces. In the Central Highlands, for example, it requires the province of Kon Tum 45 percent more to provide a given level of services relative to the efficiency frontier. Within the region, Kon Tum stands as the poorest and least densely populated province, which has a very high share of local spending in the local economy (69 percent). It is also interesting to note however, that among the five provinces in the region, Kon Tum also has the highest divergence between actual and budgeted revenue (40 percent compared to an average of 26 percent for the others); the highest divergence between actual and budgeted spending (60 percent compared to an average of 38 percent for the others); and the highest level of carry over spending (40 percent compared to 26 percent for the others). 56. In the Northern Mountains, the poorest 7 out of 14 provinces also exhibit the highest level of spending on service provision relative to the efficiency frontier. This includes the provinces of Lai Chau, Dien Bien, Ha Giang, Lao Cai, Cao Bang, Son La and Bac Kan. On average, there is 50 percent more spending to provide a given level of services in these provinces compared to 35 percent for the other 7 provinces in the Northern Mountains region. These provinces also happen to have low credibility spending plans (particularly Cao Bang, Ha Giang, and Bac Kan that spend nearly 90 percent above budget) and large expenditure carry overs. 57. In the Northern Central and Central Coast region it is interesting to note that Danang, which performs well on service delivery, has to spend 21 percent more to provide a given level of service relative to the efficiency frontier. Danang, however, is the most urbanized, densely populated province in the region with the lowest levels of poverty. The province of Danang has on average spent more than 60 percent of its budget in the past three years, and carries over nearly 49 percent of its budgeted expenditure plans. Local spending as a share of local GDP is high at 42 percent, but districts are responsible for only 16 percent of recurrent spending in the province. 58. In the Red River Delta, the provinces of Quang Ninh, Hai Phong, Vinh Phuc and Ninh Binh exhibit the highest level of spending on service delivery relative to the efficiency frontier. On average these four provinces have to spend 33 percent more to provide a given level of services relative to the efficiency frontier compared to only 9 percent for the other provinces in the Mekong River Delta region. With the exception of Hai Phong, the reason for the relatively higher level of spending is likely due to the fact that Quang Ninh, Vinh Phuc and Ninh Binh have the lowest population density of all provinces in the region. All four provinces on average carry over nearly 50 percent of the budgeted spending plans. 59. Provinces in the Mekong River Delta are quite closely clustered around the production efficiency frontier. These provinces have lower than average local spending as a share of local GDP, and with the exception of Ca Mau and Kien Giang, also have higher than average population density contributing to economies of scale in service provision. The South East region exhibits a similar pattern of relatively low level of spending for a given amount of services delivered. Although figure 3.28 shows Ba Ria Vung Tau a bit removed from the productive efficiency frontier, the deviation is not significant. Provincial performance against efficiency frontier across regions 82 Figure 3.23: Central Highlands Provinces against efficiency frontier 2.5 Hanoi 2.0 PSP scores 1.5 Hau Giang Lam Dong Gia Lai Dak Lak Kon Tum 1.0 Dak Nong 0.5 0.0 13.0 13.2 13.4 13.6 13.8 14.0 14.2 Log of per capita spending Figure 3.24: Mekong Delta Provinces against efficiency frontier 2.5 Hanoi 2.0 1.5 PSP scores Tien Giang Long An Can Tho Hau Giang 1.0 Vinh Long Ben Tre Kien Giang Dong Thap Tra Vinh Ca Mau Bac Lieu Soc Trang 0.5 0.0 13.0 13.2 13.4 13.6 13.8 14.0 14.2 Log of per capita spending Figure 3.25: NCCC Provinces against efficiency frontier 2.5 Hanoi 2.0 PSP scores 1.5 Da Nang Binh Dinh Nghe An Khanh Hoa Quang Ngai Ha Tinh Binh Thuan 1.0 Hau Giang Phu Yen Thanh Hoa Quang Tri Quang Nam Ninh Thuan Thua Thien Hue Quang Binh 0.5 0.0 13.0 13.2 13.4 13.6 13.8 14.0 14.2 Log of per capita spending Figure 3.26: Northern Mountain Provinces against efficiency frontier 83 2.5 Hanoi 2.0 PSP scores 1.5 Hau Giang Thai Nguyen Lang Son 1.0 Bac Giang Phu Tho Ha Giang Hoa Binh Son LaCao Bang Lai Chau Tuyen Quang Yen Bai Bac Kan Lao Cai Dien Bien 0.5 0.0 13.0 13.2 13.4 13.6 13.8 14.0 14.2 Log of per capita spending Figure 3.27: Red River Provinces against efficiency frontier 2.5 Hanoi 2.0 1.5 PSP scores Thai BinhNam Dinh Hai Duong Hai Phong Hau Giang 1.0 Ha Nam Vinh Phuc Quang Ninh Bac Ninh Hung Yen Ninh Binh 0.5 0.0 13.0 13.2 13.4 13.6 13.8 14.0 14.2 Log of per capita spending Figure 3.28: South East Provinces against efficiency frontier 2.5 Ho Chi Minh 2.0 Hanoi 1.5 PSP scores Dong Nai Hau Giang BRVT 1.0 Binh Phuoc An Giang Tay Ninh 0.5 0.0 13.0 13.2 13.4 13.6 13.8 14.0 14.2 Log of per capita spending 60. Understanding the determinants of productive efficiency: The above points to a number of factors 84 that may explain the relative levels of productive efficiency across different provinces. Some of the issues that seem to stand out include the size of local government in the local economy, population density, overall capacity and level of development, geography, and the quality of spending plans. The study now looks to estimate the relationship between some of these factors and the relative level of efficiency across provinces. Data on determinants of productive efficiency are grouped in five categories: (i) level of fiscal decentralization; (ii) socio-economic features of the province; (iii) fiscal trends; (iv) quality of fiscal management; (v) institutional factors. The list of variables and data sources are listed in annex C.17 61. It is important to note that there are a number of factors below that are not necessarily under the control of local authorities. This applies to structural factors such as the socio-economic features of the provinces. They can still point to policy relevant conclusions. But the focus may be on those factors that local authorities can control such as quality of fiscal management and institutions. 62. The table below presents the results of the analysis of the potential determinants of provincial productive efficiency using the censored normal Tobit regression model. This specific model is used given that there is a right-censoring in the dependent variable (i.e. the DEA efficiency scores). In this second stage analysis, the Tobit regression model was run on all provinces. Specific estimation procedures were required in an attempt to address potential econometric issues. Due to of the possible nonlinear effects of some variables and in order to ensure normally distributed residuals, they were entered in the regression in logarithm forms. The second model is estimated in order to eliminate multicollinearity among independent variables. The overall similarities in terms of the statistical significance and the sign of the estimated coefficients show that the results are somewhat robust to alternative specifications. 17 For the decentralization indicators, data on central spending in each province was not available to calculate what share of spending in the province is by central authorities and what share is by local authorities. It is therefore not possible to estimate expenditure decentralization by province. Some of the results from the PCI indicators are perception-based. This might affect the objectiveness of the indicator. There is also a lot of information available through the Provincial Governance Performance Index (PAPI) on transparency of the budget, participation in the budget process, and quality of service delivery. But this is also perception based, and the survey is done at commune level. Therefore it may not be suitable when looking at productive efficiency across all local authorities. However, this information will be very useful when looking at allocative efficiency issues and accountability of sub-provincial authorities Table 3.2: Determinants of provincial efficiency performance in Vietnam 85 (2) Provincial efficiency performance (dea (1) Correction for scores) All vars multi-collinearity Ln decentralized revenue as % of local -0.075 revenue (0.178) Level of fiscal Ln central balancing transfers as % of local -0.175*** -0.015 decentralization core expenditure (0.054) (0.041) District expenditures as % of local 0.108 -0.183 expenditures (0.173) (0.177) -0.582*** -0.402*** Ln per capita gdp (0.120) (0.121) 0.096** 0.060* Ln population density Socio-economic (0.035) (0.034) features 0.054 Ln provincial poverty rate (0.064) 0.070* 0.019 Ln urbanization rate (0.040) (0.039) -0.202 Ln decentralized revenue/local recur. Spend (0.167) -0.926*** -0.363** Ln local recurrent expenditure/local spend (0.242) (0.178) Fiscal trends Provincial admin spend as % of total local 0.272 expenditure (0.342) Ln local spending/gdp -0.630*** -0.370*** (0.120) (0.103) -0.003** Local expenditure as % of local budget (0.002) Fiscal Local revenue as % of local budgeted 0.000 -0.001 management revenue (0.001) (0.001) Ln local carry over spending as % of local -0.002 -0.046 budget (0.046) (0.050) -0.001 0.006* Qualification of local administrative staff (0.003) (0.003) -0.086 Ln of % of graduates in province (0.059) Ln % of firms paying over 10 percent of their -0.057 -0.008 Institutional revenue in extra payments (0.049) (0.051) factors 0.407** 0.289** Ease of access to provincial budgets (0.159) (0.126) Ln provincial officials’ knowledge of national -0.674** laws relating to private sector (0.288) Provincial officials’ ability to solve problems 0.005 for private sector (0.003) 2.244* 0.197 Constant (1.101) (0.560) Observations 46 46 Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1 63. According to the traditional literature, fiscal decentralization is expected to allow local government 86 officials to not only provide services in accordance with the tastes and preferences of local citizens (allocative efficiency) but also to better match expenditures with revenue needs (correspondence principle). This link between expenditure and the price of public goods is essential to achieve efficiency because it would equalize the benefit per unit of service with the cost per unit of service. As a corollary, it should impose fiscal responsibility at the margin on sub-national governments.18 Furthermore, decentralization is expected to increase accountability of local authorities to their citizens in that those paying taxes will demand accountability for service delivery and will monitor municipal performance. 64. In contrast, therefore, central balancing transfers as percentage of local core expenditures in Vietnam are expected to reduce the spending efficiency of provinces. The results of the second model in the table confirm that dependence on central transfers has the expected sign but its impact on provincial efficiency is not statistically significant. This result is probably influenced by the wide disparities across provinces in the central government transfers they receive. Similarly, district expenditures as percentage of local expenditures, a measure of decentralization within the province, is found to have no significant impact on provincial efficiency. 65. The various measures of socio-economic factors such as per capita income, population density and urbanization rate are expected to positively influence the performance outcomes of provinces in Vietnam. This came out quite strongly in the earlier analysis in this chapter. In effect, richer residents would tend to exercise greater pressure in their demands for efficient provincial services; and economies of scale could exist when providing local public services for an enlarged number of residents (high density or high urbanization rates), which would then increase efficiency. Results show that density and urbanization rates indeed have a positive effect on provincial efficiency although not statistically significant in the case of urbanization rate. This reaffirms the importance of having geographic criteria for cost norms to estimate spending needs when determining balancing transfers to local authorities. The government may also consider using population density as a criterion to determine spending needs. 66. Per capita GDP was found to reduce provincial productive efficiency. This is surprising given the earlier analysis, which showed that poorer provinces tend to have higher spending on service provision relative to the efficiency frontier. The result may give rise to the possibility that poorer residents in Vietnam tend to be more interested in more efficient services. One possible explanation for this may be due to higher level of participation and inspection by community groups in poorer provinces compared to richer provinces. This was evident in chapter 2, which noted that respondents to the PAPI survey in a number of poorer provinces noted favourable conditions for local participation and vertical accountability.19 67. With regards to institutional factors, local capacity and the transparency of planning documents were found to be positively significant in explaining efficiency. This confirms the traditional local spending efficiency empirical literature; indeed the education level of local residents is extremely relevant in the local efficiency analysis for it would provide citizens with the ability to pressure local officials to improve quality of services and to monitor their actions. Additionally, greater transparency of planning documents would foster greater accountability, thus improving both the efficiency and the effectiveness of public services. 18 A resource allocation mechanism is efficient when marginal (social) benefit is equal to marginal (social) cost. In general, over consumption of public goods is a common characteristic of lack of correspondence and fiscal autonomy. CECODES, VFF-CRT, UNDP, “The Vietnam Provincial Governance and Public Administration Performance Index, PAPI: Measuring 19  Citizens’ Experiences,” (2013) 68. The size of the local public sector as measured by local spending as a percent of GDP was found to have a negative impact on efficiency. This is in line with what other studies have also found in terms of 87 factors affecting productive efficiency (Afonso et al 2013). It is also in line with earlier discussion in chapter 2 and in this chapter on poorer provinces having higher levels of local spending to address public service delivery needs often in geographically remote areas. The lack of credibility of spending plans has a slight negative impact on productive efficiency; though in the second model estimation, the impact is not statistically significant. 69. It is interesting though that in the fiscal trends indicators there is a negative and statistically significant impact of the share of recurrent expenditures to total local expenditure on provincial performance efficiency. This does not necessarily imply that provinces with higher levels of capital spending are productively more efficient. Rather, the result may be signaling issues with the actual composition of recurrent expenditure, which in turn may link back to the lack of credibility of spending plans. It was argued earlier in the chapter that current policies of spending a set portion of over-realized revenue on salary reforms could have negative implications for the efficiency of the wage bill. This is looked at briefly in the next section. 70. Composition of recurrent spending: It is very difficult to obtain information on wages and salaries in Vietnam’s State Budget, even though this is critical to understanding fiscal sustainability. Different tiers of government selectively report on wages and salaries of administrative units within their jurisdiction. But this does not provide a consolidated picture of the overall wage bill and how it is affected by budget policies. This includes policies on channelling half of over-realized revenue to salary reforms (i.e. in-year additions to the overall wage bill), which as indicated above may adversely impact the efficiency of spending. 71. This section takes a brief look at the relationship between the share of salary spending in total recurrent spending and relative efficiency of a province as measured by its DEA score. To estimate the share of salary spending in recurrent spending, the study takes the weighted average of salary spending in education, health and transportation between 2004 and 2011.20 The purpose is not to assess the appropriate level of salary spending but to identify whether this could be an issue to look at in greater depth in selected provinces given their relative productive efficiency. 72. Figures 3.27 to 3.32 below plots each province’s DEA score against the share of wages and salaries’ in recurrent expenditure for that province. The figures group provinces by region, bearing in mind as noted above that even within a region there is diversity in the socio-economic conditions across provinces. Each figure below is split along the average DEA score and the average share of wages and salaries, dividing them into four quadrants. 20 Based on information provided by the Ministry of Finance. Data on salary share of recurrent spending in education, health and transportation were not available for Hau Giang and Son La. 73. Provinces in the top left quadrant are relatively more efficient and have low share of wages and 88 salaries relative to others in their region. Provinces in the bottom right hand quadrant are relatively less efficient and have a high share of wages and salaries relative to others in their region. For these provinces, the wage bill needs to be looked at more closely and where the policy of salary reforms may likely be having a negative impact on efficiency. Provinces in the bottom right quadrant on the other hand are relatively less efficient and have relatively low share of wages and salaries; in these cases, there may be other issues in the composition of spending or other factors discussed in the earlier section that are impacting on relative efficiency. 74. The results are not dissimilar to earlier findings on Public Sector Efficiency and the relative productive across provinces. Overall, provinces in wealthier regions in the South East and the Red River Delta have not only lower local spending as a share of the local economy, but also leaner public service when measured by the share of wages and salaries to recurrent (less than 48 percent on average) due to economies of scale. There are very few provinces that have higher than average relative efficiency and lower than average share of wages and salaries. But the ones to note are Thai Binh in the Red River Delta, and HCMC and Binh Duong in the South East. There is also higher variance in the share of wages and salaries for provinces in these regions. Ninh Binh in the Red River Delta and Tay Ninh in the South East for example both have over 50 percent of recurrent spending going to salaries and wages. Tay Ninh overall is relatively efficient compared to others, but in Ninh Binh on the other hand spending on a given level of service delivery is relatively high. This may therefore be a case in which to review application of salary reforms. 75. The poorer, less densely populated regions have a higher share of wages and salaries in recurrent spending (more than 55 percent on average) and lower relative efficiency scores. The bigger size of the public service may be linked to, as discussed earlier, the challenges of delivering services in more sparsely populated provinces. Nonetheless it is important that salaries and wages expenditure do not crowd out other critical recurrent items, which are important for the efficiency and effectiveness of public service delivery. This includes operations and maintenance costs in all three sectors considered above, but also critical goods and services, particularly in the health and education sectors. Provinces where these issues would be worth looking at in more detail include: Gia Lai but also Dak Lak and Kon Tum in the Central Highlands; Ha Giang, Dien Bien, Lai Chau, Bac Kan and Lang Son in the Northern Mountains; and possibly also Ben Tre in the Mekong River Delta. These are all provinces that are not only furthest away from the production efficiency frontier derived above, but also have a high share of salaries and wages in recurrent spending in education, health and transportation. Share of salary in recurrent expenditure and productive efficiency across provinces 89 Figure 3.29: Central Highlands Figure 3.30: Mekong Delta 0.75 1.05 Dak Nong Tra Vinh 1.00 An Giang Lam Dong 0.70 Tien Giang Hau Giang Vinh Long Kien Giang 0.95 0.65 DEA score 0.90 DEA score Long An Can Tho Dak Lak Gia Lai 0.60 0.85 Bac Lieu Ca Mau 0.80 Dong Thap 0.55 Kon Tum 0.75 Ben Tre 0.50 0.70 53% 54% 55% 56% 57% 58% 50% 52% 54% 56% 58% Salary/Recurrent Salary/Recurrent Figure 3.31: Northern Central and Central Coast Figure 3.32: Northern Mountains 1.1 0.9 Hoa Binh 1.0 0.8 Ninh Thuan 0.9 0.7 Binh Thuan Phu Yen Thai Nguyen Lao Cai Tuyen DEA score DEA score 0.8 Thua Thien Quang Yen Bai Hue 0.6 Phu Tho Khanh Hoa 0.7 Bac Giang Lang Son Thanh Hoa Quang Tri 0.5 Dien Bien Cao Bang 0.6 Nghe An Lai Chau Quang Binh Ha Giang Quang Ngai Quang Nam 0.4 0.5 Bac Kan 0.4 0.3 50% 55% 60% 65% 50% 52% 54% 56% 58% 60% 62% 64% Salary/Recurrent Salary/Recurrent Figure 3.33: Red River Delta Figure 3.34: South East 1.1 1.05 Nam Dinh Dong Nai 1.00 Ho Chi Minh 1.0 Thai Binh Hai Duong Binh Duong Binh Phuoc 0.95 0.9 0.90 Hung Yen DEA score DEA score 0.8 Bac Ninh Ha Nam 0.85 Tay Ninh 0.7 Ninh Binh 0.80 0.75 0.6 Hai Phong Vinh Phuc 0.70 0.5 Quang Ninh Ba Ria - 0.65 Vung Tau 0.4 0.60 35% 40% 45% 50% 55% 25% 35% 45% 55% 65% 75% Salary/Recurrent Salary/Recurrent CONCLUSIONS AND RECOMMENDATIONS 90 76. Local spending plans in Vietnam lack credibility whereby spending limits set in approved budgets can be relaxed without legislative review, which dilutes links between local development plans and budgets. The lack of credibility emanates from current policies and institutions in local budget management, which warrant reform, including: policies on the use of over-realized revenue; use of extra-budgetary financing; and carry over practices. These practices have not yet created fiscal sustainability issues, partly due to limits on borrowing; but local authorities need to consider the growth of non-discretionary compensation entitlements and recurrent costs of capital projects resulting from current policies on use of over realized revenue. Low credibility spending plans also impact on efficiency of spending. 77. A review of productive efficiency finds fairly wide variation in performance across provinces. Spending on delivery of a given level of services delivery in poorer provinces can on average be up to 42 percent more compared to relatively more efficient provinces. Within regions, there are also outliers, which also happen to have the least credible spending plans and highest levels of carry over. The analysis also indicates that the share of wages and salaries in recurrent spending may be impacting the efficiency of overall spending in selected provinces, which further confirms the need to review the policies on use of over realized revenue. Issues Recommendations Policies on over-realized revenues in the SBL Ensure in the SBL and associated regulations, that 2002 and associated regulations lead to lack of the approval of the use of excess revenues (by the credibility of spending plans and loss of fiscal NA or provincial Councils) for spending takes place State Budget Law 2002 discipline. in the context of a Supplementary Budget. Off-budget debt financing is likely to increase over time, which would exacerbate problems Integrate all debt and other off-budget financed of low credibility spending plans without activities into local budgets to ensure transparency stricter controls in the context of overall budget and comprehensiveness. decisions. Carry over practices lead to loss of credibility Limit carry over spending in the SBL to only in spending plans, and lack of transparency as investment spending. Delete Article 63 allowing carry over spending is not broken down in the carry over of unspent revenue. budget. Introduce population density criteria in allocation Population density has a large and significant norms to estimate spending needs of local impact on the relative cost of service delivery authorities (i.e. inversely proportional). regulations Stability The current budget classification structure, Further develop the existing Citizens’ Budget to Period budget table templates, and coverage of fiscal communicate in a more accessible manner budget activities in budget documents. policies of local authorities. Productive efficiency differentials across regions Reaffirms earlier recommendations on addressing are often related to level of development and policies that lead to loss of spending plan population density, but within regions there Follow up analysis credibility. It is also recommended to study some are outliers, many of whom suffer from low of the outlying provinces in more detail constraints credibility of spending plans and high carry to productive efficiency. overs. Several provinces in poorer regions exhibit high Review wage bill and application of salary reform share of salary in recurrent spending and low policies in outlying provinces. productive efficiency. 91 LOCAL REVENUE ARRANGEMENTS 92 04 Key issues: local authorities in Vietnam have seen increased levels of local revenue in the 2006-2011 period, but continue to have little to no autonomy over revenue policy and administration. Provinces where sharing rates declined in subsequent stability periods may have disincentives to maximize revenue effort, which points to a review of the revenue sharing arrangements. These provinces also highlight a deficit in infrastructure financing, which points to a review of potential for increased revenue autonomy. Objectives: the objectives of this chapter are to assess whether the current revenue arrangements are impacting on the revenue performance of local authorities, and how these could be addressed through reforms of the revenue sharing arrangement and the potential for more revenue autonomy in selected provinces. Key findings: compared to other countries, decentralized revenues in Vietnam constitute relatively large shares of national GDP (9-10 percent) and general government revenue (33 percent excluding extra budgetary sources). The share of decentralized revenue over total local revenue however has declined over time. This is partly because richer provinces have been making higher contributions of shared revenue to central government for the purposes of redistribution through transfers. Analysis in this chapter does not suggest that this has had a significant negative impact on revenue effort by these net contributing provinces. On average provinces in Vietnam collect roughly a little over 80 percent of their potential estimated revenues. But performance across provinces varies a lot due to factors such as local capacity and the level of revenue decentralization as measured by share of 100 percent locally retained revenue. This, together with the needs of richer, high growth potential provinces, highlights the need to consider opportunities for more revenue autonomy. The revenue sharing arrangement also warrants review to improve its transparency and equity. In particular, major taxes end up credited to provinces where major firms are headquartered rather than where the output is produced or consumed. Recommendations: (i) consider moving to sharing of vat on a formula basis rather than on a derivation basis; (ii) centralize natural resource taxes; (iii) centralize CIT; (iv) enable provinces to impose surtaxes on PIT, local businesses and excises; (v) increase autonomy over determination of user fees at local level; (vi) consider property tax in the long run. CHAPTER 4: LOCAL REVENUE ARRANGEMENTS 93 1. Background: Increased spending responsibilities for local authorities in Vietnam have been complemented over time with higher levels of local revenue collections (chapter 2). The literature on fiscal decentralization argues that decentralization of revenue responsibilities should be guided by local authorities’ spending responsibilities i.e. finance should follow function. Financing can be a combination of local authorities’ own revenue, shared revenue, transfers from upper tiers (chapter 5), and borrowing (chapter 6). This chapter looks at own and shared revenue arrangements in Vietnam. Getting the right balance of local financing sources is important for spending efficiency. Local revenue responsibilities exceeding local authorities’ spending requirements can encourage unnecessary expenditure and inequity across provinces. If on the other hand local authorities are not responsible for raising at least some of their revenues, and instead are completely dependent on central transfers, they may have too little incentive to provide local public services in a cost-effective way. 2. Aside from matching the level of local spending with the level of local revenues, it is also important to consider the actual autonomy that local authorities have over revenue policy and administration. In Vietnam, local authorities have very little formal tax autonomy or revenue assignment. There are three dimensions to this: (i) local authorities in Vietnam cannot independently choose to impose specific types of taxes within their jurisdictions as these are nationally determined; (ii) aside from limited autonomy over the level of tariffs and charges, all tax bases and rates in Vietnam are determined by the central government; (iii) the administration and collection of taxes are carried out by central tax administration offices located in provinces and districts. 3. This chapter builds on recent work (Martinez-Vazquez and Nguyen 2011) to review the current local revenue arrangements in Vietnam and possible opportunities for reform. It starts with an overview of the institutional arrangements for local revenue collections in Vietnam, including recent trends and developments. This aims to illustrate the nature of local revenue collections in Vietnam, how this compares in a broader international context, and the significance of local collections in the local economies. The second part of the chapter builds on this to empirically review how effective local authorities have been in collecting revenues, and what might explain the relative effectiveness of different authorities. 4. The third part looks more specifically at whether effectiveness of revenue collections could be strengthened through reform of the current revenue assignment and sharing arrangements. The hypothesis is that the current arrangement in which stronger performance could lead to higher transfers to central government, may act as a disincentive for those high performing local authorities to collect more revenue. In addition, more equitable sharing arrangement should promote local economic performance and revenue performance across the provinces. The final part looks specifically at possible options for more revenue autonomy, particularly for richer provinces to help offset the larger shares of revenues that they transfer to the center. As discussed in chapters 2 and 5, although the latter has contributed to redistribution objectives, there may be merit and opportunities to provide more revenue autonomy because: this could enable more investment in areas with high growth potential; and more revenues will be needed to meet higher debt servicing as discussed in chapter 6. Providing local authorities better access to some significant revenue handles at the margin will strengthen fiscal responsibility, as it promotes linkages between the financing of local spending decisions and those that benefit from these expenditures. LOCAL REVENUE SOURCES AND TRENDS 94 5. Overview of local revenue arrangements in Vietnam: Revenues collected in Vietnam can be grouped into three categories: (i) central government revenues; (ii) revenues that are retained 100 percent by local authorities; and (iii) revenues that are shared between central and local authorities. These revenues are collected throughout the entire territory by General Department of Tax offices located in sub-national jurisdictions. The local authorities’ rights over these revenues in terms of which taxes they can retain, and which ones they need to remit back to the center, are set out in the State Budget Law 2002 (SBL 2002): (i) Article 32(1) sets out which revenue types local authorities are able to retain fully; and (ii) Article 30 (2) specifies which revenue types are to be shared between central and local authorities. The SBL 2002 leaves provincial governments with considerable discretion in the design of revenue assignments to districts and communes within their borders. However, article 34 of the Law provides some general principles and minimum standards the provinces need to follow in designing the revenue assignments for their local governments. A summary breakdown of the three revenue categories is provided in figure 4.1. Figure 4.1: Overview of types of revenue retained by different tiers of government (1) Central revenues: VAT on imports; Import and Export taxes; Special Consumption Tax on Imports; CIT on enterprises with uniform accounting; Taxes and other revenue from petroleum. (2) Revenue retained 100 percent by (3) Revenue shared between central local authorities: and local authorities: Land and Housing Tax; Natural Resource VAT (except VAT on imports); CIT (except Tax (except petroleum); Tax on Transfer enterprises with uniform accounting); of LURs; Registration Fees; Licensing Personal Income Tax; Special Fees; Rental of Land and Water; Transfer Consumption Tax on domestic goods and of Land Use Rights; Sale of State services; Environment Protection Tax. Property 6. The different types of revenue allocated to different tiers of government are broadly in line with sound assignment principles. Revenue retained 100 percent is the closest match to “own source revenue” or “locally assigned revenue”. Good practice is to assign taxes with immobile bases (e.g. land and property) to local authorities because there is no scope to shift location to avoid taxation. Local authorities should also have autonomy on fees and charges for local services. But as noted above, in Vietnam these do not constitute formal revenue assignments because of local authorities’ lack of policy autonomy. With tax autonomy local authorities would have the ability to determine the rates (if not also the bases) for their own source revenues, but this is not applicable in Vietnam, as summarized in table 4.1 below. Table 4.1: Taxonomy of Subnational Revenue Arrangements 95 Local autonomy Tax base Tax rate Administration Revenue source Typical Vietnam Typical Vietnam Typical Vietnam Own-revenue assignments Own-source revenue √ χ √ χ Possibly χ Surcharges on national taxes χ N/A √ N/A Possibly N/A Revenue sharing χ χ χ χ Possibly χ Source: Adapted from Fedelino and Ter-Minassian (2010); Note: N/A = Not Applicable 7. Revenue retained 100 percent and shared revenue together make up “decentralized revenue”. These are the core, locally collected revenues. Shared revenues, as discussed in chapter 5, are similar to fiscal transfers. They constitute a pool of central revenues, which are shared with local authorities based on agreed rates. The “sharing rate” for each province is the proportion of shared revenues collected in its jurisdiction that can be retained by that province. The same sharing rate is applied on all shared taxes within a province, but the sharing rate varies across provinces. The sharing rate for a province is determined by an estimated fiscal gap (see chapter 5). The 50 provinces that have a positive fiscal gap retain 100 percent of shared taxes that they collect. Only 13 out of 63 provinces only keep a portion of the shared taxes, and transfer the rest to the center for redistribution across the country. 8. Table 4.2 shows the tax sharing rates for the “surplus” provinces over the last three Stability Periods. The changes in sharing rates between those three periods were relatively small and generally downward. There are some exceptions like Hanoi (which merged with the relatively poorer province of Ha Tay in 2009), Dong Nai and Khanh Hoa. Over the past three periods, the stability duration was extended from three to five years. The quid-pro-quo of the agreement was more stability for the provinces and the ability of the center to get higher shares from the surplus provinces. Table 4.2: Shared revenue sharing rates (%) across stability periods Province Sharing rate 04-06 Sharing rate 07-10 Sharing rate 11-15 Hanoi 32 45 42 Quang ninh 98 76 70 Hai phong 95 90 88 Vinh phuc 86 67 60 Bac ninh 100 100 93 Da nang 95 90 85 Khanh hoa 52 53 77 Quang ngai 100 100 61 Ho chi minh city 29 26 23 Dong nai 49 45 51 Binh duong 44 40 40 Ba ria – vung tau 42 46 44 Can tho 95 96 91 Source: State Budget reports 21 The sharing rate for Hanoi was 31% in 2007 and adjusted to 45% from 2008 to 2010 after Hanoi merged with Ha Tay 9. Decentralized revenue in the local economy: Decentralized revenue does not constitute a large 96 share of local economies in Vietnam. As illustrated in figure 4.2, in half of all observations between 2006 and 2009, decentralized revenue as a share of local GDP was less than 7 percent, despite the fact that most provinces retain 100 percent of the shared revenue that they collect. Only in a few cases, such as Binh Duong, Danang, Hanoi, Kon Tum, Ninh Binh and Vinh Phuc did decentralized revenue exceed 15 percent of local GDP over the 2006-2011 period. In 2008 and 2009, the distribution narrowed, as fewer provinces were collecting above 7 percent of local GDP due to economic slowdown. The distribution in 2011 was similar to that in 2006 showing little change over Stability Periods. Figure 4.2: Decentralized revenue/local GDP (%), Figure 4.3: Annual decentralized revenue/local GDP 2006-2011 (%), 06-11 -1SD Median +1SD .15 2006 2007 2008 2009 2010 2011 .15 .1 Density .1 Density .05 .05 0 0 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Decentralized revenues as % of GDP Decentralized revenues as % of GDP Source: Staff estimates based on published State Budget data 10. There are two main reasons to explain the above trends in the share of decentralized revenue in local GDP. The first is that central revenues constitute the bulk of general government receipts. All trade and petroleum-related levies accrue to the central government and not to local governments. As illustrated in figure 4.4 below, even though trade and petroleum-related receipts are falling as a share of GDP between 2006 and 2011, they are still at around 8 percent of GDP in 2011. Similarly, the bulk of Corporate Income Taxes (CIT) is paid by the largest enterprises with uniform accounting, which also accrue to the central government. As illustrated in figure 4.5 below (Light and Nguyen 2013), the top 1 percent of enterprises in terms of turnover contribute nearly 85 percent of total CIT receipts of general government (including petroleum-related CIT). CIT receipts constitute around 7 percent of overall GDP, most of which also accrues to the central government Figure 4.4: Revenue composition (% GDP) Figure 4.5: Share of Tax Burden and Output, by Companies ranked by Tax Payable VAT Non-oil CIT Company Rank Tax Payable Output Oil revenue Trade Taxes 0% 25 10% 20 20% 30% 15 40% Bottom 75% Next 15% 50% Next 5% 10 Next 4% 60% Top 1% 70% 5 80% 0 90% 2006 2007 2008 2009 2010 2011 100% Source: Based on data from IMF Article IV reports Source: Light and Nguyen (2013) 11. The trends in decentralized revenue to local GDP therefore do not necessarily reflect low revenue effort on the part of local authorities. As illustrated in figures 4.6 to 4.11 local nominal GDP growth and nominal 97 growth in decentralized revenue are positively related across most provinces. This applies particularly in the wealthier regions of the Red River Delta and the South East. There are some outliers that have higher than average economic growth but lower than average revenue growth (e.g. Vinh Phuc, Bac Ninh, Tay Ninh and Dong Nai). This could be related to the structure of the local economy, urbanization rate, or other factors that impact on the tax base and revenue potential. 12. The relatively poorer region of the Northern Central and Central Coast also show a strong link between local GDP and revenue mobilization, but the relationship is weaker in the Mekong River Delta and the Northern Mountains Region. The tax bases in these regions are naturally much smaller due lower levels of taxable assets, investments and income streams compared to the Red River Delta and the South East. There is also diversity within the regions, which might explain the disparity within regions (e.g. Son La, which has higher than average economic and revenue growth, compared to Lang Son, which higher growth but lower average revenue). Whilst beyond the scope of this study, it would be productive to analyze these outliers more closely through a separate exercise. Local nominal GDP growth and nominal decentralized revenue growth by region 98 Figure 4.6: Central Highlands Figure 4.7: Mekong Delta 60% 35% Kon Tum Can Tho 50% R² = 0.0086 30% R² = 0.042 Ca Mau Decentralized Revenue growth Decentralized revenue growth Hau Giang 25% Vinh Long 40% Long An Dak Nong 20% Ben Tre Kien Giang 30% Gia Lai Dong Thap 15% Dak Lak Soc Trang 20% Lam Dong 10% Bac Lieu 10% 5% Tra Vinh Tien Giang 0% 0% 0% 5% 10% 15% 20% 25% 30% 35% 0% 10% 20% 30% 40% Local GDP growth Local GDP growth Figure 4.8: NCCC Figure 4.9: Northern Mountains 50% 70% 45% Quang Nam 60% Ha Giang R² = 0.2615 R² = 0.0304 40% Decentralized revenue growth Quang Ngai 35% 50% Decentralized revenue growth Hoa Binh 30% Khanh Hoa 40% Son La 25% Thai Nguyen Quang Tri Quang Binh 30% Bac Giang 20% Thanh Hoa Cao Bang Bac Kan Da Nang Binh Thuan Tuyen Quang 15% Lao Cai 20% Phu Tho 10% Dien Bien Lang Son Binh Dinh 10% Lai Chau 5% 0% 0% 0% 10% 20% 30% 40% 0% 10% 20% 30% 40% Local GDP growth Local GDP growth Figure 4.10: Red River Delta Figure 4.11: South East 45% 35% 40% R² = 0.0903 Binh Duong Ha Nam Ninh Binh R² = 0.2248 30% Decentralized revenue growth 35% Hai Duong Hung Yen Decentralized revenue growth 25% 30% Ho Chi Minh Bac Ninh 20% 25% Nam Dinh Thai Binh Tay Ninh 20% Binh Phuoc 15% Vinh Phuc Ha Noi Dong Nai 15% - Vung Ba Ria 10% Tau 10% 5% 5% 0% 0% 0% 10% 20% 30% 40% 50% -40% -20% 0% 20% 40% Local GDP growth Local GDP growth 13. Compared to other countries, decentralized revenues in Vietnam constitute a relatively high share of national GDP. Figure 4.12 shows Vietnam relative to OECD countries, where Sweden has the highest local 99 revenue as a share of GDP at 15.9 percent. In Vietnam it is close to 9.6 percent of GDP (2006-2011 period average, compared to the median value of decentralized revenue to GDP of around 7 percent discussed in paragraph 8 above). Decentralized revenue to GDP in Vietnam has seen a very slight decline of 0.1 percentage points between 2006 and 2011, which is linked to an overall decline in revenue due to slower economic growth. Figure 4.12: Central and local revenue (% of GDP), 2006-2011 period average Central Local Change in local revenue/GDP (percentage point) 50 2 45 0.4 1 40 -0.1 35 15.9 0 30 -1 25 9.6 20 -2 15 30.4 -3 10 17.8 -4 5 0 -5 Czech Republic Slovak Republic New Zealand United States United Kingdom Canada Luxembourg Slovenia Switzerland Japan Sweden Greece France Netherlands Germany Estonia Hungary Spain Denmark Belgium Poland Finland Korea Iceland Portugal Australia Turkey Chile Vietnam Israel Austria Norway Mexico Ireland Italy Source: OECD Decentralization Database and Staff estimates 14. Share of local revenues: The above points to a growing share of local revenues in Vietnam. Indeed, as illustrated in figure 4.13 below, the share of decentralized revenues in general government revenue has increased over subsequent Stability Periods from 27 percent in 2006 to around 33 percent in 2011. When other sources of local revenue including off-budget are included (excluding transfers), local revenue are close to 50 percent of general government receipts (see figure 2.13, chapter 2). Figure 4.13 shows that local revenue sources have grown faster than central revenue. Central and decentralized revenue fell in 2009 due to slowing economic growth but was offset by other local receipts; in particular local authorities turned to extra budgetary source to cover capital investments as part of the fiscal stimulus. Figure 4.13: Central and local revenue as a share of total revenue (%) 100% 60% 12% 12% 13% 18% 17% 18% 80% 50% 27% 30% 28% 30% 32% 33% 40% 60% 30% 40% 61% 20% 58% 58% 52% 51% 49% 20% 10% 0% 0% 2006 2007 2008 2009 2010 2011 Central revenue Decentralized revenue Other local Central revenue (% change) Decentralized revenue (% change) Other local (% change) Source: Staff estimates based on published State Budget data 15. Decentralized revenues as a share of total revenues in Vietnam is relatively high compared to other 100 countries. As discussed in chapter 2, international comparisons need to be treated with caution due to the challenges of having full and consistent sets of comparable data across countries. Figure 4.14 below compares decentralized revenue as a share of general government revenue in Vietnam (2006-2011 period average) against local revenue over general government revenue for selected countries (1990-2000 period average). Vietnam exhibits one of the highest ratios of decentralized revenue to general government revenue after China. It is also higher than the average in other regions, where decentralization may have also increased further since 1990-2000. At the same time, however, it is also important to note that China and Vietnam do not provide subnational governments with any measure of tax autonomy. Figure 4.14: Local revenues (% of total government revenues) 51 50.0 40.0 Regional average 30 30.0 20.0 10.0 3.6 0.0 Russia Costa Rica Kyrgyz Rep. Spain China France Chile Kazakhstan Brazil Latvia Slovenia Belarus Sweden Estonia Poland Romania Bulgaria Finland Bolivia Hungary Thailand Germany Italy Malaysia Indonesia Tajikistan Denmark Portugal Azerbaijan Mexico Philippines Lithuania Norway Vietnam Switzerland Argentina Moldova Mongolia East Asia East Europe and Central Asia Europe Latin America and Caribbean Source: WB Fiscal Decentralization Indicators (data drawn from IMF Government Finance Statistics) 16. Decentralized revenue as a share of local receipts Figure 4.15: Dec rev/local receipts (including transfers and extra-budgetary), varies widely across provinces. Decentralized revenue in poorer provinces (e.g. Ha Giang, Bac Kan and Dien Bien in Northern Mountains) are less than 10 percent of local revenue (see figure 4.15). In richer provinces such as Binh Duong and Dong Nai, decentralized revenues are in excess of 75 percent of local receipts. 17. In terms of the share of decentralized revenue for provincial and district authorities, as discussed in chapters 2 and 5, provincial authorities have higher levels of decentralized revenue compared to district authorities; the latter are more dependent on transfers to meet their spending needs. This illustrated further in figure 4.16 below, which shows that decentralized revenue in district authorities constitute less than 40 percent of core receipts (i.e. including provincial transfers) in poorer regions. This is due to a combination of higher tax administration capacity and collection responsibility at provincial level compared to districts. Figure 4.16: Decentralized revenue/local core receipts at province and district levels (2006-2011 period average) 101 81% 80% Central Highlands 70% 57% Mekong River Delta 60% 50% 44% 46% Northern Central and Central 38% 40% 37% 35% Coast 40% 32% 29% Northern Mountain 30% 18% 20% 13% Red River Delta 10% South East 0% Province decentralized revenue District decentralized revenue as as % of core provincial receipts % of core district receipts Source: WB Fiscal Decentralization Indicators (data drawn from IMF Government Finance Statistics) 18. Trends in and composition of decentralized revenue: Although Vietnam has seen a gradual rise in the share of decentralized revenue in general government revenue, the share of revenue retained 100 percent in total local receipts (i.e. including shared revenue, transfers, and extra-budgetary receipts) has declined over the 2006-2011 period. For the period as a whole, the median value of all observations for revenue retained 100 percent over local receipts was around 15 percent (figure 4.17). But the median value has fallen from around 20 percent in 2006 to roughly 10 percent in 2011 when the distribution narrowed fairly significantly (figure 4.18). As noted earlier, this category is the closest that local authorities get to their “own source” revenue, though with restrictions on base and rate setting. Figure 4.17: Revenue retained 100%/local receipts Figure 4.18: Annual revenue retained 100%/local (06-11) receipts (06-11) -1SD Median +1SD 2006 2007 2008 .04 2009 2010 2011 .06 .03 .04 Density .02 Density .02 .01 0 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 100% retained revenue as % of local revenue 100% retained revenue as % of local revenue Source: Staff estimates based on published State Budget data 19. The level of revenue retained 100 percent in Vietnam is low compared to the level of own source 102 revenue in developed countries but quite similar to that in Asian countries and South East Asia and Pacific more specifically and more generally developing countries (figure 4.19). However, what is defined as “own source revenue” differs considerable across countries. Additionally, as discussed above, in a strict sense in the case of Vietnam neither the revenue retained 100 percent nor the shared revenue can be considered as “own source revenues.” Figure 4.19: Own source revenue taxes/GDP (%) 6.5 6.0 5.5 Vietnam 4.7 5.0 Percent of GDP 4.3 4.3 South, East Asia 4.5 4.1 3.7 3.8 and Pacific 4.0 Asia 3.5 3.1 3.3 Developing 3.0 3.4 countries 2.5 3.0 Developed 2.8 2.7 2.0 2.6 2.7 countries 1.5 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Source: Estimates based on published State Budget data for Vietnam, IMF GFS for other countries 20. The ratio of shared revenue to total local receipts also exhibits a falling trend over the 2006-2011 period, though not as sharp as the ratio of 100 percent retained revenue to total local receipts. For the period as a whole, for more than half of all provinces, shared revenue accounted for around 20 percent of local receipts (figure 4.20). This declined slightly by 2011 to around 16 percent. As a result of these developments, the share of decentralized revenue to local receipts also declined, as did the share of decentralized revenue in general government receipts (from 30 percent to 28 percent between 2007 and 2008 (see figure 4.13 above). However, the following year central revenues declined even more sharply, which meant that the overall share of decentralized revenue in general government receipts rose again. Figure 4.20: Shared revenue/local revenue 06-11 (%) Figure 4.21: Annual shared revenue/local revenue 06-11 (%) -1SD Median +1SD 2006 2007 2008 .04 .04 2009 2010 2011 .03 .03 Density Density .02 .02 .01 .01 0 0 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 Shared revenue as % of local revenue Shared revenue as % of local revenue Source: Staff estimates based on published State Budget data 21. The declining share of decentralized revenue in 2007-2009 was not as a result of limitations placed on any existing revenue responsibilities for local authorities. It was also suggested above that revenue 103 growth for provinces as a whole seems broadly in line with local economic growth. Nonetheless, it is worth looking more closely at revenue performance across provinces, with a particular focus on those that have seen a decline in the share of revenue they are able to retain. The main issue is whether the current sharing practice adversely impacts the revenue performance of provinces that are likely to see their sharing rates decline. 22. Declining sharing rates leading to higher contributions from wealthier provinces have of course played a critical role in the redistribution of resources as discussed in chapter 5. As illustrated in figure 4.22, falling decentralized revenue in 2007-2009 was offset by rising transfers including as a result of more shared revenue contributed from richer provinces over subsequent Stability Periods. Falling decentralized revenue were also offset by higher extra-budgetary funding (figure 4.23), particularly during the 2007-2010 crisis period. Figure 4.22: Local revenues/1000 population Figure 4.23: Annual off-budget revenue/local (avg, vnd mil) revenue (%) 3500 TOTAL REVENUE SNG TAXES 2006 2007 2008 3000 2009 2010 2011 SHARE TAXES TRANSFER .2 2500 .15 2000 Density 1500 .1 1000 .05 500 0 0 0 5 10 15 20 25 30 1997 1998 1999 2000 2002 2003 2004 2005 2006 2007 2008 2001 2009 2010 Off-budget revenue as % of local revenue Source: Staff estimates based on published State Budget data 23. These trends however raise two issues: firstly the current sharing arrangement leads to a trade-off between a province’s right to retain more revenue for its own investment needs and enabling the central government to redistribute; and secondly richer provinces rely increasingly on extra-budgetary sources of financing, which are less transparent and adversely affect the overall quality of planning and budgeting. Providing more revenue autonomy, which would benefit more directly richer provinces, could potentially help tackle both issues. But, of course, this would also call for an enhanced role of equalization or balancing transfers. Before looking at the institutional arrangements for revenue sharing and possible options for greater autonomy, the next section looks more closely at revenue performance across local authorities REVENUE PERFORMANCE ACROSS LOCAL AUTHORITIES 24. Estimating revenue potential across local authorities: There are potentially a number of incentive effects that may impact on revenue performance across provinces. One possibility, as discussed, is the current revenue sharing arrangement. Another possibility is the current system of centralized tax administration, whereby the de-concentrated system puts more effort on revenue that accrues to central government, rather than that which accrues to local authorities. This problem is mitigated in Vietnam 104 because there is a de facto dual subordination of tax administrators to both central and local authorities. This means that provincial and district officials, at least potentially, can have a recognizable influence on the decisions (e.g. how to allocate resources to enforcing different taxes) and activities (audits, collections campaigns, etc.) of tax administrators working inside their jurisdictions. Provincial authorities also provide performance bonuses and other benefits. 25. One way to measure the revenue performance of local authorities is to look at the gap between actual collections and potential collections. It is assumed that potential revenue collections are determined by certain socio-economic characteristics. Among a set of provinces that have similar characteristics, the province that raises the highest level of decentralized revenue to GDP sets the revenue potential for that group of provinces. The best performing provinces for given characteristics lie on an “efficiency frontier.” Less efficient provincial tax administrations can be seen as falling short of the frontier, and the larger the distance to the frontier the greater the degree of inefficiency. Therefore efficiency and inefficiency, like with the spending analysis in chapter 3, are relative to the best performers. This approach is illustrated in figure 4.24 below. Figure 4.24: Stochastic Frontier and Tax Effort (ii) Potential (iii) Best performance on decentralized decentralized revenue/GDP by Revenue/GDP province with characteristics “X” (iv) Estimated Stochastic Frontier (vii) Technical inefficiency (v) Decentralized revenue/GDP collected by another province (vi) Technical efficiency with characteristics “X” (i) Economic characteristics that impact on revenue X potential 26. The analysis considers the following socio-economic characteristics as potential determinants of revenue potential (see annex D): (i) economic characteristics including local GDP per capita; shares of agriculture, services and construction in the local economy; and level of informality; (ii) demographic characteristics including age dependency (ratio of population under 15 and over 65 to those aged between 15 and 64); population density; and education levels. 27. The results of the estimations using provincial panel data for 2006-11 are shown in Table 4.3. These results for the potential revenue estimation are obtained using the Stochastic Frontier analysis approach and assuming both exponential and half-normal distributions. The first two models (columns 1 and 2) include the log of construction output as an additional determinant. Because the available number of observations for this variable significantly restricts the degrees of freedom, we also show the estimation when this variable is excluded in two additional models (columns 3 and 4).22 22 Although this variable is statistically significant, the log likelihoods are not substantially different when we exclude from the specifications the construction output as a control variable. Table 4.3: Stochastic frontier estimation of potential revenue, Vietnam, 2006-2011 105 Log of decentralized revenue/ Model 1 Model 2 Model 3 Model 4 local gdp Exponential Half-normal Exponential Half-normal Log gdp per capita 0.245*** 0.232** 0.0990 0.0861 (0.0799) (0.0980) (0.0749) (0.0828) Log agriculture share in gdp -0.965*** -0.933*** -0.767*** -0.726*** (0.0781) (0.107) (0.0879) (0.1000) Log service share in gdp -0.869*** -0.823*** -0.561*** -0.540*** (0.131) (0.194) (0.160) (0.157) Log of informal sector -0.613*** -0.705*** -0.597*** -0.623*** (0.117) (0.120) (0.126) (0.127) Border effect -0.504*** -0.533*** -0.451*** -0.473*** (0.0493) (0.0595) (0.0567) (0.0677) Log of urbanization rate -0.223** -0.248* -0.0333 0.00460 (0.0873) (0.127) (0.0886) (0.113) Log of population density 0.00503 0.0218 -0.0256 -0.0168 (0.0342) (0.0375) (0.0413) (0.0462) Log of age dependency 1.712*** 1.710*** 1.176*** 1.140*** (0.192) (0.235) (0.242) (0.297) Log of general education grad 0.332** 0.366** 0.371** 0.367** rate (0.143) (0.157) (0.164) (0.184) Log of construction output -0.120*** -0.124*** (0.0353) (0.0403) Log-likelihood -135.62798 -142.40229 -171.94341 -177.22086 Observations 203 203 249 249 Note: standard errors are in parentheses. *** p<0.01, ** p<0.05, * p<0.1. 28. In summary, the results show that: a. Log of GDP per capita is significant and positive in columns 1 and 2, indicating that an increase in GDP per capita will increase potential revenue. This is the expected result since better economic development will lead to larger tax base and potential tax revenue collections.23 b. Higher share of agriculture in the local economy significantly decreases decentralized revenue as expected (1 percentage point increase in agriculture/GDP corresponding to 0.72-0.96 percentage point decrease in decentralized revenue/GDP, ceteris paribus). A larger informal sector significantly (at 1% level) lowers tax revenue of the provinces.24 c. The border effect dummy variable, which is equal to 1 if the province borders with China and/or the coast is significant at the 1% level and negative. These provinces on average collect 0.47 percentage point lower decentralized revenue/GDP, ceteris paribus. Trade taxes are assigned to central government, therefore this would not appear in the ratio of decentralized revenue/GDP. But informal border trade can lower tax revenue. 23 Although it is also important to note that GDP per capita is measured in nominal terms. Price levels, and therefore real incomes, may differ across provinces, which could yield different results. Unfortunately, this information was not available for the study. 24 Note that the inverse of retail sales as a percentage of GDP is used as a proxy for informality. The logic is that informal transactions are typically not recorded as official retail sales. d. The urbanization rate is significant and negative, indicating that highly urbanized provinces 106 collect lower tax revenues. This is surprising as urban areas experience economies of scale and scope leading to overall higher wages and earnings and therefore higher tax bases. But this may also reflect the share of the informal economy in urban areas. e. The general education graduation rate variable is significant and positive as expected. A 1 percentage point increase in the graduation rate increases decentralized revenue as a percentage of GDP by 0.33-0.37 percentage points, ceteris paribus. A higher graduation rate suggests higher earnings and employment, and potentially higher tax collection. f. inally, the age dependency variable is highly significant and positive, with magnitudes reaching over one-to-one relationship. The sign of this variable may look a bit surprising given that a higher ratio can be a sign of lower tax bases. However, studies of tax morale and tax evasion typically find that the elderly population tends to be more compliant with taxes and see more value to paying taxes for services provided. 29. Average revenue effort across provinces: With technical efficiency in Table 4.3 estimated through stochastic frontier analysis it is now possible to compute revenue effort by province as the ratio of actual revenue collections to potential revenue. The summary statistics (mean value, coefficient of variation, and maximum and minimum values) are shown in Table 4.4 below for the four different models estimated in Table 4.3 above. The efficiency scores by province are presented in annex E. A score of 1 reflects the most efficient outcome; the gap between this and a province’s actual score is the estimate of the percentage points by which the province could improve decentralized revenue/GDP given its socio-economic characteristics. Table 4.4: Summary statistics for tax effort (from models in table 4.3) Model 1 Model 2 Model 3 Model 4 Mean 0.62 0.58 0.66 0.60 Cv 0.41 0.42 0.34 0.35 Max 0.96 0.97 0.94 0.94 Min 0.05 0.06 0.12 0.12 30. The results above show that, on average, provinces in Vietnam collect roughly a little over 60 percent of their potential estimated revenues; or alternatively, on average, they suffer from technical efficiency of approximately 40 percent. However, as also shown in Table 4.4 there are significant variations in performance. Some provinces reach close to perfect technical efficiency by scoring tax efforts in the nineties while some other provinces appear to suffer from severe technical inefficiencies scoring tax efforts (less than ten). 31. Factors explaining differences in relative efficiency across provinces: To understand possible factors that might explain the variation in revenue performance, the study estimates the impact of selected variables on the latter. The variables are drawn from existing literature looking at factors impacting revenue performance (see annex E). The results summarized in table 4.5 below show that: a. The number of administrative employees is positive and significant through the different specifications, indicating that more employees leads to higher tax effort in the province. However, surprisingly, the quality of human capital measured as the share of graduate degree holders among administrative workers is negative and mainly insignificant, except for model 3 tax effort. b. The level of fiscal decentralization – measured on the revenue side – is positive and significant at the 1% confidence level. This is an expected result, since a greater proportion of tax revenues that can be used independently by the provincial governments and provides an incentive to exercise higher levels of tax effort. It is estimated that 1 percentage point increase in 100 percent retained 107 revenue as a share of total revenue will increase tax effort by 0.61-0.66 percentage points. c. The share of debt in revenues is also positive and significant. Higher debt share in the total revenue is expected to increase tax effort, since it requires consequent repayment in the short and medium terms. However, there is no significant role for the fiscal deficit and for transfers (balancing and targeted transfers). d. As expected, the larger size of the informal sector, which is used to proxy the presence of corruption, results in decreased tax effort, and the coefficient is significant at least at the 10 percent confidence level. Finally, population growth is statistically insignificant in all specifications. 32. From a policy recommendation angle, the results here explaining relative tax collection efficiency across provinces in Vietnam have several significant implications. First, investment for the improvement of human capital stock in provincial tax administrations quite clearly has an immediate payoff in terms of increased collection efficiency. This does not come as a surprise but it is positive to verify that these returns are still present and could be significant. The findings also point out the salience of incentives. The designers of central government policies need to keep the impact of incentives in mind; for example, advancing on the road toward more tax revenue autonomy at the subnational level will yield higher efficiency in tax collections, and so will clearly making subnational governments accountable for their incurred debt. This is discussed further in part 4 below and in chapter 6. Finally, it is also clear that one payoff of reducing corruption is an increased level of tax collection efficiency. Table 4.5: Explaining differences in tax effort at the provincial level, Vietnam 2006-2011 Revenue effort score from Model 1 Model 2 Model 3 Model 4 stochastic frontier analysis Tax effort Tax effort Tax effort Tax effort Fiscal decentralization 0.645*** 0.666*** 0.641*** 0.614*** (0.197) (0.191) (0.139) (0.135) Fiscal deficit 0.139 0.111 0.380 0.353 (0.414) (0.390) (0.311) (0.289) Balancing transfers 5.47e-08 5.90e-08 -1.32e-08 -3.35e-09 (9.89e-08) (9.34e-08) (6.44e-08) (6.08e-08) Targeted transfers 5.24e-08 6.32e-08 7.49e-08 7.76e-08 (6.34e-08) (5.95e-08) (5.15e-08) (4.72e-08) Administrative employees† 6.20e-06*** 6.00e-06*** 3.32e-06*** 3.15e-06*** (2.00e-06) (1.97e-06) (1.23e-06) (1.20e-06) Administrative employees -0.00380 -0.00325 -0.00556* -0.00443 education - bachelors share† (0.00411) (0.00397) (0.00312) (0.00301) Informal sector (corruption) -1.82e-06** -1.74e-06** -1.48e-06*** -1.42e-06** (8.17e-07) (7.96e-07) (5.55e-07) (5.52e-07) Debt share in revenue 0.872* 0.810 0.297 0.244 (0.525) (0.515) (0.333) (0.321) Population growth -1.096 -1.183 -0.994 -1.069 (1.246) (1.166) (0.920) (0.887) Constant 10.87 8.735 15.53*** 11.83*** (6.742) (6.701) (3.857) (3.810) Time fixed effect Yes Yes Yes Yes Regional fixed effect Yes Yes Yes Yes Observations 165 165 200 200 R-squared 0.179 0.190 0.209 0.208 Note: Standard errors are in parentheses. *** p<0.01, ** p<0.05, * p<0.1 33. Revenue effort by province: The relative efficiency score for each individual province could help 108 to identify areas where further effort may be needed. Figures 4.25 to 4.30 plot for each region, individual provinces’ relative efficiency score against decentralized revenue to GDP in that province. Each plot is divided into four quadrants based on the average revenue/GDP collection and average efficiency score in the region. For provinces that are net contributors to the central government (circled in red and green), the shared revenue portion is taken in gross terms (i.e. total shared revenue collected and not just the portion retained by the province) to ensure comparability across all provinces. Provinces in the top right hand and bottom left hand quadrants are respectively the strongest and weakest performers in their region. Relative efficiency of local authorities in mobilization of decentralized revenue Figure 4.25: Central Highlands Figure 4.26: Mekong Delta 20% 12% Decentralized revenue/local GDP 18% Decentralized revenue/local GDP Kon Tum 10% 16% Dong Thap Long An Ca Mau 14% 8% Can Tho 12% Gia Lai Tien Giang Vinh Long 10% Lam Dong 6% Soc Trang Dak Lak Hau Giang Dak Nong Bac Lieu 8% Kien Giang Ben Tre 6% 4% Tra Vinh 4% 2% 2% 0% 0% 0.00 0.20 0.40 0.60 0.80 1.00 0.00 0.20 0.40 0.60 0.80 1.00 Efficiency score Efficiency score Figure 4.27: NCCC Figure 4.28: Northern Mountains 35% 18% Lai Chau 16% 30% Decentralized revenue/GDP Decentralized revenue/GDP Da Nang 14% 25% 12% Thai Nguyen Phu Tho 20% Bac Giang 10% Lao Cai Cao Bang 15% 8% Yen Bai Son La Bac Kan Ninh Thuan Ha Tinh Quang Ngai Hoa Binh Ha Giang Khanh Hoa 6% Tuyen Quang 10% Quang Tri Binh Thuan Dien Bien Quang Nam Binh Dinh Lang Son 4% Thanh Hoa 5% Quang Binh Phu Yen 2% 0% 0% 0.00 0.20 0.40 0.60 0.80 1.00 0.00 0.20 0.40 0.60 0.80 1.00 Efficiency score Efficiency score Figure 4.29: Red River Delta Figure 4.30: South East 25% 16% 109 Binh Duong 14% Vinh Phuc 20% Binh Phuoc Decentralized revenue/GDP 12% Decentralized revenue/GDP Ha Noi 10% 15% Ninh Binh Dong Nai Hai Duong 8% Ho Chi Minh Hai Phong Tay Ninh 10% Bac Ninh Hung Yen 6% Ha Nam - Vung Ba Ria Thai Binh 4% 5% Nam Dinh Ta 2% 0% 0% 0.00 0.20 0.40 0.60 0.80 1.00 0.00 0.20 0.40 0.60 0.80 1.00 Efficiency score Efficiency score 34. There are a number of provinces that stand out in terms of lower than average revenue effort and lower than average decentralized revenue/GDP: (i) Dak Nong in the Central Highlands; (ii) Can Tho, Hau Giang and Vinh Long in the Mekong River Delta; (iii) Quang Binh and Quang Tri in the Northern Central and Central Coast; (iv) Hoa Binh, Dien Bien and Yen Bai in the Northern Mountains; (v) Nam Dinh and Thai Binh in the Red River Delta; and (vi) Ho Chi Minh City, Dong Nai and Tay Ninh in the South East. 35. Revenue sharing arrangement and revenue effort: An individual province’s efficiency scores can also provide an indication of whether the current sharing practice adversely impacts revenue performance. It was suggested earlier that provinces where sharing rates declined in subsequent Stability Periods may have disincentives to maximize revenue effort. This is because marginal costs can exceed marginal benefits if the returns from additional effort accrue mostly to the central government through higher contributions of shared revenue. The study overlaps with three Stability Periods: 2004-2006, 2007-2010 and 2011-2014. Sharing rates are constant over each Stability Period. To determine the sharing rate for each province, a fiscal gap is estimated as the difference between spending needs (formula based) and projected revenue or fiscal capacity. The latter is determined by historical performance, which may negatively impact revenue incentive as province try to keep or increase the shares of shared revenues and balancing transfers. 36. Figure 4.31 below shows the trend of average fiscal effort across all provinces based on the efficiency scores derived from the above frontier analysis. It shows that average revenue effort has fluctuated over the 2007-2010 Stability Period. The trend is similar throughout all four models, and indicates that average revenue effort increased slightly in 2007, followed by a relatively sharp decrease in 2008. However, in 2009 it increased to the previous level to continue on a steady state in 2010 and 2011. The decrease in 2007-2008 is most likely linked to economic slowdown rather than any incentive effects from the Stability Period. 107 Figure 4.31: Average tax effort, Vietnam 2006-2011 110 Model 1 Model 2 Model 3 Model 4 0.85 0.83 0.81 0.79 0.77 0.75 0.73 2006 2007 2008 2009 2010 2011 37. There is very limited evidence pointing to lower revenue effort from provinces that have seen their sharing rates decline in intervening periods (table 4.2 above and provinces circled in red in figures 4.24- 4.29). There are some exceptions, for example, BRVT, Can Tho, Ho Chi Minh City and Tay Ninh, which exhibit below average performance on revenue efficiency or decentralized revenue/GDP (or both). For provinces that have seen their sharing rates increase, namely Hanoi and Dong Nai, although the revenue efficiency score is below overall average, decentralized revenue/GDP is above average. In the six other provinces, however, revenue effort and performance has been strong. There is therefore no conclusive evidence that the current sharing arrangement impacts negatively on incentives for revenue effort. 38. Credibility of revenue forecasts: Another dimension of revenue performance, which may provide additional indication of any negative incentive effects of the current sharing arrangement is the quality of revenue forecasts. As noted in chapter 3, revenue forecasts in local budgets tend to consistently underestimate actual collections. As illustrated in figure 4.32 below, overall revenue was consistently underestimated by at least 20 percent for half of all observations between 2006 and 2011. The quality of forecasts declined over the period, from a median underestimate of 10 percent in 2006 to 20 percent in 2011 (figure 4.33). 39. The key question is whether the revenue forecasts of net contributing provinces have been significantly worse compared to those of net receiving provinces. In other words, net contributing provinces may have an incentive to underestimate revenue as these inform projections for the next period when fiscal gaps are estimated. Actual collections also inform the latter, but this information comes in with an 18-month lag when the final accounts are submitted. Therefore at the time when sharing rates are determined for the next Stability Period, central authorities only have final accounts information for the fiscal year prior to the last one. Figure 4.32: Actual/local budgeted revenue (06-11) Figure 4.33: Annual actual/local budgeted revenue (06-11) 111 -1SD Median +1SD .02 .03 2006 2007 2008 .015 2009 2010 2011 .02 Density Density .01 .01 .005 0 0 60 80 100 120 140 160 180 200 220 240 260 60 80 100 120 140 160 180 200 220 240 260 Local revenues as % of local budgeted revenue Local revenues as % of local budgeted revenue Source: Staff estimates based on published State Budget data 40. The available data suggests that shared revenue projections (figure 4.34) were in general better than projections of 100 percent locally retained revenue (figure 4.35). On average, there were no major differences between shared revenue outturn of net recipient provinces and net contributing provinces (figure 4.36). For 100 percent locally retained revenue, with the exception of Ba-Ria Vung Tau, Binh Duong, Ho Chi Minh City, and Vinh Phuc, on average net contributing provinces’ revenue forecasts are not worse than those of net recipients (figure 4.37). This reaffirms that there is no conclusive evidence that the revenue sharing arrangement have a negative impact on revenue collection efforts of net contributing provinces. Figure 4.34: local shared/local budgeted shared Figure 4.35: Local 100%/local budgeted 100% revenue (06-11) revenue (06-11) .01 .01 2006 2007 2008 2006 2007 2008 2009 2010 2011 2009 2010 2011 .008 .008 .004 .006 .004 .006 Density Density .002 .002 0 0 0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400 450 500 Local shared revenue as % of local budgeted shared revenue 100% retained revenue as % of local budgeted 100% retained revenue Figure 4.37: Local shared/local budgeted shared Figure 4.36: Local 100%/local budgeted 100% 112 (period avg 06-11) (period avg 06-11) 160 243 131 133 209 209 115 113 107 111 111 116 177 170 105 Average of net 90 recipients (145) 136 123 127 111 101 98 Ba Ria - Bac Binh Can Da Dong Hai Ho Chi Khanh Quang Vinh Ba Ria - Bac Binh Can Da Dong Hai Ho Chi Khanh Quang Vinh Vung Ninh Duong Tho Nang Nai Phong Minh Hoa Ngai Phuc Vung Ninh Duong Tho Nang Nai Phong Minh Hoa Ngai Phuc Tau Tau Source: Staff estimates based on published State Budget data 41. The quality of revenue forecasts on the other hand is most likely linked to the flexibility over the use of over realized revenue as discussed in chapter 3. Local authorities are allowed to retain, and even carry over to the following years, over-realized revenue for discretionary spending. Tax agencies get rewards from central government for higher than expected collections. There are also other political advantages for local leaders derived from showing excess revenues. The revenue impact of volatile natural resource prices, or fluctuating terms of trade and import volumes, does not have such a big impact on local receipts, as these are mostly collected by central authorities. Reform of provisions on use of over-realized revenue therefore should be an important part of the revised SBL 2002. POSSIBLE REFORMS TO REVENUE SHARING ARRANGEMENTS 42. The above analysis does not seem to suggest perverse incentive effects on revenue effort from the revenue sharing arrangement. However, there are still a number of issues to consider going forward. Similar to transfers, revenue sharing arrangements involve two main policy choices: how to define the revenue pool to share, and how to allocate that share among subnational governments. A common practice in many countries is to share taxes at the same rates to avoid distortions. In Vietnam the rate is set by province, and broadly based on an estimate of the fiscal gap, combined with negotiations between central and provincial authorities. The latter can be cumbersome and lead to sub optimal outcomes due to poor revenue forecasts discussed above and differing negotiation capacities. 43. Another issue is that shared revenues in Vietnam are split based on where revenues are actually collected rather than where the tax is incurred (the so-called “derivation principle”). This raises questions concerning the fairness of the system, especially for the Value-Added Tax (VAT) and the Corporate Income Tax (CIT). For example if an enterprise is operating in Son La province, but is headquartered in Hanoi, the enterprise’s VAT and CIT liabilities will be owed to the province of Hanoi (unless it is a firm with unified accounting, in which case CIT will be paid to the central government). 44. The major taxes therefore end up credited primarily to the few jurisdictions where the headquarters of the enterprise is located or the place of business registration, and not necessarily where the outputs are produced. This helps large cities (e.g. Hanoi or HCMC) and a few more industrially developed provinces. The latter of course are also the largest contributors to the pool of shared revenue for redistribution to poorer provinces. Some of the poorer provinces, however, have indicated that the net outcome (i.e. foregone CIT and VAT plus transfers) is not necessarily better for them. Given these issues, this section looks at possible reforms of the revenue sharing arrangements. 45. Sharing Value-Added Tax on a formula basis and international experience: VAT and income taxes are traditionally considered as poor choices for assignment to local authorities. The main difficulty lies in the fact that 113 the debiting and crediting of the VAT is likely to take place in different sub-national jurisdictions, which generally implies an arbitrary apportionment of VAT revenues across those jurisdictions. This also makes it problematic to share VAT revenues with sub-national governments on a derivation basis. However, broad-based indirect taxes including the VAT are attractive for local governments because of their revenue potential. In other countries, the VAT can be shared on the basis of population (as in Germany and Belarus), or on the basis of the regional shares in aggregate consumption (as in Canada’s Maritime Provinces, Japan or Spain).25 46. It is therefore recommended that the authorities in Vietnam consider moving closer to a formula-based approach to sharing of VAT revenues using provincial population, level of consumption per capita (available in the VHLSS) and/or GDP criteria. One problem with the latter is that provincial GDP figures at present are actually computed by the provincial authorities and therefore its use in a sharing formula would introduce a serious moral hazard issue. The idea is not to introduce a perfect formula, but a more transparent and fair one, as opposed to the current “derivation basis” which results in VAT revenues concentrated in few provinces. The introduction of a formula sharing system for these taxes will probably require a transition period in which provinces’ revenue would not be adversely affected as a consequence of the reform. 47. Given strong long-term revenue potential of the VAT, other countries have explored the possibility of implementing the VAT as a subnational tax. For example, VAT could become a central-local tax, administered by either level of government on a jointly determined base, but with each government level choosing its own rate (the so-called dual VAT). Experiences with subnational VATs in practice, however, have been mixed (Perry, 2009). The lack of harmonization increases compliance costs to taxpayers operating in different jurisdictions, and has also resulted in tax wars across states. The Canadian dual VAT appears more successful. Experience to date with the state VATs in India is too short to make definitive judgments (Fedelino and Ter-Minassian, 2010). Nevertheless, this option is not recommended in the short or even the medium-term for Vietnam. 48. currently the natural resource tax (except petroleum) in Vietnam is part of the 100 percent locally retained revenue. In general, there is a partial link between taxes on natural resource extraction and the benefit principle at the local level. Extraction activities use local infrastructure, place stress on other local infrastructure, and may pollute the environment. There has been a lively debate in the fiscal decentralization literature on the pros and cons of the assignment of natural resource revenues to sub-national governments.26 49. Notwithstanding the arguments for some form of local taxation of natural resources, there are two major arguments against. First, in the case of geographically concentrated natural resources, local taxation could cause extensive horizontal fiscal imbalances (e.g. the recent cases of Indonesia, Nigeria, and Russia). These fiscal disparities can lead to inefficient population migration and location of business. Second, given the high volatility of world commodity prices, local taxation of natural resources would not constitute a stable source of revenue. It is often believed that central government is in a better position to smooth fluctuations in revenues related to natural resources, provided that these resources are transparently managed. 50. It is therefore recommended that some balance must be reached in the case of Vietnam with uneven geographical distribution of resources. First, the government should centralize taxation of natural resources to address disparities and avoid or correct for negative economic externalities. Second, in lieu of natural resource revenue sharing or assignment, the country may consider introducing production-related excises, as in Bolivia, Indonesia, and Nigeria. Excises might be fully or partly assigned to local government as part of compensation for environmental damage of the extraction process and other costs, together with a transfer system that ensures that basic subnational spending can be adequately funded. 51. Sharing Corporate Income Tax on a formula basis and international experience: CIT is usually considered to be more appropriately assigned to the central government level because of its link to macroeconomic stabilization and, to the extent that corporations are owned by wealthy individuals, this tax also affects income 25 In the case of Canada’s Harmonized Sales Tax for the Maritime Provinces, all three provinces have a uniform rate that piggybacks on the federal VAT. 26 See, for example, McLure (1996) and Bahl and Tumennasan (2004). redistribution. At a more practical level, the assignment of profit taxes at the central level is justified by the difficulty 114 of apportioning well the profits of enterprises across sub-national jurisdictions where they operate. However, from the benefit standpoint, it is understandable that certain level of CIT could be shared with local governments as part of compensation for economic and social costs entailed with the businesses in their jurisdictions, and as an incentive for them to better promote business environment which will positively impacts on local tax bases and revenues. 52. Some countries that have CIT at the sub-national level attempt to apportion the nationwide profits of enterprises among sub-national jurisdictions using a formula. These apportionment formulas generally use a weighted index of combinations of three factors: payroll (labor), assets (capital), and sales (output). But, despite these formulas, the allocation of profits across jurisdictions tends to be quite arbitrary because of the imprecise link between the location of those factors and the generation of profits. Box 4.1: Country experience in revenue sharing Germany: The revenue from corporate income taxes is shared vertically on a 50-50 basis between the federal and the state governments. Horizontally, the state share of corporate tax revenue is allocated to the states according to the location of the establishment. In the case of multi-state companies, a formula allocates overall revenue according to the state’s share of the payroll. Russia27 : The general corporate profit tax rate is 20% (with 2% payable to the federal treasury and 18% to the regional treasuries). Regional legislative authorities can decrease the regional profits tax rate from 18% to 13.5% (effectively reducing the general profit tax rate to 15.5%). Mexico: The Shared Federal Revenue (Recaudacion Federal Participable RFP) is a pool of federal revenues that is shared with states and municipalities. It includes the VAT, income taxes (individual and corporate), and most of oil revenues. It does not include revenue from public enterprises federal government financing or other sources of non-tax revenue. The amount of money used for most of the non-earmarked funds, including the General Fund, is determined by a percentage of the RFP (Garza and Sethi, 2013). Similar arrangements exist from Brazil; and South Africa but there is no direct sharing of the CIT in these countries. The sharing is through a general pool of funds which also incorporate some CIT revenues. Canada28 and the US: Have CITs at the state/province level which are separate taxes. However not all subnational governments have a CIT. These jurisdictions use apportionment formulas based on sales, assets and payroll or just one or two of those factors to apportion company profits across different jurisdictions. Source: Authors’ consolidation. 53. There seems to be little interest in Vietnam on the ideas of sharing more VAT and centralizing the CIT even if it is hard to share the latter on an efficient and fair basis among the provinces. What seems to weigh in those considerations is the perception that the VAT is a tax in expansion while the CIT is one in contraction in relative share to total revenue. There is a general perception that sharing VAT more would adversely affect the dominant role of the central budget. There is also the view that businesses should pay taxes at the local level for the services they use. 54. The biased tax structure in Vietnam toward indirect taxation can be better perceived by looking the direct taxes versus indirect taxes ratio (table 4.7). The heavy emphasis on indirect taxation in Vietnam is of course more pronounced in comparison to averages in the OECD and the whole Asia continent. With respect to the averages in South East Asia and Developing Countries the differences have been decreasing over time because the direct to indirect ratio has been slowly increasing in Vietnam (figure 4.38) and not increasing or slightly decreasing in those two groups of countries. 27 http://crossborder.practicallaw.com/9-502-0343 28 http://www.parl.gc.ca/Content/LOP/researchpublications/prb0852-e.htm http://publications.gc.ca/Collection-R/LoPBdP/BP/bp450-e.htm Table 4.6: Direct to indirect tax ratio, 1995-2008 115 Year Oecd Asia* 29 South east asia 30 Developing Vietnam countries31 1995 2,61 1,05 1,04 1,01 0,50 1996 2,56 1,12 1,09 1,22 0,54 1997 2,56 1,14 1,27 1,02 0,59 1998 2,60 1,21 1,56 0,94 0,62 1999 2,58 1,13 1,37 0,88 0,59 2000 2,53 1,09 0,97 0,89 0,79 2001 2,55 1,05 0,94 0,92 0,86 2002 2,48 1,11 0,78 0,87 0,75 2003 2,47 1,13 0,87 0,86 0,83 2004 2,48 1,15 0,64 0,90 0,81 2005 2,51 1,20 0,53 0,99 2006 2,58 1,16 0,50 1,02 2007 2,63 0,98 0,73 0,97 2008 2,68 1,86 0,90 1,01 Notes: Property taxes treated as direct taxes. Sources: IMF GFS, CEPAL , OECD Revenue Statistics, CIAT-IADB databases. POTENTIAL FOR GREATER REVENUE AUTONOMY 55. The most significant issue with current revenue arrangement in Vietnam as noted in the first part of this chapter is the lack of any significant revenue autonomy. Effective fiscal decentralization however requires meaningful levels of revenue autonomy at the sub-national level. Although some decentralized systems in developed countries have high levels of tax autonomy, in reality it is quite rare, especially among developing countries to find large taxing powers devolved to sub-national governments at the onset of decentralization. Central governments are reluctant to devolve taxing powers for fear of having to compete with local governments for the same tax bases and/or fear of losing control of fiscal policy. At the same time, subnational governments are reluctant to take on the responsibility of making politically unpopular decisions to raise their own taxes. 56. Low levels of revenue autonomy may also be associated with low levels of administrative capacity by some sub-national governments. But this point can only explain the situation of smaller jurisdictions in developing countries. Uneven administrative capacities could in theory be addressed via asymmetric tax assignments: providing more tax autonomy to larger subnational governments and letting smaller ones “grow into this role” over time. However, asymmetric decentralization design is the exception rather than the rule, and more so in terms of tax authority.33 International experience suggests three main dimensions of tax autonomy as follows, which Vietnam can consider.34 29  Asia includes all countries on Asian continent except Russia, Turkey and Middle East Countries. 30  South East Asia includes Cambodia, Indonesia, Laos, Malaysia, Myanmar, Philippines, Singapore, Thailand and Vietnam. 31  Developing country is defined by World Band as low- and middle-income level countries. 32 http://websie.eclac.cl/sisgen/ConsultaIntegrada.asp 33  See Bird and Ebel (2007) for the possibilities and problems associated with asymmetric fiscal decentralization design in a large number of countries. 34  See Musgrave (1983), Boadway (1997), Norregard (1997), McLure (1998 and 2000a), and Bird (2000). 57. Choosing which taxes are imposed by which level of government: The first dimension relates 116 to which level of government has the right to choose the taxes that this given level can impose. There is a choice between an open list of taxes to be determined by the sub-national governments within general limits and restrictions, or instead a closed list of allowable taxes determined at the national level from which sub-national governments can choose. Overall, a closed list of sub-national taxes tends to be preferable because it avoids the introduction of nuisance taxes or higher and inefficient distortionary taxes which can easily impede local economic development and growth. 35 In the international experience, where sub- national governments are given more constitutional discretion, as in the case of some federal systems, open lists with some general restrictions are common. Closed lists are used more frequently in unitary systems of government. 58. Whether an open list or closed list approach is adopted an additional decision needs to be made as to whether the base of specific taxes should be used exclusively by one level of government or whether these bases can be used simultaneously by several levels of government. The latter approach has the disadvantage of introducing vertical tax externalities due to the fact that one level will not typically take into account the impact its policies may have on the tax base and revenues of the other level of government.36 59. International good practice suggests that the optimal assignment of taxes derives directly from the optimal assignment of government functions. Taxes with large and elastic bases, such as income taxes, should be assigned to the central government as the best instruments for both macroeconomic stabilization and income redistribution. For allocation functions, benefit taxation should be used by various government tiers. Accordingly, the provision of local services (such as public utilities and local transportation) should be subject to user charges and fees; services with local benefit zones (parks, roads) should be financed with local taxes; while goods and services with significant externalities should be financed with region-wide taxes or transfers. To prevent revenue losses or tax erosion resulting from tax competition37 , local tax bases should be relatively immobile; bases should also be evenly distributed across jurisdictions, to avoid horizontal fiscal imbalances; and their yield should be relatively stable, to allow subnational governments to rely on steady and predictable revenue streams (Fedelino and Ter-Minassian, 2010). 35 The international experience shows that providing sub-national governments with freedom to select their own taxes ( the open list approach) can easily backfire when sub-national governments introduce highly inefficient (distortionary) forms of taxation. A recent example is provided by Indonesia, which adopted an open-list approach in the 2001 decentralization reform. See Alm et al. (2004) 36 See Dahlby and Wilson (1996, 2003), Keen (1998) and Boadway et al. (1998). 37 However, not all tax competition is undesirable; a moderate tax competition gives an incentive to politicians and bureaucrats to be efficient and to provide services according to citizens’ preferences in their choice of taxes. Table 4.7: Assignments of taxes to subnational governments 117 Tax Advantages Disadvantages Tax Advantages • Buoyant revenues • May create or aggravate horizontal imbalances Disadvantages • Visible (increases accountability) In areas where average incomes are below •   ost-effective if piggybacked on national • C threshold, insufficient yield; also, as most peo- taxation ple would not pay, the price-signal effect of the tax is weakened. If levied at different rates among jurisdictions, it •  may create distortions if people are mobile. Corporate • Sometimes seen as a sort of benefit tax  obile tax based and complex administration • M income tax make it suitable for collection by the center (except taxation of small businesses) User fees and • Low-mobility tax base • Generally low yield charges  o obvious horizontal or vertical imbalance • Low cost-efficiency • N problems • Visible • Linked to benefits Property tax • Immobile tax base  ifficult administration (especially in setting up • D • Visible well-functioning cadastres), often resulting in • Stable yield low yield. • Indirectly linked to benefits Sales and • No horizontal or vertical imbalance problems  ay create cross-border shopping if levied at • M excise tax • Visible different rates among subnational jurisdictions. • Easy to administer f properly designed and administered, it Value-added tax • I • Complex tax administration could be a good subnational tax, although If applied on destination principle, border •  experience has been mixed (see text) controls between local jurisdictions required; if applied on origin principle, tax exporting and transfer pricing may arise Resource tax • Non apparent • Significant horizontal imbalances • Difficult to administer • Excessively volatile Sources: Feyrioglu and Norregaard, 2008. 60. Legislating structure of tax bases and setting tax rates: The second dimension of tax autonomy relates to which level of government can legislate over the structure of the tax bases and which level has discretion to set the tax rates. Autonomy to define the tax base is generally less desirable than autonomy to set tax rates.38 Variations in the definition of the tax base can more easily lead to complexity and lack of harmonization across jurisdictions. It is often also argued that tax rate setting autonomy may be preferred because it has a more direct impact on revenues and spending ability of sub-national jurisdictions, because it has a more transparent impact on locational decisions and fiscal competition: both households and businesses have an easier time figuring out the fiscal exchange or net benefits provided by different jurisdictions in their tax-public service packages when the differences in tax burdens are expressed in terms of differences in tax rates. 38 The ability to change either base or rate opens up the possibility of fiscal competition among sub-national governments (Wilson, 1999). Inter-jurisdictional fiscal competition can have both good aspects, such as offering choices to taxpayers and keeping public officials more accountable, and also bad aspects, such as a “run to the bottom” type of behavior actually taking place in countries with a significant degree of sub-national tax autonomy. In addition, the ability to change tax base or rate can give rise to “horizontal” fiscal externalities, whereby the policies of one jurisdiction (for example, raising tax rates) can have an effect on the tax bases of other jurisdictions (raising their tax bases related to mobile taxpayers). Intergovernmental grants and other policies can be implemented by the central government to correct horizontal fiscal externalities. See, for example, Arnott and Grieson (1981), Gordon (1983) and Wildasin (1983, 1989). 61. The most important unwanted consequence of the lack of harmonization and complexity is the higher 118 tax administration cost for all the jurisdictions involved and higher compliance costs for taxpayers who have tax obligations in several jurisdictions. On the other hand, autonomy to define the tax rate generally tends to be more desirable because it is simpler to deal with across jurisdictions for both tax administrators and taxpayers. Focusing on autonomy in tax rate setting has the additional important advantage of being perceived to generate political accountability. 62. Tax administration: The third dimension of revenue autonomy refers to which level of government is put in charge of administering the various taxes. In summary, it would generally seem to be undesirable to decentralize the administration of taxes shared between the federal (central) and local governments or subnational taxes that piggyback on national taxes. In the case of exclusively subnational taxes, their administration can be assigned to the central tax administration but this needs to be accompanied by incentive-compatible contracts so central tax administrations pay sufficient attention to the collection of subnational taxes. Subnational tax administrations are more desirable for exclusive subnational taxes for which there are information and enforcement advantages at the local level. When the system is already working with multiple levels of tax administration, a high degree of dialog and cooperation between different levels of tax administration should be institutionalized. 63. From the three dimensions of tax autonomy discussed above, the optimal option for Vietnam now would be to provide local governments a closed list of selected taxes with discretion to set rates within minimum and maximum rates established by the central authorities for most, if not all, the taxes that have been fully assigned at the subnational level. Another possible way to increase autonomy of course would be to assign new taxes at the subnational level including a unified and reformed real estate property tax. 64. Providing local governments a closed list of selected taxes with discretion to set rates within a band: Surcharges, or piggybacking, on central taxes may provide additional subnational revenue, building on the central tax administration. These arrangements confer a more limited degree of autonomy to local authorities because they simply impose a surcharge on tax bases defined by the central government. Still, there may be incentives on both sides—local authorities may have an advantage in identifying potential taxpayers, to the mutual benefit of both the center and the subnational governments. Surcharges are most typically levied on personal income taxes, but piggybacking on sales taxes and excises also occurs. Surcharge tax rates are frequently subject to both upper and lower limits set by the central government. Although surcharges are easy to administer, a possible drawback is that they may create (or deepen) horizontal disparities, because revenue tends to be geographically concentrated in richer and more- developed jurisdictions. This situation would require an appropriate equalization transfer system. Vietnam may consider introducing three types of surtaxes. 65. Surtax 1 – Centralizing Personal Income Tax and allowing surtax on PIT in selected provinces: Progressive income taxes are usually considered best assigned at the central government level, because income redistribution should be an objective pursued by the central governments because of the mobility of taxpayers and so on. Another reason for this assignment is that progressive income taxes tend to act as automatic economic stabilizers and macroeconomic stabilization which should primarily be a function of the central government. Vietnam may consider going in this direction. At the moment the share of PIT is low, but will increase as urbanization grows. 35 The international experience shows that providing sub-national governments with freedom to select their own taxes ( the open list approach) can easily backfire when sub-national governments introduce highly inefficient (distortionary) forms of taxation. A recent example is provided by Indonesia, which adopted an open-list approach in the 2001 decentralization reform. See Alm et al. (2004) 66. Nevertheless, Vietnam should consider allowing provinces to implement surcharges within a range on PIT.39 To enhance revenue autonomy, local governments may be allowed discretion in setting the flat rate 119 between minimum and maximum rates, which are centrally legislated. A flat rate local piggyback income tax easily satisfies the benefit principle and, being quite visible, it promotes political responsibility and accountability at the sub-national level. This is also an elastic tax with revenues increasing commensurate with income, so that as the demand for local services increases with income, so do tax revenues. 67. Surtax 2 – Allowing some other form of business taxes at the subnational level separate from the CIT: a number of countries have successfully introduced these kinds of surtaxes (see annex F). Certain forms of business taxes or business license fees are justified at the sub-national level as an indirect but administratively easier way to tax income of business owners (especially non-wage incomes), and as a benefit tax for the services and infrastructure provided by sub-national governments. 68. Where it is not feasible to recoup costs of local government services through user charges, some form of broad-based levy on general business activity is warranted. To avoid economic distortions, the broad- based levy should be neutral to the factor mix, applying equally to labor (payroll) and capital (assets) used by businesses. One such tax is the business value tax (BVT).41 The base of the BVT resembles that of the VAT but the two taxes function quite differently. In contrast to the usual destination-based VAT, the BVT is origin-based, therefore taxing exports (not imports) because the benefits from local services accrue at the place of production (not consumption). In addition, while the typical VAT is calculated by the subtraction method on transactions (gross receipts minus the cost of intermediate goods and services), a BVT would be calculated by adding payroll, interest, rents, and net profits on the basis of annual accounts. A potential disadvantage of a BVT is that it requires good levels of accounting and record keeping and quite advanced tax administration capacity. 69. Surtax 3 – Allowing some form of excises surtaxes: Excise taxes have potential as piggyback taxes or special taxes at the sub-national level. From an efficiency standpoint, excises should be levied on a destination basis (where the taxed good is consumed), but this may give rise to smuggling and cross- border shopping if subnational jurisdictions apply significantly different rates. Excises tend to be more politically acceptable, can be easily administered in coordination with national wholesalers as withholding agents, and allow for rates differentiated by region. For example, some OECD countries allow sub-national government surcharges on excises.42 An interesting aspect of excise taxation at the sub-national level is the taxation of public utility services. There is significant revenue potential in some of these services, as in the case of electricity and phone services. Similarly, excises of various kinds are sometimes used at the subnational level, and may be relevant at either the intermediate (regional) or local tiers of government. 70. Increasing autonomy over determination of user fees at sub-national level: Aside from the above three surtaxes, Vietnam could also consider more autonomy over the determination of user fees at local level. User charges and fees tend to represent significant shares of total revenues in the city budgets of developed countries.43 However, they tend to represent a much smaller share of total city revenues in developing countries. There are, on the other hand, some important exceptions; for example, user charges and fees represent more than one-third of total revenues in Cape Town. 39 Users include all the Nordic countries in Europe (Norway, Sweden, Denmark, Finland) and countries in Central and Southern Europe (Germany, the Netherlands, Spain Italy etc.). It has also been adopted in may Transitional countries in Eastern Europe. 40 Other forms of tax autonomy are practiced, such as the ability to modify the base of the tax by providing more or less deductions, exemptions and so on. . 41 Bird (2003) 42 For example, in the Netherlands, provinces impose a surcharge on the motor vehicle tax levied by the central government. Provinces are free to set the rate of the surcharge, subject to a ceiling imposed by the central government. 43 In the U.S. local user fees and charges represent one-fourth of own source revenues (35 percent when local public utilities are included). Canadian local governments similarly raise one-fourth of their own revenues from user fees and charges (Fox and Slack, 2010). See also Bahl (2010). 71. While user fees provide important efficiency benefits, it is important to balance the cost of collecting 120 and administering user fees with the amount of revenues collected; certain types of user fees involving many small transactions may be too costly to collect. It can make good sense to bundle the collection of a variety of fees into a single payment. For example, it is possible to collect refuse collection fees or street lighting fees as a surcharge on property taxes. 72. In Vietnam, there are a number of issues with the extent and final use of public fees and charges, and whether or not they are included in the budgets and captured within the Treasury system and if so, whether they are recorded on a gross basis (i.e. before the shared retained by local spending units). In terms of their operation, there remains too little flexibility at the subnational level although it appears that for some local fees, there is local discretion to set the fee between maximum and minimum in the public ordinance on charges and fees. As part of the Government’s devolution policy, user fees and charges by service delivery units, which sub-national governments can retain and spend at their discretion, are not all captured in the budget plan. As a consequence, total spending and receipts are understated in the accounts. 73. There is considerable room for more autonomy of local governments in Vietnam in setting fees, perhaps within a range set by central government. To the extent that the main purpose of “real” licenses and user fees is to recover the administrative costs of issuing the licenses or the cost of providing the public services, it is important to price the service right. Requiring sub-national governments to set the fee levels below the actual cost of provision imposes an unfunded mandate and it can easily lead to poor provision of services. 74. Introducing property tax in the long-term: There is ample consensus that property and land taxes and betterment levies are ideal taxes to be assigned at the sub-national level. Local governments have a comparative advantage in identifying and valuing properties because they are familiar with the housing and land available there. The degree of discretion given to local governments to manipulate the tax may vary but the thinking is that this tax belongs to local governments seems well entrenched. However, determining the value of property tax bases is often difficult, thus complicating their administration, especially in countries where market valuation is hindered by limited real estate market activity or by limited information on market transactions. Hence, determining the tax base is often a matter of judgment. International experience suggests that the yield from property taxes is usually limited44 though largely underexploited (OECD, 2004). 75. In Vietnam, there is currently intense study for the reforms of the system of property taxation possibly combining many of the different taxes related to land and structures into a modern property tax, according to the Tax Policy Department. A perceived deterrent for advancing on that line is the existing housing shortage and increased costs of housing and also how to handle distributional issues, especially in relation to low income households. The international experience shows how to meet the challenge of making the design of the property tax in Vietnam an attractive policy option for decision makers at the highest level of the central administration. Given the complexities, the introduction of a modern property tax may be delayed until 2020. 44 The share of property taxes in subnational revenue varies considerably, from as low as 5 percent in Turkey and 7 percent in Norway, to 90 percent in New Zealand and 100 percent in Australia, Ireland, and the United Kingdom. As a share of GDP, property taxes account on average for about 1 percent in unitary countries and 2 percent in federal countries (Fedelino and Ter-Minassian, 2010). Box 4.2: Land Administration Reform and Property Tax 121 Taxing land can potentially increase economic efficiency and improve equity. In 2010, the associated revenue was 2 percent of GDP and hardly sustainable given that 87 percent of the actual revenue came from land-use right allocation levies. In addition to broadening the tax base, betterment levies in the form of up-front lump-sum payments from the beneficiaries can make property tax become a useful instrument to improve public services, because it provides subnational governments with resources to invest in local infrastructure. By taxing income derived from property, this instrument can also lead to significant improvements in the quality of land use. Taxing property can also help contain land speculation, one reason for the real estate bubble, and thus contribute to improved land administration and equity. Property taxation may result in a situation where poor people are unable to pay. Exempting subsistence farms from property taxes and introducing a minimum imposable threshold are obvious solutions to poor people owning land of little value. The problem of poor people with land of significant value (particularly in cities) is that they may be reluctant to move to cheaper land for understandable social reasons, especially in the case of the elderly. A possible way to address this concern is to allow any poor person to defer his or her taxes (plus interest) until death, and then recover the debt from the value of the property. There is another concern that taxing property may be complex and costly compared to the revenues it may potentially generate. A fundamental challenge is property revaluation on a regular basis. Systematic property assessments conducted by area can bring the cost down. The self-assessment of property prices is another practical solution. Avoidance risks can be mitigated through the provision that the self- assessed value is what will be paid as compensation by government if the property is reclaimed for the purpose of infrastructure development. However, a more serious difficulty in Vietnam is the incomplete issuance of land-use right certificates, especially in urban and upland areas. Household survey data from 2006 suggest that only 76 percent of agricultural land parcels, 68 percent of urban land parcels, and 34 percent of forestland parcels have been granted land-use right certificates. In practice, this means that about two-thirds of the total land parcels of Vietnam still lack a certificate. Despite the promulgation of the Land Law in 2003, the main reason for this problem is the shortage of basic infrastructure for an effective operation of land administration, including cadastral mapping, transaction registrations, and record management to the provision of land administration services. The system itself is cumbersome, not transparent, and inefficient, and does not provide the quality services that end users can rely on. As a result, land administration is generally recognized as one of the most severe constraints to business development and transparent governance in Vietnam. Combined with limited local capacity, this calls for a step-by- step approach in introducing this new tax instrument. Source: WB, “Tax Reform in Vietnam: Toward a More Efficient and Equitable System” (2011); and Vietnam Development Report (2009): Capital Matters. CONCLUSIONS AND RECOMMENDATIONS 122 76. Local authorities in Vietnam have seen increased levels of local revenue in the 2006-2011 period, but continue to have little to no autonomy over revenue policy and administration. Compared to other countries, decentralized revenues in Vietnam constitute relatively large shares of national GDP (9-10 percent) and general government revenue (33 percent excluding extra budgetary sources). The share of decentralized revenue over total local revenue however has declined over time. This is partly because richer provinces have been making higher contributions of shared revenue to central government for the purposes of redistribution through transfers. Analysis in this chapter does not suggest that this has had a significant negative impact on revenue effort by these net contributing provinces. On average provinces in Vietnam collect roughly a little over 80 percent of their potential estimated revenues. But performance across provinces varies a lot due to factors such as local capacity and the level of revenue decentralization as measured by share of 100 percent locally retained revenue. The latter, together with the needs of richer, high growth potential provinces, highlight the need to consider opportunities for more revenue autonomy. The revenue sharing arrangement also warrants review to improve its transparency and equity. Issues Recommendations Eliminate provisions in the SBL 2002, which State Budget Law 2002 Current flexibility over spending of over realized include issues such as revenue carry over and revenue leads to negative incentives to use of over-realized revenue. The latter needs underestimate revenue in local budgets, which to be mandated by the legislature through impacts negatively on overall credibility of the plan. appropriations. Consider two options: (i) moving to sharing VAT Revenue sharing on a “derivation basis” leads on a formula basis; and (ii) centralizing natural to concentration of major revenues in selected resources tax and CIT for redistribution through the provinces. transfer system. regulations Increased revenue autonomy is needed for Consider introducing a system of surtaxes (on Stability Period better revenue effort and to help net contributing personal income, business profits, and excises) provinces meet their infrastructure financing ne and more autonomy over fee setting. Over long- eds. term consider property tax. 123 INTERGOVERNMENTAL FISCAL TRANSFERS 124 05 Key issues: revenue sharing arrangements reviewed in chapter 4 are designed to address vertical imbalances arising out of the gap between local revenues and spending needs. Balancing transfers are meant to address horizontal imbalances across provinces. One issue that arises is the extent to which the transfer system is delivery more equitable resource distribution across and within provinces. Aside from revenue sharing and balancing transfers, there are also targeted transfers that face challenges on how to link national objectives with results on the ground. Objectives: the objectives of this chapter are to assess the extent to which balancing transfers are achieving their aim to equalize resource availability not just across provinces but also across districts, and recommend opportunities for reform. Another objective of this chapter is to identify ways to reform the system of targeted transfers to more effectively deliver on their intended targets. Key findings: revenue sharing arrangements and the system of balancing transfers in Vietnam have played a significant role in narrowing vertical and horizontal fiscal imbalances respectively. There are however growing vertical imbalances due to local authorities’ increased spending responsibilities. Trading off richer provinces’ right to retain a higher share of shared revenue for growth enhancing spending to allow more redistribution is welcome though also points to potentially greater need for revenue autonomy (chapter 4) and access to debt financing (chapter 6), particularly for some of the richer provinces. Increased spending responsibility of both provincial and district authorities are covered by increased balancing transfers. The latter at the center to provincial level are underpinned by a transparent, norms-based system. At province to district level, the publication of stability period rules and regulations have also helped to improve transparency. The system of fiscal transfers has helped to promote greater equity in resource distribution both across and within provinces. There is evidence a fairly wide variation in per capita spending across districts. This may reflect the diversity in conditions of different districts, though it is difficult to say for sure what explains this remaining disparity. Target transfers continue to play an important role in local spending even though its relative share has fallen over time. The lack of predictability in targeted transfers poses serious challenges. Resourcing is not closely aligned with the targets and objectives, which are quite ambitious. Complex, input-based guidance reduces flexibility and increases burden of reporting. There is a real opportunity to review the institutional arrangements for NTPS for the coming stability period aiming to simplify the system and increasing transparency, monitoring and accountability Recommendations: (i) link stability period balancing transfers to inflation so do they do not decline in real terms; (ii) selected adjustments allocation norms to determine spending needs; (iii) ensure greater predictability of targeted transfers to provinces and districts; (iv) simplify institutional structure for NTPS and link resourcing to performance. CHAPTER 5: INTERGOVERNMENTAL FISCAL TRANSFERS 125 1. Background: Vietnam has developed a relatively transparent, formula-based system for intergovernmental fiscal transfers between central and provincial authorities. There are two main elements to this system (annex G). The first is the revenue sharing arrangement discussed in chapter 4, which allows sub-national authorities to retain a certain proportion of collected taxes to help reduce the gap in fiscal capacity across provinces caused by different revenue raising abilities. This gap is also referred to as the vertical fiscal imbalance. Whilst the revenue sharing arrangements can help to narrow the vertical imbalance, different expenditure needs across provinces can still leave a resourcing gap, referred to as the horizontal fiscal imbalance. To address this, the central government provides balancing transfers to provincial authorities. 2. Revenue sharing and balancing transfers are in turn used by each provincial authority to address vertical and horizontal imbalances across districts. For different types of revenue, resolutions issued by Provincial People’s Councils set out sharing rates between provincial and district authorities, which may be different than the rates set between central and provincial authorities. In the case of balancing transfers, different provinces estimate revenue capacity of districts and use norms to estimate spending needs; the fiscal gap provides the basis to decide the level of transfers down to districts. Some provinces may use similar spending estimation norms to those set by the center, but others may decide to use different norms. This is part of the autonomy granted for local authorities to decide sub-national fiscal arrangements by accounting for the specific conditions in their localities. 3. This chapter helps to take stock of how the current system is correcting for fiscal imbalances and how effectively it is promoting fiscal equity both across but also within provinces. Earlier studies in Vietnam (Martinez-Vazquez, 2005) have looked at these outcomes across provinces. This helped to assess the central-provincial relationship. But with the sub-national fiscal data collected for this study, this chapter is able to also analyze fiscal equity issues across districts within each province. Given the autonomy accorded to provincial authorities, this analysis is critical to assess whether the redistributive objectives in the central government’s transfer policies are being enhanced or impeded by provincial decisions over shared revenue and transfers to districts. 4. In addition to revenue sharing and balancing transfers, there are also targeted transfers from upper tiers of government to meet specific targets and objectives within lower jurisdictions. Central government provides national target transfers to provincial authorities for nationwide targets and objectives, but may also provide specific transfers for particular activities in selected provinces. Similarly provinces also provide targeted transfers down to districts and communes. This chapter reviews the institutional arrangements for National Target Programs in particular. Based on discussions with central and provincial authorities, it reviews whether the current institutional structure could be further strengthened to link objectives set by upper tiers of government to actual results on the ground. ADDRESSING FISCAL IMBALANCES ACROSS AND WITHIN PROVINCES 126 5. Vertical imbalances across provinces and center-province revenue sharing: Vertical imbalances arise due to the different revenue collection potentials and expenditure needs of different levels of government. Typically subnational governments collect from local revenues less than their needs while the opposite is true for central governments. It is difficult to measure the true extent of vertical fiscal imbalance because of challenges in estimating expenditure needs, which are affected by the quality and quantity of services. There are also different definitions of vertical fiscal balance; some use own revenue over local spending and other use transfer dependency.45 6. As noted above, in Vietnam these vertical imbalances are corrected in part by sharing of taxes between central and provincial authorities. The particular sharing rates for each province are set out in central regulations. For the districts, the sharing rates with provinces are specified in provincial level resolutions. These rates remain constant over the Stability Period. 7. This study takes 100 percent local revenue over local core expenditure as a rough approximation of vertical imbalance for a province as a whole. Normally, vertical imbalance for a given level of government is the gap between own spending (i.e. total spending minus transfers to lower tiers of government) and own revenue. In the case of Vietnam, we take local core expenditure in the denominator, which includes capital and recurrent spending by provincial authorities net of transfers to districts, and total spending by district authorities.45 The numerator includes provincial and district revenue, which are 100% retained by local authorities (i.e. does not includes shared taxes-- regardless of the sharing rate-- or transfers). 8. Available data for the 2006-2011 period shows significant vertical imbalance across provinces in Vietnam. Figure 5.1 shows that in around half of all observations over this period, less than 40 percent of local core expenditure was covered by revenue that is 100 percent retained by provinces. This is not surprising because the revenue types under the revenue retained 100 percent by local governments represent a small part of the tax base. 9. What is interesting however is that over time, more provinces are relying increasingly less on 100 percent retained revenues to finance local core spending. This is illustrated in figure 5.2, which shows that in 2011, the distribution of 100 percent retained revenue over local core spending is a lot more concentrated with a median value of around 20 percent. This means that in 2011, 50 percent of local authorities financed less than 20 percent of their core spending from 100 percent revenue. In other words, vertical imbalance across provinces has grown over time. This is due in part to the rise in spending responsibilities that are assigned to local authorities, as discussed in chapter 2; the fact that 100 percent retained revenues tend to be less elastic with respect to nominal GDP is also likely to be a contributing factor. 45 Eyraud, L, and Lusinya, L, “Vertical Fiscal Imbalances and Fiscal Performance in Advanced Economies,” International Monetary Fund (October 17, 2012 draft) 46 Core expenditures exclude spending from off-budget sources, carry over spending (which is accounted for in the year it was appropriated), spending out of contingencies and transfers to reserves. Figure 5.1: Vertical imbalance across provinces Figure 5.2: Annual vertical imbalance across (06-11) provinces (06-11) 127 -1SD Median +1SD .02 2006 2007 2008 .02 2009 2010 2011 .015 .015 .01 .01 .005 .005 0 0 0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180 Source: Staff estimates based on published State Budget data 10. The revenue sharing arrangement has played an important role in helping to narrow vertical imbalances. In Vietnam, revenue retained 100 percent by local authorities plus shared revenue (i.e. portion that is retained by local authorities) is referred to as decentralized revenue. As illustrated in figure 5.3, when shared revenues are added to 100 percent retained revenues, in around half of all observations at least 55 percent of core expenditure was covered by decentralized revenue. This compares to 40 percent coverage by 100 percent retained revenues as mentioned above. 11. The data suggests that the level of shared revenue in local authorities’ core spending has remained fairly steady over time. If anything, the distribution of decentralized revenue over core expenditure has narrowed as shown in figure 5.4, which means that more provinces are relying less on decentralized revenue to cover local spending. This suggests that fiscal transfers have grown more quickly than shared revenue to compensate for the growing vertical imbalance. This is discussed in more detail further below. Figure 5.3: Decentralized revenue/local spending Figure 5.4: Annual decentralized revenue/local (06-11) spending (06-11) -1SD Median +1SD 0 .002 .004 .006 .008 .01 .012 .014 2006 2007 2008 .015 2009 2010 2011 .01 .005 0 0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180 Source: Staff estimates based on published State Budget data 12. Vertical imbalances within provinces and province-district revenue sharing: Vertical imbalance 128 of district (and commune) authorities is measured in this study by revenue 100 percent retained by district authorities over district expenditures. This is the vertical imbalance of all district and commune authorities within a province – but we only refer to district for simplicity. 13. Available data for the 2006-2011 period shows that vertical imbalances were more severe within provinces than across provinces. Figure 5.5 shows that in around half of all observations, less than 25 percent of district expenditures are covered by district 100 percent revenue retained by the districts. For those districts therefore, 75 percent of core spending is covered from other sources. Like the result for provinces as a whole, there is also evidence that over time, districts are increasingly less reliant on 100 percent retained revenue to cover their expenditures. For 50 percent of all observations in 2011, less than 12 percent of district spending was covered by 100 percent retained revenue compared to around 20 percent in 2006. This is consistent with the finding in chapter 2 on the decentralization of spending responsibilities to district authorities, which have grown faster than 100 percent retained revenue. Figure 5.5: Vertical imbalance below province Figure 5.6: Annual vertical imbalance below (2006-2011) province (2006-2011) -1SD Median +1SD .04 .04 2006 2007 2008 2009 2010 2011 .03 .02 .02 .01 .01 0 0 0 20 40 60 80 100 120 140 160 180 0 10 20 30 40 50 60 70 80 90 100 District 100% retained revenue\district spending District 100% retained revenue\district spending Source: Staff estimates based on published State Budget data 14. Shared revenue have played an important role in helping to close the vertical imbalance within provinces but there is also evidence that decentralized revenue have not increased by as much as spending decentralization as noted in chapter 2. As illustrated in Figure 5.7, for half of all available observations from 2006 to 2011, decentralized revenue covered less than 30 percent of district spending. There is also evidence that over time districts are spending a smaller share of their expenditure out of decentralized revenue. In 2011 around half of provinces funded less a quarter of their spending out of decentralized revenue. These developments point to increased dependence over time on provincial transfers, which are discussed further below. 45 Eyraud, L, and Lusinya, L, “Vertical Fiscal Imbalances and Fiscal Performance in Advanced Economies,” International Monetary Fund (October 17, 2012 draft) 46 Core expenditures exclude spending from off-budget sources, carry over spending (which is accounted for in the year it was appropriated), spending out of contingencies and transfers to reserves. Figure 5.7: Decentralized revenue/district Figure 5.8: Annual decentralized revenue/district spending (06-11) spending (06-11) 129 -1SD Median +1SD .025 .03 2006 2007 2008 .02 2009 2010 2011 .02 .015 Density Density .01 .01 .005 0 0 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 District dencentralized revennue/district ependiture District dencentralized revennue/district ependiture Source: Staff estimates based on published State Budget data 15. Although most revenue sharing and fiscal transfer arrangements within provinces are decided in provincial level resolutions, the SBL 2002 Article 34 also has two specific requirements.47 The first is that communes and townships will receive at least 70 percent of revenues from tax on transfers of land use rights, land and housing taxes, license tax on individuals and individual households, and registration fees for land and housing. The second is that townships and cities will receive at least 50 percent of revenues from registration fees, excluding registration fees for land and housing. 16. Provincial authorities however have indicated that these are too restrictive and therefore warrant review. The reason, they argue, is that many communes and townships are not able to absorb the minimum stated shares of resources. On the other hand there are other communes that need additional resources to meet their spending needs. Funds from the sources in Article 34 cannot be reallocated from surplus to deficit areas. This can lead to inefficient spending or regular carry overs in surplus jurisdictions, and poorer service delivery results in deficit jurisdictions because of lack of available funding. For these reasons it is recommended that Article 34 of the SBL 2002 be reviewed so that district authorities are not constrained in reallocating resources across communes or townships. 17. Institutional arrangements for balancing transfers from center to provinces: After revenue sharing arrangements have helped to narrow the vertical imbalance across and within provinces, there are still horizontal imbalances that arise because of different revenue capacities and expenditure needs. Spending needs across provinces will differ for a variety of reasons. As discussed in chapter 2, costs of service provision can be higher in mountainous regions with sparsely populated areas (e.g. Northern Mountains, Central Highlands); similarly costs can be lower in more densely populated urban areas (e.g. Hanoi, Ho Chi Minh City). 47 Martinez-Vazquez, J, “Making Fiscal Decentralization Work in Vietnam,” Andrew Young School of Policy Studies, Georgia State University, Working Paper 05-13 (June 2005), p.23-24 18. To address horizontal imbalances across provinces, the central government has adopted a formula- 130 based system of balancing transfers. There are two formulas, one used to estimate recurrent spending needs across provinces, and another used to estimate capital spending needs across provinces. The recurrent and capital spending needs are aggregated to arrive at total spending needs for each province. This total is then subtracted from the estimated decentralized revenue for each province, which then constitutes the estimated horizontal fiscal gap. This in turn provides the basis for determining the level of balancing transfer to that province. In reality, there is also some negotiation between central and local authorities on the exact level of balancing transfers. 19. The formula to determine recurrent spending needs is contained in Prime Minister’s Decision No.59/2010/ QD-TTg (September 30, 2010). The way it works is that a specific per capita financial allocation, also referred to as an allocation norm, is assigned against particular criteria. For example, for the education sector, the criteria against which allocation norms are assigned might be the number of children of school going age. The allocation norm is multiplied by the number of children of school going age in a particular jurisdiction to come up with an estimate of education spending needs for that jurisdiction. 20. At provincial level, there are allocation norms for spending estimates for districts across 19 categories of expenditure, mostly different functional areas of spending such as education, health, and economic services. For each functional area, there is a per capita allocation norm that is based on geographic location (urban, plain areas, mountainous areas, and highlands/islands). For example, for the health sector, urban areas will be allocated VND 105,600 per person per year, whereas mountainous areas will be allocated VND 1,986,880 person per year. This therefore explicitly recognizes different costs of service delivery across different geographic areas. 21. Using geographic area and population is a marked improvement from past practices where allocation norms were based on physical outputs such as the number of hospital beds or length of roads. That practice encouraged higher investment in physical assets even if it meant excess capacity. Physical norms would also skew spending needs for richer provinces, which are likely to have more of these assets. Another important development change is in the education sector, where allocation norms are per person of school going age. This is better than using overall population because this would penalize areas with a proportionately higher share of children of school going age.48 22. The formula for determining capital allocation needs is set out in Decision No.60/2010/QD-TTg (2010). The allocation criteria are transparent and can be grouped into five categories: population, development, geographic area, number of district administrative units, and others. Points are awarded to each province based on these criteria and the points across all provinces are summed up. The total capital funding for the provincial governments is decided by the MOF in coordination with the MPI. Given this total available pool of funds for distribution among the provinces, each province is allocated a share depending on the ratio of its points to the total points across all provinces. The formula is summarized in table 5.1. 23. The above criteria are geared to help narrow regional gaps in economic development and also to provide capital services where they are most needed according to population, land area, etc. There are also some fiscal incentives for greater revenue collection built into the criteria. Points are awarded for “internal revenue mobilization” as well as “contribution to the central budget”. Points are proportionally higher if the internal revenue mobilizations and the contributions to the central budget are higher. While these criteria do reward fiscal effort, it is also disconnected with the fiscal capacity of the province. For instance, a more developed province will be able to automatically collect more revenues. There may be some merit because richer provinces may also have higher capital investments in absolute terms. Nonetheless an alternative method of rewarding fiscal effort linked to fiscal capacity may also be worth considering. 48 Martinez-Vazquez, J (2005), p. 55 Table 5.1: Formula to estimate capital investment needs in provinces 131 Criteria Allocation of funds by points Population (Ai) Average population of province i (hi) Ethnic population of each province (ki) population points: Ai=hi+ki Development (Bi) Poor household ratio of province i (li) Internal revenue points of province i (mi) Contribution to the central budget of province i (ni) development points: Bi=li+mi+ni Area (Ci) Natural land area of province i (oi) Rice cultivation area of province i (pi) area points: Ci=oi+pi District administration General district administrative units of province i (qi) (Di) Mountainous district administrative units of province i (ri) Highland and island district administrative units of province i (si) Border district administrative units of province i (ti) district administration points: Di=qi+ri+si+ti Additional criteria Special city points of province i is Ei Points for cities under central administration is Fi Points for provinces in economic hubs is Gi Points for regional and sub-regional centers is Hi Points of cat 1, 2, 3 cities is Ii total points of additional criteria: Max (Ei;Fi;Gi;Hi )+Ii Total (Xi) total points of province i: Xi=Ai+B+Ci+Di+Max (Ei;Fi;Gi;Hi )+Ii Allocation per province Total points of 63 provinces and cities: Y=∑i=1 63 Xi Available capital funding for provincial budgets: K Capital amount for each point: Z=K/Y Funding for each province Vi=Z* Xi 24. The recurrent allocation norms and the separate determination of recurrent and capital spending needs do not impede local authorities’ autonomy over resource allocation decisions. In other words, the recurrent allocation norms as a very useful means to benchmark spending needs across provinces, but provincial authorities are not bound by these in their decisions. Some authorities decide to apply the same norms for allocations within the province, whilst others adopt their own norms. In relation to the estimate of capital spending needs, it is important to note that this does not in any way prescribe the level of capital spending by provincial authorities. The estimate is an added input into overall spending estimates for a province based on criteria that targets local economic development. 25. Balancing transfers are highly predictable as they are fixed in nominal terms over each Stability Period. As illustrated in figure 5.9 and 5.10, actual balancing transfers at both provincial and district levels were exactly the same as budgeted balancing transfers in the vast majority of observations from 2006 to 2011. This is an important feature as low levels of own revenue and high dependence on transfers in other countries have led to greater uncertainty and less predictability in financing flows for sub-national authorities. This in turn impacts negatively on planning, budgeting, and budget implementation. Figure 5.9: Pedictability of balancing transfers to Figure 5.10: Predictability of balancing transfers 132 provinces (06-11) to districts (06-11) -1SD Median +1SD -1SD Median +1SD 30 30 25 25 20 20 Frequency Frequency 15 15 10 10 5 5 0 0 80 90 100 110 120 130 140 150 160 170 180 90 100 110 120 130 140 150 160 170 180 Actual vs. budgeted balancing transfers to provinces (%) Actual vs. budgeted balancing transfers to provinces (%) Source: Staff estimates based on published State Budget data 26. At the same time, however, a lot of provincial and district authorities have indicated that maintaining balancing transfers constant in nominal terms over a five-year period was not practical. This does not take account of rising costs over the Stability Period, which was a particularly big issue in the current period when year-on-year inflation in August 2011 peaked at 23 percent. The loss in real terms is unlikely to be offset by increased revenue mobilization because transfer dependent provinces and districts cover most of their core expenditures out of these balancing transfers. Therefore in the next Stability Period, it is worth considering some form of indexation of balancing transfers to the rate of inflation and economic growth/development. 27. Institutional arrangements for balancing transfers from provinces to districts: Provinces implement a system of balancing transfers for districts that is similar to that of central transfers. Allocation norms are assigned to specific criteria across different functions to estimate spending needs in districts. These norms are also applicable for the Stability Period and recorded transparently in Provincial People’s Council resolutions. Balancing transfers to districts remain constant in nominal terms over the Stability Period thereby ensuring predictability. Though like provincial authorities, district authorities also noted that this created difficulties in ensuring that transfers keep in line with rising costs over the Stability Period. 28. Each province has its own set of norms following the SBL 2002 principle of providing provincial government with flexibility to determine fiscal transfers in accordance with specific conditions of districts and communes. The study looked at a sample of 12 provincial resolutions to review how provinces are applying allocation norms to estimate spending needs of districts. Out of the twelve provinces, four are net contributors and the rest are net recipients. Annex H provides the full list, together with some key economic statistics. 29. All provinces use almost the same spending categories as in Decision 59, though not all categories of spending are assigned norms at provincial level. This makes sense as in some provinces districts may not be involved in particular areas like science and technology or price subsidies. These services would be provided by provincial authorities. What is interesting, however, is that there are allocation norms for functions that would ordinarily not rest with district authorities. Social protection for example, has per capita allocations based on geographic locations. As noted in chapter 2, the responsibility for social protection functions would generally fall to central, or perhaps even provincial, authorities because of inter-jurisdictional objectives and economies of scale consideration. 30. Two other interesting examples are national defense and national security, both of which have per capita allocations that vary according to the geographic location of the district. National defense and national 133 security, however, as discussed in chapter 2 are functions that should be exclusively assigned to central government. Although district authorities are not obliged to spend on these functions, the allocations norms imply that they have a role to play in the provision of national defense and security services. Although the norms are higher in areas that are more vulnerable to insecurity, these are functions that are national in scope and therefore unusual for district authorities to provide. If the earlier recommendation to clearly specify the exclusive responsibilities of central government in the revised SBL 2002 is adopted, then these norms would no longer be relevant. 31. One challenge with the allocation norms both for central and provincial authorities is that they do not adjust for the level of service provision by different orders of government within the same jurisdiction. All functions in Vietnam according to the SBL 2002 are shared across all tiers of government. Therefore in cases where for example where provincial authorities are a major supplier of services within a particular district, allocation norms should be adjusted down compared to other districts where provincial engagement is less or non-existent. Otherwise districts that have substantial provincial engagement would benefit disproportionately from higher transfers, and those that have less provincial engagement will be penalized. However, as argued in chapter 2, services for a particular function within a jurisdiction should be provided exclusively by one order of government, even if higher orders of government play a role in regulation or provision of subsidies. This would create less distortion when applying the same norms across all provinces and districts. 32. The norms also do not adjust for different levels of private sector service provision within a jurisdiction. If a particular district (or province) has significant private sector service provision within its jurisdiction, the norm for that district (or province) should be adjusted down relative to districts where private provision is less or non-existent. Particularly as the presence of private sector service providers may reflect higher local ability and willingness to pay compared to poorer provinces. Private sector provision may of course ration citizens’ access due to affordability issues e.g. because private sector fees are not affordable for citizens in that jurisdiction. In those cases upper orders of government can try to redress this through targeted transfers or social protection schemes. Norms that do not take account of how private sector provision is impacting on access to services can lead to overestimation of spending needs in richer jurisdictions, diverting resources from poorer ones. 33. Several provinces and districts have indicated that norms do not compensate for negative externalities (e.g. congestion, pressure on local capacity) created by non-residents or temporary migrants using local services. It is not recommended to compensate lower tier jurisdictions that face these externalities through higher norms and spending estimates. This may unduly complicate the system, particularly as the externalities could be temporary. Lower tiers could instead be compensated through targeted transfers. In cases where the problem is more severe or permanent, the upper tier may decide to provide the services directly. This might not feasible in all cases. In the case of health sector, for example, Central Hospitals in HCM City or Hanoi may not be used to treat migrants as some of them are specialized ones. 34. Earlier assessments of provincial norms (Martinez-Vazquez, 2005) noted concerns over the use of physical assets to estimate spending needs. Provinces seem to have moved away from this in most sectors, which are now instead allocating on a per capita basis. In the health sector, however, most provinces reviewed regardless of levels of development continue to apply allocations norms per hospital bed. As noted earlier, this can create perverse incentives to overinvest in hospital capacity. The central government can provide guidance to discourage the use of physical criteria, in the same way as it has eliminated such criteria from the determination of balancing transfers to provinces. 35. Several provinces are using norms based on school going age population to estimate spending on education by district authorities. The norms are generally higher for rural and mountainous districts. The use of school age population as noted above is better than estimating on the basis of total population because the 134 latter would overestimate spending in districts that have large populations with a low share of people in the school going age category. It would also penalize districts with small population a large share of which is of school going age. 36. However, a number of provinces also have norms based on shares of recurrent spending going to salaries for teachers versus other recurrent spending. In most cases, the norms state that 80 percent of recurrent costs will go to salaries and 20 percent to non-salary recurrent spending. The education spending allocation norms are based on the school age population. When budget proposals are discussed with provincial authorities, if due to high number of teachers, the salary spending is more than 80% of education spending estimate, the center will provide additional allocation to ensure non-salary spending allocation is 20%. This is to protect non-salary spending. In the past this share was 30% so it seems that the salary spending had clouded out further non-salary spending. This surely will have implication for efficiency. 37. The budget share norms are applied mostly by transfer dependent provinces, but this practice poses two difficulties. The first is that 80 percent of recurrent spending going to salaries is high and leaves little for important things like school supplies and maintenance of school buildings. This norm may be trying to ensure adequate compensation to avoid teacher absenteeism issues. But the latter should be addressed through pay and grading, not by fixing budget allocation shares. Secondly, this practice may create an incentive for recruiting more teachers than needed. Of course the resources would be tied to financing the fixed of salaries, but districts would also benefit from higher absolute levels of non-salary recurrent estimates. 38. A similar issue arises for norms to estimate spending by districts on training and vocational training. Net contributing provinces have norms that are based on population, which is sensible. The norms are higher at provincial level compared to those set in Decision No. 59, signaling local authorities’ desire to play a bigger role. But four of the net recipient provinces reviewed use output based norms, namely the number of enrolled students. Martinez-Vazquez (2005) noted that this form of funding has not encouraged enough privatization of technical training in other countries and that it has led to mass training with low quality output. A number of local authorities met during the preparation of this report indicated that both provincial and district authorities were involved in vocational training, which led to overlaps and inefficiency. A switch to population based criteria would be more desirable. 39. Of the 12 provincial resolutions reviewed, 7 of them had allocation norms to estimate capital expenditure needs of district authorities. Capital projects assignment within the province is based on the size of the projects i.e. the capacity of the district to manage projects. Beyond this, in most cases, the norms are assigned against criteria that reflect the level of development in the districts including for example: poor household ratio, number of schools that are not up to national standards. The allocation norms are inversely related to the level of development leading to higher capital spending estimates for less developed districts. 40. There are some capital spending norms however that create a bias towards better off districts. For example, similar to provisions in Decision 60, capital spending estimates are positively linked to revenue mobilization and contribution to provincial budgets. This will be higher in the case of richer districts. Similarly, as noted above in the case of health spending norms, some provinces also use physical asset criteria (e.g. length of roads, the number of administrative units, number of cultural centers), which will also favor better off provinces. Of course the districts do not need to stick to the capital spending needs estimate derived from these norms – the estimates are used only notionally to determine balancing transfers. But the principles still matter, and it is therefore recommended not to use this for capital spending estimate. 41. Horizontal imbalance across provinces and central balancing transfers: Central balancing transfer policies for the 2011-2015 Stability Period, embedded in Decisions No. 59 and 60 and provincial resolutions, explicitly aim to reduce horizontal imbalances. As noted in chapter 1, central transfers have helped to increase per capita spending in poorer parts of the country particularly to help meet social service delivery needs. Another way to illustrate the redistributive impact of balancing transfer is by comparing it to revenue autonomy 135 at provincial level. Figure 5.11 shows revenue autonomy as measured by decentralized revenue as a share of local receipts across provinces (2006-2011 average). Figure 5.12 on the other hand shows dependence on balancing transfers to cover core spending across provinces. The asymmetry clearly highlights that poorer provinces that have less revenue autonomy are being compensated by balancing transfers to meet their spending needs. Figure 5.11: Decentralized revenue as % of local Figure 5.12: Balancing transfers as % of local receipts core spending 42. To assess the impact of central balancing transfers on horizontal imbalances, balancing transfers are added to local decentralized revenue and divided by local core expenditure. This is a rough estimate of horizontal imbalance because core spending is used as a proxy for spending needs, which as discussed earlier is difficult to accurately estimate. Figure 5.13 shows that balancing transfers have more than helped to fill horizontal imbalances for nearly all observations between 2006 and 2011. For over half the observations, decentralized revenue and core balancing transfers cover more than 150 percent of core spending needs. This is because core spending excludes spending from off-budget sources, carry over spending, contingency spending, and transfers to reserves. Figure 5.13: Coverage of provincial horizontal Figure 5.14: Annual coverage of horizontal 136 imbalance (06-11) imbalances (06-11) -1SD Median +1SD .01 .015 2006 2007 2008 2009 2010 2011 .008 .01 .004 .006 Density Density .005 .002 0 0 50 100 150 200 250 300 350 50 100 150 200 250 300 350 Local dencetralized revenue and central transfers/Local core spending Local dencetralized revenue and central transfers/Local core spending Source: Staff estimates based on published State Budget data 43. Horizontal imbalance within provinces and provincial balancing transfers: To assess the impact of central balancing transfers on horizontal imbalance across districts, provincial balancing transfers are added to district decentralized revenue and divided by district expenditure. In half of all observations, decentralized revenue and balancing transfers covered more than around 115 percent of the districts’ spending needs. Figure 5.16 shows that balancing transfers become more important over time in terms of covering district spending. However, it was not possible to exclude carry over or other non-core spending from district expenditure. Non-core spending is less at district level than at province level, particularly because districts do not borrow. But it still exists, which is in one third of all observations provincial balancing transfers do not cover all district spending. If non-core spending was excluded, all districts’ budgets would be balanced as required in the SBL 2002. Figure 5.15: Coverage of district horizontal Figure 5.16: Annual coverage of horizontal imbalance (06-11) imbalances (06-11) -1SD Median +1SD .015 2006 2007 2008 .015 2009 2010 2011 .01 .01 Density Density .005 .005 0 0 50 60 80 100 120 140 160 180 200 220 240 40 60 80 100 120 140 160 180 200 220 240 Dictrict dencetralized revenue and provincial transfers/Dictrict spending Dictrict dencetralized revenue and provincial transfers/Dictrict spending Source: Staff estimates based on published State Budget data 44. Dependence on transfers versus own revenues: The above trends at both provincial and district level suggest increased dependence on transfers for provinces as a whole and at district level. This is confirmed by 137 figures 5.17 and 5.18, which show that for half of all observations, central and provincial balancing transfers constituted more than 30 percent of local core and district spending respectively. What is interesting though is that the dependence on balancing transfers has also increased over subsequent Stability Periods for provinces as a whole and districts. This is illustrated by figures 5.19 and 5.20, which show that the median value for all observations on balancing transfer dependence has increased to around 45 percent compared to roughly 30 percent in 2006 for provinces as a whole and districts. This is consistent with the above discussion on the declining role of decentralized revenue to close vertical imbalance across and within provinces. At the same time however given the lack of revenue autonomy in Vietnam, in essence all revenues are transfers including 100 retained revenues and of course revenue sharing. Figure 5.17: Central balancing transfers/local core Figure 5.18: Annual central balancing transfers/ spending (06-11) core spending (06-11) -1SD Median +1SD -1SD Median +1SD .025 .02 .02 .015 .015 Density Density .01 .01 .005 .005 0 0 0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 Central balancing transfers/local core expenditure Balancing transfers/district expenditure Figure 5.19: Province balancing transfers/district Figure 5.20: Annual prov. Balancing transfers/ spending (06-11) district spending (06-11) 2006 2007 2008 2006 2007 2008 2009 2010 2011 .03 2009 2010 2011 .03 .02 .02 Density Density .01 .01 0 0 0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 Central bancing transfers/local core expenditure Bancing transfers/district expenditure Source: Staff estimates based on published State Budget data 45. The above shows signs of a growing misalignment between expenditure assignments and decentralized 138 (or “own”) revenue at local level. The fiscal decentralization literature argues that a close match between spending assignments and revenue autonomy can lead to higher allocative efficiency (Oates 1972). A recent study (Gemmel et al 2013) empirically tests this hypothesis for a group of OECD countries. The study finds that a closer link between spending and revenue decentralization can indeed enhance economic growth. The reasoning could be that revenue autonomy can help to increase accountability of local authorities. The conditions in these OECD countries are of course very different to Vietnam, including the fact that those countries do have true revenue autonomy. 46. In Vietnam, however, the increased misalignment seems to be more due to the rising spending needs of local authorities than the reduction of revenue raising powers of local authorities. The increased spending needs are related both to rising levels of access for previously underserved populations, but also because the demand for higher quality services increases as provinces become more developed. On the own revenue side, poorer local authorities are already able to retain all the shared taxes collected within their jurisdiction. In other words, notwithstanding the lack of autonomy over tax rates and bases (and thereby lack of any actual revenue assignment), most local authorities are already collecting as much as they potentially can for a given set of tax policies. It would be undesirable, for poorer authorities to increase the revenue burden (e.g. through higher fees) for local citizens in their jurisdiction. At the same time, local authorities also have little room to affect what is collected due to lack of autonomy over revenue policy. 47. Net contributing provinces are of course transferring an increasing share of their receipts to central authorities over subsequent Stability Periods, which have created concerns over the growing imbalance between own revenue and spending needs in those provinces. This is however primarily what has enabled redistribution to poorer provinces. Trading off richer provinces’ right to retain a higher share of shared revenue for growth enhancing spending to allow more redistribution for greater equity seems like the correct policy choice in Vietnam. As discussed in chapter 4, there are options to grant greater autonomy over revenue for some of the richer provinces. Chapter 6 also highlights that from a fiscal sustainability perspective, richer provinces have the potential to access more debt financing. The alternative of transferring less from richer provinces to the center thereby leading to potentially less redistribution could lead to increased disparity between growing and lagging regions, and even greater regional concentration of poverty. EQUITY ACROSS AND WITHIN PROVINCES 48. To assess whether the above trade off (i.e. between provinces’ right to retain more shared revenue for growth enhancing investments versus enabling more redistribution) is actually delivering on its stated objectives, this section looks at the impact of balancing transfers on the distribution of resources across and within provinces. As noted in Martinez-Vazquez (2005) delegating the responsibility for arranging local fiscal relations to an intermediate provincial authority carries a number of risks. In particular, the experience in a number of other countries shows that provinces often end up centralizing authority by delegating less revenue responsibility to lower jurisdictions but also by “capturing” central transfers for local authorities. They may at the same time be delegating more spending responsibilities, leading to a growing gap between spending responsibility and financing. 49. Most critically, this can undo any efforts by central authorities to ensure more equitable distribution of resources across the entire country. Central authorities’ equalization policies ultimately are concerned with the quality of and access to services delivered for citizens at all level of government. In other words, although Decisions 59 and 60 reviewed above are premised on these objectives, the outcomes can be undone if the provincial authorities’ policies towards its districts are less redistributive. Vietnam’s hierarchical structure of vertical government means that central authorities do not deal directly with district or commune authorities. Provincial authorities act as the conduit for implementing any central policies. But as discussed in this and 139 earlier chapters, central authorities do not dictate provincial choices, which instead are quite autonomous. It is therefore important for central government equalization policy to take into account any possible offsetting or reinforcing effects of these provincial choices. 50. The prior in this study is that in Vietnam, provincial policies are likely reinforcing central authorities’ redistributive objectives. This is based on findings in chapter 2 that show increased decentralization of spending responsibilities since 2006; findings in this chapter on the gradual increase in transfers to meet those responsibilities; and the review in this chapter of the overall transparency and institutional arrangements for determining provincial transfers to districts. This section therefore assesses the evidence on: (i) the overall trends in terms of per capita resourcing across different provinces; (ii) the impact that these trends have had on fiscal disparities across provinces; (iii) how provincial policies have impacted fiscal within provinces (i.e. across districts); and (iv) the relative impact of different revenue sources (i.e. decentralized revenue, balancing and targeted transfers) on disparities across and within provinces. 51. Overall trends in terms of per capita resourcing across provinces: Both “shared revenues” and “100 percent retained revenues,” especially the former category, have been elastic over time with steady annual increases (figure 5.21). This suggests as discussed earlier that all other things equal, central authorities have not reduced the revenue raising abilities of provincial authorities over subsequent Stability Periods. It has also not increased it from the perspective of autonomy over tax rates and bases. But the main point is to reaffirm that the growing gap between spending and revenue responsibilities is more due to policies on increased decentralized spending rather than reduced revenue responsibility. Figure 5. 21: Average nominal per capita budget revenues of provinces, Vietnam 2006-2011 1500.0 Retained revenue Shared revenue Balancing transfers Targeted transfer Borrowing Off budget revenue 1000.0 500.0 0.0 2006 2007 2008 2009 2010 2011 Source: Staff estimates based on published State Budget data 52. The trends in different sources of financing in 2006-2011 reaffirm the importance of balancing transfers for provincial authorities. There is a very noticeable jump in balancing transfers in 2011, when these transfers were adjusted according to PM Decisions 59 and 60. There is also a jump in balancing transfers in 2007, which was the start of the previous stability period. As expected per capita balancing transfers remain fairly stable with the stability periods. It is also noticeable that targeted transfers per capita significantly surpassed balancing transfers in importance starting in 2006 and it is only in 2011 that balancing transfers become slightly larger. This reflects the fact that targeted transfers are adjusted annually while balancing transfers are mostly adjusted only once during the entire stability period. Finally, borrowing proceeds have remained low and steady while off budget revenues, although also relatively small, have experienced a growing trend. 53. After adjusting for inflation, the trends in different sources of financing in 2006-2011 in real per capita 140 terms are much flatter (figure 5.22). Average shared revenue per capita increased steadily while average retained revenues fluctuated around VND 300,000 per capita over the entire period. For balancing transfers it is easier to observe the significant jump in resources at the start of the stability periods but also how those resources tend to decrease in real terms within each stability period. However, what is important is that in 2011 the average per capita balancing transfers were more than double the value in 2006. Targeted transfers in per capita real terms still increased steadily over the period but at a lower rate and they experience a slight drop in 2011. Finally, note that in real terms per capita borrowing has all but declined over the observation period. Figure 5.22: Average real per capita budget revenues of provinces, Vietnam 2006-2011 900.0 800.0 per capita thousands VND 700.0 Retained revenue 600.0 Shared revenue 500.0 Balancing transfers 400.0 Targeted transfer 300.0 Borrowing 200.0 Off budget revenue 100.0 0.0 2006 2007 2008 2009 2010 2011 Source: Estimates based on published State Budget data. Real values calculated using overall Vietnam CPI from IMF with 2005 as a base year 54. Adjusting for inflation also puts a different perspective of what has happened over the 2006 to 2011 period in terms of per capita expenditures at the provincial level. Average per capita nominal and real expenditures are shown in Figure 5.23. While per capita nominal expenditures increased by 133 percent, in real terms the increase was only a more modest—although not insignificant—of 33 percent. Figure 5.23: Average per capita nominal and real expenditures of provinces, Vietnam 2006-2011 4500.0 4000.0 per capita thousands VND Provincial 3500.0 expenditure 3000.0 (nominal) Provincial 2500.0 expenditure (real) 2000.0 1500.0 1000.0 2006 2007 2008 2009 2010 2011 Source: Estimates based on published State Budget data. Real values calculated using overall Vietnam CPI from IMF with 2005 as a base year 55. As noted earlier, the different financing sources at provincial level more than cover provincial spending requirements. Figure 5.24 presents cumulative per capita revenues and average per capita expenditure. 141 Adding both balancing and targeted transfers to shared and “own” revenues pretty much covers the per capita expenditure each year, while adding borrowing and off-budget revenues per capita the cumulative values slightly exceed the average per capita expenditures for all years in the observation period 2006-2011. The surplus is due primarily to carry over revenues. Figure 5.24: Average per capita cumulative budget revenues and average per capita expenditure of provinces, Vietnam 2006-2011 5000.00 4500.00 Retained revenue per capita thousands VND 4000.00 + Shared revenue 3500.00 ++ Balancing transfers 3000.00 2500.00 +++Targeted transfer 2000.00 ++++ Borrowing 1500.00 +++++ Off budget revenue 1000.00 500.00 Provincial expenditure 0.00 2006 2007 2008 2009 2010 2011 Source: Estimates based on published State Budget data. 56. Central policies and fiscal disparities across provinces: Analysis of available data for 2006- 2011 confirms the hypothesis that the system of intergovernmental finance at the central-provincial level in Vietnam is equalizing. This is measured by looking at four budget measures all in per capita terms: (i) decentralized revenue, where we would expect large disparity across provinces; (ii) decentralized revenue plus balancing transfers, which should significantly reduce disparities across provinces; (iii) decentralized revenue and balancing transfers plus targeted transfers, which should further reduce disparities; and (iv) per capita expenditure by districts, which show the disparities that ultimately matter since they are the closest reflection of the public services actually received by residents. 57. The degree to which transfers are helping to equalize levels of resourcing across provinces is measured by what happens to the coefficient of variation for per capita resourcing as transfers (and other resources) are added to decentralized revenues. The coefficient of variation helps to measure how widely values for per capita resourcing across all provinces are distributed relative to the mean value of the whole distribution. In other words, the bigger the coefficient of variation, the more widely dispersed the population is around the mean value, and the greater inequality there is in per capita resourcing across provinces. Therefore the expectation is that the coefficient of variation reduces as transfers are added to decentralized revenue across the different provinces. 58. Figure 5.25 presents the coefficients of variation of provincial per capita cumulative budget revenues and per capita expenditure for the period 2006-2011 (see annex I for details). The coefficients of variation of per capita retained tax revenues and shared tax revenues are quite high. This variation sharply drops after accounting for the balancing and targeted transfers. Except for 2011, targeted transfers contribute to a more equal distribution of per capita resources. Note that by accounting for transfers the coefficient of variation decreases to the level of coefficient of variation for expenditures. This highlights the importance of transfers for fiscal equalization at the provincial level in Vietnam. Figure 5.25: Coefficient of variation of per capita cumulative budget revenues and per capita 142 expenditure of provinces, Vietnam 2006-2011 2.00 Retained revenue + Shared revenue 1.50 ++ Balancing transfers +++Targeted transfer 1.00 ++++ Borrowing +++++ Off budget revenue 0.50 Provincial expenditure 0.00 2006 2007 2008 2009 2010 2011 Source: estimates based on published state budget data. 59. Finally, accounting for borrowing slightly increases the coefficient of variation, which indicates that relatively richer provinces have more capacity and ability to borrow. In all, the coefficient of variation for expenditures per capita oscillates just above 50 percent. This indicates that despite the strong equalization process exerted by transfers, disparities in spending per capita at the provincial level remain relatively high. Part of this is due to higher per capita spending by poorer provinces as discussed in chapter 2. 60. Another way to look at disparities is to calculate the ratio between the maximum and minimum values per capita in resourcing. This measure is a different way of emphasizing the difference in resourcing across provinces. For example, the high ratio for retained revenue shows the big difference between the maximum value of per capita retained revenue, which would most likely be collected in the richest province, and the minimum value of per capita retained revenue, which would most likely be collected in the poorest province. The ratio would of course be expected to fall as shared revenue, transfers, off budget revenues, and borrowing are added to retained revenues. 61. The ratio of retained revenue and borrowing reveal some interesting trends. The ratio for retained revenue is very high and increasing up to 2008, but it sharply drops in 2009, most likely due to the effects of the world financial crisis. The borrowing ratio on the other hand sharply jumps up in 2009 and especially in 2010, seemingly indicating that richer provinces that were hit hard by the crisis turned increasingly to borrowing particularly to meet infrastructure spending needs as part of their fiscal stimulus package. But most importantly, in figure 5.26 we can see that after accounting for transfers, the ratio for per capita cumulative resourcing sharply drops. This emphasizes again the equalization role of fiscal transfers in Vietnam’s intergovernmental finance system. 140 Figure 5.26: Ratio of maximum to minimum of per capita cumulative budget revenues and per capita expenditure of provinces, Vietnam 2006-2011 143 400.00 Retained revenue 350.00 + Shared revenue 300.00 ++ Balancing transfers 250.00 +++Targeted transfer 200.00 ++++ Borrowing 150.00 +++++ Off budget revenue 100.00 50.00 Provincial expenditure 0.00 2006 2007 2008 2009 2010 2011 Source: estimates based on published state budget data. 62. Provincial policies and fiscal disparities within provinces: The analysis of disparity in resourcing across districts uses the same methodology as above. In particular, it looks at what happens to the coefficient of variation on per capita resourcing as balancing transfers and targeted transfer are added to decentralized revenue in each district in 2011 where data were available. The study used published district level fiscal information on decentralized revenue, provincial transfers, and district level expenditure in this analysis. As can be seen from annex J adding balancing and targeted transfers naturally increases the overall mean and median resourcing per capita at the district level. These measures exceed the overall mean and median of expenditures in all years, pointing to potential revenue carry over, discussed in chapter 3. 63. In terms of disparities in per capita resourcing, as expected the coefficient of variation for all years decreases quite significantly when moving from decentralized revenue per capita to decentralized revenue plus balancing transfer (figure 5.27). So the balancing transfers implemented by the provincial governments toward their districts are indeed equalizing, as they are supposed to be. After adding the targeted transfers in all cases the coefficient of variation decreases in all years, although less so than when just adding the balancing transfer. This means that the targeted transfers as expected have an overall equalizing effect. Nonetheless it is important to note that although provincial transfers have equalized resourcing across districts, an average coefficient of variation of nearly 80 percent is not insignificant. Figure 5. 27: Coefficient of variation of per capita cumulative budget revenues and per capita expenditure of districts, Vietnam 2006-2011 1.4 1.2 1 "Own" revenue 0.8 + Balancing Transfer 0.6 ++ Targeted Transfer 0.4 District Spending 0.2 0 2006 2007 2008 2009 2010 2011 Source: estimates based on published state budget data. 64. The disparities in expenditure per capita are less than the disparities in total resourcing per capita at 144 district level, but there still appear to be significant disparities in per capita spending across districts. In reality, there is a lot of dispersion in the coefficients of variation for different provinces. There are many with very low coefficients of variation for expenditures per capita (0.1-0.2), but there are others significantly above the mean. Four provinces (Dong Nai, Kien Giang, Kon Tum and Soc Trang) for which there is complete data in 2011 have very high coefficients of variation with values close to 3.0. These provincial governments may value more heavily other economic objectives, such as economic development. It would be useful to look more closely at intergovernmental finances in those provinces. Figure 5.28: Coefficients of variation of per capita resourcing and expenditures across districts 3 2.5 2 CV of per capita Decentralized Revenue and Transfers CV of per capita district spending 1.5 1 0.5 0 Nam Dinh Long An Dong Thap Quang Ngai Kien Giang Soc Trang Dong Nai Ba Ria - Vung Ta Hoa Binh Binh Duong Dak Lak Lai Chau Phu Tho Quang Tri Bac Giang Bac Kan Ninh Thuan Ha Giang Ha Tinh Kon Tum Gia Lai Hau Giang Lang Son Ninh Binh Dien Bien Quang Ninh Binh Phuoc Ha Noi Da Nang Quang Nam Can Tho Quang Binh Ca Mau Nghe An Ho Chi Minh Hung Yen Khanh Hoa Tay Ninh Vinh Phuc Lam Dong Son La Ben Tre Binh Dinh Bac Ninh Dak Nong Lao Cai Yen Bai Ha Nam Phu Yen Thai Nguyen Cao Bang Hai Phong Vinh Long Tuyen Quang Tra Vinh Source: estimates based on published state budget data. 65. The disparity in per capita spending across districts could of course simply reflect the diversity of conditions in those districts. As discussed in chapter 2, the variation in local per capita spending across provinces seems to be linked inversely to local per capita GDP and population density within the province. Figures 5.29 and 5.30 below suggest that there may be a similar trend across districts also as dispersion in per capita spending is negatively correlated to both per capita GDP and population density – but the relationship is weak. It is likely to be stronger if per capita income were correlated against population density for each district. This result is also consistent with the earlier analysis of allocation norms for estimating needs, as these are positively linked to population and inversely linked to levels of development. Figure 5.29: District per capita spending and per Figure 5.30: District per capita spending and capita GDP population density 60000 1400 1200 Population density (persons/km2) 50000 Per capita GDP (VND m) R² = 0.0375 1000 40000 R² = 0.0552 800 30000 600 20000 400 10000 200 0 0 0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 CV for per capita spending across districts CV for per capita spending across districts Source: Staff estimates based on published State Budget data 66. One important qualification in the above analysis of per capita spending across and within provinces is that these do not include spending by upper tiers within an entire province or within a district. In other 145 words, per capita spending in a province includes spending by local authorities but not spending by central government in that province. Similarly, per capita spending in a district includes spending district and commune authorities but not by provincial authorities in that district. As discussed in chapter 2, the current published budget reports do not give a full picture of spending within a jurisdiction. If upper tier spending were included, the above results are likely to be reaffirmed because upper tiers of government are likely to spend mostly in poorer provinces and districts. 67. In summary, it is possible to conclude that overall the intergovernmental finance systems at the central- provincial and provincial-local levels work relatively effectively to reduce fiscal disparities across jurisdictions. The balancing transfers implemented both at the central and provincial levels are especially effective in reducing the fiscal disparities observed in terms of decentralized revenues. Overall, targeted transfers are less effective in terms of equalization but that is also to be expected because these transfers have other main objectives. The final distributions of expenditures per capita both across and within provinces are fairly equalized, although there are remaining disparities. This may reflect the diversity of conditions across these jurisdictions, including higher cost of service provision in poor and sparsely populated areas. This could be looked at further by looking more deeply at selected provinces. TARGETED TRANSFER PROGRAMS 68. Overview of target programs: This final section looks at the system of central targeted transfers to local authorities, with special focus on National Target Programs (NTPs). Targeted transfers are conditional as resources are provided to local authorities based on provincial characteristics and to achieve socio-economic development targets set by the central government. The targets are set by central government, but local authorities are responsible for implementation. Target programs are therefore reflected in local budgets. There are two types of target programs. The first constitute a set of 16 programs grouped under NTPs, which are reviewed in more detail below. The second are conditional transfers aimed at specific provinces or types of provinces. 69. Decision 60 lists 28 target transfers under this second type (see annex K), which are in addition to the 16 NTPs. The 28 target transfers cover a wide range of objectives including for example regional development through capital investments; investment in the 62 poorest districts; specific infrastructure investment (e.g. dykes, research and development centers); economic development programs in East Sea islands and border areas; and investments in provincial and district hospitals. 70. Although the relative share of targeted transfers in local spending has fallen over time, it still constituted on average 25 percent of local spending between 2006 and 2011. Figure 5.31 shows that for half of all observations between 2006 and 2011, target transfers constituted more than 30 percent of local core spending. Figure 5.32 however also shows that the relative importance of target transfers has fallen slightly across two Stability Periods, which could be because some of the target programs became absorbed into regular government spending. The distribution becomes more concentrated over time with target transfers constituting around 20 percent of local spending. A similar, though even more pronounced trend is visible in the case of provincial transfers to districts. Target transfers over district spending was more widely distributed in 2006, and over the 2007-2010 Stability Period before converging around 20 percent in 2011. Figure 5.31: Central target transfers/local core Figure 5.32: Annual central target transfers/local 146 spending (06-11) core spending (06-11) -1SD Median +1SD .025 2006 2007 2008 .025 2009 2010 2011 .02 .02 .015 .01 .015 Density Density .01 .005 .005 0 0 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 Targeted transfers/local core expenditure Targeted transfers/local core expenditure Figure 5.33: Province target transfers/district Figure 5.34: Annual province target transfers/ spending (06-11) district spending (06-11) -1SD Median +1SD .025 2006 2007 2008 .04 2009 2010 2011 .02 .03 .015 Density Density .02 .01 .01 .005 0 0 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 District targeted transfers/district expenditure District targeted transfers/district expenditure Source: Staff estimates based on published State Budget data 71. The above trends suggest less dependence of local authorities on ‘non-discretionary’ resources, but the large number of overlapping target programs create high transaction costs. The number of non-NTP target transfer schemes has fallen from 50 in the 2007-2011 Stability Period to 28 in the 2011 Stability Period. But adding the 16 NTPs brings the total number of target transfer schemes to 44. This does not include the provincial target transfer schemes for districts. From the list of central target programs, it is clear that there is scope to simplify and consolidated some of these schemes. This is discussed further below on the NTPs. 72. A second issue that impacts on the effectiveness of target transfers is the lack of predictability. This is evident from figure 5.35, which shows that for half of all observations between 2006 and 2011, actual targeted transfers exceeded what was budgeted by more than 175 percent. In other words, actual target transfers were 1.75 times more than what was initially budgeted for half of all observation. As discussed in chapter 3, this lack of predictability can impact negatively on the quality of fiscal management, especially if these transfers make up nearly a quarter of local receipts. Provincial target transfers to districts are even more unpredictable. Figure 5.36 shows that for half of all observations between 2006 and 2011, target transfers were more than 4 times what was initially budgeted. Figure 5.35: Actual vs. Budget central target Figure 5.36: Actual vs. Budget central target transfers (06-11) transfers (06-11) 147 -1SD Median +1SD -1SD Median +1SD 14 8 10 12 14 16 18 12 10 Frequency 8 6 6 4 4 2 2 0 0 100 150 200 150 300 350 200 400 600 800 1000 1200 1400 1600 1800 Actual vs budgeted targeted transfers from centre to province (%) Actual vs budgeted targeted transfers from province to district (%) Source: Staff estimates based on published State Budget data 73. Local authorities have indicated that the lack of predictability in target transfers can put significant pressures on their budgets and lead to proliferation of unfunded mandates. These are spending responsibilities set under target transfer schemes, or under new national policies promulgated by the central government, but without (full) funding for implementation by local authorities. Whilst the SBL 2002 explicitly states that there should be no unfunded mandates, in practice national programs and policies are not always accompanied by adequate or timely financing. This makes it difficult to implement those programs or policies at the local level. Local authorities indicated that they are encouraged to initiate implementation with own finances and seek reimbursement later. But these reimbursements often come either late in the fiscal year or early part of the next fiscal year, creating pressures on local budgets. 74. Overall objectives of and institutional arrangements for NTPs: NTPs aim to accelerate progress against urgent objectives of national significance. Prime Minister’s Decision 135 (November 2009) sets out the guidance for developing and managing NTPs. For the period 2011-2015, central line ministries prepared 16 NTPs as noted above. These were endorsed by the National Assembly and approved by the Prime Minister (PM) through Decision 2406 (December 2011). Each NTP has a PM Decision setting out its objectives, targets, beneficiaries, timeframe, components, funding arrangements, and roles and responsibilities. Each PM Decision has detailed circulars issued by line ministries on implementation arrangements (annex L). 75. The 16 NTPs cover a wide range of objectives targeting poverty, education, health, livelihoods, rural development, culture, energy use and climate change. Each NTP has several sub-components, which total close to 70 across all NTPs. Central line ministries, referred to as “NTP Authorities” are responsible for budget allocations to and oversight of NTPs. Sub-components may be overseen by other line ministries, referred to as “Management Agencies”, in cases where an NTP covers more than one sector. Individual activities under the NTPs are proposed by sub-national (and in some limited cases central) authorities who are ultimately responsible for implementation. They are referred to as the NTP “Delivery Units.” The Ministry of Finance (MOF) and the Ministry of Planning and Investment (MPI) have the overall responsibility for financing decisions and monitoring across all NTPs. 76. Priorities are discussed between Delivery Units and NTP Authorities, but the latter have little control 148 over the detailed allocation to and implementation of activities. Delivery Units prepare proposals, which are discussed with NTP Authorities, who in turn propose to MOF and MPI the overall financial envelopes for their NTPs, including allocations for individual provinces. Budget allocations are then made to Delivery Units as part of the local budget process. The Delivery Units, which can be sub-national line departments, Departments of Planning and Investment, or departments within the Peoples’ Committee, are responsible for implementing specific activities that will help them meet the targets set out in the NTP. NTPs are therefore not projects under the control of central line ministries/NTP Authorities, which they implement across the country. Delivery Units report on progress to NTP Authorities, who compile results and report to MPI. 77. Guidelines on implementation of NTPs are set out in the individual guiding circulars issued by the NTP Authorities. The circulars cover a mix of issues including but not limited to eligible expenditures, procurement regulations and cost norms. Provincial and district authorities generally go through a bottom-up planning process to identify specific activities under each NTP. The details on this are not prescribed in the circulars. Activities are agreed with the relevant People’s Committees and sub-national line Departments, before they are submitted to the Departments of Finance and Planning and Investment at the provincial level. Detailed budget appropriations are by the People’s Council. 78. The allocation of NTP resources is based on a set of eligibility criteria set out in the relevant PM Decisions and accompanying circulars. The criteria constitute mainly socio-economic indicators. When Delivery Units apply for NTP funding from Management Agencies and NTP Authorities, they need to specify in their proposal how the conditions in their jurisdiction meet the NTP eligibility criteria. Funding is conditional on this and not on outputs or outcomes achieved with the NTP funding, which as discussed below is an area that warrants review. Figure 5.37 shows that the least developed regions, namely the Central Highlands, the Northern Mountains and the Northern Central and Central Coast have benefited most from NTPs in the 2006-2011 period. This helps to explain some of the redistributive benefits of targeted transfers discussed in the earlier section. Figure 5.37: Per capita spending on NTPs by region (2006-2011) 35 2006 30 2007 25 2008 2009 20 2010 15 2011 10 5 0 Central Highlands Mekong River Delta Northern Central and Northern Mountain Red River Delta South East Central Coast Source: estimates based on published state budget data. 79. Funding arrangements for NTPs and impact on NTP management: Nearly 75 percent of spending on NTPs in 2012-2013 went on four programs (table 5.2):49 Job Creation and Vocational Training (12 percent); 149 Sustainable Poverty Reduction (15 percent); Education and Training (17 percent); and Development of a New Rural Life (29 percent). Around 56 percent of funding for all NTPs over the period 2012-2013 came from the central authorities; 26 percent from local authorities; 5 percent from external donors; 4 percent from borrowing; and 9 percent from community contributions (table 5.2). 80. The different funding sources for NTPs are subject to different financial management procedures. The NTP circulars and guidelines mentioned above only apply to NTPs implemented using State Budget funds. The execution of these resources in terms of fund allotments, payment procedures, accounting and reporting are subject to the same treasury rules and procedures as the rest of the government budget. Rules and procedures are however different in the case of donor and community-funded resources for the same NTP. Although non state funding constitutes a relatively small share of NTP financing, they can impose disproportionate management costs. Table 5.2: NTP spending 2012-2013 (VND million) No. Name of program Total (VND m) NTP as % Central Local Others of total budget budget 1 Job creation and vocational training 10,432 12% 87% 10% 3% 2 Sustainable poverty reduction 12,948 15% 81% 15% 3% 3 Water supply and rural environment 7,850 9% 29% 11% 61% 4 Health 4,423 5% 79% 4% 17% 5 Population and family planning 2,879 3% 90% 5% 5% 6 Food hygiene and safety 976 1% 84% 13% 3% 7 Program in culture 2,825 3% 63% 28% 9% 8 Education and training 14,929 17% 75% 16% 9% 10 Drug abuse prevention and control 899 1% 91% 7% 1% 11 Efficient and economic use of 292 0% 52% 15% 33% energy 12 Response to the climate change 846 1% 47% 5% 48% 13 Development of a new rural life 25,775 29% 19% 64% 17% 14 Prevention and control of hiv/aids 3,023 3% 35% 3% 62% 15 Information flows to remote areas 478 1% 69% 29% 2% 16 Pollution remediation and envt. 299 0% 92% 8% 0% Total 88,874 100% 56% 28% 17% Sources: GOV report to National Assembly on implementation of NTPs (2011-2013) 81. A number of local authorities indicated that the level of funding was not adequately linked to the overall targets set under the NTPs. In a number of cases, local authorities felt that the targets were too ambitious. They were developed with little consideration for the cost of meeting the targets and whether this would be fiscally sustainable. Though on the other hand, funding is also not conditional on outputs under the program. In addition, capital projects developed under NTPs lack operations and maintenance resources. This is impacting on the sustainability of the NTP targets and objectives. 49 GOV report to National Assembly: “Mid-term Review of the 3-Year (2011 – 2013) Implementation of National Target Programs,” (P. 2-3), October 2013 82. Moving to output-based conditional grants: This last point, and earlier discussion in paragraph 80, 150 suggests that the basis for allocating NTP funding warrants review. At the moment, NTPs use a mix of input and output-based indicators to set targets and objectives. A summary of those indicators is provided in annex M. Allocation of NTP funding is partly based on those input targets (i.e. allocation norms) and partly based on socio-economic eligibility criteria. 83. Socio-economic eligibility criteria have contributed to the equalizing effects of target transfers, and input-based allocation norms help to control spending – but these criteria pose several difficulties. Firstly as noted above there is a misalignment between targets and resourcing. Secondly, the system does not necessarily allow adjustments to allocations based on actual progress against targets i.e. higher allocations where more progress is being achieved and vice versa. Thirdly, although local authorities do report on overall progress against NTP objectives, these are high level making it difficult to assess attribution. These are high priority areas that would need to be addressed in any redesign of the NTP arrangements. 84. One option to address some of the above issues is for Vietnam to move to a system of output- based conditional grants. The socio-economic eligibility criteria would still be important to ensure that better performing, high capacity provinces do not capture all the transfers. But there is also merit in linking transfers to results while providing full flexibility in the design of the program and associated spending levels to achieve those objectives. Table 5.3 below (Shah 2007) summarizes some of the expected benefits from an output-based conditional transfer system versus the more traditional input- based arrangement. 85. A number of countries have had good experience with combining socio-economic eligibility criteria with performance conditionality in the context of target transfers programs (Shah 2007). In Uganda and Kenya, in poverty targeted program, allocations made on the basis of socio-economic indicators were adjusted on the basis of performance. Some of the poorer provinces actually achieved better results than the more wealthy ones. Tanzania followed a similar approach, where minimum access conditions were set using both quantitative and qualitative criteria. These were used to determine size of conditional grants, but the latter was adjusted based on ex post assessment of progress against objectives. Ethiopia also successfully piloted a capacity building project for local service delivery. The program was managed at the federal level but resources were allocated based on socio-economic criteria across regions. Program disbursements were conditional upon performance of the regions. Annual reports were able to show clearer attribution between program funds and performance. Table 5.3: - Features of traditional and output-based conditional grants 151 Feature Traditional grant Output-based grant Grant objectives Spending levels Quality and access to public services Grant design, Complex Simple and transparent administration Recipient government departments/ Recipient government provides funds to all Eligibility agencies government and nongovernment providers Expenditures on authorized functions and Conditions Outputs-service delivery results objects Program or project proposals approvals Allocation criteria Demographic data on potential clients with expenditure details Client feedback and redress, comparison of Compliance Higher level inspections and audits baseline and post grant data on quality and verification access Public censure, competitive pressures, voice Penalties Audit observations on financial compliance and exit options for clients Little or none. No tolerance for risk and no Absolute. Rewards for risks but penalties for Managerial flexibility accountability for failure persistent failures Autonomy and Little Absolute flexibility Transparency Little Absolute Focus External, competition, innovation and Internal benchmarking Hierarchical to higher-level government, Accountability controls on inputs and process with little or Results-based, bottom-up, client-driven no concern for result Source: Shah, 2007 86. However, moving to more output or performance-based transfers has to be a gradual process. It is not advisable to move the whole system to performance-based one. Performance-based systems have their own challenges in terms of incentives and the quality of reporting, which impact on their effectiveness. The government may therefore consider a sequenced approach in rolling out a performance-based transfer system. The new system could for example target first those provinces that have higher capacity, with poorer provinces still receiving transfers on the basis of socio-economic characteristics or inputs. As the new system gets tested and piloted, it could be gradually expanded and tailored (e.g. in terms of performance measures) to poorer provinces. In the meantime, for the latter, it is important to generate as much information as possible on what existing targeted transfers are delivering in terms of outputs and development outcomes. This will provide important inputs in the design of grants that are tailored to the specific challenges different categories of provinces. There are example from other countries as discussed above, including from Indonesia, which has established a number of conditional transfer schemes for local development. 87. Coordination across NTPs to strengthen local planning and budgeting: Moving to output- based conditional grants could be combined with some consolidation of NTPs to help strengthen local planning and budgeting. At the moment it is difficult for local authorities to coordinate across so many target programs. Although there are national, and in some cases sub-national, steering committees that 152 have been established to help coordinate NTP planning and budgeting, in reality these have not been very effective. This has resulted in fragmentation and weak monitoring and evaluation. This is the case even within provinces, where despite participatory approaches to planning for selected NTPs and the use of local Socio-Economic Development Plans to guide resource allocation, the end result is a set of un-strategic projects. 88. The current NTP regulations encourage sub-national authorities to integrate NTPs and other projects. This was one of the expected benefits of decentralizing the responsibility for managing and implementing NTPs. But in practice, it has not been very successful. The government has reported that a number of provinces including Kon Tum, Ha Nam, Ha Tinh, Thanh Hoa, Da Nang, and Dien Bien, have issued Decisions to integrate NTPs other local programs and projects in the same jurisdictions. 89. Once provinces have been assigned ceilings for each NTP, province and district authorities have discretion to make detailed activity level allocations. NTPs are budget line items in province or district level implementing authorities. The latter have discretion on how they plan and budget for NTP activities. This offers them a real opportunity to integrate planning and budgeting across Target Program. It is almost implicit that there will be community involvement in preparing NTP proposals and budget submissions. But the very detailed implementation guidance and the specificity of targets in reality leads to a silo approach to planning, budgeting, execution and reporting on Target Programs. 90. The study looked at four NTPs with similar objectives to explore opportunities and constraints for integration of planning and budgeting: Job Creation and Vocational Training; Sustainable Poverty Reduction; New Rural Life; and Program 135. The objectives across all these Programs broadly target income poverty through job creation and capital expenditure, but there are also important differences. The NTP on Job Creation and Vocational Training is divided across six sub-projects that aim to improve labor market mobility through for example training, manpower exports, and direct job creation. The NTP on Sustainable Poverty Reduction is focused mostly on capital spending for rural infrastructure and business capacity. The NTP on New Rural Life also targets infrastructure, but covers a very wide range of other areas including education, health, water and sanitation and other social services. Finally, Program 135 is focused on infrastructure spending but in the poorest communes in the country. 91. The differences in objectives should not be an impediment to integrating planning and budgeting but the detailed implementation guidelines and targets make coordination difficult. For example, each of the four NTPs above has very specific program beneficiaries, sometimes focused in specific geographic areas. They also have different targets for funding sources, in some cases restrictions on spending on certain items from specific sources, different arrangements for ensuring oversight of implementation of investment projects, and different line agencies responsible for implementing (or simply providing guidance on) different elements of the NTP. This means that planning and budgeting for one NTP is already very challenging let alone coordinating the planning and budgeting across different NTPs. 92. In response to this issue, local governments are required to establish Steering Committees for coordination of NTPs. But in reality because the actual implementation of an NTP can be spread across a range of different administrative units (i.e. all of whom receive a budget allocation for that NTP), the committees have had low to mixed success in coordinating the planning and budgeting for NTPs. The government review of NTPs found that a lot of the Steering Committees operate formally, but with little authority to make strong decisions on resource allocation. The approval authorities for allocations to NTPs work in the same way as that for other parts of the budget, with some oversight from the NTP Authorities. 153 93. Another issue is the difficulty that the current model must pose in terms of compiling reports on NTP implementation. Reporting on implementation is critical to inform planning and budgeting for the next fiscal year. However, if the implementation of a given NTP is spread across so many administrative units, each having their own budget allocation for that NTP, it will be very difficult to consolidate reports on implementation of any one NTP. 94. Sub-national authorities have the flexibility to integrate planning and budgeting for Target Programs. NTP Decisions and Circulars do not prescribe how each Program needs to be planned or budgeted for. In theory therefore, there is scope to integrate planning and budgeting across programs on the basis of consolidated local development plans. This was in fact the intention of decentralizing the management and implementation of NTPs to provinces and establishing Steering Committees to oversee coordination. 95. Way forward on NTPs: The coming period offers a real opportunity to revisit the institutional arrangements for NTPs to inform target transfers programs for the 2016-2020 Stability Period. There are several ways that the government can help to strengthen NTPs so that they can act as an effective instrument for delivering on central objectives. Firstly, it is important to ensure greater predictability in targeted transfers. Secondly, the government should explore with Development Partners the options for more alignment and harmonization of different funding sources through greater use of country systems. Thirdly, the government should look at using both socio-economic criteria, but also output-based performance, to determine allocation of NTP resources, even if on a pilot basis to start with. Finally, a consolidation of NTPs, with greater focus on a limited set of output and outcome indicators, rather than complex input-based indicators, could help to strengthen planning and budgeting of NTPs within a jurisdiction. Underpinning all this would be a more simplified architectural design for these transfers, similar to a more conventional conditional grant scheme. Like in most other countries one single agency would have the role for setting eligibility conditions and performance-based outcomes/outputs and be responsible for monitoring and reporting. CONCLUSIONS AND RECOMMENDATIONS 96. Revenue sharing arrangements and the system of balancing transfers in Vietnam have played a significant role in narrowing vertical and horizontal fiscal imbalances respectively. There are however growing vertical imbalances due to local authorities’ increased spending responsibilities. It is argued here that trading off richer provinces’ right to retain a higher share of shared revenue for growth enhancing spending to allow more redistribution for greater equity seems like the correct policy choice in Vietnam. Increased spending responsibility of both provincial and district authorities are being covered by increased levels of balancing transfers. The latter at the center to provincial level are underpinned by a transparent, norms-based system. At province to district level, the publication of Stability Period rules and regulations have also helped to improve transparency. 97. The system of fiscal transfers has helped to promote greater equity in resource distribution both across and within provinces. The evidence shows that balancing transfers are helping to equalize levels of resourcing across provinces and across districts. There is evidence a fairly wide variation in per capita spending across districts. This may reflect the diversity in conditions of different districts, though it is difficult to say for sure what explains this remaining disparity. 98. Target transfers continue to play an important role in local spending even though its relative share 154 has fallen over time. The lack of predictability in targeted transfers poses serious challenges for both provincial and district authorities. It impedes on planning, budgeting, and delivery of NTP targets. The level of resourcing however is not closely aligned with the targets and objectives, which are quite ambitious. Complex, input-based guidance on NTP implementation reduces flexibility and increases burden of reporting. There is a real opportunity to review the institutional arrangements for NTPs for the coming Stability Period aiming to simplify the system and increasing transparency, monitoring and accountability. Issues Recommendations Law 2002 Allocation norms to estimate spending needs Amend Articles 31 and 33 in the SBL 2002 to Budget State and determine balancing transfers include clarify the exclusive responsibilities for central areas that should be exclusively the central government, thereby eliminating the need for government’s responsibility. norms in these areas. Maintaining balancing transfers for provinces Index balancing transfers to the rate of inflation in and districts constant over the Stability Period the formula for the next Stability Period. does not take account of rising costs. Allocation norms for central and provincial Services for a particular function within a authorities do not adjust for the level of service jurisdiction should be provided exclusively by provision by different orders of government one order of government, even if higher orders of within the same jurisdiction, which could government play a role in regulation or provision of penalize areas where upper tiers are not subsidies Stability Period regulations 2002 providing any services. Adjust norms down for provinces with significant Allocation norms do not adjust for different private sector provision. In case this leads to levels of private sector service provision within lower access, upper orders of government should a jurisdiction address through targeted transfers or social protection schemes. Allocation norms do not compensate for Compensated affected provinces and districts negative externalities (e.g. congestion, pressure through upper order transfers or direct provision of on local capacity) created by non-residents or services, not through higher allocation norms. temporary migrants using local services Use of allocation norms based on physical Ministry of Finance can issue guidance on the assets or the share of recurrent spending leads appropriate use of allocation norms to determine to distorted incentives spending estimates. Lack of predictability in targeted transfers Institute more discipline and predictability in target puts pressure on local budgets and creates transfers from center to province and province to unfunded mandates. districts. Different funding sources for NTPs add Explore greater use of country systems by transaction costs due to differing financial Development Partners supporting NTPs. management and procurement procedures. NTP targets are not adequately linked to level of resourcing, the system does not link Move to a system of output-based conditional National Assembly resources to performance, and high level grants, complemented by the existing socio- regulations targets make it difficult to meaningfully monitor economic criteria used to determine eligibility. NTP performance. Coordination across NTPs is difficult due to Review options to consolidate NTPs and other fragmented management and complex, input- target programs and focus on fewer output and based guidelines. outcome indicators. 155 LOCAL BORROWING 156 06 Key issues: the system of intergovernmental fiscal relations in Vietnam, as illustrated in the previous two chapters, is geared to redistributing locally collected revenues, the bulk of which are contributed by a handful of provinces that point to a growing infrastructure financing deficit. Local borrowing has accordingly emerged as an important topic in Vietnam particularly for those provinces that are not able to satisfy their capital spending needs through the existing local revenue and transfer arrangements Objectives: to assess whether selected provinces could potentially borrow more to address this deficit, and highlight reforms to the existing institutional framework to enable this. Key findings: borrowing by local authorities over the 2006-2011 period has remained very low (less than 3 percent of GDP). The five largest cities account for just over 40 percent of total local debt. Local authorities have several sources of debt financing including development banks (38 percent of total in 2011), state treasury (29 percent), local bonds (22 percent), central government on-lending (8 percent). The SBL 2002 requires local authorities to maintain total outstanding debt from these sources at below 30 percent of annual capital budgets, except for HCMC and Hanoi for whom the ceiling is 100 percent of annual capital budget (to be raised to 150 percent for HCMC from 2015). 13 provinces have already breached this ceiling in recent years, which in the first place is not a good indicator of local authorities’ actual borrowing capacity. An illustrative debt sustainability analysis of HCMC shows that under specific assumptions, the province could borrow more than the current limits prescribed in the SBL 2002 and maintain sustainable levels of debt. However, without greater access to decentralized revenue, the province would face liquidity pressures, thereby reiterating recommendations on revenue autonomy in chapter 5. Aside from direct debt liabilities, local authorities also need to look more closely at indirect liabilities and contingent liabilities, potential sources of which include: public financial funds; local SOEs; and banking sector stress from payment arrears to local contractors. Recommendations: (i) restate fiscal rules related to local debt in the SBL 2002 and adopt more standard qualitative fiscal rules in the revised law; (ii) adopt specific thresholds on local debt stock and debt servicing in line with debt sustainability targets; (iii) improve transparency and reporting on public debt, including through statistical debt bulletins in more advanced provinces; (iv) transition to a two-tier local debt system with central government facilitating more market access for advanced provinces and greater ODA on-lending for less advanced provinces. CHAPTER 6: LOCAL BORROWING 157 1. Local borrowing has emerged as an important topic in Vietnam particularly for provinces that are not able to satisfy their capital spending needs through the existing local revenue and transfers. As discussed in chapters 4 and 5, the current system is geared to redistributing locally collected revenues, the bulk of which are contributed by a handful of provinces such as Ho Chi Minh City, Hanoi and Binh Duong. The latter in turn have highlighted significant infrastructure financing deficits. It was suggested in chapters 2, 4 and 5 that the redistributive nature of intergovernmental fiscal policies in Vietnam has made important contributions to development particularly in poorer parts of the country. Although this has meant a trade off in terms of richer provinces’ ability to retain a higher share of revenue for local investments, it was also suggested that it may in fact also be desirable for these provinces to explore higher levels of debt financing. This is particularly the case for richer provinces that are looking to higher investments in physical rather than social infrastructure, which are meant to be growth enhancing. Debt financing can also afford the necessary cash flow for lumpy capital investments and potentially promote greater discipline in capital budget management as suggested earlier in chapter 3. Debt financing is also intergenerational fair because future generations will get benefits from the infrastructure but also will contribute to paying its costs through the future servicing of the debt 2. This chapter therefore looks at the potential for sub-national borrowing to address the infrastructure financing deficit. The first part provides an overview of the current institutional framework for local borrowing in Vietnam, including applicable fiscal rules and limitations on local debt. It looks at the overall trends in local level debt and the different types of debt financing available to local authorities in Vietnam. The second part introduces local level debt sustainability analysis as a tool to help assess actual borrowing capacity of individual provinces. In particular it takes the case of Ho Chi Minh City to see how increased levels of capital spending might impact on debt levels and servicing requirements. This opens up the discussion on appropriate fiscal rules to help monitor debt sustainability. The third part introduces issues of fiscal risks that might arise from higher levels of local borrowing. It illustrates the experience of dealing with fiscal risks in other countries, and applies this to the context in Vietnam. The final part pulls the different sections together to make recommendations on fiscal rules, comprehensiveness and transparency in subnational debt monitoring and reporting, and the sound management of subnational debt from both national and subnational perspectives. It also makes recommendations on the central government’s role in strengthening debt management capacity of more advanced (or frontier) and less advanced provinces. INSTITUTIONAL FRAMEWORK AND TRENDS IN LOCAL BORROWING 3. Legal framework for local borrowing: The State Budget Law 2002 (SBL 2002) and the Public Debt Management Law 2009 (PDML 2009) stipulate the golden rule that provincial governments cannot borrow to meet recurrent expenditures. Borrowing is for capital investment projects that can generate returns to service the debt. These laws also place a ceiling on local outstanding at 30 percent of the province’s annual capital budget; Hanoi and Ho Chi Minh City are exceptions, where the ceiling is set at 100 percent of the annual capital budget. The Government’s Public Debt Management Strategy (2012)50 sets a ceiling of 3 percent of GDP for all local government debt; and a ceiling of 65 percent of GDP for total public and publicly guaranteed debt including local debt. 50 Government of Vietnam, Prime Minister’s Decision No. 958/QD-TTg, “Approval of the Public Debt and National External Debt Strategy over 2011-2020 and Vision toward 2030,” July 27, 2012. 4. The legal framework pertaining to subnational debt management in Vietnam is detailed (annex N). The 158 framework specifies the authorization to borrow, purpose of borrowing,51 types of debt instruments allowed, procedural rules for debt management, and mandatory reporting to the local assembly or similar body on debt management activities. In addition, the State Audit of Vietnam (SAV) is empowered to conduct external audit of the subnational government debt as part of local budget when needed. LGs do not have direct access to financing from the central bank. These rules are largely in line with international sound practices. 5. Local authorities are authorized by the PDML to borrow from domestic sources through bond issuance and borrowing from other financial sources. All external borrowing, however, is to be undertaken only through central government on-lending. Provincial borrowings are subject to various approval procedures depending on the utilization of fund and instruments, stipulated by PDML and other secondary regulations. In general all borrowings have to be inspected and approved by the Ministry of Finance (MOF) and other central government agencies where applicable. 6. Trends in sub-national borrowing: Although nearly 51 provinces have engaged in some form of debt financing, borrowing by local authorities over the 2006-2011 period remains very low. During the past ten years, subnational debt was always below 3 percent of GDP (figure 6.1). Sub-national borrowing increased in 2009-2010 due to the global financial crisis. It rose sharply again by around VND 8 trillion (around $400 million) in 2012 due to consolidation of on-budget capital spending following the adoption of Directive 1792. Figure 6.1: Local government debt 2006-2012 40,000 3.0% 35,000 2.77% 2.5% 30,000 Disbursement Repayment 2.0% 25,000 2.17% Outstanding debt Outstanding debt in % of GDP 20,000 1.78% 1.5% 1.62% 15,000 1.47% 1.0% 1.18% 1.22% 10,000 0.5% 5,000 0 0.0% 2006 2007 2008 2009 2010 2011 2012 Source: Ministry of Finance 7. Volatility in local debt flows is partially explained by their concentration the five large cities (figure 6.2). As the total amount of borrowing is relatively low, any decision to borrow by any of these cities tends to cause significant fluctuations in the total debt flows and levels of indebtedness. The five large cities represented more than half of sub-national borrowing until 2011. The total sub-national debt is also concentrated in these cities (42 percent of total local debt in 2012), but even within this group HCMC dominates with an accumulated 38 percent of the total local debt. 51 Examples of borrowing purposes are to finance budget and cash balance deficits; finance investment projects approved by the local assembly outside the budget process; refinance and pre-finance outstanding debt; and to finance honoring of outstanding guarantees (Source: World Bank, Subnational DeMPA tool). Figure 6.2:Borrowing in the five large cities 159 25,000 63.8% 70% 58.2% 59.2% 56.0% 56.6% 60% 20,000 64.8% 50% 15,000 49.9% 52.0% 51.6% 46.5% 40% 41.8% 10,000 30% 20% 5,000 7.7% 2.8% 10% 4.4% 0 0% 2006 2007 2008 2009 2010 2011 2012 Disbursement Outstanding debt Share in total Share in total disbursement outstanding debt Source: Ministry of Finance 8. In 2011, subnational debt only financed 4 percent of the development expenditures. This ratio was lower in the wealthier Red River Delta and South East region, but higher in the poorer Mekong River Delta and Central Highlands regions (figure 6.3). However, in 2011 around 13 provinces exceeded their outstanding debt stock limits of 30 percent of annual capital budget. Debt in some provinces was twice as high as the limit (Dong Thap, Lang Son, Bac Lieu, and Ca Mau). The regional average of the Mekong River Delta was above the limit (108 percent) and the South East region was close to it (99 percent). In this case HCMC pushed the regional average upwards, because the city itself had debt 19 percent higher than its capital investment in that year (figure 6.4). Figure 6.3: Average Local debt (VND million) Figure 6.4: Local government debt (VND million) 6.3 6.4 1,500,000 15% Average borrowing by province, 2011 Average borrowing by province, 2011 Average borrowing by province, 2012 140,000 1.2 Debt in % of borrowing limit, 2011 Borrowing in % of development investment, 2011 120,000 1 1,000,000 11.6% 10% 100,000 10.8% 0.8 6.3% 80,000 0.6 4.3% 60,000 500,000 5% 0.4 40,000 1.8% 0.9% 0.2 20,000 0 0% 0 0 Mekong Central North, Northern Red River South East Central Mekong North, Northern Country Red River South East River Delta Highlands Central Midlands Delta Highlands River Delta Central Midlands total Delta Coast Coast Source: Ministry of Finance 9. Borrowing as a share of local receipts as illustrated in figures 6.5 to 6.10 is higher in provinces that are: (i) more developed (figure 6.5); (ii) are more fiscally autonomous as measured by decentralized revenue/ total local revenue (figure 6.6), which is important for debt servicing; (iii) more fiscally sustainable (though weak link) as measured by the share decentralized revenue to recurrent expenditure (figure 6.7). The data shows, as expected, direct and strong links between borrowing levels and the share of capital spending in local expenditure (figure 6.8). It also shows that more urbanized provinces engaged more in debt financing (figure 6.9). There is however no link between budget credibility (as measured by outturn over plan) and the level of borrowing as discussed in chapter 3 (figure 6.10). In summary therefore, borrowing levels 160 are low but also seem consistent with the level of local development, capital spending needs, and fiscal sustainability trends. Figure 6.5: Borrowing/local receipts and Figure 6.6: Borrowing/local receipts and fiscal development autonomy 50 Decentralized revenue as % of local 95% 45 R² = 0.1609 85% Per Capita GDP (VND m) 40 75% 35 65% R² = 0.11 revenue 30 55% 25 45% 20 35% 15 25% 10 15% 5 5% 0% 2% 4% 6% 8% 10% 12% 14% 0% 5% 10% 15% Borrowing as share of local receipts (%) Borrowing as a share of local receipts (%) Figure 6.7: Borrowing/local receipts and fiscal Figure 6.8: Borrowing/local receipts and share of sustainability recurrent spending 400% 65% Decentralized revenue as % of Local recurrent expenditure as % local recurrent expenditure 350% 60% R² = 0.0591 of total local expenditure 300% 55% 250% 50% 45% 200% 40% 150% 35% 100% 30% 50% 25% 0% 20% 0% 5% 10% 15% 0% 5% 10% Borrowing as a share of local receipts (%) Borrowing as share of local receipts (%) Figure 6.9: Borrowing/local receipts and fiscal Figure 6.10: Borrowing/local receipts and share sustainability of recurrent spending 2.00 220 Local expenditure as % of 1.80 1.60 200 Urbanization rate 1.40 local budget 180 1.20 1.00 160 0.80 140 0.60 0.40 120 0.20 0.00 100 0% 5% 10% 15% 0% 2% 4% 6% 8% 10% 12% 14% Borrowing as share of local receipts (%) Borrowing as a share of local receipts (%) Source: Staff estimates based on published State Budget data 10. Sources and types of local borrowing: Local authorities have a variety of debt financing options available to them including: (a) the domestic capital market (local bonds, loans from commercial banks); 161 (b) the State Treasury; (c) development banks; and (d) on-lending from the central government of external funds. Such borrowing is monitored closely by the Ministry of Finance under the ceilings prescribed in the SBL 2002; these categories of borrowing are often referred to as borrowing as per Article 8.3 of SBL 2002. Due to the ceilings imposed under SBL 2002, local authorities also turn to other forms of borrowing e.g. Local Infrastructure Development Fund, which are not subject to the same limits. In some cases of Overseas Development Assistance on-lent by central government, the limits also did not apply. 11. Composition of local government debt went through a structural change during the past years. The share of VDB loans in financing sub-national capital investment increased and is the highest share of the local debt portfolio (38 percent). At the same time, the amount of local bond obligations also decreased, especially in FY 2011 (14 percent). However, this was also the year when the absolute level of loan disbursement was low and borrowing in the large five cities dropped to almost zero. So the debt to central government on-lending has increased sharply (30.8 percent of the total debt). Figure 6. 11: Composition of local debt, 2006-2012 4.8% 7.4% 12.0% 14.4% 2.3% 100% 17.0% 2.6% 0.1% 0.7% 8.4% 90% 1.5% 2.6% Other 80% 33.2% 3.9% 30.8% 21.7% 70% 43.4% 30.8% 21.2% Onlent central govt. loan 41.8% 60% 13.8% 50% 29.4% 25.9% Local bond 22.5% 27.5% 16.9% 40% 22.3% 30% 21.7% State Treasury 20% 36.0% 35.9% 38.3% 26.2% 29.7% 30.4% Vietnam Bank for Development 10% 23.0% 0% 2006 2007 2008 2009 2010 2011 2012 Source: Ministry of Finance 12. Local Bonds: The bond market in Vietnam is largely underdeveloped, with central government bonds dominating the market. So far, the only sub-national governments that have accessed financing in the capital markets through bond issues are Hanoi, HCMC, Da Nang, and Dong Nai Province. Of these, HCMC was the first to issue bonds, and the only one to have done so more than once. The Provincial People’s Council (PPC) has the right to issue bonds in the domestic market for investment in social and economic infrastructure projects or projects that have the ability of capital recovery. All local authorities must obtain approval from the People’s Council for a specific project and it’s financing through the issuance of a bond. Subsequently, the Finance Department of the local government must send the proposal to the MOF for consideration. 13. Sub-national bonds are required to have a term or duration of one year or above and each bond must have a face value of VND 100,000 and minimum denomination of VND 100,000. The PPC determines the interest rate for the bond, but it must be within the range set by MOF. Sub-national bonds can be bid through the Hanoi stock exchange and/or dealers. The PPC is required to adopt the procedures that apply for issuing Government bonds to all subnational bond issuance subject to bidding through the Hanoi stock exchange and guarantees. If the sub-national bond is issued through dealers, it is the responsibility of the PPC to select competent dealers to carry out the process. These dealers may be commercial banks, security firms, or provincial development and investment funds. Here again the PPC is expected to adhere to selection criteria and procedures compatible to those applicable for Government bonds. 14. Although an attempt is being made to move towards a market-based borrowing method, at present 162 this is constrained. Both the amount of borrowings and the rates are controlled. There is also lack of transparency and a degree of moral suasion for the banks to invest in bonds of the local government where they are located. In the case of HCMC, the lead underwriters are BIDV and Vietinbank. However the process of the negotiated syndicate borrowing has added to the transaction costs and has made it harder for the investor to subscribe to the LG’s bonds. 15. Loans from the commercial banking sector: if local authorities cannot go to the bond market, they borrow from banks, which can be a difficult and costly channel. Commercial lending to local authorities in Vietnam has been very limited. Commercial banks are reluctant to finance local authorities, and view lending for local development purposes as too risky, particularly in the absence of a clear recourse collateral mechanism. Past policies of ‘directed lending’ for infrastructure and other sectors have meant that banks conducted little or no technical and risk assessments, with few incentives to develop the capacity to evaluate and monitor projects according to commercial principles. 16. The repayment capacity of projects and borrowers was rarely considered, because of the assumption of implicit guarantees by the local or central government. Loans were rarely secured with proper collateral, and those loans that did have collateral were in effect equivalent to project financing without recourse to the government sponsor. The result has been the accumulation of non-performing loans (NPLs), with loans often having to be restructured and extended. Lending by commercial banks is also constrained by the limited capital base of the banks, and maturity mismatches between short-term deposits and the long maturity needed for development loans. 17. Borrowing from the State Treasury (ST): A PPC has the option to borrow through the provincial Treasury for infrastructure projects that are allocated from the budget. If the PPC cannot repay its debt to the State Treasury, the latter has the authority to implement and intercept and deduct payment from the provincial budget. When a PPC borrows from the State Treasury to accelerate progress on a project(s) it is required to document specifics of the project(s), borrowing and payment schedules, and the extent of total borrowing, including borrowing from bond issuance and commercial banks. This documentation must be endorsed by the People’s Council and submitted to the MOF for approval. Similarly, for projects designed to attract investments, the PPC is required to go through the same approval process. 18. The maximum term of borrowing from the State Treasury is 12 months for projects that are allocated from the State budget. The MOF determines the borrowing term for projects that aim to attract investments. The State Treasury applies a uniform borrowing fee of 0.15 percent per month (30 days) of the total loan and imposes a late fee on the basis of the key interest rate announced by the State Bank of Vietnam. When a payment is 30 days overdue the State Treasury reports it to MOF and will deduct an amount equivalent to the overdue payment plus any late fees from the provincial budget. 19. Information on amounts borrowed, borrowing fee and late fees are posted according to the chart of account. The provincial Treasury reports the total amount to the State Treasury (VST) on a monthly, quarterly and annual basis. In turn, VST reports to MOF semi-annually. This is by far the most preferable type of borrowing for LGs given the low costs and streamlined approval process (by the MOF only). However, this type of borrowing will significantly decrease with the coming introduction of the Treasury Single Account as part of the expected improvements in cash management. 20. Borrowing from domestic development banks (VDB and VBSP): The Vietnam Development Bank (VDB) and the Vietnam Social Policy Bank (VSPB) are non-profit institutions that mobilize capital from different sources and lend to targeted groups. From the perspective of the sub-national borrower, the more important of the two is the VDB, which provides credit for developmental investment of local authorities and export-oriented projects of SOEs and other economic organizations. The former has to be pre-approved by 163 the Central Government. The focus of VSBP is to provide credit to the poor and other policy beneficiaries. 21. The VDB issues bonds and lends to the LGs at zero interest rate. The central government provides the subsidy for the rate of interest of the VDB. Lending by VDB is conditioned by the borrowers’ ability to cover 20 percent of the capital investment in a project from own resources. VDB lending is focused on social-economic infrastructure, agriculture, and industry. The duration of each loan is determined by the chief executive officer of VDB, on a case-by-case basis, up to a maximum of 12 years. The interest rate is set in accordance with the borrowing and operating costs of VDB and is adjustable, with the relevant rate applied to each loan disbursement. 22. According to a State Audit Report the non-performing loans of VDB is very high compared to other commercial banks: more than 12 percent in 2010. This year restructuring proposals were submitted to the Government in 193 cases, with a value of VND 1.3 trillion. The typical measure is debt freezing (VND 845 billion), cancelling the principal (VND 216 billion) or the interest (VND 217 billion) and rescheduling could be used only to limited extent (VND 15.9 billion)52 . VDB was authorized to use its contingency funds and to sell bad debt to other parties. 23. On-lending from CG of external funds: Currently all on-lending to LGs are from Official Development Assistance (ODA) sources53 . Current regulations54 stipulate ODA on-lending in the same terms to enterprises and Local Government (LGs) with repayment capacity. However there have not been any explicit policy framework and guidelines on the allocation and management of ODA on-lending. Current on-lending arrangements are negotiation-based (between MOF and PPCs)—there is no formula for the proportion of ODA grant and on-lending to Local Governments. There also lacks proper incentives for LGs to borrow from ODA sources. Timing for on-lending decision was often after official approval of Development Partners (DPs). This has directly affected the operations of both government DPs. 24. The preparation of ODA on-lending projects is part of capital investment planning process. All programs and projects funded by ODA loan above USD 1 million should be approved by the Prime Minister. ODA funds have been on-lent though three channels: directly to the provincial budget, indirectly through the Local Development Investment Funds (LDIFs), and through the umbrella projects led by central ministries. The shares of ODA on-lending in subnational borrowing and in total ODA to local levels remain low (less than 10 percent and 1 percent respectively). However, given the changing ODA landscape and debt profile, the shares are expected to gradually increase in local development financing in the medium-term. It is expected that the Government will adopt a policy to promote this trend. 52 See Nguyen Thi Thanh, 2013, based on investment newspaper online dated August 28, 2013 under MPI 53 There term ODA used in this report follows the definition of OECD Development Assistance Committee, which is not entirely similar to that of Vietnamese regulation: ODA by OECD-DAC definition includes ODA and concessional loans by Vietnam’s definition. 53 Including Public Debt Management Law (2009), Decree No. 78 (2010), and Decree No. 38 (2013). Table 6.1: Summary of different sources and types of local borrowing 164 Borrowing Controlled and monitored by cg Constraints Both amount and rate of borrowings are Ceiling rates determined by bfid, and this controlled. Amount is monitored under Bonds results in under-subscription as rates to threshold; rate is kept below ceiling fixed investors are unattractive by bfid. Loans from banks Not controlled Banks ask collateral; higher rates Short term borrowing for cash management Borrowings from ST Amount monitored under threshold purposes Borrowings from Only for specific projects and purposes VDB Quantum depend on the overall ODA available All external on-lending is through the On-lending to Vietnam; also graduating from the highly center concessional loans in the medium term Hierarchical to higher-level government, Accountability controls on inputs and process with little or Results-based, bottom-up, client-driven no concern for result LOCAL DEBT SUSTAINABILITY ANALYSIS: HCMC CASE STUDY 25. As noted above, cities such as Hanoi and HCMC’s debt have a higher debt stock threshold of 100 percent of annual capital budget, compared to others that are limited by 30 percent. Despite this both Hanoi and HCMC argue that even higher levels of borrowing are required to finance their capital spending needs. According to the authorities in HCMC, the city’s urban annual infrastructure development needs amount to around VND 60-70 trillion, compared to current annual investment levels of between VND 10-20 trillion. They argue, as a result, that most of the capital expenditure goes to land clearance, resettlement, and ODA loan repayments rather than construction activities. 26. Fiscal situation in HCMC: To better understand borrowing capacity at the local level, the study undertakes an illustrative debt sustainability analysis for HCMC. Under the SBL (2002), borrowing is treated outside the balance sheet, and therefore not accounted for in the local budget balance. Provinces are required to maintain a balanced budget. The actual budget balance for HCMC (i.e. inclusive of debt financing) can be approximated using Government Finance Statistics norms. On this basis, HCMC has maintained an average budget deficit of roughly 0.1 percent of local GDP between 2006 and 2011. This has helped to keep borrowing at relatively low levels. Total outstanding debt from bond issuances at the end of 2013 was an estimated VND 11.6 trillion, which is around 1.5 percent of local GDP and 49 percent of the capital budget in 2013. 27. An important factor contributing to overall fiscal discipline in HCMC is relatively strong local revenue mobilization. This is in part because a large number of enterprises are headquartered in HCMC, which means that even if the enterprises’ production base is elsewhere, all CIT and VAT payments accrue to HCMC. More than half of local financing (inclusive of new debt flows) comes from decentralized revenues (figure 6.12). The latter have constituted on average 8 percent of local GDP in the 2006-2011 period. Given that this includes only 23 percent of the shared revenue collected by HCMC (i.e. net of transfers back to center), and none of the levies on trade, gross local revenue as a share of GDP in HCMC is likely to be above 30 percent. The ratio of decentralized revenue to local recurrent expenditure has averaged around 200 percent over the course of 2006-2011, which is another good indication of fiscal sustainability. Figure 6.12: HCMC Local Financing (% of local GDP) Figure 6.13: HCMC Local Spending (% of local GDP) 165 16.0 12.0 14.0 10.0 12.0 2.7 1.8 2.1 8.0 10.0 3.2 2.0 4.0 3.8 8.0 1.4 6.0 3.8 1.2 4.1 3.8 3.8 3.7 3.9 6.0 0.9 4.0 4.0 8.6 8.6 7.4 7.2 7.6 7.2 5.4 6.4 5.1 5.2 2.0 3.5 3.5 3.4 4.2 3.8 2.0 3.1 0.0 0.0 2006 2007 2008 2009 2010 2011 2012 e 2013 e 2006 2007 2008 2009 2010 2011 2012 e 2013 e Decentralized revenue Off budget Capital Expenditure Recurrent Expenditure Central Transfers Other Debt servicing Off-budget Expenditure Other Source: Staff estimates based on published State Budget data 28. An estimated 40 to 50 percent of local spending in HCMC between 2006 and 2011 went on capital investments (figure 6.13). This is high when compared to a national average of around 30 percent. Capital spending grew very rapidly in 2008-2009 (over 70 percent in real terms), most likely reflecting a fiscal stimulus in response to the effects of the global economic crisis. However, total local spending in HCMC averaged around 11 percent of GDP in 2006-2011, which is much below the national average of around 28-30 percent, reflecting the large share of non-state activities in HCMC. Capital spending fell quite sharply after 2010; it is even estimated to have contracted in real terms, as provincial authorities started consolidating investments in response to Directive 1792 (figure 6.14). Recurrent spending growth has been more stable. Figure 6.14: HCMC Local Revenue and Spending Figure 6.15: HCMC Local Fiscal Outturns Aggregates (% change real) 251 250 80.0 206 190 70.0 200 60.0 143 141 50.0 150 129 118 110 105 107 105 40.0 101 100 30.0 20.0 50 10.0 0.0 0 -10.0 2006 2007 2008 2009 2010 2011 2012 e 2006 2007 2008 2009 2010 2011 -20.0 Local capital expenditure as % local capital budget Capital Recurrent Local recurrent expenditure as % of local recurrent budget Source: Staff estimates based on published State Budget data 29. Although recurrent expenditure outturn is reasonably good (averaging 108 percent of budgeted recurrent 166 spending in 2006-2011), capital spending outturn averaged 177 percent of budgeted capital spending in 2006- 2011 (figure 6.15). This shows a clear revealed preference for capital investments. As discussed in chapters 3 and 4, higher levels of actual spending on capital investment than what was budgeted are due to a range of institutional factors that lead to loss of credibility in spending plans. 30. Debt sustainability analysis: If HCMC authorities now want to raise more debt to meet further capital investment needs, it is necessary to assess its borrowing capacity. This can be done through an illustrative Debt Sustainability Analysis (DSA). A DSA would help assess how the HCMC authorities’ current public debt levels and prospective new borrowing will affect its ability to service its debt in the future. Debt levels are sustainable if the authorities can service the debt without major corrections to its ongoing expenditures or without having to take recourse to debt cancellations. For the purposes of the HCMC DSA, public debt does not include: (i) borrowing from idle cash balances in the Treasury; and (ii) Overseas Development Assistance funds that are on-lent by the central government. 31. The first step in the DSA is to set a baseline scenario on how selected macroeconomic and debt variables might evolve based on some realistic assumptions. The assumptions for the HCMC DSA are set out in Box 6.1 below. These are largely based on historical trends and any important changes expected over the medium-term e.g. new Stability Period 2016-2020. Fiscal information up to 2011 is based on HCMC’s final accounts; fiscal information for 2012 and 2013 are estimates, and 2014 is the first year of projections. Box 6.1: Key Macroeconomic Assumptions for Baseline Scenario (2013-2025) Real GDP growth over the 2006-2011 averaged around 10 percent per year. Over the medium-term, real growth is expected to remain slightly below the period average at 9 percent. Structural reforms of SOEs and the banking sector may lead to some slowdown in the short-term before real growth starts picking up again and reaching 10 percent over the long-term. Inflation is projected at 6 percent over the medium-term. This is considerably below the 2006-2011 period average of 10 percent when expansionary policies and commodity price shocks led to rapid increase in price pressures. Over the long-term price pressures may pick up slightly with growth, but are expected to stabilize at around 5 percent. Local revenue growth in 2012 and 2013 is estimated to be lower than the historical average partly due to slower growth but also because of counter-cyclical tax breaks (figure x). However, it is expected to pick up again gradually over the medium to long-term due to higher growth and possible revenue assignment reforms for the Stability Period 2016-2020. Off-budget financing (including bond issuances) are assumed to have fallen from the period average because there were no major bond issuances between 2007 and 2011. The baseline assumes lower ‘unspent revenues’, which are carried over from one year to the next. These over realized revenues are likely due to earmarked sources or cash flow issues, which leads to excess receipts at the end of the year. This however is not regarded as good practice and may be addressed in the revision of the SBL (2002). Local expenditure growth slows down significantly in 2011 and 2012, which was a period of 167 consolidation, before picking up again in 2013. Thereafter capital spending spikes up going from 3.1 percent of GDP to 4 percent of GDP over the medium-term (figure 6.16) to meet HCMC’s capital spending needs. Recurrent spending also increases slightly from 6.1 percent of GDP to around 6.8 percent of GDP before stabilizing over the long-term. The budget deficit and thereby HCMC’s net financing requirements are projected to grow over the medium-term in line with growing capital spending. Maturity and interest rate assumptions show better terms than what HCMC has been able to borrow at historically. Three categories of loans: (i) 3-year maturity with 7 percent interest; (ii) 5 year-maturity with 6 percent interest; and (iii) 10-year maturity with 5 percent interest. Figure 6.16: HCMC Local Revenue (% of GDP) Figure 6.17: HCMC Local Expenditure (% of GDP) 12.0% 7.1% 6.7% 6.8% 11.6% 6.6% 11.5% 6.1% 6.3% 6.4% 6.1% 11.0% 10.8% 10.8% 10.5% 10.5% 10.0% 10.1% 9.9% 9.5% 5.4% 9.4% 9.0% 9.0% 9.1% 8.9% 4.4% 8.5% 8.5% 4.2% 4.2% 4.0% 8.0% 3.5% 8.3% 7.5% 3.1% 7.0% 2006 2007 2008 2009 2010 2011 2012 e 2013 e 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2006 2008 2010 2012 e 2014 2016 2018 2020 2022 2024 Local Revenue Capital Recurrent Source: Staff estimates 32. The second step of the DSA is to set debt burden indicators to compare HCMC’s debt service and debt stock levels against various measures of its repayment capacity. The following ratios are used as measures of debt burden: (i) HCMC debt stock to GDP (threshold of 25 percent to assess solvency); and (ii) HCMC debt service to local revenue (threshold of 25 percent to assess liquidity). The thresholds are benchmarks and not strict ceilings. Other indicators of fiscal sustainability include: (i) interest payments as a share of recurrent spending; (ii) the primary fiscal deficit to assess the level of fiscal consolidation efforts; and (iii) local revenue to recurrent expenditure to review maintenance of the “golden rule” such that borrowing is restricted to investments only. 33. Under the baseline scenario for the HCMC DSA there are no apparent issues in terms of solvency. The debt stock to GDP ratio remains well within the applicable threshold (figure 6.18). Debt stock to local GDP peaks at 7.7 percent under the baseline scenario. Therefore the anticipated debt stock for HCMC remains affordable when comparing to local GDP as the indicator for repayment capacity. 165 34. On the other hand, the authorities would need to be mindful of liquidity pressures. Although the debt 168 service to revenue ratio remains broadly affordable, the applicable threshold is breached in 2018 and 2021 (figure 6.19). The liquidity ratio fluctuates a lot because the DSA assumes “balloon payments” whereby the entire principal on a bond issuance is due at the end of the maturity period. This explains the periodic sharp increases in the debt service to revenue ratio. Another factor is that the revenue denominator includes only those receipts that are retained by the province, which is the correct variable when assessing the sustainability of local debt. But what this shows is that it would be difficult for HCMC to borrow more and have a further reduction in its revenue sharing ratio in the next Stability Period without priority spending being crowded out by debt service burden. Figure 6.18: HCMC Debt-to GDP ratio Figure 6.19: HCMC Debt service-to-revenue ratio 30.0% 35.0% 25.0% 30.0% 25.0% 20.0% 20.0% 15.0% 15.0% 10.0% 10.0% 5.0% 5.0% 0.0% 0.0% 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 HCMC Debt-to GDP ratio Threshold HCMC Debt service-to-revenue ratio Threshold Figure 6.20: HCMC Interest payment/Recurrent Figure 6.21: HCMC Overall and Primary Balance (% Expenditure of GDP) 8% 0.5% 7% 0.0% 2014 2015 2016 2017 2019 2020 2021 2022 2023 2024 2018 6% 2025 -0.5% 5% -1.0% 4% -1.5% 3% -2.0% 2% -2.5% 1% 0% -3.0% 2014 2015 2016 2017 2018 2019 2020 2021 2022 2024 2025 2023 -3.5% HCMC Interest Payment/Recurrent Expenditure Balance (% of GDP) Primary Balance (% of GDP) Source: Staff estimates 35. Another issue to watch under the baseline scenario is the growing share of interest payments in total recurrent spending (figure 6.20). Interest payments go from just over 1 percent of recurrent spending in 2014 169 and peak at 7 percent in 2021. This is a fairly substantial share, even assuming relatively favorable borrowing terms. It will be important to ensure that it does not crowd out high priority recurrent spending needs. The overall balance tracks the primary deficit closely (figure x) thanks to assumptions on favorable borrowing terms relative to borrowing terms in the past. Under the baseline scenario, HCMC is able to gradually consolidate its primary deficit thanks in large part to assumptions on stable recurrent spending (figure 6.21). HCMC is also able to maintain the golden rule of borrowing only for investment purposes. Decentralized revenue cover recurrent expenditure throughout the projection period. Finally, under the baseline scenario, the debt to capital expenditure/budget ratio breaches the HCMC threshold of 100 percent throughout most of the projection period. However, as noted above the authorities may wish to review this threshold going forward. Figure 6.22: HCMC Revenue/Recurrent Expenditure Figure 6. 23: HCMC Debt-to-capital expenditure (%) 180.0% 200.0% 160.0% 140.0% 180.0% 120.0% 160.0% 100.0% 80.0% 140.0% 60.0% 40.0% 120.0% 20.0% 100.0% 0.0% 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2006 2008 2010 2012 e 2014 2016 2018 2020 2022 2024 HCMC Debt-to capital expenditure ratio Revenue/Recurrent Expenditure Threshold Source: Staff estimates 36. To assess the potential impact of different macroeconomic and policy developments on HCMC’s fiscal sustainability, two alternative scenarios are considered. The first is a low case scenario in which: (i) growth slows to 5 percent per year over the medium-term and gradually stabilizes to below historical average at 8 percent over the long-term; (ii) revenue growth falls by nearly a third; (iii) expenditure remains the same as in the baseline. The second is a high case scenario in which: (i) growth reaches 11 percent per year over the medium-term before stabilizing at 10 percent over the longer term; (ii) government revenue buoyancy as a result picks up; and (iii) this affords an increase in government spending over the medium term at over 2 percentage points of GDP above the historical average. 37. Not surprisingly, debt servicing requirements under the low case scenario would create unsustainable liquidity pressures, and would crowd out recurrent expenditure. Although the debt to GDP ratio remains within the applicable threshold, the debt-service-to-revenue ratio exceeds the threshold for almost the entire projection period. Interest payments reach close to 14 percent of recurrent spending. In addition, it would not be possible under this scenario to maintain the golden rules over the long-term. 38. The high case scenario, on the other hand, shows that with stronger growth and revenue mobilization, HCMC would more than meet its capital spending needs and maintain sustainable levels of debt. However, there would still be liquidity challenges in the initial years, and the assumptions on economic growth and revenue buoyancy are quite optimistic. Revenue reforms may help to meet some of these targets, but this will be a function of what is agreed on revenue sharing arrangements for the next Stability Period. 39. Fiscal rules for local debt sustainability: The above analysis points to the need for reviewing sub- 170 national fiscal rules in Vietnam. There are four main types of fiscal rules, which can be distinguished based on the type of budgetary aggregate that they seek to constrain: (a) Debt rules set an explicit limit or target for public debt in percent of GDP; (b) Budget balance, or structural budget balance rules constrain the variable that primarily influences the debt ratio and are largely under the control of policy makers; (c) Expenditure rules set limits on total, primary, or current spending; (d) Revenue rules set ceilings or floors on revenues and are aimed at boosting revenue collection and/or preventing an excessive tax burden. 40. The most common fiscal rule is the budget balance requirement, which predominantly targets the outturn of an annual budget. Most sub-central governments also face a restriction on borrowing. There has been a move in a number of countries away from micro-management through a prior approval system on a case-by-case basis towards aggregate and numerical targets. Cross country practices on subnational borrowing rules and patterns of financing is elaborated to draw peer examples for Vietnam. For instance, possible improvements with reference to the rules and practices that are more commonly used internationally as follows: Table 6.2: International Sub-National Borrowing Practices Type of restriction Description Countries Ceilings on (i) debt service / local Argentina, Brazil, Italy, Japan, Spain, Affordability formulae revenues; (ii) debt service / local current Lithuania, Romania, Poland, Colombia saving Indebtedness Limit on total outstanding debt / net Brazil, Colombia, Italy, EU formulae revenues “Golden rule” Brazil, Canada, USA, Austria, South Africa, Borrowing for capital expenditures provision Switzerland, India Local councils are required to pass Balanced budget Brazil, Canada, Germany, Netherlands, USA balanced budgets Local councils are required to approve Local approval Canada, Switzerland, USA borrowing for individual projects 41. There are generally three types of indicators for controlling local government debt: limits on (i) debt service, (ii) new borrowing in a fiscal year and (iii) on total outstanding debt. Examples from nineteen countries summarized in Annex O show that most typically the debt service and the debt stock are the subjects of regulations on local government borrowing. The limits are defined as ratios, where the total or a subset of local revenues (e.g. typically own source revenues) are in the denominator. The debt stock is usually compared to the total budgets. The new (net) loan is rarely regulated, as it does not really show the borrowing capacity of a local government. All the actual values of the borrowing ceiling are influenced by the local government functions and by the local revenues within the system of intergovernmental finances. 42. As these international practices show, the borrowing limits might be regulated by combined indicators. So local governments should meet multiple criteria, e.g. limits on both debt service and debt stock. The other important question is which year should be the basis of comparison. Typically the revenues from the previous year are used, because these data are available at the moment of borrowing decisions. However, in some countries (Croatia, Romania) the average of the previous three years is the basis of the debt limits. This longer term average makes the revenue estimates more reliable. The borrowing regulations usually require to taken into account the guarantees, as well, when the local government borrowing limit is calculated. In countries, where local borrowing requires central government approval, assessment of all the indicators is supplemented with the evaluation of the local government creditworthiness on a project-by-project basis. MANAGING FISCAL RISKS FROM LOCAL BORROWING 171 43. Explicit and implicit contingent liabilities: Appropriate fiscal rules can help to manage and mitigate fiscal risks that emanate from higher levels of sub-national borrowing discussed above. The latter constitute direct liabilities of local authorities, which as discussed earlier are still fairly limited in Vietnam at the moment, but could potentially grow especially as debt thresholds are being raised. Aside from direct liabilities, however, it is also important for local authorities to have a strategy in place to address possible indirect, or what is commonly referred to as contingent liabilities. Represent potential financial claims against the government. But they can crystallize into actual liabilities following events such as debt default, insolvency, or insufficient revenues. Contingent liabilities can be grouped into two categories (table 6.3): (i) explicit liabilities, which are based on law and contracts; and (ii) implicit liabilities, are commitments that are based on political announcements, public expectations and possible interest group pressures. Table 6.3: Sources of Fiscal Risks Source of obligation Direct liabilities Countries n  uarantees for sub-national government G n Foreign sovereign borrowing and soe obligations n Domestic sovereign borrowing n Guarantees for soe fiscal performance (e.g.,  n  ther direct borrowing (e.g., from the O electricity uptake) social security fund) n  Guarantees for policy banks’ borrowing Explicit liabilities n  urrent committed budget expenditures C (development bank) (legal obligation) (may include direct subsidies/transfers n  Guarantees for trade and exchange rate to soes) risks n  egally mandatory future expenditures L n Guarantees for private investments (ppps) (e.g., social and health insurance n State insurance schemes (e.g., on bank  schemes) deposits, private pension funds, crop failure, flood, war) n  efaults of sub-national governments and D n Soe arrears to the government soes on non-guaranteed debt and other n  uture public pensions, social security F obligations Implicit liabilities schemes, health care financing (if not n Default of banks (support beyond state  (public/ required by law) insurance) Political choice) n  uture recurrent cost of public F n  Failures of non-guaranteed pension funds, investments or other social security funds n Natural disaster relief, environmental  recovery Source: The World Bank 44. Cross country experience suggests that explicit contingent liabilities, especially loan guarantees, may be the most frequent type of indirect liability but implicit ones are often the most costly. The latter include bailouts of SOEs, banks and local governments. It also includes natural disasters contingent liabilities, which can significantly add to governments’ balance sheet risk in as much as they imply additional leveraging and tend to be triggered in times of financial stress, with realized costs having a major impact on a country’s fiscal position and debt sustainability. These are all potentially relevant issues for both central and local authorities in Vietnam. This is illustrated in part by the experience in China where despite restrictions on local borrowing, the level of contingent liabilities of local authorities is reported to have grown substantially through the latters’ use of special-purpose vehicle to borrow on the market (box 6.2). 172 Box 6.2: Contingent liabilities of local authorities in China Even though China's budget law forbids local governments from taking on debt directly, localities have borrowed heavily through special-purpose vehicles known as local-government financing vehicles (LGFV). Most recently, Zhang and Burnett (2014)55 show that in China contingent liabilities include potential costs associated with nonfinancial SOE debt (excluding LGFVs); policy banks’ liabilities; fiscal costs of recapitalizing banks, for example that could stem from losses related to nonperforming loans (NPLs) (such as NPLs from LGFVs); and liabilities of state-owned asset management companies. Yang et al (2012)56 estimated total 2010 contingent liabilities at just above 100 percent of GDP (around USD 5.88 trillion). Their estimate includes debt of nonfinancial SOEs but excludes LGFVs (¥ 35.6 trillion), policy banks’ outstanding financial debt (¥ 5.2 trillion), old NPLs assumed by the asset management companies (¥ 4.2 trillion), nonperforming loans (NPLs) in the current banking sector (¥ 0.4 trillion), and liabilities from social security funds (¥ 3.5 trillion). China's National Audit Office (NAO), which published a detailed survey of local debt in 2011, recently estimated the 2012 local government debt outstanding at between 6.73 trillion and 9.7 trillion yuan (about USD 1 – 1.5 trillion).57 Earlier, the NAO's widely-cited 2011 report estimated local debt at 10.7 trillion yuan by end-2010 (about USD 1.5 trillion). Loans to LGFVs have been growing in recent years as local governments raced to meet the central government’s growth goals, and market observers have been raising the red flag on what they see as increasing bank exposure to poorly designed projects that have no hope of repaying their debt. As a result, China’s regulator had to impose stricter oversight and control requirements on LGFVs. The China Banking Regulatory Commission (CBRC) distributed guidance that strengthened the supervision, such as setting up of new risk management systems, banning bank guarantees on LGFV bonds, and preventing banks from lending to weaker LGFVs that are more likely to default on payment.58 45. Contingent liabilities in Vietnam: Recent developments in Vietnam, particularly in the banking sector and SOEs underscore the importance of monitoring contingent liabilities. Rapid credit growth in 2008-2009 with insufficient due diligence, coupled with the effects of the global economic crisis, has led a build-up of Non-Performing Loans and a rapid deterioration in the balance sheets of domestic banks in Vietnam. The crisis also brought to the fore the vulnerability of SOEs and the lack of transparency around their operations and finances. MOF/MPI have estimated total SOE debt as at September 2011 to be closer to US$20 billion, or roughly 15 percent of GDP. More specifically for local authorities in Vietnam, there are potentially three important sources of contingent liabilities: (i) Public Financial Funds; (ii) local SOEs; and (iii) banking sector stress as a result of payment arrears to capital construction projects. 46. Public Financial Funds (PFF): Vietnam has more than 30 state PFFs established at both central and sub-national level, with diverse scope, nature and scale of operations. Some PFFs are established by and for the central government, others by and for sub-national governments, with some funds established at both central and sub-national levels. Only certain central level funds (e.g. the compulsory social insurance fund, health insurance fund, unemployment fund, job creation assistance credit fund, central SOE restructuring and development fund, the sinking fund), and the sub-national development investment funds have substantial own revenue sources and expenditure mandates; together these funds account for 95% of total expenditure of State PFFs. The others are small-sized funds with narrow scope of operation within certain localities only. 55 http://www.imf.org/external/pubs/ft/wp/2014/wp1404.pdf 56 http://www.erj.cn/en/IssueInfo.aspx?m=20120821110357200398 57 http://uk.reuters.com/article/2013/04/16/uk-china-debt-local-idUKBRE93F05Y20130416 58 http://asianbankingandfinance.net/sites/default/files/asianbankingfinance/print/ABF_Sept2013_LR%209.pdf 47. In general, although some financial funds can mobilize revenues from contributions of beneficiaries (such as the compulsory social insurance fund, health insurance fund, unemployment insurance fund.), the majority 173 of State financial funds are dependent on financing from the State budget (i.e. in the initial establishment and during operations), and their non-budgetary sources of revenues are limited. Although budget plans, budget execution reports and end-year financial statements of these funds are submitted to the sponsor Ministries (or their counterparts at local level), and audit reports on their activities are published, there has been no effort to consolidate the revenues and expenditures plans of these funds into the annual budget and budget final account documents (PEFA, 2012). Figure 6.24: Largest Local Public Financial Funds 4,622 2,433 864 575 412 330 302 150 39 Investment Land Job Creation Credit Farmer Housing Environment Cooperative Gratitude Fund Fund Development Fund Insurance Fund Assistance Development Protection Development for National Fund Fund Fund Fund Supporting Services and Fund Source: Ministry of Finance 48. The largest PFFs at the local level are Local Development Investment Funds (LDIF). There are 28 LDIFs now – these are special purpose vehicles (SPVs) set up to mobilize capital and enter into contracts with the private sector for the development of municipal infrastructure.59 LDIFs are expected to operate as commercial- oriented entities, raising medium and long term capital and investing in cost-recovery infrastructure projects. They have the legal status of a sub-national State Financial Institution, undertaking financial and development investments. 49. LDIFs have expanded considerably in Vietnam over the past decade, mobilizing important capital for infrastructure investment. Between 2005 and 2011 the average annual growth rate of the operational capital in the LDIFs system was high at 40 percent. In 2004, the total charter capital of LDIFs in Vietnam was approximately US$ 200 million; in 2011 it was estimated to be approximately US$450 million, with the top- seven LDIFs investing approximately US$ 100 million per year, accounting for an investment increase of 118 percent for these funds over the 2002 level.60 50. In the case of the HCMC Investment Fund for Urban Development (HIFU), established in 1996 and recently upgraded into an Investment and Financial Corporation (HIFC), ODA lending was used to provide the base capital. HIFU/HIFC then borrows from the market including commercial loans from foreign banks, and lends for infrastructure projects. These loans have been backed by government guarantees; their number and volume are small and so far there have been no defaults. HIFU is recognized as a model LDIF in terms of internal policy and procedures, subproject appraisal, social and environmental safeguards and partnership with the private sector. 59 LDIFs organization and operation are stipulated under Government Decree 138/2007 and Decree 37/2013. 60 WB, “Assessment of the financing framework for municipal infrastructure in Vietnam,” (2013) 51. The LDIFs that have qualified for ODA funding have good financial positions, including very low NPL 174 ratios and acceptable profitability and balance sheet strength. The Ministry of Finance carries out an annual qualification process to ensure the financial, operational and institutional soundness of LDIFs before they can receive ODA funding. The qualified LDIFs have a proven track record of building major infrastructure. LDIFs operate under a legally binding Credit Statute of the Provincial Government that mandates investment analysis to ensure cost recovery and provides guidelines for investment monitoring and workouts when necessary. LDIFs also have an internal appraisal and investment monitoring system, with the decision-making authority resting with the LDIF staff. These LDIFs price their loans at market rates, maintain annual audited financial statements based on accepted accounting standards, and are financially self-sufficient. 52. LDIFs also engage in short-term borrowing, on a roll-overs basis from State-owned Commercial Banks and other State-Owned Enterprises. These short-term borrowings can lead to short-term oriented investments, potentially re-allocating LDIF capital away from long-term infrastructure development. Inappropriate borrowing by the LDIFs can also have a negative effect on the developing banking sector. 53. Local State Owned Enterprises: There are 1,506 State Owned Enterprises (SOE) and 982 Public Service Enterprises (PSE) under local governments. They have capital of VND 302 trillion and around 460,000 employees. PSEs could be established by state (local) decisions or operate under the Enterprise Law. In the latter case there is a contractual relationship between the client PPC and the service organization, which determines the financial responsibilities as well. 54. From a purely legal point of view the SOEs and PSEs are financially independent from the Provincial People’s Committees; local authorities are therefore not explicitly responsible for the enterprise debt. However, in practice when the SOEs and PSEs face debt repayment, the responsible state agency that exercises ownership rights helps to coordinate the debt workout. At the local level PSEs are dependent on the PPCs. PPC decisions determine basic conditions of these public enterprises through the local tariff/user charges policies, co-financed capital investment programs by grants or through on-lending, and by influencing the PSE management. 55. Various forms of private sector participation will further complicate this relationship between local governments, as co-owners, clients and the service organizations, as partially equitized entities, operating under a service contract. Financial risk sharing should be also assessed in these cases. Only in the water management and sanitary services PSEs have accumulated debt from on-lent government external loans of USD 360 million (VND 7.56 trillion) by the end of 2012 (MOF data). Local SOEs borrowing for municipal infrastructure is substantial – the World Bank has estimated that of about USD 3.5 billion of infrastructure financing from SOCBs in 2007, more than half of these loans were made to SOEs, and more than one-quarter was for municipal infrastructure. 56. Banking sector stress caused by payment arrears to construction companies: Local authorities 175 in recent years have accumulated payment arrears to construction firms, which in turn has prevented the latter from servicing their debts and led to growing Non-Performing Loans in local banking systems.This is in part due to loose commitment controls, which led local authorities to enter into arrangements with civil works contractors without adequate budget authority. Contractors completed public works, but were not able compensated, which led to a build of arrears, peaking at close to US$4 billion in 2011. In 15 provinces, the volumes of arrears exceeded their annual capital budget.61 57. Policies have been adopted to impose more discipline on local capital budgets.62 Capital allocations since 2011 have only been agreed for ongoing projects, clearance of arrears, or ODA counterpart funding. New projects have only been agreed on very exceptional circumstances. These efforts have helped to bring arrears down to around US$2 billion as at the end of 2012. In addition to the pressures that these arrears have placed on the banking sector, there are also indications of overdue wage payments and an increasing number of workers are being laid off. 58. Managing contingent liabilities and fiscal risks: Actions to better identify the fiscal risks from contingent liabilities are critical, and should be seen within a broad set of key policy processes the government might consider in better managing contingent liabilities: n Creating a general policy for government exposure to contingent liabilities: this includes identifying the types of risk to cover, the circumstances in which a guarantee, rather than a loan or subsidy, is justified and who is best placed to manage risk. Guidelines should be prepared on what types of risk would be accepted by the government, minimum requirements for project support, risk-sharing and fees policies. n Ensuring budgetary transparency and discipline: explicit contingent liabilities should be identified, registered, and disclosed. The expected cost or maximum probable loss should also be quantified. Good practice is to include a provision within the annual budget to meet estimated sums falling due for payment in the year. Some of the costs and risk of CLs will be correlated: problems for one SOE will generate problems for another and this interaction can be analyzed within a portfolio approach. n Applying financial risk management: the main risks affecting CLs are similar to those affecting other  assets and direct liabilities (e.g., macroeconomic volatility). Consequently the techniques for quantifying CL risk can be similar to those used for estimating risk of other balance sheet items. Through the same processes, methodologies can be designed for pricing guarantees and charging risk premiums. 59 Map of capital arrears is from: http://tuoitre.vn/Kinh-te/567953/giat-minh-voi-91-000-ti-dong-no-dong.html 60 Directive 27 (October 2012) and Prime Minister’s Directive Number 14/CT-TTg (June 28, 2013) were issued to accelerate clearance of capital expenditure arrears and stem the accumulation of new arrears. INSTITUTIONAL REFORMS FOR LOCAL BORROWING 176 59. Revision to local borrowing provisions in SBL 2002: The types of legislation for subnational debt management vary across countries but usually consist of two components: (a) ex-ante fiscal rules for local authorities, stipulating purposes and types of and limits for debt instruments, and issuance procedures; and (b) ex-post debt restructuring in the event that local authorities become insolvent.63 The regulatory frameworks in many countries are evolving, and the pace of developing regulatory elements varies. 60. Firstly it is recommended that the overall fiscal rules related to local debt in Vietnam be restated. Under the SBL 2002, provinces are required to maintain a balanced budget. Meanwhile the SBL allows them to borrow more for capital spending. At present, sub-national borrowing is treated outside the balance sheet, and therefore not accounted for in the local budget balance. These conflicting provisions should be removed and local de-facto budget deficit should be included as part of the overall State Budget deficit. The aim of such a cap is to restrict borrowings only for investment purposes, which could be more appropriately captured by imposing the “golden rule” of keeping the current account in balance (current revenues to meet current expenditures) and borrowing undertaken only for capital expenditures. There should be no borrowings to meet current expenditures. 61. The existing regulations relating to subnational borrowings are based on quantitative numerical ceilings — which do not necessarily reflect the ability of the local authorities to repay debts. Comparing the debt stock with annual capital budget does not show the long term creditworthiness of a province or a city. There are also issues with related definition to and compliance with the ceilings. This ceiling seems to be hardly enforceable, as 13 PPCs were above the limit in 2011. 62. The ‘accumulated borrowing level’ is capped against the ‘annual capital budget’ which is in general unstable as highlighted in the analysis of capital budget credibility. The capital investment cost might also increase during the project implementation, because PPCs first usually want to “step in the door” with a lower cost estimate. The delays and carried forward expenditures make the indicator unreliable and less enforceable. The other component of the ratio, the total local government debt, does not measure the entire pool of fiscal obligations. Local governments also pledge the loans of their public service enterprises. These explicit guarantees and other contingent liabilities should be included in the total local government debt. 63. This ceiling on local government debt does not create proper incentives for the borrowers. Maximizing the debt stock in proportion to the annual capital expenditures does not support PPCs to balance their long term expenditure and revenue forecasts. The total amount of debt stock does not indicate the actual obligations for repaying the principal and the interest. It is vaguely related to the debt service due in a particular year. So PPCs cannot foresee their actual debt burden and more importantly it does not help to predict the debt obligations in the future. 64. For the above reasons it is recommended that current rules for capping sub-national borrowing be revamped, and debt stock thresholds that are more commonly used internationally are applied. These fiscal rules could be included in the revision to the SBL 2002 and/or future Debt Strategy. Based on the earlier analysis of fiscal and debt sustainability for HCMC and keeping in congruence the targets set in the national debt management strategy (see Box 6.3), suggestions for fiscal rules are given below. The focus is to design rules that are simple, workable and enforceable within the legislative framework of Vietnam and the infrastructure requirements of the provinces. It is suggested that differentiated rules be applied to several groups of provinces which are in line with assessment on their fiscal and debt situation. 63 Insolvency mechanisms help strengthening preventive rules by increasing the pain of circumventing ex ante rules for both lenders and borrowers. Box 6.3: Current debt management strategy in Vietnam 64 177 The current debt strategy for Vietnam - the Medium-term Debt Management Program for 2013-2015 (MTDMP) is prepared in the framework of the Long term National Strategy on Public Debt and National External Debt for the period 2011-2020. It is supported by a robust legal framework and defines the objectives in terms of lowering funding costs at acceptable risk. It assess the portfolio risks including the exposure to contingent liabilities such as guarantees and on-lending; and sets long term debt management targets such as reducing external debt to 40% from the current 60% and gradually lengthening ATM of domestic debt. It also sets the following targets: (i)  Public debt (including government debt, government guaranteed debt and local government debt) not to exceed 65 percent of GDP by 2015, of which government debt not to exceed 55 percent of GDP and external debt not to exceed 50 percent of GDP; (ii) Share of public external debt to GDP: 45% - 50%; (iii) Share of government external debt to GDP: 50 – 55%; (iv) Reduced market risks, credit risks and liquidity risks (no targets provided); (v)  Direct government debt service obligations (excluding on-lending obligations) as percentage of total annual state budget revenue not to exceed 25 percent; (vi) Annual external debt service obligations maintained at 25 percent of export value; Foreign exchange reserve as percentage of total short-term external debt to be above 200 percent. (vii)  Of the above targets that will impact rules at the provincial level are (i), (ii), (iii) and (vi) given that funds are on-lent to provinces. 65. Thresholds on debt stock and debt service: it is suggested that subnational debt be controlled by two types of thresholds, similar to Vietnam’s national level debt thresholds (box 6.3) and widely used in other countries: (i) debt service and total outstanding debt stock. Debt service could be monitored against decentralized revenue (i.e. 100 percent local revenue and shared revenue, excluding capital or extraordinary revenue such as land sales). This is a good liquidity indicator of the repayment capacity of the province and also helps to understand the fiscal space available for the province to meet other expenditures after debt servicing. Cross country experiences suggest a ratio of interest payments against own revenues of about 10–15 percent, leaving fiscal room for other expenditures. Debt stock could be taken as a ratio of local (or national) GDP to measure solvency. 66. Debt thresholds however should be complemented by analysis of other indicators that can point to potential fiscal vulnerabilities and risks. Compliance with debt limits in a particular year does not automatically ensure that a local authority will be able to repay its debt in the long-term. Therefore closer analysis is needed on the evolution of fiscal aggregates such as for example the local authorities’ ability to meet recurrent spending from decentralized revenue; the share of interest payments in recurrent spending; the share of non- discretionary items (e.g. debt servicing, wages and salaries) in recurrent spending; and the level of fiscal effort (e.g. primary balance or fiscal balance excluding interest payments). 64 See “Medium Term Debt Management Program 2013-2015” 67. Comprehensiveness and transparency monitoring and reporting local debt: The successful 178 implementation of any rules requires transparency and reporting arrangements, with a review on frequent basis along with appropriate corrective measures. It is important to also undertake an independent evaluation of performance of the prescribed fiscal indicators (as is being conducted by the SAV). In addition there should be exclusion clauses if the province cannot meet the rules on account of any exogenous factors, natural calamities and other exceptional grounds to be specified by the central government. 68. An important consideration is the monitoring of contingent or off-budget borrowing. It is recommended that borrowings by local authorities for cost-recovery projects should be strictly included under the thresholds proposed above. Also, at present, the public debt management law provides for on-lending to local governments to be excluded from total local debt. However, it is recommended that on-lending should also be monitored under the proposed thresholds. 69. As the main sources of contingent liabilities, borrowings of Public Services Enterprises (PSEs) and other SOEs, particularly for projects to provide public services, and of Local Development Investment Funds (LDIFs) should be more closely monitored. It is important that periodic reports on total portfolio risks are presented to the PPCs and to the central government. If such steps are not taken, then defaults could well siphon off budgetary resources and upset the enforcement of fiscal rules. International good practice suggests monitoring of SOE debt under the threshold. A the same time, short-term borrowing from Treasury balances can be excluded from total borrowing, as this is more of a short-term cash management practice than accumulation of public debt. Table 6.4: Coverage of public debt (Vietnam and IMF definitions) Coverage Vietnam IMF 1. Government debt a) Domestic debt √ √ - T-bills, T-bonds √ √ - Borrowing from SIF, SCIC √ χ - Borrowing from Treasury balances, the Sinking Fund √ √ b) External debt (incl, the borrowing for on-lending to SOEs) √ √ 2. Government guaran teed debt (incl, SOEs') a) Domestic debt √ √ b) External debt √ √ 3. Sub-national government debt - Sub-national/municipal bonds √ √ - Borrowings from VDB - Borrowings from Treasury balances √ χ - On-lending loans of Government's external borrowings √ √ - Others √ √ 4. SOE debt under self-borrowing self-repayment mechanism χ √ 70. Although local authorities are required to follow a large number of steps and regulations to contract debt, there is little formal requirement on reporting and disclosure of local debt. PPCs report on bond issuance to 179 the Banking and Finance Department at the Ministry of Finance, which in turn prepares quarterly reports for the Minister. But these reports are not regularly shared with the Debt Management and Budget Departments at the Ministry of Finance. The feedback from the Ministry of Finance is that local authorities’ reports on local borrowing are often late, not detailed enough (e.g. on the terms of actual bond issuance), and sometimes not even submitted. 71. There is a national Public Debt Bulletin that is disclosed on semi-annual basis, though with a six months to one year time lag. Thanks to the implementation of the Debt Management and Financial Analysis System (DMFAS) for domestic debt management, the Ministry of Finance has recently been able to produce, for the first time, a consolidated Bulletin including both domestic and external government and government guaranteed debts. However, information on sub-national debt is still missing. Given that the data is available, the Bulletin should ideally include this information, among other things, to provide a comprehensive picture on the total public debt. Currently the DMFAS system does not cover sub-national debt. One possible way to collect subnational debt information from local authorities is to develop an interactive Web Portal, which requires local authorities to submit consistent and standardized information on debt flows, stocks, and terms and conditions. 72. Increased transparency is important not only for the government’s internal management purpose but also for private investors to take timely and adequate decisions. Currently investors have little information on the financial positions of the local authorities, making it hard for them to assess the creditworthiness of both local authorities. In terms of sound international practice, local authorities should meet all its statutory or contractual reporting obligations relating to both its debt and loan guarantees to external parties. This will normally include reporting to the central government, stock exchanges and market supervisory bodies, and foreign regulatory authorities, where applicable. 73. For the more advanced provinces, the need for transparency is particularly important. These are provinces like HCMC that plan to go increasingly to debt markets for infrastructure financing. Secondly, SOEs in these provinces are also turning increasingly to the market on account of falling budgetary subsidies. As discussed earlier, these could be a potential source of contingent liabilities. This should require them to publish detailed information on their assets, liabilities and financial and operational performance to inform the market and credit rating entities. 74. The requirement for greater transparency and disclosure should apply across all provinces, but in terms of sequencing, the more advanced provinces should already start preparing debt statistical bulletin covering its domestic and external (on-lent) debt. In parallel, other provinces should also gear up to do the same. The bulletin should be published at least annually and provide information on the provinces’ debt stocks (by creditor, instrument, currency, interest rate basis, and residual maturity); debt flows (principal and interest payments); debt ratios and indicators; and basic risk measures of the debt portfolio. The bulletin should cover the share of fixed rate to floating rate debt, share of short-term to long-term debt, currency composition of debt, average maturity of the debt and the maturity profile of the debt. 75. Evolving local financing landscape and moving to a two-tier local debt system: A number of changes are expected over the medium-term on access to local debt. The introduction of a Treasury Single Account (TSA) will tighten or gradually eliminate borrowing from the idle cash at local State Treasury offices, which currently accounts for almost 30 percent of subnational debt. Instead the TSA will consolidate information across local State Treasury offices and have better oversight on cash flows and balances to be able to redirect resources from surplus to deficit areas. The ODA landscape is also changing where, at the central level, the share of grants and low-cost resources will be replaced by less concessional sources as Vietnam continues its transition to middle income country status. Although most of the ODA transfer to local authorities to date has been in the form of grants, the share of on-lending is expected to gradually increase. 76. Given these developments, it is recommended that the central government consider a parallel approach 180 between provinces that already have market access and others that are looking to access debt financing. The approach suggests central government will need to facilitate appropriate ‘enabling environment’ to ensure smooth supply of loanable funds for the big cities with market access while keeping more simple ‘on-lending’ options for those dependent on central transfers. The parallel approach should maximize flexibility and optimize the channels of bond financing.65 77. Subnational debt in more advanced provinces: So far only the largest PPCs are able to issue bonds; the costs of raising debt on the domestic market is high compared to ODA on-lending. For provinces that have market access (HCMC, Hanoi, Danang and Haiphong) it would be desirable to adopt a more performance based approach with enhanced transparency and credibility to facilitate market based borrowings. For effective bond financing, the provinces will need effective debt management, voluntary disclosure norms and practices, transparency, reporting, accountability and audits, credit ratings, credit assessments of projects, and adequate debt data recording systems. 78. The Public Debt Development Strategy envisages a shift towards increased domestic debt in the national debt portfolio. In 2013, the Prime Minister issued Decision 261 approving a Roadmap for Bond Market Development. This Roadmap has outlined the targets for both government (national and municipal) and corporate bond market development, setting priorities to address the market challenges. The Roadmap called for coordinated actions by stakeholders, both government agencies and private stakeholders. These include units within the MOF (i.e. Banking and Finance Department, Debt Management Department), the State Bank of Vietnam , the State Treasury (which plays the role of government bond issuer, thus being responsible for conducting the issuance processes), the Hanoi Stock Exchange (which provide a secondary market platform for bonds, and where government bond trades must be reported), the Vietnam Securities Depository (where settlement for bond trades take place), and Vietnam Bond Market Association (which not only plays a significant role in improving standards of market practices, but is also tasked with the establishment of bond market information center). 79. The reform agenda is ambitious and will require careful stakeholder planning a strong champion to drive the process from the Ministry of Finance. The planned reforms include modernizing bond trading infrastructure that will provide a connection between Hanoi Stock Exchange and the Bloomberg trading system. Reforms will also address the bond custody and settlement that will be transferred from commercial banks to the State Bank of Vietnam to reduce settlement risk and achieve a full delivery versus payments (DvP) and will also look at the prospect to develop a central counterparty that will help secondary market transactions. 80. Although, the reforms are aimed at the national government and corporate bond markets and there are no separate action markers for developing the municipal bond market, the reforms will aim to set in standards that will benefit the municipal bond market too. Specific aspects such developing market infrastructure, custody and settlements systems with DvP, appropriate legal framework and the diversification of investor base through primary dealers will have beneficial and complimentary effect in the municipal bond market too. Moreover, banks holdings of municipal bonds are allowed for meeting their mandatory liquidity requirements. This coupled with the developments in market infrastructure will impact positively on the municipal bond market too. 81. On specific targets the Roadmap aims that the value of the municipal bonds reaches 1 percent of GDP by 2020 (currently it is about 0.25 percent), while the total value of bonds outstanding (including government66 and corporate) is aimed to 38 percent in 2020. It aims to increase the lending maturities of the bonds to 6-8 65 A similar approach is adopted in India where certain states are classified as “special category states” on account of their terrain, geographical location, etc. These states are given higher grants and fewer loans, and also depend more on central transfers to meet budgetary and developmental expenditures rather than through market access. 66 Of which the value of the Government bonds is aimed to account for 22 per cent of GDP years for 2016-20 from the average of 4-6 years now and also increase in the average bond trading volumes. In volume terms the increase anticipated in municipal bonds is, however, not large but it needs to be realized 181 that these will only be concentrated for the provinces of HCMC, Hanoi and some other advanced provinces. 82. Subnational debt in less advanced provinces: For provinces dependent significantly on transfers and grants from the center, the approach will be more focused on on-lending, and on-grants to lower transaction costs and maximize the lower cost of borrowings from the center. For using the ODA funds more effectively regulatory changes are needed. Regulations should establish a rules-based and transparent system both for grant allocation and for on-lending, which would also promote risk management and debt sustainability. A transparent planning and regulatory framework should open up equal opportunities for the provinces in accessing additional concessional funds for development. Greater local discretion in management of ODA will catalyze more effective and strategic use of available resources and improve public investment efficiency. 83. The first step of central government is therefore to set priorities for allocation of ODA to local governments. Local ODA grants and on-lending mechanisms should be designed within the framework of national strategies and objectives. The socio-economic development plan specifies the national development priorities: the preferred sectors (e.g. urban development, infrastructure, and poverty reduction), the priority regions (Highlands vs. Coast) or the supported activities. There should be several criteria for decision-making on allocation of ODA as grant versus loan financing. These may include by beneficiaries (groups of provinces), by program areas (ODA grants would be higher in the preferred service areas such as social and poverty reduction vs. more on-lending in infrastructure areas). 84. The PPCs could be grouped on the basis of a combination of objective indicators of economic development (e.g. GDP per capita) and local budget capacity (net contributor/recipient to central budget), or even demographic characteristics (e.g. ratio of urban population) and geographic conditions (e.g. mountain or remote areas). These ODA funding regulations should be set for a longer period, so the applicant PPCs are able to plan ahead their development projects. 85. Given the prevailing use of ODA as grants to local governments, it is important to make on-lending more appealing to PPCs. Through the co-financing requirements fiscal incentives can be created for borrowing and raising supplementary funds by local government applying for ODA grants. Central government might even set on-lending as the default mechanism for ODA to local governments, and in this case ODA grants should be considered as supplementary sources for preferred service areas or groups of provinces, and/or for well- performing PPCs. Also, grant allocation should be more outputs-based, where provinces receiving grants should be obliged to monitor and report in higher details on the outcomes of the supported activities. Further strengthening of the hard budget constrain in capital investments will also make PPCs more inclined to use ODA loans. 86. There should also be consideration to review the regulation on on-lending terms. At present, the Public Debt Management Law and Decree 78 stipulate that ODA be on-lent to local authorities on the same terms. Meanwhile, international practices suggest that (i) local authorities’ exposure to exchange rate risk should be limited, and where there is exposure they should receive support on how to manage it; (ii) On-lending charges might cover the credit risk and any other administrative costs, connected to the intermediary organization, such as banking charges, loan application study and project supervision costs67 . Assessment of credit risk will be instrumental in the adequate pricing of on-lending and one policy option is to charge different prices for borrowers with different credit risks. The charges, if any, should not make ODA loans untenable for local governments, which would defeat the purpose of local ODA on-lending. As debt management capacity at central and local levels is further strengthened, it might even be considered that the maturity of ODA on- lending be different to that of the original borrowing from the development financial institutions at the central level for maturity matching purposes. 67 They would require regulatory changes, as on-lending fees are not allowed under the present ODA regulations. CONCLUSIONS AND RECOMMENDATIONS 182 88. Access to debt financing is becoming an increasingly important topic particularly for local authorities that are facing an infrastructure financing deficit including because of higher contributions of shared revenue to central authorities for redistribution across provinces. This chapter finds that the current ceilings in the SBL 2002 are not an effective means of either limiting borrowing levels or linking debt financing to borrowing capacity. A case study debt sustainability analysis of HCMC illustrates that the province could potentially borrow more than the limits prescribed in the SBL 2002, but that without more access to decentralized revenue (i.e. increased revenue autonomy), more borrowing would lead to liquidity pressures. Aside from direct liabilities, local authorities should also monitor indirect liabilities more closely. The chapter recommends to: (i) restate fiscal rules related to local debt in the SBL 2002 and adopt more standard qualitative fiscal rules in the revised law; (ii) adopt specific thresholds on local debt stock and debt servicing in line with debt sustainability targets; (iii) improve transparency and reporting on public debt, including through statistical debt bulletins in more advanced provinces; (iv) transition to a two-tier local debt system with central government facilitating more market access for advanced provinces and greater ODA on-lending for less advanced provinces. Issues Recommendations Conflicting rules in SBL 2002, requiring Revising rules in SBL 2002 to require current provinces to maintain balanced budget while account balance only, and account subnational allowing them to borrow for capital spending. budget deficit in the overall national budget deficit. State Budget Law 2002 Revamping the thresholds to better monitor both Existing quantitative ceiling of subnational debt debt service (e.g. debt service/local revenues) does not reflect provinces’ debt sustainability, and debt stock (e.g. outstanding debt stock/local solvency or repayment capacity. budget). Considering balancing the operational and capital budgets in the longer-run. Including borrowings for cost-recovery projects and on-lending in the total subnational debt amount. Concerns over “off-budget” borrowings Monitoring closely borrowings of SOEs and Local Development Investment Funds. Including subnational debt data in the national Public Debt Bulletin. Considering interactive way to collect subnational debt information from LGs. Missing subnational debt data in the national Encouraging frontier provinces to develop debt Public Debt Bulletin. No provincial public debt Strategy for strengthening debt statistical bulletin covering both domestic and bulletins despite interest of potential investors. external (on-lent) debt and other fiscal information management capacity that is of interest to credit rating agencies and potential investors. At the central level, developing comprehensive methods for local borrowing capacity assessment, Limited capacity in public debt management at and strengthening MOF capacity. At the local level, both central and local governments. strengthening DOFs capacity to develop medium- term fiscal plan and public debt management program, and closely monitor contingent liabilities. Central government facilitating appropriate “enabling environment” to ensure smooth supply Evolving fiscal landscape and debt profile at of loanable funds for the frontier provinces, while both central and local levels. promoting ODA on-lending in net budget recipient provinces. REFERENCES 183 Adam Fforde and Associates Pty Ltd (2003): Decentralization on Vietnam / working effectively at provincial and local government level. AusAID Afonso, A, Romero, A, Monsalve, E, “Public Sector Efficiency: Evidence for Latin America,” (May 2013) Afonso, A., & Fernandes, S. (2008). Assessing and explaining the relative efficiency of local government. The Journal of Socio-Economics, 37, 1946–1979. Albrecht, David – Hocquard, Hervé – Papinu, Philippe, (2010): Urban Development in Vietnam: the Rise of Local Authorities. AFD, Paris Alm, J., & Torgler, B. (2006) “Culture differences and tax morale in the United States and in Europe” Journal of Economic Psychology, 27(2), 224–246 Bahl, R, and Martinez-Vasquez, J, “Sequencing Fiscal Decentralization,” Annals of Economics and Finance 14-2 (B) (2013) Banker, R. D., Charnes, A., & Cooper, W. W. (1984, September ). Some Models for Estimating Technical and Scale Inefficiencies in Data Envelopment Analysis. Management Science, 30(9), 1078-1092. Bjornestad, L, “Fiscal Decentralization, Fiscal Incentives, and Pro Poor Outcomes: Evidence from Vietnam,” ADB Economics Working Paper Series No. 168 (July 2009) Boadway, Robin (1992). The Constitutional Division of Powers : An Economic Perspective. Ottawa: Economic Council of Canada. Boadway, Robin and Anwar Shah (2009). Fiscal Federalism. Principles and the Practice of Multi-order Governance. London and New York: Cambridge University Press Boetti, L., Piacenza, M., & Turati, G. (2010). Decentralization and Local Governments’ Performance: How Does Fiscal Autonomy Affect Spending Efficiency? 66th Congress of the International Institute of Public Finance (IIPF). Uppsala, Sweden. Buchanan, J. & Allen, C. (2000) ‘The Growth of Contractors in the Construction Industry: Implications for Tax Revenue’, Economic and Labour Relations Review, 11(1), pp.46–75. Canuto, Otaviano-Liu, Lili (Editors), (2013). Until debt do us part: subnational debt, insolvency, and markets. IBRD/The World Bank, Washington, DC, doi: 10.1596/978-0-8213-9766-4 CECODES, VFF-CRT, UNDP, “The Vietnam Provincial Governance and Public Administration Performance Index PAPI 2013: Measuring Citizens’ Experiences,” (2013) Charnes, A., Cooper, W. W., & Rhodes, E. (1978, November). Measuring the efficiency of decision making units. European Journal of Operational Research, 2(6), 429-444. Cyan, M., J. Martinez-Vazquez and V. Vulovic (2014) “New Approaches to Measuring Tax Effort” in Taxation and Development: The Weakest Link? R. Bird and J. Martinez-Vazquez (eds.) Cheltham: Edward Elgar. Damjanovic, Dusan.-Pavlovic-Krizanic, Tatjana- Péteri, Gábor (Editors), (2010): Public-Private Partnership. 184 Successes and failures in Central and Southern Eastern Europe. (Co-editor with) PALGO Center, Beograd Davey, Kenneth (1988): Municipal development funds and intermediaries. WPS 32, The World Bank Dexia (2008): Sub-national governments in the European Union. Dexia Editions, Paris Ebel, D.R-Péteri, Gábor (Editors), (2007): Kosovo Decentralization Briefing Book Forum 2015/KFOS/LGI, Prishtina Eyraud, L, and Lusinya, L, “Vertical Fiscal Imbalances and Fiscal Performance in Advanced Economies,” International Monetary Fund (October 17, 2012 draft) Farvacque-Vitkovic, Catherine-Kopanyi, Mihaly (Editors) (2013): Municipal finance. A handbook for local governments. The World Bank Institute, Washington, DC Fedelino, A, and Ter-Minasian, T, “Making Fiscal Decentralization Work: Cross-Country Experiences,” International Monetary Fund, Occasional Paper 271 (2010) Freire, Maria Emilia (2013): Managing external resources. in: Farvacque-Vitkovic, Catherine-Kopanyi, Mihaly (Editors) (2013): Municipal finance: A handbook for local governments. The World Bank Institute, Washington, DC Fukuda-Parr, S, Guyer, P, Lawson-Remer, T, “Does Budget Transparency Lead to Stronger Human Development Outcomes and Commitments to Economic and Social Rights?” IBP Working Paper, December 2011 Garza, Homero Garza and Sethi, Nitika, “Fiscal Federalism in Mexico: Recommendations to Move the Agenda Forward”, (May 2013). Gemmel, N, Kneller, R, and Sanz, I, “Fiscal Decentralization and Economic Growth: Spending Decentralization versus Revenue Decentralization,” Economic Inquiry, Vol 51, No 4 (October 2013) Government of Vietnam and the World Bank, “Vietnam: Managing Public Expenditure for Poverty Reduction and Growth – Public Expenditure Review and Integrated Fiduciary Assessment,” (April 2005) Government of Vietnam and the World Bank, “Vietnam: Managing Public Resources Better: Public Expenditure Review 2000,” (December 13, 2000) Government of Vietnam, “Vietnam Public Expenditure and Financial Accountability: Public Financial Management Performance Assessment 2012,” (September 2013) Government of Vietnam, Prime Minister’s Decision No.59/2010/QD-TTg on “Promulgation of the allocation norm for the allocation of State Budget recurrent expenditure in 2011.” Hormazábal, R. S., & Wyngard, H. R. (2007). Frontier methodologies for the determination of efficiencies in distribution costs. Ingeniare. Revista chilena de ingeniería, 15(3), 220-226. Huang, J.-T., K.-T. Lo, and P.-W. She. (2012) "The impact of fiscal decentralization on tax effort of China's local governments after tax sharing system " The Singapore Economic Review 57 (01), 2012.1250005. International Monetary Fund, Vietnam Article IV Reports 185 Jokay, Charles (2011): Framework for Sub-Sovereign Debt in Macedonia. Issues Paper: Municipal Services Improvement Project (MSIP). Skopje Khagram, S, Fung, A, De Renzio, P, “Open Budgets: The Political Economy of Transparency, Participation and Accountability,” Brookings Institution Press (2013) Kokkinou, A. (2009). Stochastic frontier analysis: Empirical evidence on Greek productivity. Glasgow, United Kingdom: University of Glasgow, United Kingdom. Lee, C., & Ji, Y.-b. (2010). Data Envelopment Analysis. The Stata Journal, 10(2), 267-280. Light, M, Nguyen, NA, “Impact of CIT and VAT Reforms: Results from the Vietnam Tax Impact Model (VTIM),” in a TA assignment funded by the European Union (February 2013) Liu, Lili - Pradelli, Juan (2012): Financing Infrastructure and Monitoring Fiscal Risks at the Subnational Level. The World Bank, WPS 6069 Liu, Lili- Skrok, Emilia (2013): Poland, Managing Subnational Debt Sustainability. The World Bank, Draft manuscript Malesky, E, Nguyen, CV, Tran, A, “The Impact of Recentralization on Public Services: A Difference-in- Difference Analysis of the Abolition of Elected Councils in Vietnam,” (2013) Martinez-Vasquez, Jorge (2005). Making Fiscal Decentralization Work in Vietnam. Working Paper 05-13. Georgia State University. Martinez-Vazquez, J, “Making Fiscal Decentralization Work in Vietnam,” Andrew Young School of Policy Studies, Georgia State University, Working Paper 05-13 (June 2005) Martinez-Vazquez, J, and McNab, RM, (2003) “Fiscal Decentralization and Economic Growth,” in World Development, 31 (9), p. 1597-1616 Martinez-Vazquez, J, and McNab, RM, (2003) “Fiscal Decentralization and Economic Growth,” in World Development, 31 (9), p. 1597-1616 Martinez-Vazquez, J, and Nguyen, VM, “Tax Reform and Revenue Assignments in Vietnam,” in Shukla, GP, Pham, MD, Engelschalk, M, and Le, TM (eds.), “Tax Reform in Vietnam: Toward a More Efficient and Equitable System,” (September 2011) Martinez-Vazquez, J, Qiao, B, and Zhang, L, “The Role of Provincial Policies in Fiscal Equalization Outcomes in China,” The China Review, Vol 8, No 2 (Fall 2008) Martinez-Vazquez, J, Qiao, B, and Zhang, L, “The Role of Provincial Policies in Fiscal Equalization Outcomes in China,” The China Review, Vol 8, No 2 (Fall 2008) Mkandawire, T. (2010) "On Tax Efforts and Colonial Heritage in Africa." The Journal of Development Studies 46 (10), 2010.1647-1669. Murillo-Zamorano, L. R. (2004, February). Economic Efficiency and Frontier Techniques. Journal of Economic Surveys, 18(1), 33-77. NALAS (2011): Guidelines On Local Government Borrowing And Recent Developments In NALAS Countries. 186 Skopje National Assembly (2002). The State Budget Law 2002. National Assembly (2013). The Constitution of the Socialist Republic of Vietnam. Adopted by the National Assembly on 28 November 2013. (English translation by IDEA) . Nguyen, Hoang-Phuong, “What is in it for the poor? Evidence from Fiscal Decentralization in Vietnam,” Journal of Public and International Affairs No. 19 (June 2008): pp. 69-90 Nguyen Thi Hong, Yen (2013): Evaluation of local ODA. Project background paper. Ha Noi Nguyen Thi Thanh, Ha (2013): Capital investment planning and implementation process. Project background paper. Ha Noi Nguyen Thi Thanh, Ha (2013a): Provincial government contingent liabilities. Project background paper. Ha Noi Nguyen, Van Chinh, and Vu, Quang Viet, “Vietnam economy in the years of reform” (2002). Nguyen-Hoang, P, Schroeder, L, ”An Analysis of Quasi-Decentralized Budgeting in Vietnam,” in International Journal of Public Administration (October 20, 2010) Oates , Wallace (1972). Fiscal Federalism. New York: Harcourt Brace Jovanovich. Oates, W.E., “Fiscal Federalism,” New York: Harcourt Brace Jovanovich (1972) Palackova, Hana (1998): Contingent government liabilities. The World Bank PRWP. Péteri, Gábor (2013): Fiscal relations between municipal companies and local governments. Manuscript, Prepared for the World Bank Péteri, Gábor (2004): Municipal debt control: experiences of Central-Eastern European countries. Report for Club “Economika 2000”, PHRD Grant # TF051882. Sofia Peterson, George E.(1996) : Using municipal development funds to build municipal credit markets. Manuscript, Government of India, The World Bank Prime Minister of the SRV. (2010). Decision 59. Promulgation of the Allocation Norms for the Allocation of State Budget Recurrent Expenditure in 2011. Prime Minister of the SRV. (2010). Decision 60. Promulgation of principles, criteria and norms for the allocation of development capital from the State Budget Source of Funds over the period 2011-2015 . Public debt strategy (2012): Approving the public debt and national foreign debt strategy in the period 2011- 2010 and the orientation towards 2030. Decision of the Prime Minister No. 958/QD-TTg Qiao, B., L. Liu, J. Zhao, L. Zhang, and Y. Xu. (2009) “Report on public finance in China: tax revenue and tax capacity Beijing, China” Financial & Economic Publishing House. 2009. Ramanathan, R. (2003). An Introduction to Data Envelopment Analysis: A tool for Performance Measurement. New Delhi, Thousand Oaks, London: Sage Publications 187 Shah, Anwar (1994). The Reform of Intergovernmental Fiscal Relations in Developing and Emerging Market Economies, 1994 Policy and Research Series 23, 1994, Washington, D.C.: World Bank. Shah, Anwar and Sana Shah (2006). The New Vision of Local Governance and the Evolving Roles of Local Governments. In Local Governance in Developing Countries, edited by Anwar Shah, 2006, Chapter 1:1-46, Washington, DC: World Bank. Shah, Anwar: “Intergovernmental fiscal transfers – principles and practice.” 2007 Standing Committee of National Assembly, “Ordinance on Exercise of Democracy in Communes, Wards, and Townships,” No 34/2007/PL-UBTVQH11 (April 20, 2007) Swianiewicz, Pawel (Editor) (2004): Risks and rewards: local government borrowing in Central Europe. OSI/ LGI, Budapest, Tanzi, V.(1987) “Quantitative characteristics of the tax systems of developing countries.” The theory of taxation for developing countries, Oxford University Press for the World Bank, D. Newberry, N.H. Stern (Eds.), New York (1987) The World Bank, “Vietnam Fiscal Decentralization and Delivery of Rural Services: An Economic Report,” October 31, 1996 The World Bank, “Vietnam Fiscal Transparency Review: Analysis and Stakeholder Feedback on State Budget Information in the Public Domain,” (May 2013) Torgler, B. (2004) “Tax morale in Asian countries” Journal of Asian Economics, 15(2), 237–266. Town Development Fund (2011): Main Report and Business Plan, GIZ-KfW-ADB-PPIAF/World Bank, Nepal Trinh Hong-Loan - McCluskey, William J. (2012): Property Tax Reform in Vietnam: A Work in Progress. IMFG Papers on Municipal Finance and Governance. University of Toronto, Toronto Uchimura, H, and Kono H, “Intergovernmental Fiscal Relationships in Vietnam after the 2002 State Budget Law: The Center-Province and the Province-District/Commune Dimensions,” in “Fiscal Decentralization and Development: Experience of Three Developing Countries in South East Asia,” Uchimura, H (ed.). Institute of Development Economies, JETRO (2012) Vo, DH, “Fiscal Decentralization in Vietnam: A Preliminary Investigation,” School of Economics and Commerce, The University of Western Australia (March, 2005) Vu, Thanh Thuy- Zouikri,Messaoud, (2011): Decentralization and sub-national governance. Manuscript, Paris X University Wheeler, Graeme (2004): Sound practice in government debt management. IBRD/The World Bank, Washington, DC World Bank (2011). Vietnam Urbanization Review. Technical Assistance Report. Hanoi. World Bank (2012), Vietnam Development Report: Market Economy for a Middle Income Vietnam World Bank (2013), “Water Supply and Sanitation in Vietnam. Service Delivery Assessment. Turning Finance 188 into Services for the Future.” Draft 31 October 2013. World Bank (2013). Socialist Republic of Vietnam. Irrigated Agriculture and Irrigation Systems Management Reforms. Report no. AC54619, 23 September 2013.EASVS , East Asia and Pacific Region. World Bank (2013): Assessment of the financing framework for municipal infrastructure in Vietnam Urbanization Review. World Bank (2013): Assessment of the Financing Framework for Municipal Infrastructure in Vietnam. Final Report World Bank, “Vietnam Fiscal Transparency Review: Analysis and Stakeholder Feedback on State Budget Transparency” (2013) World Bank, “Vietnam: Review of State Budget Law, 2002” (April 2014) Wu, A and Wang, W, (2013) “Determinants of Expenditure Decentralization: Evidence from China,” in World Development, Vol 46, p.176-184 Yang, L., Z. Xiaojing, C. Xin, T. Duoduo, and L. Cheng, 2012, “China’s Sovereign Balance Sheet and Its Risk Assessment,” Economic Research Journal, Vol. 47, No.6, (June). Zhang, Yuanyan Sophia Zhang and Barnett, Steven (2014): Fiscal Vulnerabilities and Risks from Local Government Finance in China. IMF Working Paper W/14/4 Annex A: Input and output/outcome variables used in expenditure efficiency analysis 189 Input Outputs 68 • Graduates as % of total candidates (upper secondary) (GSO) • Pupils per teacher (pre-school, primary, lower secondary and upper secondary) (GSO) Total education  ross enrolment (primary, lower secondary) as a share of school going population (GSO • G recurrent and FDA) expenditure • Mathematics attainment (grades 1-5, % excellent) (FDA) assigned to local • Vietnamese attainment (grades 1-5, % excellent) (FDA) authorities (2011) • Grades 1-5 repetition rate • Percentage of classrooms in good condition (FDA) • Teachers with 12+2 qualifications (FDA) • Infant mortality rate (HSY) • Share of children under 5 underweight (HSY) Total health recurrent • Share of children under 5 stunted (HSY) expenditure • Immunization coverage for children under 1 (fully vaccinated) (HSY) assigned to local • Pre-natal care rate (HSY) authorities (2011) • Births attended by midwives (HSY) • Post-natal care rate (HSY) • In-patient/Out-patient rates (normalized by population) (HSY) • Ratio of passengers traveling by road to total passengers (GSO) Total road • Number of passengers per km of road (GSO) construction • Ratio of freight traveling by road to total freight travel (GSO) recurrent • Weight of freight per km of road (GSO) expenditure • % of schools that have car access (FDA) assigned to local • Percentage of roads in provincially managed roads that are paved with asphalt (PCI) authorities (2011) • Assessment of road quality (% good or very good) (PCI) Local approval Local councils are required to approve borrowing for individual projects 68 The output and outcome data are taken from: (i) Government Statistics Office (GSO) website; (ii) Health Statistics Yearbook (2011); (iii) the education sector FDA dataset; and (iv) the provincial competitiveness index Annex B: 190 Background on the non-parametric Data Envelopment Analysis (DEA) The analysis of the productive or technical efficiency of provinces in Vietnam and factors that may potentially determine the efficiency levels and thus define the optimal characteristics of an efficient provision of local public services, relies on a non-parametric frontier technique. The fundamental idea behind a frontier approach is to calculate or estimate an efficient frontier function against which the performance of decision making units (DMUs), in this case, provinces in Vietnam, is measured. A production frontier reflects the maximum output attainable by a given sets of inputs and existing production technologies. Technical efficiency is defined in terms of a maximum level of output produced from a given set of inputs (output orientation) or of a minimum amount of inputs used to produce a given level of output (input orientation)69 (Kokkinou 2009). In the literature, the measurement of the relative productive efficiency of a set of DMUs in general requires a variety of feasible and identical inputs and identical outputs used by these DMUs. In general, the best-performing DMUs in the set would have an efficiency score of one while the performance of others DMUs would vary between zero and one relative to this best performance (Murillo-Zamorano 2004, Ramanathan 2003).70 In the non-parametric approach or the Data Envelopment Analysis (DEA), the functional form of the efficient production frontier is not pre-established, that is, the functional relationship between inputs and outputs (or the production function) needs not to be predefined, but is calculated via a mathematical programming model or an econometric technique applied to a sample of observed data. From this sample of observed data, a frontier envelopment surface is defined and the DMUs that lie on that surface are termed productively efficient and are assigned a value or an efficiency score of one. Alternatively, the DMUs that do not lie on that surface are considered as productively inefficient and an inefficiency score of less than one will then be calculated for each one of them (Murillo-Zamorano 2004, Ramanathan 2003). Unlike parametric stochastic frontier techniques which allow for a clear distinction between the effects of random noise and the effects of inefficiency, the DEA technique does not take into consideration the usual random noise or rather considers random noise as part of the efficiency score, assuming that all deviations from the efficient production frontier are under the control of the DMU. The main disadvantage of the non-parametric Data Envelopment Analysis most cited in the literate is this above-mentioned deterministic nature. Additionally, the DEA technique compares how well a DMU is doing compared to other DMUs in the set but not compared to a theoretical maximum. Boetti, Piacenza and Turati (2010), Kokkinou (2009), Afonso and Fernandes (2008), and Murillo-Zamorano (2004) provide a comprehensive overview of the mathematical foundations of the non-parametric Data Envelopment Analysis (DEA) used in the present study. Initially, the analysis will focus on estimating the level of productive efficiency for 63 provinces in Vietnam using the DEA technique. Subsequently, efficiency scores are explained in a second stage analysis by means of a censored normal Tobit regression model. 69 With the input orientation of an efficiency measure, an inefficient DMU is one that can produce the same level of output as an efficient DMU with less input. With the output orientation of an efficiency measure, an inefficient DMU is one that, compare to an efficient DMU, produces a lower level of output using the same amount of inputs (Financial and Fiscal Commission 2011). 70 The efficiency scores estimated will be relative to the best performing or the most efficient decision-making unit (s) (Ramanathan 2003). The use of the DEA as mathematical linear programming technique for measuring the performance efficiency of DMUs originated from Charnes, Cooper and Rhodes (1978). These authors defined the 191 performance efficiency of a DMU among a set of DMUs as the ratio of the linear weighted sum of outputs to the linear weighted sum of inputs (all inputs and outputs being positive), where the weights for each DMU are determined by means of mathematical programming with the assumption of constant returns to scale (CRS) and subject to the condition that the efficiency of every DMU be less than or equal to unity (i.e. no DMU can operate beyond the production possibility set) and that the weights are nonnegative (Ramanathan 2003, Lee and Ji 2010). This is called the CCR/CRS DEA model; with CCR standing for Charnes, Cooper and Rhodes (1978). The basic linear programming technique they proposed to envelop observed input-output data is as follows: Assuming there are n DMUs (in this case, the DMUs are the municipalities), each one with m inputs and s outputs, the efficiency score of DMU0 relative to other DMUs is obtained by solving the following linear programming model: Min θ0 Subject to: θ0 xi0-∑j=1 λj xij ≥0; i=1,…,m (1) -yr0+∑j=1 λj xrj ≥0; r=1,…,s λj ≥0; j=1,…,n Where θ0 is the technical efficiency score of DMU0 with inputs xi0 and outputs yr0 that satisfies θ0 ≤1, which measures the distance between a DMU and the efficient frontier. A DMU or province is efficient when =1 , which means that the municipality is located on the frontier. On the other hand, a provinve is inefficient when θ<1, which means that it is located inside the frontier. The first constraint states that weighted combination of inputs of all DMUs must be less than or equal to the input for the reference DMU0 multiplied by its efficiency. The second constraint states that weighted combination of outputs of all DMUs must be greater than or equal to the output of the reference DMU0. λj represents the weights of the DMUs (Hormazábal and Wyngard 2007). The model (1) above is an input minimization DEA program where the weighted sum of inputs is maximized and where the goal is to produce the observed outputs with minimum inputs. The output maximization DEA program maximizes the weighted sum of outputs and aims to maximize output production subject to a given level of inputs. Given that the assumption of constant returns to scale (CRS), where DMUs operate at their optimal scale, has often been considered over restrictive in the literature, the CCR model was initially modified in 1984 by Banker, Charnes and Cooper (1984) to incorporate an additional convexity constraint that would account for variable returns to scale (VRS). This modified DEA model was called the BCC DEA model. Specifically, the convexity constraint ∑j=1 λj =1 was imposed on the model (1) above to ensure that DMUs operating at different return to scale are considered efficient. 192 Region Province DEA PSP PSE Mekong River Delta An Giang 1 1.48 1.95 Mekong River Delta Hau Giang 1 1.41 1.82 Mekong River Delta Tien Giang 1 1.49 1.97 Mekong River Delta Tra Vinh 1 1.15 1.31 Mekong River Delta Vinh Long 1 1.34 1.69 Northern Central and Central Coast Binh Dinh 1 1.31 1.61 Red River Delta Ha Noi 1 1.52 2.04 Red River Delta Nam Dinh 1 1.30 1.60 Red River Delta Thai Binh 1 1.42 1.84 South East Binh Duong 1 1.91 2.83 South East Binh Phuoc 1 1.36 1.72 South East Dong Nai 1 1.72 2.44 South East Ho Chi Minh 1 1.85 2.70 Red River Delta Hai Duong 1.00 1.32 1.65 Mekong River Delta Kien Giang 0.96 1.38 1.80 Mekong River Delta Long An 0.96 1.35 1.73 Northern Central and Central Coast Ninh Thuan 0.94 1.28 1.61 Mekong River Delta Can Tho 0.90 1.36 1.82 Mekong River Delta Bac Lieu 0.89 1.18 1.46 South East Tay Ninh 0.88 1.41 1.93 Northern Mountain Hoa Binh 0.86 0.89 0.92 Northern Central and Central Coast Phu Yen 0.85 1.18 1.50 Mekong River Delta Ca Mau 0.83 1.08 1.33 Northern Central and Central Coast Binh Thuan 0.83 1.26 1.68 Mekong River Delta Dong Thap 0.82 1.17 1.51 Red River Delta Ha Nam 0.81 1.12 1.44 Mekong River Delta Ben Tre 0.80 1.17 1.53 Northern Central and Central Coast Da Nang 0.79 0.90 1.01 Red River Delta Bac Ninh 0.79 1.15 1.51 Red River Delta Hung Yen 0.79 1.16 1.53 Northern Central and Central Coast Thua Thien Hue 0.77 1.07 1.36 Mekong River Delta Soc Trang 0.74 1.06 1.39 Central Highlands Dak Nong 0.73 0.91 1.09 Northern Central and Central Coast Khanh Hoa 0.73 1.20 1.67 South East BRVT 0.71 1.18 1.65 Central Highlands Lam Dong 0.71 1.09 1.47 Northern Central and Central Coast Thanh Hoa 0.69 0.98 1.27 Northern Mountain Thai Nguyen 0.68 0.89 1.11 Northern Mountain Lao Cai 0.66 0.76 0.85 Red River Delta Ninh Binh 0.65 0.91 1.17 Northern Mountain Tuyen Quang 0.65 0.90 1.15 Central Highlands Dak Lak 0.62 0.93 1.24 Northern Mountain Yen Bai 0.62 0.83 1.05 Central Highlands Gia Lai 0.61 0.91 1.21 193 Northern Mountain Lang Son 0.61 0.76 0.90 Northern Mountain Bac Giang 0.60 0.93 1.26 Northern Central and Central Coast Quang Tri 0.60 0.79 0.98 Northern Mountain Phu Tho 0.59 0.90 1.20 Northern Central and Central Coast Nghe An 0.59 0.95 1.32 Red River Delta Hai Phong 0.59 1.02 1.46 Northern Central and Central Coast Quang Binh 0.58 0.82 1.06 Northern Central and Central Coast Quang Ngai 0.58 0.91 1.24 Northern Central and Central Coast Quang Nam 0.57 0.83 1.10 Northern Mountain Son La 0.56 0.71 0.87 Central Highlands Kon Tum 0.55 0.73 0.92 Red River Delta Vinh Phuc 0.52 0.86 1.20 Northern Mountain Dien Bien 0.51 0.57 0.62 Red River Delta Quang Ninh 0.50 0.79 1.09 Northern Central and Central Coast Ha Tinh 0.50 0.79 1.09 Northern Mountain Cao Bang 0.49 0.59 0.69 Northern Mountain Lai Chau 0.47 0.55 0.63 Northern Mountain Ha Giang 0.45 0.56 0.67 Northern Mountain Bac Kan 0.38 0.53 0.68 Annex C: 194 Productive efficiency analysis – explanatory variables Expected Category Indicators Source impact Budget dataset Positive Decentralized revenue as % of local revenue Level of fiscal Local revenue (exc. Transfers) as % of local revenue Budget dataset Positive decentralization Central balancing transfers as % of local core expenditure Budget dataset Negative District expenditures as % of local expenditures Budget dataset Positive Per capita GDP GSO Positive Socio-Economic Population density GSO Positive features Provincial poverty rate GSO Negative Local decentralized revenue as % of local recurrent Budget dataset Positive expenditure Fiscal trends Local recurrent expenditure as % of total local expenditure Budget dataset Positive Provincial Administrative expenditure as % of total local expenditure Budget dataset Negative Local expenditure as % of local budget Budget dataset Negative Fiscal Local revenue as % of local budgeted revenue Budget dataset Negative management Local carry over spending as % of local budget Budget dataset Negative Institutional Qualification of administrative staff in provinces Establishment Positive factors % of firms paying over 10 percent of their revenue in extra Survey Negative payments PCI D 11 Ease of access to provincial budgets Positive Provincial officials' knowledge of national laws relating to PCI (F1.1-1.3) PCI (H7.2) Positive private sector Provincial officials' ability to solve problems for private sector PCI (H7.3) Positive Annex D: Stochastic Frontier Analysis of Tax Effort by Province in Vietnam 2006-2011 195 1. Defining efficiency through Stochastic Frontier Analysis71 In a world where there is no inefficiency, the tax administration in a province collects tax revenues qi=f(zi,β). Stochastic frontier analysis, however, assumes that tax administrations potentially collect less revenue than it might due to a degree of inefficiency, that is qi=f(zi,β) ξi where ξi=(0,1] is the level of inefficiency in its revenue collection. If ξi=1, the tax administration is collecting the optimal amount of tax revenues, using the available inputs zi defining the tax bases, and the production function f(zi,β). When ξi<1, the tax administration is not making the most of the available inputs zi. Since tax collection qi is assumed to be strictly positive (qi>0), the degree of technical inefficiency is also assumed to be strictly positive (ξi>0). Tax revenue collection qi is also assumed to be subject to random shocks, implying that qi=f(zi,β) ξi exp (υi ) (2) Taking the natural log of equation (1) yields ln(qi )= ln[f(zi,β)]+ ln(ξi )+υi (3) Assuming that function f(zi,β) is linear in logs, that there are k inputs defining the country’s or province tax bases, and defining ui=-ln(ξi ) yields ln(qji )=β0+∑ βj ln(zji )+ υji- uji (4) where qi represents a ratio of total revenues (sum of tax and non-tax revenues) to GDP, while zji represents a matrix of variables affecting the country’s potential revenues.72 We assume that the idiosyncratic error component, υi, is independently N(0,συ ) distributed over the observations. Since ξi=(0,1], it implies that ln(ξi )≤0 and, therefore, ui≥0. In other words, the inefficiency effect ui lowers the tax collection from its potential level. We assume two alternative specifications of the inefficiency term, ui. In the first one, the ui is independently half-normally N+ (0,σu2 ) distributed, and in the second one, the ui is independently exponentially distributed with variance, σu2. 71 Stochastic frontier models became a popular subfield in econometrics after they were first introduced by Aigner, Lovell, and Schmidt (1977) and Meeusen and van den Broeck (1977). The discussion in Annex 1 and 2 is partly based on Cyan, Martinez-Vazquez and Vulovic (2013) 72 The regressions may also account for regional fixed effects and time effects. However, with our data when we include regional dummies the estimation does not converge. The inclusion of time effects leads essentially to the same log-likelihood. 2. Explaining Technical Inefficiency 196 As we mentioned above, the stochastic frontier analysis allows us to estimate the level of technical inefficiency and its determinants in revenue collection systems. Basically, after estimating equation (4) k ln(qji )=〖β0+∑ βj ln(zji ) υji- uji j=1 we predict the technical inefficiency term, (uji ) , and then we estimate the following equation k (uji )=∑ θj wji+αi+μt+εji (5) j=1 where wji represents a set of variables that may explain technical inefficiency in revenue collection; αi is the unobserved individual province (country) effect, while μt is the time effect. 3. Provincial tax efforts Province 2006 2007 2008 2009 2010 Ba Ria - Vung Ta 0.18 0.16 0.28 0.58 0.51 Bac Giang 0.23 0.25 0.23 0.23 0.20 Bac Kan . 0.87 0.81 0.85 0.87 Bac Lieu 0.77 0.87 0.82 0.79 0.77 Bac Ninh 0.41 0.32 0.27 . . Ben Tre . 0.83 0.82 0.79 0.82 Binh Dinh 0.70 0.74 0.76 0.75 0.76 Binh Duong 0.86 0.85 0.82 0.73 0.65 Binh Phuoc . 0.90 . 0.81 0.87 Binh Thuan 0.75 0.80 0.78 0.76 0.81 Ca Mau . 0.86 0.86 0.86 0.87 Can Tho 0.63 0.56 0.54 0.51 0.47 Cao Bang . 0.89 0.82 0.85 0.81 Da Nang . 0.64 0.55 0.34 0.32 Dak Lak 0.80 0.89 0.84 0.82 0.76 Dak Nong 0.63 0.28 0.12 0.16 0.67 Dien Bien 0.66 0.62 0.57 0.60 0.64 Dong Nai . 0.66 0.57 0.54 . Dong Thap 0.42 0.40 0.76 0.40 0.40 Gia Lai 0.68 0.81 0.80 0.75 0.72 Ha Giang . . 0.82 0.88 . Ha Nam . . . . 0.35 Ha Noi 0.87 . . . 0.83 Ha Tinh . . . 0.59 . Hai Duong 0.88 0.91 0.91 . 0.87 Hai Phong 0.89 . . . . Hau Giang 0.49 0.58 0.38 . 0.46 Ho Chi Minh 0.26 0.29 0.31 . 0.24 Hoa Binh 0.37 0.45 0.45 0.79 0.80 Hung Yen 0.45 0.62 0.64 0.67 0.91 197 Khanh Hoa 0.91 0.88 0.86 0.85 0.85 Kien Giang 0.76 0.77 0.77 . . Kon Tum 0.84 0.85 0.83 0.88 0.88 Lai Chau 0.74 0.80 0.79 0.66 0.70 Lam Dong . 0.82 0.78 0.70 0.65 Lang Son 0.88 0.88 0.77 0.85 0.81 Lao Cai 0.79 0.83 0.78 . . Long An . 0.78 0.81 0.74 0.80 Nam Dinh 0.62 0.71 0.55 0.56 0.58 Nghe An 0.54 . 0.17 0.24 0.23 Ninh Binh 0.77 0.85 0.76 0.48 . Ninh Thuan 0.82 0.80 0.84 . . Phu Tho . 0.82 . 0.73 0.92 Phu Yen . . . . 0.83 Quang Binh 0.25 0.18 0.14 . 0.16 Quang Nam 0.66 0.72 0.71 0.82 0.84 Quang Ngai 0.90 0.91 0.82 0.79 0.94 Quang Ninh 0.84 0.79 0.73 0.81 0.86 Quang Tri 0.37 0.40 0.18 0.25 0.19 Soc Trang 0.84 0.83 0.73 0.70 0.59 Son La 0.62 0.70 . 0.76 0.61 Tay Ninh 0.69 0.60 0.57 0.56 0.66 Thai Binh 0.26 0.25 0.20 0.18 0.18 Thai Nguyen . 0.25 0.78 0.72 0.75 Thanh Hoa 0.87 0.86 0.83 0.79 0.78 Thua Thien Hue . . . . . Tien Giang 0.72 . . . 0.69 Tra Vinh 0.75 0.79 0.74 0.73 . Tuyen Quang 0.68 0.77 . 0.64 0.63 Vinh Long 0.34 0.34 0.74 0.76 0.84 Vinh Phuc 0.93 0.92 0.94 0.94 0.93 Yen Bai 0.68 0.75 0.60 0.64 0.17 Average 0.65 0.68 0.65 0.66 0.66 When identifying determinants of the provinces’ revenue potential that can be derived from the stochastic frontier regression analysis, it is assumed that a province’s revenue capacity depends on economic, demographic and institutional factors. These factors are less numerous when the analysis is conducted at the subnational level (as compared to the country level), because some factors will tend to be homogenous for the entire country, such as the inflation rate among economic factors or the level of corruption among institutional factors. Among the economic factors, typically regional GDP per capita is included as a control variable as a 198 proxy for the level of economic development, and generally one may expect a positive relationship with revenue collection because of higher ability to pay taxes. On the other hand, richer areas may exert lower tax effort than their resource-scarce counterparts. Another determinant of the ability to collect taxes is the sectoral structure of the economy. Certain sectors in the economy have been traditionally hard to tax, such as agriculture, services, and construction; in particular agriculture typically receives quite favorable tax treatment. The construction and services sectors have higher percentages of output produced informally and therefore hard to reach by tax administrators. Higher levels of informality directly measured are also expected to lead to less revenue potential. There are demographic factors that can also play a significant role in determining a province’s tax revenue potential. These include Age dependency, measured as the ratio of dependents (the population under age 15 and above age 65) to the working-age population (those aged 15-64) and population density, whose final impact on revenue potential is ambiguous because of both potential positive and negative effects. For example, the elderly may have lower incomes but they may exhibit greater tax morale or willingness to pay taxes. Or, higher concentrations of people could make taxation easier but it may also encourage informal activities that are difficult to tax. The level of education has been frequently used as another important demographic component of revenue capacity. The effect of education is also ambiguous. On the one hand, the more educated people are the better they can understand the relationship between public goods provision and the importance of paying taxes to finance them. On the other side, the more educated people are the more knowledgeable they become regarding how to avoid paying taxes, in which case we would expect a negative effect of education on revenue collection. Annex E: Explaining the difference in tax effort performance across provinces 199 A set of variables has been used in the previous literature explaining variations in technical efficiency of tax collections across (national or subnational) jurisdictions. These include the following: a. The number and education levels of the tax administration employees should make a big difference in how efficiently tax collection and enforcement behave. Several variables may capture differential incentives provincial tax administrations may have b.  to collect taxes. For example, higher levels of government debt may have a positive effect on government efficiency in collecting taxes because of the pressure put on governments to repay the debt in the future. The level of the budget deficit should operate in similar ways to debt level. The presence and relative importance of transfers from the central government may put a damper on incentives to collect own revenues. The level of fiscal decentralization may also provide incentives to collect better and more. c. Among other variables, corruption may increase technical inefficiency in the tax system directly by deviating funds and indirectly by introducing permanent instability in the political system; we may expect a negative relationship between this variable and technical efficiency. Population growth rate is associated with higher inefficiency in the tax system because it is difficult d.  to administer a rapidly rising population of taxpayers. e. At the country level, the complexity of the tax system has been argued to increase administration and compliance costs and therefore reduce efficiency. At the provincial level, like in the case of Vietnam, this variable will not make a difference if the legal structure of taxes is the same because it is fully determined at the central level. f. Another variable that has been used in explaining differences in technical efficiency is tax morale, which is measured in a series of surveys by the percentage of the population who declare cheating on taxes as never justifiable. Since higher tax morale makes it easier for the government to collect taxes, it could be interpreted to be an input contributing to the larger tax base. However, in high-tax-morale societies, tax administration may be more relaxed in collecting taxes and have lower audit rates (and therefore, all other things equal, be relatively more inefficient in extracting revenue for a given tax base), which may give way to higher tax evasion.73 73 Other variables in the past literature, which again are not relevant at the provincial level in Vietnam, include the extent of democratic institutions (the expectation being that democracies tend to have more efficient tax systems), level of fractionalization in the government (as a proxy for better representation of citizens and more efficient provision of services, and seignorage revenues, proxied by increases in the monetary base which may be expected to discourage governments from collecting taxes. Annex F: 200 Local and Regional Business Taxes* in Selected OECD Countries Country Designation Taxpayer Tax base Tax rate Local operating profit plus half the 9% - 20%, average: Gewerbesteuer Business enterprises, interest expenses on long-term 16.3% lower rates for Germany (local business excluding farmers, debt. small firms tax) professionals Allowance of Euro 24,500 for non- incorporated firms. Taxe Business enterprises professionnelle Local fixed assets rental value, Limited to 3.5% of France and professionals, (local business reduced by 16% gross value added excluding farmers tax) Entrepreneurs or Kommunalsteuer Wage expenses, low threshold for Austria other employers 3% (municipality tax) small firms subject to VAT Local operating profit. Allowance Impôt commercial Business enterprises, of Euro 40,000 for non- Luxembourg (local business excluding farmers, 6% - 9% incorporated firms and Euro tax) professionals 17,500 for incorporated firms Imposta regionale Local net value added from sulle attività Entrepreneurs, non- the provision of goods and Standard rate 4.25%, Italy produttive - profit organizations services (subtraction method), region’s discretion of IRAP (regional and public bodies wage expenses for non-profit +/- 1%-point business tax) organizations Surcharge on Corporations subject Portugal corporation Local share of CIT liability 0% - 10% to CIT income tax (CIT) Helyi iparűzési Local gross value added Hungary adó (local Entrepreneurs (subtraction method). Allowance 0% - 2% business tax) of Euro 10,500 (optional) Switzerland Taxe Business enterprises Local business sales, rental Canton de professionnelle and professionals, value of fixed assets, number of Genève (local business excluding farmers employees tax) Cantonal corporation Cantonal/local share of CIT Corporations subject income tax (CIT), liability Cantons: 2 % - 10% to national CIT local surcharge local: 2 % - 10% All cantons Enterprises subject to Vermögenssteuer Equity capital 0.05 % - 0.5% cantonal PIT or CIT (cantonal net worth tax, local surcharge) Impuesto sobre Industry sector and floor space actividades Business enterprises used, number of employees, Spain económicas and professionals, electricity consumption. (local business excluding farmers Exemption up to a turnover of tax) Euro 1,000,000 Canada Incorporated Capital tax Equity capital 0.3% - 0.5% (provinces) enterprises Business enterprises Japan Enterprise Tax and professionals, Local operating profit 3% - 12% 201 excluding farmers Notes: *Business taxes with considerable discretion over the tax revenue assigned to the local government, in particular the right to set the tax rates at least in certain limits. Sources: Mennel and Foerster (2006), OECD (2006), European Commission (2007), IBFD (2007). How Should Local Governments Tax Local Business? Lessons from an International Comparison and a Micro-simulation Analysis for Germany http://www.diw.de/documents/publikationen/73/diw_01.c.62887.de/dp717.pdf Germany74 Municipalities levy a local business tax (the Gewerbesteuer). The tax base is the same across municipalities but the rates can be set independently. The main source of the local business tax base is the operating profit attributed to the local jurisdiction. Therefore, received dividends are not subject to the tax unless they stem from shareholdings of less than 10 %, and, correspondingly, losses from shareholdings are not allowed to be set off against taxable income. Moreover, the tax base is augmented by half of the interest expenses on long-term debt. Based on the resulting taxable income, the local business tax is determined in two steps. In the first step, the taxable income is multiplied by a basic federal tax rate (“Messzahl”) of 5 % (in 2007) in order to obtain the uniform basic tax. Unincorporated firms, in particular SMEs, benefit from an allowance of €24,500 and reduced basic federal tax rates up to a taxable income of €72,500. The uniform basic tax is allocated to the local jurisdictions involved. In the second step, the local jurisdictions apply a multiplier (“Hebesatz”), which they are entitled to determine, to their allocated share of the uniform basic tax. These multipliers range from a minimum rate of 200 % to almost 500 % in high performing agglomerations such as Munich, Hamburg, or Frankfurt. Taking into account that the local business tax liability reduces its own tax base as deductible expense, the effective local tax rates range from a minimum rate of 9 % to almost 20 % in 2007. The average rate is about 16 %. Sole proprietors and partners of non-incorporated firms can credit at least parts of the local business tax against their personal income tax liability (the credit is a multiple of the uniform basic tax). Italy75 IRAP: It’s a regional tax on productive activities (Imposta regionale sulle Attività Produttive). IRAP is levied on the net value of the production derived in each Italian region. The standard rate is 4.25%. Regional authorities may increase or decrease the standard rate by up to one percentage point. Taxpayers carrying on business activities in more than one region by employing personnel in each region for more than 3 months, must apportion their taxable base among the regions on the basis of the remuneration paid to personnel employed in each region. 74 http://www.uni-graz.at/socialpolitik/papers/Bach.pdf https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=5&cad=rja&ved=0CFEQFjAE&url=http%3A%2F%2Fwww.lex. 75  uniba.it%2Fta%2FITALIAN_CORPORATE_TAXATION.ppt&ei=D0vIUYTOGoLc9ASZ64G4DQ&usg=AFQjCNF0ea347dkAmV3vAUNg VXy6_5-fow&sig2=IJi6hWFsRs_c_yatyVRotw&bvm=bv.48293060,d.eWU Annex G: 202 Intergovernmental Fiscal Transfers Map REVENUE EXPENDITURE Taxes 100% assigned STATE BUDGET Revenue = A + r + O Central Government Exp(C ) to Center (A) Any formula for Target programs T = Tc + Tp how much of Shared Taxes (r ) Tc = By central line the state ministries budget allotted Recurrent exp transfers to for target Other sources (Oc) provinces: Expenditure norms in programs (T) Decision 59 (N) and capital expenditure (K) Capital Transfers to provinces: Provincial Budgets Decision 60 (K) Shared Taxes (R-r ) Revenue = B + (R- r ) + Tp + Op F I Provincial Government Exp (P) S C B- (R-r) Balancing Transfers (B) Transfers to Target Other A B = Recurrent transfers (N Districts and L Programs (Tt) Expenditure (Op) Communes (X) G A Target Programs (Tp) Provincial Target National Target P Programs (Td) Programs (Tp) Other sources (Op) District and Commune Budgets Target Other Capital Capital Recurrent Programs sources of Transfers Transfers exp Expenditures (both revenue tranfers national Annex H: List of provinces for which balancing transfer arrangements were reviewed 203 Key indicators for review of provincial level allocation norms Provinces Net Net Per capita Population Decentralized Balancing Local contributor recipient GDP density revenue/local transfers/ decentralized (VND m) (person/km2) revenue local core revenue/local expenditure recurrent expenditure Can Tho x 49.19 853.58 59.6% .. 180.7% Da Nang x 41.03 739.93 61.7% .. 340.0% Hanoi x 42.19 2,023.62 63.2% .. 215.4% Quang Ninh x 46.91 191.24 .. .. .. Tuyen Quang x 18.48 124.91 15.6% 42.3% 29.5% Kien Giang x 36.09 269.69 33.3% 24.4% 74.0% Binh Phuoc x 27.53 131.21 46.7% 13.0% 106.1% Cao bang x 13.05 76.63 11.3% 45.5% 22.1% Lam Dong x 25.57 124.69 41.5% 18.8% 87.8% Hai Duong x 22.64 1,040.76 65.7% 6.1% 136.8% Kontum x 17.82 46.61 25.1% 30.0% 63.1% Quang nam x 22.30 137.73 24.3% 32.7% 65.1% Source: Staff estimates based on published State Budget data 204 Annex I: Disparities in nominal per capita revenue, plus transfers and expenditures of provinces, Vietnam 2006-2011 2006 2007 Retained + Shared ++ Balancing +++ Target Expenditure Retained + Shared ++ Balancing +++ Expenditure revenue revenue transfers transfers revenue revenue transfers Target transfers Mean 318.6 679.6 984.3 1568.7 1529.6 401.3 794.9 1356.9 1996.0 1745.8 SD 373.5 663.0 528.2 724.7 862.6 844.3 1094.2 949.2 1160.6 864.3 CV 1.17 0.98 0.54 0.46 0.56 2.10 1.38 0.70 0.58 0.50 Min 21.4 141.7 458.0 274.8 461.1 23.2 175.4 375.3 680.3 563.2 Max 2331.9 3609.3 3609.3 3970.5 4932.4 5347.4 6742.3 6742.3 7019.4 4642.1 Max/Min 109.1 25.5 7.9 14.4 10.7 230.7 38.4 18.0 10.3 8.2 2008 2009 Retained + Shared ++ Balancing +++ Target Expenditure Retained + Shared ++ Balancing +++ Expenditure revenue revenue transfers transfers revenue revenue transfers Target transfers Mean 403.7 1017.5 1584.1 2440.6 2212.0 432.6 1180.3 1768.1 2847.3 2835.0 SD 687.7 1271.4 1060.5 1341.9 1417.8 558.5 1264.4 1063.8 1488.8 1762.0 CV 1.70 1.25 0.67 0.55 0.64 1.29 1.07 0.60 0.52 0.62 Min 11.0 33.6 371.4 862.1 761.7 50.4 232.4 417.8 675.5 909.6 Max 4262.1 6045.4 6045.4 6261.1 8790.8 3222.4 5282.6 5282.6 6361.7 9818.1 Max/Min 387.8 180.1 16.3 7.3 11.5 63.9 22.7 12.6 9.4 10.8 2010 2011 Retained + Shared ++ Balancing +++ Target Expenditure Retained + Shared ++ Balancing +++ Expenditure revenue revenue transfers transfers revenue revenue transfers Target transfers Mean 625.0 1573.6 2150.2 3444.8 3472.2 681.6 1791.7 3124.6 4431.4 4043.1 SD 938.6 1749.5 1537.6 1856.5 2001.9 1110.5 1726.5 1544.3 2334.4 2179.9 CV 1.50 1.11 0.72 0.54 0.58 1.63 0.96 0.49 0.53 0.54 Min 41.9 278.2 412.7 488.4 1392.3 31.9 155.1 764.1 568.2 1501.0 Max 5867.7 8752.0 8752.0 9470.7 13279.2 6659.1 9728.0 9728.0 10587.2 14517.4 Max/Min 139.9 31.5 21.2 19.4 9.5 208.5 62.7 12.7 18.6 9.7 Source: Author calculation Annex J: Disparities in nominal per capita revenue, plus transfers and expenditures of districts, Vietnam 2006-2011 2006 2007 "Own" + ++ Targeted District "Own" + Balancing ++ District +++ Expenditure revenue Balancing Transfer Spending revenue Transfer Targeted Spending Target Transfer Transfer transfers Grand mean 0.80 1.12 1.39 0.99 1.06 1.59 1.95 1.28 1996.0 1745.8 Grand median 0.46 0.82 1.05 0.86 0.76 1.25 1.59 1.09 1160.6 864.3 Grand max 12.78 12.78 12.84 5.55 13.00 13.02 13.06 8.16 0.58 0.50 Grand min 0.03 0.00 0.13 0.22 0.02 0.01 0.32 0.04 680.3 563.2 Max/min 426.00 98.77 25.23 650.00 1302.00 40.81 204.00 7019.4 4642.1 Coef. Of Var. 1.37 1.00 0.86 0.59 1.14 0.80 0.73 0.67 Std. Dev. 1.10 1.11 1.19 0.58 1.20 1.28 1.41 0.86 2008 2009 "Own" + ++ Targeted District "Own" + Balancing ++ District +++ Expenditure revenue Balancing Transfer Spending revenue Transfer Targeted Spending Target Transfer Transfer transfers Grand mean 1.04 1.59 2.10 1.72 1.48 2.13 2.83 2.21 1996.0 1745.8 Grand median 0.64 1.32 1.68 1.42 0.94 1.60 2.14 1.75 1160.6 864.3 Grand max 9.03 10.49 12.46 9.65 16.77 19.05 23.59 16.68 0.58 0.50 Grand min 0.05 0.00 0.06 0.34 0.05 0.15 0.43 0.16 680.3 563.2 Max/min 180.60 207.67 28.38 335.40 127.00 54.86 104.25 7019.4 4642.1 Coef. Of Var. 1.07 0.79 0.75 0.63 1.10 0.84 0.75 0.69 Std. Dev. 1.12 1.27 1.57 1.08 1.63 1.79 2.12 1.53 2010 2011 "Own" + ++ Targeted District "Own" + Balancing ++ District +++ Expenditure revenue Balancing Transfer Spending revenue Transfer Targeted Spending Target Transfer Transfer transfers Grand mean 2.06 2.63 3.56 2.77 2.29 3.42 4.22 3.41 1996.0 1745.8 Grand median 1.40 1.99 2.66 2.28 1.48 2.64 3.27 2.88 1160.6 864.3 Grand max 21.08 23.19 25.50 19.18 25.30 25.30 27.02 24.41 0.58 0.50 Grand min 0.05 0.00 0.12 0.96 0.01 0.00 0.08 0.01 680.3 563.2 Max/min 421.60 212.50 19.98 2530.00 337.75 2441.00 7019.4 4642.1 Coef. Of Var. 1.15 0.95 0.83 0.62 1.12 0.81 0.76 0.67 Std. Dev. 2.37 2.49 2.96 1.72 2.56 2.78 3.21 2.30 Note: Population values were available only for 1999 and 2009, therefore the values for in between years were interpolated, while 2010 and 2011 were estimated using the trend. Moreover, for districts for which only 2009 population data were available, the overall provincial population growth rate was used to estimate the values for 2006-2011 data. Finally, given the extreme values registered for some observations which could not be independently verified lead us to remove the following districts as outliers: Van Ninh (Khanh Hoa province), District 1 (Ho Chi Minh City), Con Co (Quang Tri), Con Dao and Tan Thanh (both Ba Ria – Vung Tau), Muong Lay (Dien Bien), Phuc Yen and Vinh Yen (both Vinh Phuc). 205 Annex K: 206 Targeted funding from the central budget for the stability period 2011-2015 S.No. Name Criteria and comments 1 National target There are currently 16 NTPs in Vietnam for the stability period 2011-2015. programs Appendix A3 lists these NTPs and the allocation criteria for these. 2 Regional socio- These are programs to develop the Northern highland and mountainous region; economic development north-central region and central coastal region; Red river delta; Central highland; program and the Mekong delta. They currently run for the period 2011-2015 and have specific criteria and quotas for investment allocations that include points awarded for population, land area, and whether in mountainous regions. 3 Strengthening and These are programs to strengthen and upgrade the sea dike system from Quang upgrading program for Ngai to Kien Giang. There are no specific criteria. Funds are allocated depending the sea dike system on available central funds. This works like a subsidy program with the central Quang Ninh – Kien government subsidizing upto 50% of total investment in provinces and cities that Giang and upgrading contribute to the central budget and upto 90% for provinces and cities that do not. program for river dike system 4 Seeds, breeds This program funds administrative and research institutions as well as investment development program in infrastructure. There are no criteria for allocation, though once a project has been approved, projects under central administration receive 100% funding while the percentage varies across provinces and cities depending on whether they receive additional central funding or not. 5 Refuge sites for fishing This program funds projects that have been approved and are in the sub-national boats and ships plan. Regional projects are eligible to be funded upto a maximum limit of VND 100 billion and provincial projects upto a maximum limit of VND 80 billion per project. The balance remaining would need to be sources from the local budgets. 6 Construction or Investments eligible under this program include the construction and upgrading upgrading of fresh of irrigation system, fresh water reservoirs and dikes. There are no criteria for water ponds and allocation and funding by the central government varies depending on their irrigation system on contribution to the central state budget and the ratio of their local revenues to densely populated local expenditure. islands 7 Investment in 62 poor There are specific criteria and points for allocating additional target funding districts allocation from the central budget to these areas that include total population, ethnic population, land area, number of communes, percentage of poor households and percentage of forested area. 8 Relocation and There are no specific criteria for these programs. Funding from the central budget resettlement program is based on the annual available state budget resources with priority being given for ethnic minorities to poor districts with low budget revenues and incomplete or ongoing projects. 9 Relocation and This program pertains to regions that prone to natural disasters, regions in resettlement program extremely difficult conditions, in mountainous, highland, island or border areas as well as in regions of uncontrolled migration, and restricted protection of forests. Only poor districts with low budget revenue are eligible for funding. There are no criteria and funding percentages can vary from a maximum of 50% in the Red river delta and Mekong delta to 90% in the Northern mountainous provinces. 10 Support to Criteria and quotas for budget allocation as contained in Decision 126/2009/QĐ- infrastructure TTg dated 26/10/2009 (get details). investments on the coast line economic zone 11 Support to industrial Criteria and quotas for budget allocation as contained in Decision 43/2009/QĐ- zones in provinces with TTg dated 19/3/2009 (get details). 207 difficult socio-economic conditions 12 Support to Funding is available only to provinces having a GDP from industry lower than infrastructure the national average by at least 10%. Provinces that are net contributors to the development in central budget period are not eligible under this scheme. There are no specific industrial hubs in criteria though maximum limits for funding have been specified. These range from provinces with difficult VND 6 billion per industrial hub and VND 70 billion per province in the Northern socio-economic highland and mountainous regions and the Central highlands to VND 5 billion per conditions industrial hub and VND 50 billion per province in other areas. 13 Support to Priority is given to projects that contribute to the area’s socio-economic infrastructure development. Funding depends on the available state budget and funding investment in border percentage varies 20% to 90% depending on the contribution of the province to economic zones central government budget. 14 East Sea-Islands Priority is given to poor and less developed provinces. The amount of central program support depends on available funds and contribution of the provinces to the central budget. 15 Support to forestation Priority is given to projects in North Western region, Central Highland and Central and forest protection Coastal regions with vast areas of barren hills that need forestation. The amount of support varies from VND 200,000/ha/year to VND 15 million/ha/year. 16 Socio-economic Priority is given to existing projects. There is a specific scoring system based development along the on six criteria that border areas of provinces; number of border inhabitants of borders with China, provinces; the length of the border; number of communes on the border; number Laos and Cambodia of districts on the border; and the number of existing and to be built border posts. 17 Support to The scope and investment of each ATK is determined by the prime minister. infrastructure There are specific allocation criteria that include the complexity of the terrain, the development in “Safety area, population, number of communes in the ATK, the ratio of poor households Zone” region (ATK) in ATK, the ratio of households with access to clean water, and other investments in ATK. 18 Border management Central funds are allocated to projects only in approved plans. No specific criteria and protection program listed. 19 Support to new Central funding supports projects in newly separated provinces since 2004. No provinces and districts specific criteria listed except that it would depend on the ratio of contribution to the central budget, own revenues, and criteria based on area, population, ratio of poor households, and the basic infrastructure of districts. 20 Counter funding to ODA Priority is given to counterpart funding to ODA projects in the area of hunger projects under local eradication and poverty reduction, transport, rural development, healthcare and administration environment amelioration. The funding varies 50% to 90% depending on the how dependent the province is on fiscal transfers. 21 Support to investment Priority is given to ongoing projects. Provinces contributing to the central budget in social education are not eligible for support. Other provinces receive a maximum of 30% to 80% centers of counterpart funding depending on the percentage of overall budgetary support that they receive. 22 Support to investments Priority is given to ongoing projects. No specific criteria but funding support in district and provincial ranges from 50 to 70% of construction cost and 70 to 100% of equipment cost hospitals depending on whether the province is the Northern mountainous region, Central highland region, or other regions. S.No. Name Criteria and comments 208 23 Support to urgent The amount of central support varies according to available resources and can projects that local vary from 50% to 90% of each project. budget is unable to finance 24 Investment in overseas Priority is given to countries and territories of strategic importance and having offices and housing close relationship with Vietnam and new purchases. No specific criteria outlined. of Vietnamese representations and other agencies 25 Search and rescue Funding depends on the available annual budget, with a focus on transition program projects. 26 Investments as directed Priority is given to the downgraded buildings and offices that have become unsafe by Resolution 49-NQ/ and offer less than 70% utilization capacity. TW of the Politburo on judicial reform 27 Special support to Funding support is given to provinces with big revenues and supports investment provinces with big programs in infrastructure, and other structures serving public interest, social revenues order and safety, environment protection and sustainable development. Funding is provided to provinces that contribute to the central budget for the period 2011- 2015; or to provinces without contribution to the central budget for the period 2011-2015 but have annual planned revenue of VND 10,000 billion or more. The criteria are specifically laid out. Provinces get points for net contribution to the central budget, the amount of actual revenue collected and the export revenues. The amount of targeted support from the central budget is based on the available budget for development investment as well as the points as per the criteria laid out in this program. 28 Investments in office Priority is given to newly separated communes and townships that do not have building of commune their own offices yet or whose office buildings have downgraded and become people's committees unsafe for use. Communes in extremely difficult regions, mountainous, remote, island regions and coastal warps get 100% support while the rest get 70%. 29 Investment in tourism Support is given to develop tourism infrastructure to not more than 3 projects per infrastructure year per province provided there are no ongoing projects. Funding upto 80% of investment is provided depending on the location of the tourist area. Annex L: List of National Target Programs for 2011-2015 Stability Period 209 Annex M: 210 Examples of indicators used in NTPs Name of program Input indicators Output indicators Livelihoods and income poverty Job Creation and Vocational Provide training to people Number of jobs created Training Build training school Increase in employment rate Sustainable Poverty Reduction Build roads in poor communes Number of households escape poverty Water Supply and Rural Provide 60 litre/person of clean Number of people in rural area have access Environment water to hygienic water Number of schools and hospitals have hygienic latrines Development of a New Rural Number of communes reach countryside Life Program 135 standards Reduction in poor household rate Number of provinces with roads and electricity Health and population Health Number of doctors and health Reduction in number of patients with leprosy, specialists are trained to treat the tuberculosis, malaria, dengue fever diseases Number of children get vaccinated Reduction in birth mortality rate Population and Family Increase rate of using Reduction in number of population, number Planning contraceptive of children per each family Food Hygiene and Safety Number of markets and cities that Number of centers that meet the ISO provide food safety standard Develop the department of food safety and hygiene Examine factories for food safety and hygiene Train staff on food safety and hygiene Raise awareness of people Prevention and Control of HIV/ between age 15-49 on the HIV/ Reduce the number of HIV infected patients AIDS AIDS Culture Culture Preserve historical places, cultural Registration to UNESCO heritages Education Education and Training Number of children enroll in Literacy rate primary schools and kindergartens Teach foreign language to students Public Security Drug Abuse Prevention and Number of criminals arrested Reduction in heroin addicts Control 100% of illegal marijuana eliminated Crime Prevention and Control Number of serious crimes investigated Prevent emerging crimes from happening Prisoners are trained Develop database of criminals Energy Efficiency Efficient and Economic Use of Provide energy management Reduce 5-7% of energy consumption during 211 Energy training to staff in the construction 2012-2015 and industrial sectors Environment and Climate Change Response to the Climate Change Pollution Remediation and Improve environment of selected Environmental Improvement craft villages Revitalize areas that are polluted by chemical Manage waste water before discharging to river Information and Communication Information Flows to Remote Train staff in communes to All communes receive radio coverage Areas operate radio broadcasting stationsImprove radio equipment and system Annex N: 212 The Main Secondary Legislations Concerning Government Debt Management 1. Government Decree No. 60/2003/ND-CP Dealing and Guiding the Implementation of the State Budget Law 2. Government Decree No. 78/2010/ND-CP on On-lending of Government’s external Borrowing 3. Government Decree No. 79/2010/ND-CP on Public Debt Management Operations Government Decree No. 01/2011/ND-CP on the Issuance of Government Bond, Government 4.  Guaranteed Bond and Local Government Bond 5. Government Decree No. 15/2011/ND-CP on Provision and Management of Government Guarantee 6. Government Decree 38/2013/ND-CP on Management and Use of Official Development Assistance (ODA) and Concessional Loans 7. Ministry of Finance Circular No. 81/2012/TT-BTC on Guideline for Issuing Local Government Bonds in the Domestic Market 8. Government Decree No. 75/2011/ND-CP on State Investment Credit and Export Credit 9. Prime Minister Decision No. 108/2006/QD-TTg on the Establishment of the Vietnam Development Bank 10. Government Decree 138/2007/ND-CP on the organization and operation of Local Development Investment Funds 11. Government Decree 37/2013/ND-CP which amends and supplements Decree 138/2007 Annex O: The choice of approaches across countries on subnational debt regulation 213 SNG is SNG borrowing SNG borrowing activities are (at least partially) SNG is eligible SNG is not autonomous controls such regulated by the rules imposed by central(or upper for borrowing eligible for to borrow or as fiscal targets level) government with prior borrowing incur debt. and debt ceiling approval of except Constraints Numerical Numerical The borrowing are results of central(or on-lending on purpose of constraints on constraints on activities of negotiation upper level) through central borrowing fiscal balance, new debt, debt SNG is mainly process government government expenditure level and debt monitored and between central service controlled by and local the market and/ governments or regulated by local level regulations Canada Australia state Germany region Germany local Germany local Japan local Mexico state province Austria local and local (the Italy state and Spain local (above certain (external) Japan local (up Belgium local Constitution) local Brazil local limit) Armenia local to certain limit) Germany region Italy state France local Peru local Korea local (external) Switzerland Spain region and local (the Brazil local Estonia local Germany local Kazakhstan state Constitution) Columbia local Hungary local Spain region local (external) US state France local Peru local Lithuania local and local Kosovo Argentina Brazil local Lithuania local Latvia local (outside EMU) Russia local province Mexico state Russia local Poland local Turkey local (external) (domestic) Peru local Indonesia local Russia local UK local Cambodia local Czech Republic (external) Vietnam local Serbia local (external) India state local Estonia local Slovakia local Argentina (external) Zimbabwe local Romania local Philippine local province Indonesia local Russia local Thailand local (external) (external) Slovakia local Nigeria state Brazil local Nigeria state Indonesia local (external) (external) South Africa Columbia local local (bond issuance) Peru local (external) Serbia local Slovakia local (above certain limit) India state (domestic) Indonesia local (domestic) Thailand local Vietnam local (external) Source: Subnational DeMPA Guide 214 Regulation of local government borrowing by limiting Country annual debt service total outstanding debt total annual new borrowing other Albania <72% of operational surplus <77% of recurrent revenues <20% of average total revenues for three years Brazil <11.5% of net current revenue (without transfers) <200% net current revenue Bulgaria <25% of own source guarantees < 5% of own source revenues, equalization grants from revenues, equalization grants the previous year (previous year) Croatia <20% of the actual revenues overall LG’s debt < 3% of total from the previous year LG recurrent revenues Estonia <20% of total budget <60% of total planned budget France <50% of operating revenues debt service includes guarantees paid Hungary <50% of own revenues government approval of borrowing, guarantees India <20% of revenue budget deficit<3% of State GDP Ireland EUR 200 M per local authority annually Italy <15% of current revenues Lithuania <35% of total revenue, excluding specific grant <35% of total budget (minus specific <20% of annual revenue; grants) (<10% for short term borrowing) Macedonia < 30% of the total current operational budget of (including guarantees) <100% of the previous year total revenues Poland <15% forecasted annual revenue debt service including guarantees Poland <15% forecasted annual revenue Romania <30% of average of own revenues in the past 3 years Slovakia <25% of current revenues in the previous year <60% of current revenues in the previous year Slovenia <8% of revenues of the previous year <20% of total revenues from previous year Serbia <50% of total recurrent revenues Turkey <100% of total budget (150% for <10% of realized revenues metropolitan cities) from the previous year Source: Dexia, 2003, Liu-Pradelli, 2013, NALAS, 2011, Swianiewicz, 2004 The World Bank in Vietnam 63 Ly Thai To Street, Hanoi Tel: (84-4) 3934 6600 Fax: (84-4) 3935 0752 Website: www.worldbank.org.vn