Philippines: Public Expenditure Review Strengthening Public Finance 55695 for More Inclusive Growth June 2011 es The World Bank Group in the Philippines or Making Growth Work for the Poor ©2011 The International Bank for Reconstruction and Development / The World Bank 1818 H Street NW World Bank Ofce Manila Washington DC 20433 23rd Floor, The Taipan Place Telephone: 202-473-1000 F. Ortigas, Jr. Road, Ortigas Center Internet: www.worldbank.org Pasig City, Philippines Telephone: (63-2) 637-5855 Internet: www.worldbank.org/ph/cas DISCLAIMER This volume is a product of the staff of the International Bank for Reconstruction and Development/The World Bank with support from the Australian Agency for International Development (AusAID). The ndings, interpretations, and conclusions expressed in this volume do not necessarily reect the views of the Executive Directors of The World Bank or the governments they represent or of AusAID. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement of acceptance of such boundaries. This report was co-funded by the World Bank and AusAID Trust Fund on Partnership to Support Philippine Development. Cover photo credits: 1) Nicole Arellano (“Macroeconomy Starts Here”); 2) Dave Llorito (“Education”); 3) Vincent Kho (“Health Care”) and 4) Eric Le Borgne (“Cheerful Kids”). Report No. 55695 - PH. Philippines: Public Expenditure Review Strengthening Public Finance for More Inclusive Growth June 2011 Poverty Reduction and Economic Management Sector Unit East Asia and Pacific Region CURRENCY AND EQUIVALENT UNITS Exchange Rate Effective June 30, 2010 1 US$ = Philippines Peso 46.4 FISCAL YEAR January 1 – December 31 Vice President James W. Adams Country Director Bert Hofman Sector Director Vikram Nehru Sector Manager Linda Van Gelder Task Team Leader Rosa Alonso I Terme ii Acronyms and Abbreviations ADB Asian Development Bank CPI Consumer Price Index AFMA Agriculture and Fisheries CPSD Consolidated Public Sector Deficit Modernization Act CRS Constant Returns to Scale AIDS Acquired Immune Deficiency DAR Department of Agrarian Reform Syndrome DAU Dana Alokasi Umum ALGU Assistance to Local Government (General Allocation Grant) Units DBCC Development Budget Coordination ANCAI Advice of Notice of Cash Committee Allocation DBM Department of Budget and Issued Management APIS Annual Poverty Indicators Survey DEA Data Envelopment Analysis ARMM Autonomous Region in Muslim DOF Department of Finance Mindanao DOH Department of Health AUSAID Australian Agency for International DPWH Department of Public Works and Development Highways BEIS Basic Education Information System DPWH-CA Department of Public Works and BHS Barangay Health Station Highways - Congressional BIA Benefit Incidence Analysis Allocations BIR Bureau of Internal Revenue FC Fiscal Capacity BLGF Bureau of Local Government FDI Foreign Direct Investment Finance FIES Family Income and Expenditure BOC Bureau of Customs Survey BOT Build Operate Transfer GDP Gross Domestic Product BPO Business Process Outsourcing GOCC Government-Owned and -Controlled BSP Bangko Sentral ng Pilipinas Corporation BTr Bureau of the Treasury GRDP Gross Regional Domestic Product CA Congressional Allocation HDI Human Development Index CAC Coalition Against Corruption HDM-4 Highway Development and CALABARZON Region IVA, including the Management Model 4 provinces of Cavite, Laguna, IDA International Development Batangas, Rizal, and Quezon Association CAR Cordillera Adm. Region IRA Internal Revenue Allotment CCT Conditional Cash Transfer JBIC Japan Bank for International CE Compensation Earners Cooperation CDA Cooperatives Development LGU Local Government Unit Authority LTS Large Taxpayer Service CF Congressional Fund MDG Millennium Development Goals CHED Commission on Higher Education MOOE Maintenance and other Operating CI Congressional Initiative Expenses CIT Corporate Income Tax MSMEs Micro, Small and Medium Enterprises CLD Center for Legislative Development MTPDP Medium Term Philippine COA Commission on Audit Development Plan COFOG Classification of Functions of NCR National Capital Region Government iii NDHS National Demographic and Health PO People's Organization Survey PPBS Planning, Programming, and NEDA National Economic and Budgeting System Development Authority PPPs Public-Private Partnerships NFA National Food Authority RATE Run Against Tax Evaders Program NG National Government RATS Run After The Smugglers Program NGA National Government Agency RHU Rural Health Unit NICA National Intelligence Coordinating SAAOB Statement of Appropriations, Agency Allotments, Obligations and NKI National Kidney Institute Balances NSCB National Statistical Coordination SARO Special Allotment Release Order Board SEF Special Education Fund NSO National Statistics Office SEP Self-Employed and Professional NTRC National Tax Research Center SNITS Simplified Net Income Taxation OECD Organization for Economic Scheme Co-operation and Development SSL Salary Standardization Law PDAF Priority Development Assistance SUCs State Universities and Colleges Fund TLRC Technology and Livelihood PEFA Public Expenditure and Financial Research Center Accountability UNESCO United National Educational, PER Public Expenditure Review Scientific and Cultural Organization PHC Philippine Heart Center VAT Value Added Tax PHIC Philippine Health Insurance VRS Variable Returns to Scale Corporation WDR World Development Report PS Personnel Services WEF World Economic Forum PIT Personal Income Tax WHO World Health Organization iv Table of Contents Acknowledgements............................................................................................................................................1 Preface. ...............................................................................................................................................................3 Executive Summary...................................................................................................................................5 Top 10 Policy Reforms Emerging from This Public Expenditure and Revenue Reviews. ....16 Top 10 Data to be Produced, Published, Posted on the Internet and Disseminated Annually......18 The 2010 PER Recommendations and the 2011 Budget.................................................................19 Chapter 1.Trends in Public Expenditures, Key Outputs and Outcomes..............................27 Introduction......................................................................................................................................................29 A. Key Development Performance Gaps..................................................................................................30 .............................................................32 B. Trends in the Economic Distribution of Public Expenditures. C. Trends in the Sector Distribution of Public Expenditures, Key Outputs and Outcomes.....................37 • Basic Education................................................................................................................................38 • Health Sector....................................................................................................................................39 • Transport...........................................................................................................................................42 • Agriculture........................................................................................................................................46 • The Judiciary. ...................................................................................................................................48 • Government Owned and Controlled Corporations (GOCCs). .........................................................49 Conclusions......................................................................................................................................................50 Recommendations............................................................................................................................................52 References........................................................................................................................................................54 Chapter 2. Efficiency of Public Spending..............................................................................................55 Introduction......................................................................................................................................................57 A. Education Spending Efficiency............................................................................................................57 B. Health Spending Efficiency. ................................................................................................................62 C. Transport Sector Efficiency. ................................................................................................................66 Conclusions......................................................................................................................................................68 Recommendations............................................................................................................................................69 References........................................................................................................................................................69 v Chapter 3. Decentralization and Regional Inequality.................................................................71 Introduction......................................................................................................................................................73 A. Contribution of Fiscal Decentralization to Regional Divergence. ........................................................82 B. Two decades into decentralization, regional inequality remains stark in incomes, living standards, and acces to public services.................................................................................................................88 Conclusions . ...................................................................................................................................................90 Recommendations............................................................................................................................................91 References........................................................................................................................................................92 Chapter 4. Equity of Government Spending on Education and Health...............................95 Introduction . ...................................................................................................................................................97 A. Equity in Access to Education . ...........................................................................................................97 • Participation Rates............................................................................................................................97 • Completion Rates. ............................................................................................................................99 • School Leavers. ..............................................................................................................................100 • Service Utilization, Government Spending and Benefit Incidence in the Education Sector..........102 ..................................................................................................104 B. Equity in Access to Health Services. • Households’ Access to Health Services..........................................................................................106 • Service Utilization, Government Spending and Benefit Incidence in the Health Sector...............108 Conclusions....................................................................................................................................................111 Recommendations..........................................................................................................................................112 References......................................................................................................................................................113 Chapter 5. Sufficiency, Efficiency and Equity of Public Revenues.......................................115 Introduction....................................................................................................................................................117 A. Sufficiency.........................................................................................................................................117 • Trends in Tax Collections...............................................................................................................118 • Revenue Performance vis-à-vis Targets. ........................................................................................120 B. Collection Efficiency. ........................................................................................................................123 • Comparative Tax Collection Efficiency. ........................................................................................123 • Decomposition Analysis.................................................................................................................125 • Tax Gap Analysis............................................................................................................................126 • Equity. ............................................................................................................................................127 • Evolution of the PIT.......................................................................................................................127 vi • Value-Added Tax............................................................................................................................130 • Excise Taxes...................................................................................................................................132 C. Overall Tax Burden and Income Inequality.......................................................................................134 • Corporate Income Tax....................................................................................................................136 Conclusions....................................................................................................................................................138 Recommendations . .......................................................................................................................................139 A. Tax Policy..........................................................................................................................................139 B. Tax Administration . ..........................................................................................................................141 References......................................................................................................................................................144 Chapter 6. Medium-Term Macroeconomic Outlook.................................................................145 Introduction....................................................................................................................................................147 ...............................................................................................................................147 A. Baseline Scenario. B. Reform Scenario.................................................................................................................................152 References......................................................................................................................................................155 Tables Table 1: Fiscal Performance (percent of GDP) ......................................................................................................... 30 Table 2: Per Capita Real GDP Growth in East Asia and Pacific................................................................................ 31 Table 3: Selected Health and Education Indicators in East Asia; 2007...................................................................... 31 Table 4: Student Performance on International Tests on Math and Science.............................................................. 31 Table 5: Gross Fixed Capital Formation. .................................................................................................................... 32 Table 6: National Government Capital Expenditures and Net Lending..................................................................... 33 Table 7: Government Spending on Education and Health. ......................................................................................... 33 Table 8: Total National Government Primary Spending............................................................................................ 34 Table 9: Agricultural productivity in Selected East Asian Countries......................................................................... 48 Table 10: Student Performance on International Tests on Math and Science.............................................................. 58 Table 11: Average Public Education Spending and Efficiency Estimates. .................................................................. 59 Table 12: Education Sector Efficiency and Output Indicators..................................................................................... 61 Table 13: Health Spending, Indicators and Spending Efficiency Estimates. ............................................................... 63 Table 14: Health Spending Efficiency and Intermediate Input Indicators. .................................................................. 65 Table 15: Ranking of Transport Infrastructure............................................................................................................. 67 Table 16: Transport Outcome and Spending Indicators and Efficiency Estimates...................................................... 68 Table 17: Philippines: Poverty by Administrative Region, 2000-2006........................................................................ 73 Table 18 : Government Spending on Elementary and Secondary Education in 2007................................................... 87 Table 19: Government Spending on Health in 2007.................................................................................................... 88 Table 20: Net Participation Rates, by Level, by Gender and by Poverty Status, 1999, 2004 & 2007......................... 98 Table 21: Distribution of School Leavers by Sex and Poverty Status, 1999, 2004 and 2007.................................... 101 vii Table 22: Distribution of Public School Enrollment. ................................................................................................... 103 Table 23: Early Childhood Mortality Rates by Wealth Quintiles, 2003-2008............................................................. 105 Table 24: Place of Delivery, 2003-2008. ...................................................................................................................... 106 Table 25: Percentage of children vaccinated, 2003-2008............................................................................................ 107 Table 26: Problems in accessing health care, 2003-2008............................................................................................ 108 Table 27. Distribution of Households Utilizing Various Health Facilities.................................................................. 109 Table 28: Revenue performance vis-à-vis targets (percent of GDP)........................................................................... 121 Table 29: Efficiency of tax collection across selected countries................................................................................. 124 Table 30: Income inequality before and after taxes and transfers............................................................................... 136 Table 31: Reform Scenario Revenue Options. ............................................................................................................. 141 Table 32: Philippines: Baseline Growth Projection, 2006-2016. ................................................................................. 148 Table 33: Growth Projections in the Reform Scenario, 2006-2016............................................................................. 155 Figures Figure 1. Percentage of Capital Outlays and Infrastructure in National Government Expenditures..........................35 Figure 2. Distribution of Total Public Sector Expenditures, 2001-2008. ....................................................................36 Figure 3. Share of Public Sector Spending on Economic Services, Social Services and Debt .................................36 Figure 4. National Government Expenditures by Expense Class as percent of GDP, 1997-2010 proposed ....................37 Figure 5. Real per Pupil National Government Expenditure on Basic Education in 2000 Pesos. ..............................38 Figure 6. Basic Education Completion Rates (1998-2008)........................................................................................39 Figure 7. Real Per Capita Public Health Expenditures (1997-2008)..........................................................................40 Figure 8. National Government Health Expenditure by Economic Classification.....................................................40 Figure 9. Prevalence of Underweight Children Under 5 Years Old (1997-2008). ......................................................42 Figure 10. DPWH Expenditures as a Percentage of GDP (1997-2010). ....................................................................... 43 Figure 11. Philippine National Roads Indicators, Selected Years. ................................................................................ 44 Figure 12. National Government Expenditures on Agriculture as a percent of GDP (1997-2008).............................. 47 Figure 13. Philippines: Government Budget Support to GOCCs................................................................................. 50 Figure 14 . Average Growth Rate by Region in Selected East Asian Countries . ......................................................... 74 Figure 15. Transfers to LGUs rose markedly after the 1991 Decentralization Law . .................................................. 75 Figure 16. IRA transfers account for a growing share of NG tax effort....................................................................... 75 Figure 17. Non-agricultural wages have tended to diverge since 2006........................................................................ 77 Figure 18. Poverty incidence has tended to diverge..................................................................................................... 77 Figure 19. Regional incomes have diverged since 2001 . ............................................................................................ 78 Figure 20. But there has been some modest convergence in regional household spending......................................... 78 Figure 21. Strong β- and σ-convergence across regions in literacy rates…................................................................. 78 Figure 22. but recent divergence in net primary enrollment rates................................................................................ 78 Figure 23. Primary completion rates have diverged..................................................................................................... 79 Figure 24. Immunization rates also show some divergence......................................................................................... 79 Figure 25. Strong β- and σ-convergence across regions in infant mortality rates........................................................ 79 Figure 26. as well as in child mortality rates................................................................................................................ 79 Figure 27. International Comparison: Infant Mortality Rate........................................................................................ 80 Figure 28. Access to a safe water supply...................................................................................................................... 82 Figure 29. The same seems to be the case for access to sanitary toilets....................................................................... 82 viii Figure 30. Concentration curves for LGU revenue, 2001............................................................................................. 85 Figure 31. Concentration curves for LGU revenue, 2008............................................................................................. 85 Figure 32. Concentration curves for LGU sectoral spending, 2001............................................................................... 85 Figure 33. Concentration curves for LGU sectoral spending, 2008............................................................................... 85 Figure 34. Education and Housing Spending Shares by Regions, 2001-2008.............................................................. 86 Figure 35. The richer the region, the higher the exclusive use of private health facilities............................................ 89 Figure 36. The poorer the region, the more likely it is that households do not visit any health facility when sick...... 89 Figure 37. Educational Attainment of Population Aged 17-24 . ........................................................................ 99 Figure 38. Educational Attainment of Population Aged 17-24 . ...................................................................... 100 Figure 39. Incidence of General Government Spending on Education....................................................................... 103 Figure 40. Incidence of Government Spending in Primary and Secondary Education............................................... 105 Figure 41. Incidence of National and LGU Health Expenditure by Type of Service ................................................ 110 Figure 42. Incidence of General Government Spending on Health..............................................................................111 Figure 43. Tax effort (percent of GDP). ....................................................................................................................... 118 Figure 44. Share of taxes in BIR (average for 2006-08). ............................................................................................. 118 Figure 45. Excise tax (percent of GDP)....................................................................................................................... 119 Figure 46. Corporate income tax (percent of GDP). .................................................................................................... 119 Figure 47. Personal income tax (percent of GDP)....................................................................................................... 119 Figure 48. VAT collection (percent of GDP)............................................................................................................... 119 Figure 49. BOC collection (percent of GDP).............................................................................................................. 120 Figure 50. VAT Efficiency Ratio. ................................................................................................................................ 125 Figure 51. Tax decomposition average growth in percent for 2000-05....................................................................... 126 Figure 52. Tax decomposition average growth in percent for 2006-08....................................................................... 126 Figure 53. Tax gap for BIR (percent of GDP(Average in 2000-08)............................................................................ 127 Figure 54. Tax gap for BOC (percent of GDP)............................................................................................................ 127 Figure 55. Income Gini coefficient before and after income tax................................................................................. 130 Figure 56. Income Gini coefficient before and after income tax(sex disaggregated). ................................................ 130 Figure 57. VAT tax burden by percentile (in percent). ................................................................................................. 131 Figure 58. Income Gini coefficient before and after VAT. .......................................................................................... 132 Figure 59. Income Gini coefficient before and after VAT (sex disaggregated)........................................................... 132 Figure 60. Excise tax burden of tobacco and alcohol products by decile (in percent) .............................................. 133 Figure 61. Excise tax burden of gasoline/diesel by decile (in percent)....................................................................... 133 Figure 62. Total excise tax burden by decile (in percent)............................................................................................ 133 Figure 63. Income Gini coefficient before and after total excise taxes....................................................................... 134 Figure 64. Income Gini coefficient before and after total excise taxes (sex disaggregated)....................................... 134 Figure 65. Total tax burden by percentile (in percent) Scenario 1. ............................................................................. 134 Figure 66. Total tax burden by percentile (in percent) Scenarios 2 & 3...................................................................... 134 Figure 67. Income Gini before and after income tax................................................................................................... 135 Figure 68. Income Gini before and after income tax (sex disaggregated)................................................................... 135 Figure 69. Corporate income tax rates (in percent)..................................................................................................... 137 Figure 70. Remittances are projected to perform well as deployments increase and the peso weakens..................... 149 Figure 71. Debt dynamics would worsen as a result of high deficits, weak growth, and rising spreads. ................... 149 Figure 72. Spreads are projected to rise in the no reform-no fiscal consolidation scenario........................................ 149 ix Figure 73. Deficit financing could crowd out domestic investment............................................................................ 149 Figure 74. Domestic interest rates are projected to rise due to increased risk aversion.............................................. 150 Figure 75. Capital spending is set to be negatively affected by the weaker macro environment................................ 150 Figure 76. Share to Exports, Average 2008-2009 ....................................................................................................... 150 Figure 77. Share to Imports, Average 2008-2009........................................................................................................ 150 Figure 78. Terms of trade projected to be stable.......................................................................................................... 151 Figure 79. Growth path, Transition and Recovery. ...................................................................................................... 151 Figure 80. Growth Path under the reform scenario. ..................................................................................................... 154 Figure 81. Reform scenario growth path, transition and recovery.............................................................................. 154 Boxes: Box 1. Vietnam’s Fiscal Decentralization System—Supporting Effective Service and Infrastructure Delivery Nationwide........................................................................................................................................... 83 Box 2. Political Economy of Comprehensive Tax Policy and Administration Reform. .............................................. 122 Annexes: 159 Annex 1......................................................................................................................................................................... 171 Annex 2......................................................................................................................................................................... 175 Annex 3......................................................................................................................................................................... 183 Annex 4......................................................................................................................................................................... 193 Annex 5......................................................................................................................................................................... x Acknowledgements The team of this Public Expenditure Review comprised Rosa Alonso I Terme (Senior Public Sector Specialist and Task Team Leader), Eric LeBorgne (Senior Economist), Karl Kendrick Chua (Country Economist), Qiang Cui (Economist), Justine Diokno-Sicat (Consultant), Rosario Manasan (Consultant), Romulo Miral (Consultant), Sheryll Namingit (Consultant), Erwin Tiongson (Consultant), and Nenette Santero and Mildred Gonzalves (Program Assistants). Research assistance was provided by Jason Alinsunurin, Grendell Vie Magoncia, Rex Willie Semeno and Rashiel Velarde. The team has received overall guidance from Vikram Nehru (EASPR Sector Director and Chief Economist), Bert Hofman (Country Director), Linda Van Gelder (EASPR Sector Manager) and Ulrich Lachler (Lead Economist). Peer reviewers were James Brumby (PREM-PR Sector Manager) Jose Edgardo Campos (Lead Governance Specialist), Tuan Minh Le (Senior Economist) and Truman Packard (Lead Economist). AusAID Manila colleagues participated as members of the overall study team. The team worked closely with the Government of the Philippines. Its principal counterpart was the Department of Budget and Management, led by Undersecretary Laura Pascua. We would like to express our appreciation to DBM for its excellent cooperation and guidance during this process. We would also like to thank Undersecretary of Department of Finance Gil Beltran and his team as well as the Departments of Education, Health and Public Works and Highways. The team is also indebted to AusAID for the kind provision of funding for this report as well as excellent cooperation throughout. 1 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 2 Preface The inauguration of a new Government and Congress in July 2010 offers a window of opportunity to improve development outcomes in the Philippines. The Philippines, a vibrant island nation with enormous potential as a strong economic and social presence in the East Asia region, has not lived up to its growth and development potential. This is due at least in part to weak fiscal policy and management. The new administration has the opportunity to change that by improving the level, efficiency and distribution of public expenditures and revenues, and by strengthening overall governance to achieve faster, more inclusive growth. This Public Expenditure Review (PER) aims to provide information, analysis and guidance to support the Philippines in this important endeavor. Its primary target audience is the incoming government and, especially, its economic and social policy teams. We also hope the document will prove useful to the members of Congress, academia, civil society, the media and the Filipino public at large, as well as other external cooperation partners. The report aims to be overarching, but not comprehensive, focusing on the impact of government revenues and expenditures on development outcomes. Key issues are the amounts, efficiency and distribution of government revenues and expenditures, including both national and local public resources. The report also highlights data gaps that limit the scope of analysis, monitoring and evaluation of public revenues and expenditures. The report builds on the 2003 PER and more recent economic and sector work in the World Bank, including the Philippines Development Reports (2007, 2008) and the Inclusive Growth Report (2010). It also draws from previous sector PERs on agriculture (2007), transport (2009) basic education (2010), and health (forthcoming 2010), and on the Philippines Discussion Notes series (2010). The report does not address issues of public financial management in any detail, given that substantial work is already being carried out separately, including a Public Expenditure and Financial Accountability (PEFA) review in 2009. The team has devoted careful attention to the political economy of policy change. As a result of this focus, it puts forth three proposed avenues that the government and other actors can consider undertaking to facilitate implementation of the report’s policy recommendations. The first is the establishment of a “fiscal pact” between the Executive and Congress on revenue and deficit targets as well as priority expenditures, the second is the establishment of a Data Forum to advocate and monitor improved production and dissemination of data, and the third is a Tax Watch organization to monitor issues related to tax policy and administration. 3 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 4 PHILIPPINES: Public Expenditure Review Executive Summary1 1. The Philippines exhibits important gaps in its growth and development performance vis-à-vis other East Asian countries. The Philippines succeeded in maintaining macroeconomic stability in the aftermath of the 1997 East Asian crisis, but was unable to keep up with other countries in the region in terms of growth and poverty reduction. The country was able to muster positive per capita growth of 2.9 percent per annum during 2000-08, but the other large East Asian economies grew much faster, averaging over 4 percent. Similarly, education and health outcomes in the Philippines have on balance improved over the last decade, but the progress made fell short of that observed in the rest of the region, including lower middle-income countries such as Indonesia and Vietnam. One noteworthy consequence of this relative performance pattern is that the poverty rate reported for the East Asia and Pacific region as a whole is now below the Philippine rate, even though it had been nearly twice as high just two decades ago. 2. The performance gaps exhibited by the Philippines are due, at least in part, to expenditure shortcomings in priority areas. The country is either not devoting enough resources, or not spending them efficiently enough, in priority areas that are closely associated with growth and poverty reduction. Two generally recognized determinants of growth and poverty reduction during the developmental stage are the rates of physical and human capital accumulation. These rates typically depend on the amount of total spending on public infrastructure and on health and education services, as well as on the quality of such spending. Public spending has an important role to play in these areas, which generally comprise the lion’s share of the total public sector primary budget. 3. A key finding of this PER is that shortfalls in the size of public spending generally appear to be more important than shortfalls in the quality or efficiency of public spending in explaining these performance gaps. In certain critical sectors, however, the Philippines also exhibits shortfalls in the efficiency of public spending, defined broadly as the performance outcomes achieved per unit of public spending per annum. An important factor contributing to a lower efficiency in public spending are the disparities in the distribution of public expenditures, observed both across income groups and across geographic regions. That is, public expenditures that are concentrated on fewer needy beneficiaries tend to generate weaker development impacts. These findings suggest that the Philippines cannot hope to achieve the same development outcomes as the better-performing countries in the region through efficiency improvements alone, but will have to make a greater effort in stepping up the quantity of public spending in priority areas. Improvements in the efficiency of public spending can, of course, help in reducing the performance gaps, and this can be achieved by ensuring that the benefits of public spending are distributed more equitably across regions and income groups The first page of this executive summary has been drafted by U. Lachler. 1 5 Philippines Public Strengthening Public Expenditure Finance Review for More Inclusive : Strengthening Public Finance for More Inclusive Growth Growth The Overall Size of Public Spending 4. Public spending in priority areas --basic education, health and transport-- is significantly lower in the Philippines than in comparator countries. Expenditures on basic education, health and transport have been declining as a percentage of GDP and as a share of government expenditures. They are also significantly lower than in comparator countries. The basic education and health sectors exhibit a gap with regional comparators of 2.5 and 0.7 percent GDP respectively while the gap in national government (NG) capital investment is almost 4 percentage points of GDP. This gap is compounded by an equally large shortfall in private investment. All combined, gross fixed capital formation in the Philippines is over 10 percentage points of GDP below that of regional comparators. The expenditure gaps in these three priority areas sum up to an overall spending gap of between 5 and 7 percent of GDP. 5. Lower expenditures on priority sectors mostly reflect a lower overall level of public expenditures rather than differences in the sector distribution of public spending. However, the shares of government expenditures devoted to priority sectors have also declined. Between 1997 and 2010, the shares of the national budget going to the Department of Education, the Department of Health (DOH) and to capital outlays declined from 18 to 14 percent, from 3 to 2 percent and from 20 to 10 percent respectively. Local government expenditures have followed a similar trend, with allocations to education, health and “economic services” (including infrastructure) declining and personnel services and “other purposes” rising. Another reason why priority sector expenditures have declined is that, until 2005, they have been increasingly crowded out by rising debt service payments. This trend was reversed with the fiscal adjustment of 2005-2008. The Evolution of Public Expenditures and Revenues 6. Priority sector expenditures declined between 1997 and 2005 and increased between 2006 and 2009. Primary public spending on basic education, health, and transport all show this pattern, which mirrors the evolution of public revenues. Tax revenues had been eroding continuously since 1997, until a package of tax reforms in 2005 temporarily halted this decline, resulting lower fiscal deficits and a decline of the public debt ratio. This also allowed primary public expenditures to recover slightly over the subsequent four years. In 2009, the macro-economic situation worsened again due to the combined impact of two major typhoons, the global financial crisis and introduction of new revenue-eroding measures by Congress. Revenues declined as a result, but expenditures continued to expand as the government implemented countercyclical policies in response to the global crisis and typhoons, causing the deficit to soar to unsustainable levels. In the absence of new revenue-raising measures, therefore, expenditures in the priority areas, particularly infrastructure, are expected to decline again beginning in 2010 and widen further the existing spending gaps. 7. In education, trends in public spending and key sector outcomes are closely related. As public expenditures on basic education declined during 1997-2005, key education outcome indicators – enrollment rates, completion rates and test scores – deteriorated. Once sectoral expenditures picked up in 2006, these outcomes began to improve again. Overall, however, the Philippines lags behind regional comparator countries in the development of human capital. School enrollment and completion rates are below the levels of other regional middle income countries, including Indonesia and Vietnam. Similarly, the ranking of the 6 PHILIPPINES: Public Expenditure Review Philippines in math and science test scores, a key indicator of educational quality, is well below that of comparator countries, including Indonesia. This weak performance bodes poorly for the future of inclusive growth in the Philippines. 8. The health sector exhibits a similarly close relationship between public expenditures and sector outcomes. As real per capita health expenditures declined between 1997 and 2006, health outcomes in the Philippines improved less than in other East Asian countries with similar levels of sector efficiency but higher levels of expenditure. This suggests that low levels of public expenditures in health constitute a bottleneck. Meanwhile, out-of-pocket health spending has been rising and financial constraints are one of the top reasons cited by lower income groups for not accessing health facilities. As a result, although infant and maternal mortality rates have improved, they continue to be much higher than in comparator countries. Progress toward meeting health Millennium Development Goals (MDGs) is mixed: prospects are good for achieving the infant and child mortality MDG and in combating HIV/AIDs and malaria, but only moderate for reducing the prevalence of underweight children, and poor for reducing maternal mortality. 9. In the transport sector, the quality of infrastructure and size of public spending also seem related. The percentage of national roads in good and fair condition and the percentage of paved national roads have broadly stagnated since 1982, reflecting under-funding of the roads sector and, in particular, of road maintenance expenditures. Road density is on par with comparator countries, but road quality is sub-par. Department of Public Works and Highways (DPWH) expenditures peaked in 2000, but were cut in half between 2001 and 2005, dropping from 1.8 percent to 0.9 percent of GDP. Other indicators of overall infrastructure expenditures, such as levels of capital outlays, public investment and local government expenditures on “economic services,” followed the same trend. DPWH expenditures increased again after 2006 and by 2007, there had been an increase in the percentage in roads in good and fair condition. In 2008, however, DPWH expenditures were still lower than in the peak year of 2000 and are poised to decline again in 2010. 10. The same pattern of public spending since 1997 is also observed in other priority sectors, such as agriculture and the judiciary. Even though they only represent a minor portion of the NG budget, the agriculture sector plays a critical role in poverty reduction as the main source of unskilled labor employment, and the judiciary is essential to shaping the country’s governance environment. From 1997 to 2005, public spending on agriculture declined from around 0.1 percent to 0.04 percent of GDP, and public spending on the judiciary fell from 0.22 percent to 0.15 percent of GDP. In the judiciary, reduced resource availability has skewed the composition of sector expenditures toward personnel services, squeezing out maintenance and operating expenses and impairing the sector’s capacity to provide judicial services, especially for poor and marginalized groups. In agriculture, the negligible amount of NG spending on agriculture has been dominated by the rice subsidies financed through the National Food Authority (NFA), whose off-budget losses rose to over 1 percent of GDP in 2008. The bias generated by the NFA’s policies to promote rice self- sufficiency has been compounded by NG spending on irrigation and production subsidies favoring traditional, mainly import-competing sectors. On the other hand, the provision of public goods, such as R&D, market information, food safety and quality, and market development support has not received sufficient resources. The resulting distortion in production incentives helps explain the Philippines’ lackluster agriculture sector performance over the past decade. 7 Strengthening Public Strengthening Philippines Public Finance Public More Inclusive for Finance Expenditure ReviewforGrowth More Inclusive : Strengthening Growth Public Finance for More Inclusive Growth The Efficiency of Public Spending 11. Evidence on the relative efficiency of public spending in the Philippines vis-à-vis other countries in the region is less clear-cut than evidence on relative spending levels.2 The preceding sections noted that there are significant differences in sector spending levels across countries, and that these differences appear to be correlated with sector outcomes. In contrast, the Philippines does not stand out vis-à-vis other countries at similar levels of development, both regionally and worldwide, in regard to the efficiency of public spending in the social sectors. The findings of this PER corroborate earlier work on the comparative efficiency of social expenditure in the Philippines from 1995 to 2002 on a worldwide basis.3 We do find that the Philippines exhibits comparatively low efficiency in the transport sector. 12. The Philippines exhibits similar efficiency scores as the regional averages in public spending on basic education. The Philippines ranked number three among a group of eight countries in the region in terms of the efficiency with which public spending in basic education translates into primary and secondary net enrollment ratios, primary completion rates, literacy rates and several other sector outcome indicators. Relative to the Republic of Korea as the best performer in this group, the efficiency score of the Philippines is 0.98, above the group average of 0.88. 13. Efficiency scores of education spending across geographic regions in the Philippines exhibit an inverse U-shape form when plotted against the level of sector spending. This suggests the existence of a non- linear relationship between spending and outcomes according to which a minimal level of expenditure is needed to achieve high efficiency while diminishing marginal returns set in after a certain spending threshold has been reached. In basic education, correlations between efficiency scores and intermediate indicators point to physical capacity constraints as the main bottleneck. Considering the large differences in the amount of basic education expenditures across regions, there appears to be some room for efficiency gains by redistributing expenditures from well-resourced regions to less well-resourced ones. Using the average regional group efficiency as a benchmark, however, the magnitude of potential efficiency improvements through reallocation across regions is only around 7 percent.4 These findings suggest that, although some improvements in sector outcomes can be achieved by reallocating existing spending per capita more evenly across regions within the Philippines, the country cannot hope to achieve the same basic education outcomes as other countries in East Asia without significantly increasing the amount of total sector spending. 2 The measure of efficiency in public spending being used here refers to the relationship between the amount of sector spending and sector outputs/outcomes, such that countries exhibiting better outcomes per unit of sector spending are considered more efficient. The methodology used for this estimation is Data Envelopment Analysis (DEA). This measure differs from other commonly used indicators of efficiency, such as the budget execution ratio, which is more a measure of a government’s capacity to spend than of the impact of such spending. 3 Using a data envelopment analysis approach to measure efficiency and data from 1995-2002 (World Bank, 2005), the Philippines achieved an average score of 0.65 in terms of the efficiency with which public spending in basic education translates into primary and secondary net enrollment ratios, primary completion rates, literacy rates and several other sector outcome indicators. This score is very close to the East Asian & Pacific regional average of 0.63 and better than the worldwide average of 0.56. 4 The 7 percent is calculated as the average efficiency improvement needed by regions with efficiency scores below the average efficiency rate to achieve that average efficiency. 8 PHILIPPINES: Public Expenditure Review 14. The efficiency of public health spending in achieving sector outcomes is also in line with comparator countries, but the disparity in health spending across regions is more pronounced than in education. The larger inter-regional differences in efficiency may be due to the fact that the health sector is much more decentralized than the education sector. The inter-regional disparity in public health spending per capita is striking, with a ratio of 5 to 1 from the highest to the lowest. As in education, reallocating health spending across regions could help to achieve better health outcomes. Using the average efficiency of the regions as a benchmark, the potential magnitude of the improvement is around 11 percent. Moreover, our findings indicate that higher efficiency gains are likely to be generated by increasing the share of health sector expenditures devoted to public health, bridging the physical capacity gap in rural health facilities and placing a greater focus on front-line primary health care delivery. As in education, however, a reallocation of existing expenditures is unlikely to bridge the overall resource gap. To close the gap, the resource allocations for the sector would also have to be stepped up. 15. The efficiency of transport expenditures in achieving sector outcomes appears to be comparatively low. Unlike basic education and health, the transport sector exhibits considerable efficiency gaps. Using kilometers of paved road and kilometers of railway per square kilometer of arable land as outputs and national transport spending as the input, we find that the efficiency of public expenditures in the Philippines in delivering those outputs is well below that of comparator countries. This analysis coincides with the recent Transport Public Expenditure Review (World Bank, 2009) in concluding that the process for allocating public resources in the transport sector needs urgent improvement and that higher public spending in the sector would only be effective under an improved institutional and policy environment. 16. The public investment management system is fragmented, politicized and poorly governed. An important reason why the transport sector exhibits low spending efficiency is weak public investment management. Congressional allocations, through which legislators choose projects outside the regular planning and budgeting system, account for about a fourth of total NG transport allocations. These projects tend to be particularly fragmented, ill-executed and unequally distributed across districts. Transparency guarantees are low and allegations of graft and corruption abound. The technical planning process of central government agencies is often over-ridden from above with political motivations. As a result, a large proportion of transport projects are not identified according to their technical efficiency or their effectiveness in achieving development objectives. In addition, many public-private partnership (PPP) arrangements are unsolicited and have not turned out as effective as expected. The Equity of Public Spending in Education and Health 17. General government expenditures are progressive for primary education, less so for secondary education, and regressive for tertiary education. This is due to the fact that, for the poor, school enrollment declines with age and grade level. Overall, general government spending on education was marginally more progressive in 2007 than in 1999. Over this time period, Local Government Unit (LGU) spending on basic education has become more regressive, while NG spending has become more progressive. The distribution of government expenditures on both elementary and secondary education does not seem to be biased in favor of either males or females. 18. Household access to basic health services improved for all income groups between 2003 and 2008, although the access gap across income groups widened. The proportion of women who did not receive prenatal care declined and the percentage of fully immunized children increased, but the poverty gap is large 9 Strengthening Public Philippines Public Finance for More Inclusive Expenditure Review Growth : Strengthening Public Finance for More Inclusive Growth and it has widened. This is due to physical inaccessibility of facilities, financial constraints, and weak demand for health care among the poor, among other reasons. Seventy five percent of mothers with children under five reported having problems in accessing health care in 2008 and cited cost as the main constraint. 19. Poor households rely more on public health facilities than wealthier households. The utilization of LGU-operated provincial/district hospitals is also progressive, although less so than that of barangay/rural health units. In contrast, the use of hospitals retained by the DOH is mildly regressive, as the poor prefer to access local health facilities. 20. NG spending on public health has been progressive, while spending on hospital services is regressive. In the aggregate, NG health spending is regressive, since hospital services account for over three-quarters of total NG health spending. Between 2003 and 2007, total NG spending on the health sector has become less regressive because of the increase in the share of public health expenditures in the overall DOH sector budget. At the same time, LGU health spending has become less progressive because the distribution of benefits from hospital care has become marginally regressive. Regional Decentralization and the Efficiency and Equity of Public Spending 21. The distribution of revenues, expenditures and some key development outcomes across regions in the Philippines is becoming more unequal. Since 2000, per capita incomes, wage levels and poverty rates in the different regions have been evolving along diverging paths. Only household expenditures exhibit some modest convergence. Basic education and access to basic infrastructure indicators are also diverging across regions. Health indicators that can be directly attributed to the health sector, such as immunization rates, have not been converging. The only case of clear and strong convergence can be found in infant and child mortality rates. 22. LGU spending is strongly regressive and has become more so, especially in education. Two of the 17 regions—National Capital Region (NCR) and CALABARZON—account for over 60 percent of total LGU spending from 2001 to 2008, even though they only account for 44 percent of GDP and 26 percent of the total population. The most regressive LGU expenditure category is education. NCR alone accounts for over half of total LGU spending on both education and housing. The stark inequality in education expenditures across regions is largely due to the highly regressive real property tax, which is earmarked for the Special Education Fund (SEF). 23. LGU expenditures account for almost half of total general government non-wage expenditures on basic education. This means that access to learning materials, for instance, is highly unequal across regions. NG per pupil spending on basic education across regions is less unequal, but does not systematically or fully compensate for differences across LGUs. This inequality in public expenditures is accentuating regional differences in human capital accumulation, with serious implications for regional income distribution and development potential in the future. 24. Richer LGUs have higher levels of both LGU and NG health expenditures. There is a positive and statistically significant relationship between per capita household income and per capita LGU spending on both public health and hospital services. The reason is that per capita LGU spending on health is largely 10 PHILIPPINES: Public Expenditure Review driven by the resource envelope of each LGU. Per capita NG spending on hospital services is also found to be positively correlated with per capita household income. 25. The increasingly unequal distribution of public expenditures across regions reflects an increasingly unequal distribution of LGU revenues. The increased inequality in distribution of total LGU revenues is explained by increased inequality in own-revenue mobilization and a less progressive distribution of NG transfers. That is, from 2001 to 2008, total transfers from the NG to LGUs went from being slightly progressive to becoming neutral. These transfers consist mostly, but not entirely, of the Intergovernmental Revenue Allotment (IRA). LGUs own-revenue sources exhibit great disparities across regions and these are becoming more pronounced. LGUs in the NCR collected about half of the total LGU tax revenues from the country’s 17 regions, or about the same amount as the other 16 regions combined. 26. The growing gap in performance across regions is of particular concern because overall service delivery standards seem to have deteriorated since decentralization started. While positive examples of good local governance can be found, they appear to be exceptions rather than the norm. Moreover, data on local government expenditure, outputs and outcomes is particularly poor and local accountability especially weak. These weaknesses, together with low LGU capacity and the absence of systematic information on LGU performance, have resulted in poor local service delivery. Revenue Sufficiency, Collection Efficiency, and Equity 27. Since the late 1990s, the country has been caught in a low-expenditure, low-revenue trap, with tax collections chronically below the levels targeted in government plans and budgets. A key reason why these targets have not been met is the enactment of revenue-eroding measures by Congress. Between 1997 and 2009, the tax effort declined from 17 percent of GDP to 12.8 percent, with a brief interruption in 2005- 07, where tax reforms temporarily halted the decline. Two-thirds of this revenue decline is attributable to the erosion of excise tax collections (which are not indexed to inflation), while the remaining third is due to declining corporate and personal income tax revenues. Leakages are highest among the self-employed at 70 percent and lowest for wage earners at 16 percent. The total tax gap for the Value Added Tax (VAT) and income taxes – defined as potential tax liability minus actual tax collection – is currently estimated at 4 percent of GDP. 28. The current level of public revenues is not sufficient to close the development gap between the Philippines and East Asian comparator countries Declining tax collections have resulted in lower public expenditures, as the government has sought to maintain fiscal balance. This has led to poor public services and faulty infrastructure. The Philippines already exhibited development outcome gaps and sector spending gaps vis-à-vis comparator countries in the mid-1990s. The decline in public spending and revenues since then has widened these gaps. To close them, higher levels of public spending are needed. As argued below, this will require significantly higher tax revenues. 29. Besides not generating enough revenues to satisfy its development needs, the Philippines’ tax system is also highly inequitable across sectors and employment categories. Although the tax system in the Philippines is slightly vertically progressive, the wedge between the taxation of companies benefiting from the maximum available tax incentives and those that do not – at 20 percentage points – is one of the 11 Philippines Public Strengthening Public Expenditure Finance Review for More Inclusive : Strengthening Public Finance for More Inclusive Growth Growth largest in the world. This impairs good governance. It also distorts allocation (when incentives are not redundant) or provides windfall profits to benefiting businesses (when they are redundant). Similarly, tax evasion among the self-employed and professionals is so high that they only pay 20 percent of income tax collections, even though they account for 60 percent of non-corporate income. The remaining 80 percent is being borne by wage-earners. The top 10 percent of income earners in the Philippines reportedly only pay around 1.7 percent of their income in income taxes. 30. Improving the collection efficiency of the tax system would enhance its equity. Improving collection efficiency would require in-depth reforms, particularly the reversal of revenue-eroding measures introduced by Congress that have also increased the horizontal and vertical inequity of the tax system. Even so, the current system is mildly progressive, as the Gini coefficient after taxes (47.7) is lower than before taxes (50.0). If tax payments were made according to people’s actual earnings, the Gini coefficient would likely decline further to 46.0. Key Policy Choices Ahead 31. Closing the Philippines’ performance gaps vis-à-vis other East Asian countries will require a significant reform effort, as well as stepping up public expenditures in priority areas. Aligning public spending in priority areas with the average levels spent in East Asian comparator countries is estimated to require an increase in expenditures of around 5-7 percent of GDP. Of this total increase, between 2 and 3 additional percentage points of GDP are needed for the social sectors, while the other 3 to 4 percentage points of GDP are needed to close the gap in public infrastructure spending. Reforms are needed to improve the investment climate (and thereby help to close an even larger private investment gap), as well as to strengthen the public investment management system. Other priority expenditures discussed in the PER include additional funding to expand the conditional cash transfer program (requiring around 0.3 percent of GDP) and an additional 0.1 percent of GDP for the judiciary sector. 32. The options for financing a 5 to 7 percent of GDP increase in public spending in priority areas are limited and would require a significantly higher tax collection effort. There are three broad options for financing this increase in priority spending: reduced non-priority spending, increased debt financing and higher tax revenues. Because total public expenditures in the Philippines are already quite low, the scope for achieving fiscal savings by cutting back on non-priority programs is limited. Potential candidates for such expenditure-cutting efforts are the rice subsidies granted through the NFA, as well as subsidies granted to some of the other Government Owned and Controlled Corporations (GOCCs) without a clear public goods rationale. The potential fiscal savings to be had from these two sources are estimated at up to one percent of GDP, or far below the expenditure increase needed to close the expenditure gap in priority areas. Similarly, with an estimated public sector deficit of 3.9 percent of GDP in 2009 and total public debt in excess of 60 percent of GDP, the Philippines is in no position to expand the fiscal deficit without substantially raising the risk of destabilization. If anything, this fiscal deficit needs to come down. The only remaining option for financing the bulk of the envisaged increase in priority areas, therefore, is through increased tax revenues. 33. Sustainable reform of tax policies and tax administration will require a firm commitment at the highest political level and sound strategic management at the bureaucratic level. It will also require facing corruption head-on. Given the depth of the governance problems currently affecting the Bureau of Internal 12 PHILIPPINES: Public Expenditure Review Revenue (BIR) and the Bureau of Customs (BOC), an effective institutional reform needs to be accompanied by a broader anti-corruption campaign encompassing other public sector institutions, such as the justice system. Congressional support will also be necessary. A continued focus by civil society, the private sector and the media on good governance in tax administration is also likely to be critical in sustaining the reform process. A “Tax Watch” organization designed to include them as participants could usefully play this role. Political commitment to reform will need to be complemented by sustained leadership in implementation, avoiding the frequent turnover of BIR Commissioners that has bedeviled reform efforts in the past. Tax administration reforms should include monitoring BIR performance through Key Performance Indicators and an amended Attrition Law and strengthened registration, large taxpayers’ service, audit and IT systems as well as improved customer service. We estimate that improved tax administration can yield up to 1.4 percentage points of GDP in additional collection over the medium-term. 34. There are important measures that the government can take to raise the impact of public spending in priority areas, without raising total spending. In particular, the reallocation of education expenditures toward basic education and of health expenditures toward public health would enhance equity and efficiency. This reallocation would also align the function of the public sector more closely toward the production of public goods and services with high positive externalities. On the other hand, higher subsidies to tertiary education are difficult to justify given the large spending gap in basic education, the regressive incidence of tertiary education and the fact that the returns to that expenditure are largely private. Instead, it would be preferable to replace these subsidies by student loans. Scaling up public health expenditures and reforming the health insurance system to expand coverage, especially of the Sponsored Program for Indigents by improving the targeting of the program’s beneficiaries, increasing its support value and improving payments mechanisms, would also contribute to better sector outcomes. 35. Another critical pathway for improving development outcomes is redistributing expenditures across regions and strengthening the transparency and incentive system of LGUs. By ensuring a more equitable distribution of public expenditures across regions, the government can lay the groundwork for provision of equal access to education and health services across the country. Greater equality of access to services would in turn reduce the human capital gap between richer and poorer regions and LGUs, helping to reduce overall income inequality. Ideally, greater equity in revenue allocation would be brought about by introducing a strong revenue-equalization factor to the IRA formula for intergovernmental transfers. Meanwhile, the NG can contribute by systematically compensating for inter-regional and inter-LGU resource differences through the targeted distribution of national government expenditures. Moreover, establishing a transparent and publicly accessible system to monitor and evaluate the performance of LGUs would be of great help in improving service delivery and LGU accountability. Insofar as LGU spending accounts for a rising share of total public spending in the Philippines, the performance of the LGUs plays an increasingly important role in determining the efficiency and equity of total public spending. 36. The Philippine government is at a crossroads. It can continue along a similar path followed over the last decade. This would likely lead to marginal improvements in some areas, resulting in a low-revenue, low-expenditure equilibrium, modest growth, slow poverty reduction and high income inequality and out-migration. Or it can opt for a more forceful reform effort that holds the promise of faster growth, faster poverty reduction and improved governance and income distribution. These alternative options and their consequences are shown in the two scenario simulations in Table I. Both scenarios involve the same assumptions with regard to external developments. 13 Philippines Public Strengthening Public Expenditure Finance Review for More Inclusive : Strengthening Public Finance for More Inclusive Growth Growth Table I. Philippines: Alternative Simulation Scenarios Source: World Bank Projections. 1/ Includes net lending 37. In the absence of policy reforms, the baseline scenario projects growth to be lower than pre-crisis levels, reflecting a less hospitable global environment. In the absence of tax policy reforms, total government revenues would remain broadly constant as a share of GDP, undermining the government’s capacity to step up public spending in priority areas. Moreover, with the increased public debt burden incurred during 2009-2010 and tighter financial markets in the wake of the global financial crisis, average public debt service is projected to increase in comparison to the pre-crisis situation. This implies an increase in fiscal risk and a decline in primary public spending. The absence of more public investment in human capital and public infrastructure, together with a riskier macroeconomic environment, would also deter private investors so that total investment spending in the economy would remain at least as constrained as before, offering little hope for accelerating growth and reducing poverty. 14 PHILIPPINES: Public Expenditure Review 38. The reform scenario is characterized by a comprehensive tax reform, increased investment in human and physical capital, and continued stable macroeconomic management. This scenario assumes the implementation of a tax policy and tax administration reform program in 2010. The reform would aim to increase revenue collections by 3 to 4 percent of GDP, widening the overall fiscal space while preserving macroeconomic stability. Policy measures could include increases in the excises on tobacco, alcohol, and petroleum products (as well as their indexation to inflation) and a broadening of the tax base of the VAT. In the outer years, revenue gains would also be derived from the phasing out of tax incentives, which will be critical to the effectiveness of tax administration reforms in enhancing the efficiency of tax collections. Other reforms include the simplification of the corporate and personal income tax structure. Strengthened tax effort would reduce perceptions of fiscal risk, lower risk premia, and expand the fiscal space available for investment in human capital and public infrastructure. These two developments, together with governance reforms, would significantly contribute to improving the investment climate, resulting in significantly higher domestic investment levels. 39. In spite of a higher public spending profile, the reform scenario involves smaller fiscal deficits and a faster return to a sustainable debt path in the medium term. By 2016, the fiscal deficit (in the government’s definition) is projected to come down to 1.4 percent of GDP (from 3.9 percent in 2009) and the NG debt is projected to reach 44 percent of GDP (from 57.3 percent in 2009). 40. Opting for the reform scenario would permit the Philippines to substantially reduce its spending gap in priority areas vis-à-vis the faster-growing East Asian countries. This offers the hope of significantly reducing the performance gap that has heretofore separated the Philippines from its regional neighbors, ushering in a new era of faster growth and poverty reduction. 15 Strengthening Philippines Strengthening Public Public Public Finance Finance Expenditure for More Review for Inclusive More Inclusive : Strengthening Growth Growth Public Finance for More Inclusive Growth Top 10 Policy Reforms Emerging from this Public Expenditure and Revenue Reviews5 1. Enact a comprehensive tax policy and administration reform package to raise revenues, improve efficiency and equity and improve governance (Chapter 5): a) Tax policy reforms should include, at least, the rationalization of tax exemptions, the broadening the VAT tax base and an increase in rates and indexation of excise taxes; b) Tax administration reforms should focus on strengthening governance by appointing a Commissioner at the head of BIR for a minimum period of 3 years and monitoring BIR performance through Key Performance Indicators; strengthen registration, large taxpayers’ service, audit system, IT systems and customer service. 2. Significantly increase expenditures on basic education and public health (Chapter 1): a) Extend basic education to a 12 year-cycle, plug the shortage in classroom infrastructure and the inequality across regions in access to books and other learning materials; consider introducing pre- school as part of formal education; b) Step up MDG-related health interventions and expand the NG funding of the sponsored program for indigents. 3. Overhaul the public investment management system and step up infrastructure spending (Chapter 1): a) Reduce and consider eliminating Congressional allocations (DPWH and PDAF); failing that, review their regulation to enhance alignment with government objectives and enhance their transparency and accountability mechanisms; b) Systematically scrutinize new projects by subjecting them to the regular planning and budgeting system and enhance their transparency, monitoring and evaluation; c) Review framework for public private partnerships and enhance their scrutiny; d) Once that is done, increase spending on infrastructure and communications. 4. Wind down subsidies to NFA and transfer gains to expansion of CCTs/4Ps program (Chapter 1): a) CCT funds should be increased, especially those for attendance in secondary education; b) Within agriculture, increase expenditures on public goods and market development support, like wholesale markets, market information, R&D, and food safety and quality. 5 These recommendations can be found in the recommendations sections of the indicated chapters. 16 PHILIPPINES: Public Expenditure Review 5. Revise the IRA formula and introduce a strong revenue equalization component; either (Chapter 3): a) Include a measure that captures differences in revenue generating capacity in the allocation formula, such as poverty incidence; b) Or use a set of minimum expenditure levels for the provision of a basic level of services to the population across the country and establish an equalizing transfer to LGUs unable to meet those minimum expenditure levels; c) In the interim, use NG expenditure to compensate for inequality of LGU resources. 6. Strengthen the incentive system, transparency and accountability of LGUs (Chapter 3). Establish a transparent and publicly-accessible system for systematically monitoring and evaluating LGU performance which may be used as the basis for performance incentive systems. 7. Review the management and monitoring system of GOCCs and consider winding down their subsidies (Chapter 1). 8. Improve data production and transparency and consider creating a “Data Forum” to advocate for and monitor progress. (See our recommended Top Ten Data priorities). 9. Strengthen the monitoring and evaluation of key government policies, outputs and outcomes. Use the next MTPDP, sector strategies and the budget’s Organizational Performance Indicators Framework as the basis. (Chapter 1) 10. Consider using a “fiscal pact” with Congress and the next MTPDP to make all this happen. (Chapter 5) 17 Philippines Public Strengthening Public Expenditure Finance Review for More Inclusive : Strengthening Public Finance for More Inclusive Growth Growth Top 10 Data to be Produced, Published, Posted on the Internet and Disseminated Annually Information is critical to the ability of the government, Congress, and civil society to monitor and assess public sector performance. Without relevant, timely and accessible information, one cannot monitor government activity or evaluate its efficiency and effectiveness in achieving development results. Moreover, transparency guarantees are a core element of a sound democratic system. The Philippines has significant gaps in the information needed to assess government performance. Significant gaps in government expenditures, and key outputs and outcomes are simply not produced at both the national and the local government level. The Philippines has the bureaucratic capacity to produce this data and the political system to use to it to enhance performance and accountability. Here are our top 10 picks on critical development data that ought to be produced and publicized: 1. Executed budget expenditures with internationally-recognized classifications (by the end of the first quarter of the calendar year): • National and Local Government (executed) expenditures by full economic and functional classifications (IMF-GFS1986) and by region and province 2. COA accounts matching the budget’s classification 3. Annual monitoring reports of the new Philippine Medium-Term Development Plan 4. Annual report on local government executed budgets and a set of key outputs and outcomes in core areas of LGU responsibility disaggregated to the regional, provincial, city, and municipal level 5. Core infrastructure expenditure, outputs and outcomes • Including DPWH spending on investment and maintenance of national roads by province, kilometers of roads built and maintained by province; LGU spending on roads and kilometers of roads built and maintained by province; aggregate data on the length of national, secondary and local roads by province. 6. Household survey data including enrollment rates in public and private schools and private spending on education and health by province, as well as data on health services use based on self-reported morbidity and type of health expenditure. 7. PDAF and DPWH allocated funds, projects, releases and outputs, including maps 8. Tax collections by tax and taxpayer type and revenue district 9. Key Performance Indicators of BIR 10. Improving National Accounts at the national and regional level and producing selected macro-economic indicators by province We recommend the creation of a Data Forum network to monitor data production. This network could include representatives from academia, government, Congress, the private sector, civil society and the donor community and should ideally be housed at an academic institution. The network could facilitate the identification of data gaps and advocate for their being filled, proposing time- lines and a system for monitoring progress. It could host a website, publish a newsletter and organize periodic workshops on specific topics of interest, from fiscal data to social and infrastructure outputs and outcome indicators, household surveys, local government performance, tax data, and macro-economic data. (Top 10 Policy Recommendations Number 8). 18 PHILIPPINES: Public Expenditure Review The 2010 PER Recommendations and the 2011 Budget The budget for 2011 was submitted to Congress on August 24th 2010 by President Benigno S. Aquino III and it was approved by Congress on December 30, 2010–the first on-time approval in 11 years.6 In his budget speech, President Aquino presented this budget as a tool for social reform with the underlying philosophy of supporting: (1) transparency and accountability, (2) the poor and vulnerable, (3) fiscal responsibility, (4) public-private partnerships, and, (5) zero-based budgeting. The 2011 budget and its announced accompanying public expenditure management policies reflect a greater focus on the efficiency, transparency and accountability of public expenditures. As detailed below, the budget also includes a notable re-distribution of expenditures toward the social sectors, a particularly arduous task in a flat overall budget.7 Resource reallocations have also been effected within sectors and across programs to enhance the effectiveness of public expenditures in achieving development outcomes. However, the increases in expenditure allocations contained in the 2011 budget account for only a small portion of what the 2010 Public Expenditure Review identified as needed to begin narrowing the gap with better-performing East Asian countries. Moreover, the increase in social expenditures was achieved at the cost of a small reduction in allocations to infrastructure. This decline is understandable in the short run. In the medium-term, however, the PER recommended an increase of 3-4 percent of GDP in the public infrastructure budget so as to begin reducing the country’s infrastructure gap. Therefore, despite significant resource re-allocation, expenditure increases in priority areas are insufficient due to the low level of overall resources available in budget. Resource availability will continue to be a constraint until the revenue-enhancing tax policy and administration reform also advised by this PER is enacted. Overall, the expenditure re-distribution and transparency measures reflected in the 2011 budget are fully consistent with the PER’s advice. The enactment of a revenue-enhancing tax policy and administration reform, on the other hand, is only beginning to be discussed. The macro-economic framework for the 2011-2016 Medium-Term Philippine Development Plan (MTPDP) projects an increase in the tax-to- GDP ratio of almost 3 percentage points of GDP. Since tax administration efficiency increases have not yet yielded any increases in the revenue-to-GDP ratio, it is becoming ever clearer that revenue policy and administration reforms are critical to achieving the MTPDP’s revenue, expenditure and overall development outcome objectives. The matrix below presents the Top 10 Policy Recommendations of this PER next to the 2011 reform budget and its accompanying measures. Due to joint work with Government throughout this PER and a coincidence of views with the new administration, many recommendations regarding expenditure allocations and budget transparency have already been taken up while those in public investment management and data production have begun and need deepening. The greatest challenge remains in our number one recommendation –comprehensive tax policy and administration reform. 6 Due to the delay in the budget’s approval in previous years, the budget of the preceding year was automatically re-enacted. These delays caused uncertainty and created problems for resource allocation and budget execution, especially for newly enacted programs, which had to wait till the new budget was approved before spending could commence. 7 The 2011 budget includes 1.645 Trillion pesos in expenditures compared to almost 1.541 Trillion pesos in 2010, signifying an increase of 6.8 percent in nominal and 2.54 percent in real terms. 19 Philippines Public Strengthening Public Expenditure Finance Review for More Inclusive : Strengthening Public Finance for More Inclusive Growth Growth PER 2010 Policy Recommendations 2011 Reform Budget Principles and Policy Measures 1. Enact a comprehensive tax policy and • By mid-2011, the Government was beginning to administration reform package to raise consider potential tax increases, but no tax policy revenues, improve efficiency and equity reform program had been elaborated yet. A bill on and improve governance the rationalization of fiscal incentives continues to await approval in Congress. A 2011-2015 strategic plan for BIR is about to be issued including a set of agency-level key performance indicators according to good international practice. 2. Significantly increase expenditures on EDUCATION basic education and public health • The 2011 budget allocation to the Department of Education is 14% higher in real terms than in 2010. This, however, is only an increase of 0.16% of GDP compared to a PER recommended increase of 1.7%. In line with PER recommendations, this budget is meant to fund the construction of 13,147 classrooms and the creation of 10,000 teacher positions. • Some of the education programs that received an increase in budget include: a) Every Child a Reader Program; b) Education Service Contracting; c) School-based Management Grants. Some were scaled down, such as the Education Voucher System, and others terminated, like the Food for Schools Program, based on a DBM review that highlighted its poor targeting compared to DSWD’s National Household Targeting System. • DepEd’s Education Service Contracting (ESC) program wherein the government contracts with private schools to enroll students in areas where there is a shortage of places in public high schools. HEALTH • The Department of Health received a budget increase of 8% in real terms. This increase, however, represents a 0.01% of GDP, compared with the PER recommended 0.7%. The increased budget allocation is to finance the expansion of some of DoH’s key public and primary health programs, exactly along the PER’s recommendations, including: ◊ Immunization Program; 20 PHILIPPINES: Public Expenditure Review PER 2010 Policy Recommendations 2011 Reform Budget Principles and Policy Measures ◊ Upgrading of health facilities and establishment of 1,278 Basic and Comprehensive Emergency Obstetrics and Newborn Facilities; ◊ Tuberculosis Control Program- Directly Observed Therapy Short Course; and ◊ Provision of low-cost medicines through establishment of an additional 3,931 Botika ng Barangay neighborhood pharmacies. • Improved targeting of government health insurance premium subsidies for poor families by focusing the allocated P3 billion premium subsidies to the 4.7 million poor families as determined by the National Household Targeting System. An additional 500 million was allocated to subsidize the health insurance enrolment of 1.4 million informal sector families in the second lowest income quintile. • National Health Insurance Program (NHIP): The focus in the 2011 budget is to improve the targeting of national government funds for NHIP through the use of the National Household Targeting System to target the poor. 3. Overhaul the public investment TOTAL INFRASTRUCTURE ALLOCATIONS management system and step up • The PER recommends an increase in infrastructure infrastructure spending spending of 3-4% of GDP in the medium term. In 2011, total infrastructure allocations declined by 12% or 0.27% of GDP. Allocations to DPWH declined by 24% in nominal terms, but those to DOTC increased by 81%. For DPWH, the decline was partly due to the transfer of its CAs from the regular DPWH budget to the PDAF. • Additional infrastructure spending is expected from PPPs. To this effect, P12.5 billion have been allocated to DPWH, DOTC and DA as catalytic or support funds for road right of way acquisitions and related infrastructure support for PPPs. CONGRESSIONAL ALLOCATIONS (CAs) • All Congressional Allocations to DPWH have been consolidated under the Priority Development 21 Strengthening Public Philippines Strengthening Expenditure Public Public Finance Review for Finance for More Inclusive : Strengthening More Inclusive Growth Public Finance for More Inclusive Growth Growth PER 2010 Policy Recommendations 2011 Reform Budget Principles and Policy Measures PER 2010 Policy Recommendations 2011 Reform Budget Principles and Policy Measures Assistance Fund (PDAF)—in order to enhance transparency and accountability; • A menu of strategic projects for PDAF in line with Government priorities has been introduced for Congressmen and women to choose from, but these projects still do not undergo the same scrutiny of the regular planning & budgeting system. • The Department of Budget and Management (DBM) is required to post on its official website all releases and realignments of the Priority Development Assistance Funds (PDAF) (Special Provision 5, PDAF); PLANNING, BUDGETING & MONITORING OF PUBLIC INVESTMENT • Tighter scrutiny of DA’s farm-to-market roads; construction to be cleared by DBM based on submission of a design document including: a) standard design of the project (whether it is gravel, cement, asphalt, etc); (b) specific project location with roadmap (if possible, up to the barrio level); (c) identified priority areas; (d) connection of the proposed FMR project to the existing local roads; (e) identified beneficiaries. The design will also consider coordination with key agencies and stakeholders such as the DPWH, DILG, LGUs, CSO, communities, etc. (Special Provision 3, Department of Agriculture); • A new provision mandating each department, bureau, office or agency, especially the Constitutional Commissions, branches of government and offices enjoying fiscal autonomy to publish on their respective websites the following (General Provisions 97 Transparency and Accountability, GAA 2011): ◊ respective approved budgets; ◊ performance measures and targets set in the organizational performance indicator framework approved by the DBM; ◊ major programs and projects to be implemented; ◊ annual procurement plan; 22 PHILIPPINES: Public Expenditure Review PER 2010 Policy Recommendations 2011 Reform Budget Principles and Policy Measures PER 2010 Policy Recommendations 2011 Reform Budget Principles and Policy Measures ◊ contracts awarded and the names of the contractors/suppliers/consultants ◊ targeted and actual project beneficiaries; ◊ fund utilization; ◊ status of implementation; and ◊ program/project evaluation and/or assessment reports. PUBLIC-PRIVATE PARTNERSHIPS • PPP Center was restructured and attached to NEDA (EO No. 8), and a permanent CEO has been appointed. • To promote the development of new PPP projects, the 2011 Budget (GAA) includes P12.5 billion for the PPP Strategic Support Fund, and EO 8 allocates P300 million for the Project Development and Monitoring Fund. • Government will put up a client-friendly one-stop shop and information centers for business permit application and processing; and • Government has identified 10 priority projects primarily in transport sectors and will allocate funds to DPWH and DOTC to adequately prepare technical and financial feasibility studies and for engaging financial advisors; 4. Wind down subsidies to the National NFA SUBSIDIES AND ROLE Food Authority (NFA) and transfer • Allocations to NFA declined by almost 70% in real gains to expansion of Conditional Cash terms or 0.07% of GDP. Moreover, To reduce its Transfers (CCTs) program NFA losses, the Government will: ◊ Reduce rice importation and increase private sector participation; ◊ Limit the selling of low-priced rice to the poorest of the poor; and ◊ Gradually increase the selling price of rice to other consumers. • Push for a Congressional amendment of the NFA charter to separate the proprietary function of the NFA from its regulatory and buffer-stocking function. 23 Strengthening Philippines Public Strengthening Public Public Finance Expenditure Finance for More Review for Inclusive More Inclusive : Strengthening Growth Growth Public Finance for More Inclusive Growth PER 2010 Policy Recommendations 2011 Reform Budget Principles and Policy Measures PER 2010 Policy Recommendations 2011 Reform Budget Principles and Policy Measures SCALING UP CCTs and OTHER SOCIAL PROTECTION • Allocations to CCTs have increased by 103% in real terms or 0.11% of GDP to reach 2.3 million households; • DSWD’S total budget increased by 115% in real terms or by 0.2% of GDP. Aside from CCTs, other programs receiving higher budget allocations are: • 2nd phase implementation of the KALAHI- CIDSS community-development project tin 97 municipalities; • Rice subsidy to farmers and pensions for indigent senior citizens. • Improved public expenditure targeting of social protection programs by identifying the DSWD’s National Household Targeting System as the central database for direct subsidies. 5. Revise the IRA formula and introduce a • No progress yet in revising the IRA strong revenue equalization component intergovernmental transfer system. 6. Strengthen the incentive system, • A new program called the Performance- transparency and accountability of local Based Challenge Fund (P500 million) will be government units (LGUs) implemented by the Department of Interior and Local Government (DILG) to reward good practices in financial reporting and governance by LGUs (Special Provision 1, DILG). • The Bureau of Local Government Finance will now be producing LGU expenditure data according to the budget classification & the “other expenditures” residual category was eliminated. 7. Review the management and monitoring • Allocations to GOCCs, in terms of equity and system of GOCCs and considering subsidy contribution and net lending decreased by winding down their subsidies 58% in real terms or by 0.17% of GDP; • A review of the need for government support to questionable GOCC programs, particularly those which have been shown to be inefficient or not sustainable, such as those for NFA, LRTA, MRTC and other GOCCs. • In November 2010, the first fiscal risk statement ever was produced and published. 24 PHILIPPINES: Public Expenditure Review PER 2010 Policy Recommendations 2011 Reform Budget Principles and Policy Measures 8. Improve data production and transparency • Required website publication of the status of project and consider creating a “Data Forum” to implementation and fund utilization for national advocate and monitor progress government agencies (NGA) (General Provisions Section 97 Transparency and Accountability); • Required website publication by DBM of lump sum fund releases like the PDAF, the DPWH nationwide projects, the DepEd Schoolbuilding Fund and the Internal Revenue Allotment (IRA); • DBM committed to produce and publish on its website the budget execution report for 2011 including all obligation data submitted by line agencies and remaining gaps by end March 2012. • Reduce lump-sum budgetary items under special purpose funds by transferring to the appropriate line agencies; this is the case for the: ◊ Pension of retired military personnel; and ◊ Premium Subsidy to Indigents. Un-programmed funds were reduced by 46% in real terms, declining from 7.2 to 3.9% of the budget. • To enhance transparency, the share of Special Purpose Funds (SPFs) to the total budget decreased from 17.12% to 14.13%. • Improved estimation of the Philippine system of national accounts by revising and rebasing them according to the latest international standards and the conduct of household surveys by increasing the level of disaggregation and frequency of the survey. 9. Strengthen the monitoring and evaluation • NEDA and DBM are working to link the new of key government policies, outputs and Philippine Development Plan and its results matrices outcomes. to sector strategies in key sectors and annual output budgets. 10. Consider using a “fiscal pact” with • There is incipient interest in the development of Congress and the next MTPDP to make such a pact. For the moment, however, there are all this happen two other ongoing initiatives which could totally or partially fulfill the role of the fiscal pact were it not to pan out. First, the Philippine Medium-Term Development Plan which is being elaborated could serve as a tool to engage Congress in a fiscal pact featuring higher revenues and higher development expenditures. Second, the Administration is 25 Philippines Public Strengthening Public Expenditure Finance Review for More Inclusive : Strengthening Public Finance for More Inclusive Growth Growth PER 2010 Policy Recommendations 2011 Reform Budget Principles and Policy Measures advocating for the passage by Congress of a Fiscal Responsibility Bill to reduce the country’s debt to sustainable levels.  Its enactment into law, coupled with prudent spending and a better tax effort would help Government to contain the budget deficit to 2.0 percent of GDP by 2013. The fiscal bill, however, could also under-write a low- tax, low-expenditure, low-deficit fiscal scenario which would not address the shortcomings in development  outcomes outlined in this PER. References: http://www.dbm.gov.ph/index.php?pid=9&xid=30&id=1306 http://www.dbm.gov.ph/index.php?pid=9&xid=31&id=1308 Republic of the Philippines, National Expenditure Program Fiscal Year 2011. Republic of the Philippines, General Appropriations Act Fiscal Year 2011 (Republic Act No. 10147). The President’s Budget Message Fiscal Year 2011 26 Chapter 1. Trends in Public Expenditures, Key Outputs and Outcomes 27 Trends in Public Expenditures, Key Outputs and Outcomes 1. Trends in Public Expenditures, Key Outputs and Outcomes Introduction 1. During the past ten years, the Philippines has been caught in a low-investment, low-growth, high- inequality trap. Public revenue and investment levels have been well below those of comparator economies in the region and elsewhere, including in the key areas of human and physical capital. Its ratings in investment climate and governance have trailed those of competitors. The result has been a rate of GDP growth of less than half the regional average, the highest income inequality in the region and sub-par performance in social indicators. Persistently low levels of investment and high inequality in the distribution of human and physical capital across income groups and regions do not bode well for the future. 2. In the fiscal area, this trap is largely due to cycles of weakening government revenues, growing deficits, and declining productive expenditures. There seems to be a pattern to the macro-fiscal management of Philippine governments over the past two decades. According to this pattern, revenue-eroding measures are enacted, typically by Congressional initiative. These revenue shortfalls first lead to rising deficits and, when a primary deficit emerges and the fiscal stance becomes unsustainable, expenditure is compressed (Table 1).8 3. This pattern can be seen in the fiscal adjustments subsequent to the 1987 crisis. Between 1979 and 1986, government revenues declined by 1.5 percent of GDP while expenditures increased by 3.5 percent. As a result, by 1986, the government was faced with a primary deficit of 1.6 percent and a fiscal deficit of over five percent of GDP. In 1987, the government adjusted by reducing primary expenditure by 2.4 percent of GDP. Because of a simultaneous increase in the wage bill, capital spending was cut from 4.7 to 2.7 percent of GDP. Thereafter, Philippine governments have reacted to revenue shortfalls by adjusting primary expenditures, particularly after reaching a primary deficit. For instance, beginning in 1998, the tax-to-GDP ratio began to decline. The government allowed the deficit to rise until 2002, when a primary deficit emerged rendering the fiscal stance unsustainable. To avoid a fiscal crisis, the government cut primary spending by 1.2 percent of GDP in 2003-2004; half of the reduction fell on infrastructure expenditures. 4. With a primary deficit of 0.3 and a fiscal deficit of 3.9 percent of GDP in 2009, the government is expected to cut primary expenditure in 2010. Given recently enacted wage increases, the weight of the adjustment will likely fall on productive physical investment and non-wage current expenditures in the social and infrastructure sectors. To avoid these expenditure cuts, revenue mobilization efforts combined with vigorous investment climate and governance reforms are needed. The future growth path of the country is at stake. 5. Without policy actions to reduce the fiscal deficit, public finances would increasingly become a source of risk to macroeconomic stability. Growing deficits, along with low growth and rising spreads, would cause public debt to continue rising as a share of GDP over the medium term. Combined with the public sector’s large gross financing requirements, currently at about 19 percent of GDP, this increased public debt would have the potential for creating macroeconomic instability. 8 This may have been a lesson from the early 1980s. Between 1980 and 1983, the government did not respond quickly enough to the rising primary deficit, which led to a fiscal crash in 1984-85. 29 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Table 1: Fiscal Performance (percent of GDP) Source: DOF, BTR 1/ Government definition of revenue and expenditure. This differs from the internationally recognized GFSM. A. Key Development Performance Gaps 6. The Philippines exhibits important gaps in economic growth, poverty reduction and income equality compared with other East Asian countries. In the 1980s and 1990s, GDP growth per capita in the Philippines was less than one-seventh of the regional average (Table 2). Although performance improved during 2000-08, per capita growth was still less than one third of the average achieved elsewhere in East Asia. Furthermore, while the rest of the region recorded strong progress in poverty reduction, poverty indicators improved much more modestly in the Philippines. Meanwhile, income and consumption inequality continues to be among the highest in the region, and it appears to be increasing.9 7. Social indicators show similar gaps. For example, the incidence of tuberculosis and maternal and under-5 child mortality rates is much higher than in comparator countries in the region (Table 3). The country is falling behind even in education, an area in which it had traditionally recorded good performance. Primary enrollment and completion rates are below all the countries in our sample, including Indonesia and Vietnam. Moreover, the country’s ranking in math and science test scores, a key indicator of educational quality, is well below that of other middle income countries in the region, including poorer middle income countries such as Indonesia (Table 4). Because the state of human capital takes the temperature of a country’s economic condition and contributes to its overall economic health, this bodes poorly for the future of economic growth in the Philippines. See World Bank (2010), Philippines Discussion Note on “Achieving Sustained Poverty Reduction.” 9 30 Trends in Public Expenditures, Key Outputs and Outcomes Table 2: Per Capita Real GDP Growth in East Asia and Pacific (average percent per annum) Source: World Bank, World Development Indicators. Table 3: Selected Health and Education Indicators in East Asia; 2007 Source: World Bank, World Development Indicators and MDG database. Table 4: Student Performance on International Tests on Math and Science Source: Trends in International Mathematics and Science Study 2003 and 1999. 31 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 8. These performance gaps are due, at least in part, to shortcomings in public spending. That is, the Philippines is either spending less or spending less efficiently than its neighbors in these priority areas that are closely associated with growth and poverty reduction. Rates of physical and human capital accumulation are generally recognized determinants of growth and poverty reduction. In turn, the level and efficiency of public expenditure in education, health and infrastructure is a key determinant of such investment rates. This chapter will examine trends, outputs and outcomes in the level of public expenditure in these critical areas. The next chapter will examine the efficiency of these expenditures. B. Trends in the Economic Distribution of Public Expenditures10 9. Inadequate investment in human and physical capital seems to be a bottleneck preventing higher and more inclusive growth. The Philippines stands out vis-à-vis other countries in East Asia, exhibiting slower economic growth and a much lower level of total investment. During 2002-2008, the average annual rate of capital formation in comparator East Asian economies was 26.2 percent of GDP, while in the Philippines it was 15.6 percent (Table 5). Moreover, the total level of investment in the Philippines has been declining since the end of the 1990s, raising questions about the sustainability of growth. Table 5: Gross Fixed Capital Formation (as % of GDP per annum) Source: World Bank, World Development Indicators. 10. The gap in total investment is due to differences in levels of both public and private investment. Table 6 shows that the average GDP share of general government investment in the Philippines during 2002-08 (2.7 percent) was less than half the share in comparator East Asian economies (6.6 percent). A large gap has also developed on the side of private investment, which amounted to an average of only 12.5 percent of GDP in the Philippines during 2002-08, compared to 20 percent in the comparator East Asian economies. 10 We do not use the standard GFS definition of economic distribution of expenditures so as to carry out an analysis of “investment” that includes investment in both physical and human capital. 32 Trends in Public Expenditures, Key Outputs and Outcomes Table 6: National Government Capital Expenditures and Net Lending (average annual percent of GDP) Source: World Bank, LDB and World Development Indicators. 11. Declining total investment in the Philippines is mostly due to decreasing private investment,11 though public investment also shows a decline (Table 6). Private investment also declined in comparator East Asian countries, but to a lesser extent. This trend may be reflecting a less attractive investment climate in the aftermath of the East Asian crisis of 1997-98. In contrast to the Philippines, however, most of the comparator countries have resisted declines in public investment spending. 12. The Philippines exhibits similar public spending gaps compared with its East Asian neighbors in human capital expenditures. It devoted an average of 2.5 percent of GDP to public spending on education during 2002-07, and another 1.3 percent on public health in 2006, for a total of 3.8 percent of GDP (Table 7). In contrast, other countries in East Asia spent around 6.1 percent of GDP on average on both sectors, or 2-3 percent of GDP more than the Philippines. Table 7: Government Spending on Education and Health (Percent of GDP) Source: World Bank, World Development Indicators and UNESCO and WHO databases. Note: 1/ Latest year in the period. 11 Table 6 shows that private investment in the Philippines declined by roughly 5 percent of GDP between 1996-2000 and 2002- 2008. This decline is even larger considering that the average investment over 1996-2000 was strongly influenced by the East Asian financial crisis of 1997-98, which led to a large drop of private investment. Before then, private investment had been on the order of 20 percent of GDP. Taking this as the point of reference, private investment has dropped by almost 8 percent of GDP. 33 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 13. Social spending in the Philippines is not only low, but has been declining since the late 1990s. Between 1997 and 2008, total government spending (national plus local government) as a share of GDP declined by 50 percent in the health sector and 36 percent in the education sector. This pattern is contrary to the one observed elsewhere in the region and does not bode well for closing the gap in human capital accumulation that separates the Philippines from the other countries. As will be discussed in the sector-specific sections below, education, health and infrastructure expenditure levels increased between 2007 and 2009, but began declining again in 2010. 14. Low levels of public spending on education, health and infrastructure mainly reflect a low level of overall government expenditure (Table 8). Namely, the expenditure gap in these sectors is mainly due to an overall gap in public expenditure levels between the Philippines and comparator countries. This expenditure gap is as large as 4 to 6 percent of GDP.12 Table 8: Total National Government Primary Spending (average annual percent of GDP) Source: World Bank, LDB and Unified Survey. 15. But there has also been a decline in the share of government expenditures devoted to these sectors. The share of the non-interest NG budget devoted to the priority sectors is only moderately lower in the Philippines than in comparator countries.13 However, some of these shares have been declining. For instance, the share of capital outlays and infrastructure expenditures in the NG’s budget declined by a half between 1997 and 2005 (Figure 1). Although these shares increased over 2006-09 with the overall increase in government revenues and expenditures, they are expected to resume their decline in 2010. This is due to the enactment of the Salary Standardization Law III (SSL III), which is projected to increase the public wage bill by 1.7 percent of GDP between 2008 and 2016. This will raise the share of personnel expenditures at the expense of capital outlays and non-wage current expenditures, which would need to decline by 1.4 percent of GDP by 2016 in the absence of revenue increases. 12 The differences in primary (or non-interest) national government spending are largely attributable to differences in interest payments on the public debt. While total interest payments averaged 1.9 percent of GDP in the four comparator East Asian countries during 2002-08, they averaged 5.1 percent in the Philippines. 13 Over 2002-2007, total government spending on public infrastructure, health and education averaged 6.5 percent of GDP in the Philippines versus 12.4 percent in the four comparator counties in East Asia. These spending amounts translate into 49 percent of total non-interest government spending in the Philippines versus 57 percent in the comparator countries. 34 Trends in Public Expenditures, Key Outputs and Outcomes Figure 1. Percentage of Capital Outlays and Infrastructure in National Government Expenditures Source: Department of Budget & Management (DBM) 16. The share of priority expenditures in both the national and local governments also declined. The share in the NG budget devoted to the Department of Education and the DOH consistently declined between 1997 and 2010, going from 18 to 14 and from 3 to 2 percent respectively. The share of local government expenditures devoted to health, education and infrastructure also declined over 2001-2008 while the share of personnel services and “other purposes” increased.14 Like in the NG, the share of the wage bill in LGU expenditures should continue to rise over the next four years due to the SSL III. 17. In the aggregate, the share of social expenditures in general government expenditures has been on the decline since 2004 (Figure 2). Looking at the consolidated data between national and local government expenditures, we find that the share of social expenditures has continuously declined since 2004. 14 The classification of sectoral classification of local government expenditures does not allow us to distinguish infrastructure. “Economic services” includes agriculture, communication, roads & other transportation. At the elementary level, schools experiencing acute classroom shortages were instructed to hold as many as three or four shifts, with the first shift starting as early as 6 a.m. and the last shift ending as late as 8 p.m. Although the institution of multiple shifts is an understandable stop-gap measure, it cannot substitute for bridging the gap in school shortage if learning outcomes are to be significantly improved. Department of Education Order No. 62, S.2004. 35 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Figure 2. Distribution of Total Public Sector Expenditures, 2001-2008 Source: DBM 18. Debt service has crowded out investment in human and physical capital. Debt service averaged 23 percent of NG expenditures. Figure 3 shows that social and economic services (including infrastructure) as a percent of GDP took the plunge in the past decade while debt service rose. The increase in debt service payments, in turn, is due to the government’s failure to raise sufficient revenues and the resulting emergence of debt-financed deficits (see Chapters 5 and 6). Debt payments declined significantly between 2006 and 2009, but they are expected to resume their increase in 2010 because of the 2008-2009 fiscal stimulus package and the introduction of revenue-eroding measures. The medium-term path of public debt will depend on the government’s macro-fiscal policies as well as its management of the investment and governance climate. As elaborated in Chapter 6, this PER projects a continuation in the rise of the public sector debt-to-GDP ratio in the baseline (non-reform) scenario. In the reform scenario, it projects a significant reduction in the public debt-to-GDP ratio due to a declining deficit following an increase in the tax effort. These measures will be critical both to strengthening macroeconomic stability and providing the needed fiscal space for investment in human and physical capital. Figure 3. Share of Public Sector Spending on Economic Services, Social Services and Debt (as a percentage of GDP) Source: DBM 36 Trends in Public Expenditures, Key Outputs and Outcomes 19. All economic categories of NG expenditure as a percent of GDP declined from 1999 till 2004-2006. This reflected reduced resource availability and rising debt service payments. As a result, between 1998 and 2005, the wage bill, maintenance and other operating expenditures and capital outlays as a percent of GDP declined by 30, 22 and 49 percent respectively (Figure 4). Figure 4. National Government Expenditures by Expense Class as percent of GDP, 1997-2010 proposed Source: DBM 20. Civil servants’ wages and the overall NG wage bill have been on the decline for most of the period. Between 1998 and 2008, NG civil servants’ wages declined from 38 to 28 percent of total NG expenditures and from 7 to 5 percent of GDP. This decline was mainly due to declines in civil servants’ real wages. In particular, nominal wages were completely frozen between 2002 and 2006. As a result, the real wages of civil servants fell by 13 percent between 1998 and 2008. SSL III was passed in 2009 to begin addressing this shortcoming. Under SSL III, civil servants’ salaries are due to increase by a weighted average of 47 percent over a four-year period, with higher increases for higher salary grades. This increase seeks to bring the level of public compensation closer to private sector levels, especially for professional and managerial staff.. To achieve this goal, the SSL also includes an upgrade of the salary grades of lawyers, doctors, teachers, nurses and accountants. Some other work streams needing reclassification of pay grades include statisticians, computer programmers, and weather forecasters. These are some of the remaining areas in which civil servants are most underpaid vis-à-vis private sector comparators, which makes it difficult for government to attract and keep needed professional staff. C. Trends in the Sector Distribution of Public Expenditures, Key Outputs and Outcomes 21. This section reviews the evolution of public expenditures and, to the extent possible, outputs and outcomes, in basic (elementary and secondary) education, health, transport and agriculture. These sectors have been selected because they are particularly important in determining growth and poverty reduction.They They also constitute a significant part of government expenditures. Finally, the World Bank has recently carried out detailed diagnostic work in each of these sectors. The following sub-sections are largely drawn from this work. 37 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Basic Education 22. NG real per pupil spending on basic education declined for most of the past decade. The decline was continuous between 1997 and 2005, at which point recovery began (Figure 5). Trends in consolidated public expenditure per pupil in real terms follow those of the national government as it accounts for 93 percent of total public sector expenditure on basic education. In 2007, basic education accounted for 84 and 87 percent of overall NG and LGU education spending, respectively. Figure 5. Real per Pupil National Government Expenditure on Basic Education in 2000 Pesos Source: Department of Education (DepEd) 23. There seems to be a relationship between public expenditure levels and key education outputs in both primary and secondary education. Pupil-teacher and pupil-classroom ratios increased between 2002 and 2006 and declined thereafter. As expenditures decreased between 2002 and 2006, these ratios tended to increase and they began to decline as public expenditures increased again starting in 2006. After that year, all categories of public expenditure increased, from personnel services to non-wage current expenditures (MOOE) and, to a much lesser extent, capital outlays. However, higher expenditures were not the only reason for these changing ratios. In 2004, the Department of Education estimated a shortage of almost 52,000 classrooms. As a stop-gap measure to address this shortage, it instituted double-shifting in public schools beginning in school year 2005-06.15 Chapter 2 finds physical capacity constraints (reflected in pupil per classroom ratios) to be a key determinant of the efficiency of education expenditures. This is consistent with the finding of the Basic Education PER that less crowded schools tend to be more efficient.16 15 At the elementary level, schools experiencing acute classroom shortages were instructed to hold as many as three or four shifts, with the first shift starting as early as 6 a.m. and the last shift ending as late as 8 p.m. Although the institution of multiple shifts is an understandable stop-gap measure, it cannot substitute for bridging the gap in school shortage if learning outcomes are to be significantly improved. Department of Education Order No. 62, S.2004. 16 Basic Education PER. World Bank, 2010. Chapter V. 38 Trends in Public Expenditures, Key Outputs and Outcomes 24. Enrollment rates, completion rates and test scores mirrored the evolution in public expenditures. Net enrollment rates in both primary and secondary education have declined almost continuously over the past eight years.17 This trend, however, began to be reversed in 2007-08 for secondary and in 2008-09 for primary education. Indicators that are closely related to the quality of education, such as completion rates (Figure 6) and test scores18, closely follow the evolution in public expenditures, deteriorating until 2005- 06 and improving thereafter. The same is the case for indicators of internal efficiency of the sector such as promotion and dropout rates.19 Figure 6. Basic Education Completion Rates (1998-2008) Source: Philippine Statistical Yearbook (PSY) by the National Statistics Coordination Board (NSCB) 25. There seems to be a strong relationship between levels of real per pupil public expenditure on basic education and sector outcomes. As this section has shown, declines in public expenditure on basic education over 2002-2005 were accompanied by declining enrollment and completion rates and worsening test scores. The beginning of the recovery in public expenditures in 2006 started to reverse this decline. Chapter 2 will present an analysis of key determinants of efficiency in basic education while Chapters 3 and 4 will assess the equity of this expenditure across regions and income groups.20 Health Sector 26. Public health spending in real per capita terms has followed a downward trend in the past decade (Figure 7). Although the largest part of the decline is due to the decrease in NG spending, local government expenditures on health also declined. Between 1997 and 2004, NG real health expenditure per capita was cut by half and LG expenditure by one fourth. They both increased in 2007, but declined again slightly in 17 The aggregate net enrollment rate in primary education fell from 90.7 percent in 2002-03 to 83.6 percent in 2007-08 and increased slightly to 83.75 in 2008-09; the net enrollment rate in public schools fell from 83.8 in 2002-03 to 76.9 in 2007-08 and, similarly, increased slightly to 77.1 percent in 2008-09. 18 At the elementary level, the mean percentage test score fell in SY 2005-06 (Figure 9) the year when double-shifting was implemented and real per pupil NG spending was lowest. Starting in 2007-08, all test scores, except for science improved. At the secondary level, test scores were also lowest in 2005-06 (except for math) (Figure 10). 19 Both promotion and dropout rates increased from 2002-03 until 2005-06, and began decreasing thereafter. 20 For a detailed review of the sector, see the Basic Education Public Expenditure Review (World Bank 2010). 39 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 2008. Given the devolution of health services, LGUs contribute the bulk of general government spending on both public health (77 percent) and hospital services (56 percent). In 2007, public health accounted for 24 percent of NG, 51 percent of LGU and 41 percent of general government total health expenditures. Figure 7. Real Per Capita Public Health Expenditures (1997-2008) (In 2000 Pesos) Source: DBM and the Statement of Income and Expenditures (SIE) of the Bureau of Local Government Finance (BLGF) 27. Non-wage current expenditures and capital outlays were hit particularly hard. While the share of NG health expenditures devoted to the wage bill increased from roughly 40 to 60 percent, the shares of non-wage current expenditures and capital outlays dramatically declined. (Figure 8). This decline in their share compounded the absolute decline in real per capita sector expenditures. Given the falling share of education and health in LG budgets and the growing share of wage expenditures, it seems that non-wage current expenditures and capital outlays on the sector also declined at the local level. Figure 8. National Government Health Expenditures by Economic Classification Source: DBM 40 Trends in Public Expenditures, Key Outputs and Outcomes 28. As public expenditures on health decreased, private and out-of-pocket expenditures rose. Between 2000 and 2006, real household medical expenditures rose by 50 percent, placing great financial pressure on household budgets, especially those of lower income groups. As the Health Sector Review documents21, the overall trend throughout the past decade has been one of declining public expenditures and rising private and out-of-pocket expenditures. 29. The reduction in public expenditures has been accompanied by declining key health outputs. Between 1999 and 2005, per capita government health manpower remained stagnant. The same was true of physical facilities. Hospital bed capacity and the number of community clinics per capita both failed to keep up with population growth and declined slightly over the same time period.22 The measles immunization rate also performed poorly, declining from 89 percent in 1997 to 74 percent in 2001. Though it started to recover in 2002, in 2006 it stood at 83 percent, still below the 1997 level. This is worrying, as both immunization rates and barangay health stations per person show a strong positive correlation with estimates of the efficiency of public health expenditure in attaining key sector outcomes, such as lowering infant mortality rates (See Chapter 2). Moreover, the shortage particularly affects the poor, who are especially likely to using rural health facilities (Chapter 3). 30. Progress toward achieving health MDGs is mixed. The Philippines is on track to reach the MDG goal on child mortality rates by 2015. The maternal mortality rate is slowly declining, but the chances of reducing it by three fourths by 2015 are low. The prevalence of underweight children under 5 years of age—classified under MDG 1 of eradicating extreme poverty and hunger—decreased by 25 percent between 1998 and 2006, but it had increased again by 2008. The National Statistical Coordination Board (NSCB) rates progress in this indicator as moderate and the percentage of children underweight remains very high by regional standards (Figure 9). Finally, the 2008 Accomplishment Report of DOH notes that the Tuberculosis target detection rate of 70 percent (an indicator of MDG 6 on combating HIV/AIDS, malaria and other diseases) has already been surpassed and currently stands at 76 percent, while the cure rate stands at 83 percent, already close to the 85 percent target rate to be attained by 2015. World Bank Health Sector Review (2010). 21 Beds per capita declined from 11.4 to 11.0. Barangay health stations per capita declined from 183 to 181 and rural health units per 22 capita declined from 34 to 27. 41 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Figure 9. Prevalence of Underweight Children Under 5 Years Old (1997-2008) Source: World Development Indicators (WDI) 31. Overall, both the absolute level of health indicators in the Philippines and their rate of progress are well below those of comparator countries in the region. Infant and child mortality rates are almost twice the regional average and the incidence of tuberculosis is 2.5 times as high (Table 3). Moreover, as pointed out by the 2010 Health Sector Review, declines in infant mortality rates and increases in life expectancy are less steep than those witnessed in other countries in the region. In addition, the pace of improvement in those indicators in the Philippines –such as in infant mortality rates-- visibly decelerated in the past decade as public expenditures on the sector declined. Higher levels of public expenditure on health are needed to support faster progress and begin reducing the growing gap with East Asian comparators. Poorer regions and lower income groups are in particular need of support (See Chapters 3 and 4). Transport 32. Infrastructure spending as a percentage of GDP has been on the decline over the past decade.23 The expenditures of the Department of Public Works and Highways (DPWH) decreased in the immediate aftermath of the East Asian crisis (1997-99), recovered and peaked in 2000, and then declined continuously until 2005 (Figure 10). A Motor Vehicle‘s User Charge (MVUC) was introduced in the year 2001 to raise additional resources for road maintenance, safety and pollution control. Rising financing from MVUC, however, was countered by a decline in general appropriations for transport.24 In the aggregate, between 23 Because of the absence of a sufficiently disaggregated budget classification, we cannot identify infrastructure expenditures per se. What we refer to in this section are the obligations (again in the absence of actual expenditure data) of the Department of Public Works and Highways. Other departments and programs also include infrastructure expenditures, such as the Department of Transportation and Communication (DOTC) and special purpose funds such as the Priority Development Assistance Fund (PDAF). In addition, government may engage in Public-Private Partnerships (PPPs) or Build-Operate and Build-Operate-Transfer (BOT) projects. Finally, LGUs also undertake small infrastructure projects. Nevertheless, the bulk of public infrastructure spending is carried out through the DPWH. 24 Since 2009, there was no release of MVUC funds from DBM under NRIMP2 project. In the 2011 budget, however, it was re- assessed in line with the new Government’s priorities and the allotment from MVUC is included in the GAA. 42 Trends in Public Expenditures, Key Outputs and Outcomes 2001 and 2005, DPWH expenditures as a percentage of GDP dropped by almost 50 percent, going from 1.8 to 0.9 percent. Other indicators of overall infrastructure expenditures, such as levels of capital outlays, public investment and local government expenditures on “economic services,” all followed the same trend.25 Although DPWH expenditures increased again since 2006, in 2008 they were still lower than in the peak year of 2000. The 2009 Transport PER26 found that overall infrastructure investments stood at 3.1 percent of GDP in 2006, well below the 5 percent level the PER recommended. Moreover, due to the projected increase in the wage bill and the continued erosion of revenues, infrastructure expenditures are projected to decline again over 2009-2012. Figure 10. DPWH Expenditures as a Percentage of GDP (1997-2010) Source: DBM 33. The quality of infrastructure has stagnated over the past ten years. As the Transport PER pointed out, in the case of the Philippines, investments in the quality of transport are even more important than those on capacity expansion (World Bank 2010, 2). The percentage of national roads in good and fair condition and the percentage of national roads paved have broadly stagnated since 1982 (Figure 11).27 On the other hand, as public expenditures on the sector picked up in 2006, the percentage of national roads in good and fair condition rose and the percentage of national roads paved increased slightly. Overall, the low percentage of roads in good condition shows under-funding of the roads sector and, in particular, of road maintenance expenditures (Transport PER). 25 Local government expenditures on transport infrastructure followed the same trend as NG expenditures, growing from 1998 to 2000 and declining up to 2006, when they were roughly half the level of 2000. 2009 Transport PER: 23. 26 Philippines: Transport for Growth. An Institutional Assessment of Transport Infrastructure. World Bank 2009. 27 World Bank Infrastructure Data-base. 43 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Figure 11. Philippine National Roads Indicators, Selected Years Source: World Bank Report No. 4728-PH, Philippines Transport for Growth: An Institutional Assessment of Transport Infrastructure, Feb. 2009. 34. The transport infrastructure network of the Philippines compares poorly with other countries in the region. Road density is on par with comparator countries, but road quality is sub-par. Other transport indicators are also wanting. The Transport PER found that the quality of railroad tracks and services was poor. In the aggregate, the 2007 logistics performance index (LPI) for the Philippines was number 65 out of 150 countries, below Thailand (31), China (30), Indonesia (43) and Vietnam (53). In the 2010 LPI, the Philippines’ ranking improved significantly, moving to number 44 out of 155 countries, ahead of Vietnam and Indonesia. The country’s ranking improved in all dimensions of the LPI except customs. 35. Fragmentation, politicization and poor governance of public investment management processes are at the core of weak sector performance. The level of expenditures is not the only contributor to poor infrastructure outcomes. As the Transport PER documented, institutional issues are critical to explaining the particularly weak performance of the infrastructure sector. Chapter 2 of this PER corroborates that conclusion, finding that, unlike in health and education, the efficiency of public expenditures in achieving transport sector outcomes in the Philippines is particularly poor with respect to comparator countries. This result is due to a number of factors, including the excessive fragmentation of infrastructure expenditures. First, there is fragmentation and poor coordination across various levels of government. Second, public private partnerships (PPPs), especially those that are unsolicited, tend to be poorly planned and coordinated with other government interventions and make inefficient use of limited government resources. 36. Congressional allocations on infrastructure account for about a fourth of total sector allocations. Total congressional funds, including Priority Development Assistance Fund (PDAF) projects and other congressional allocations, are the second highest expenditure item in the DPWH budget, accounting for more than 30 percent of the total (The World Bank, 2009). This means that almost a third of the department’s 44 Trends in Public Expenditures, Key Outputs and Outcomes expenditures are selected outside of the main strategic planning and budgeting process. This significantly weakens the ability of NEDA, DPWH and DBM to plan and budget scarce resources in a coherent and strategic manner. In addition, projects funded out of congressional allocations are identified only after the enactment of the General Appropriations Act. As a result, recipient agencies may not have enough lead time to prepare for their implementation. Moreover, politics does not stop at the project selection phase. There is some evidence that, starting in 2006, fund releases for congressional projects tended to be selective and subject to partisan politics (Boncodin, 2010).28 This may help explain the particularly low execution rate of PDAF and DPWH projects.29 37. There is some evidence that projects funded with Congressional allocations tend to be particularly fragmented and ill-executed (Sicat, 2007). For example, these projects often deliver segments instead of complete stretches of roads. Because these funds are viewed as electoral tools, they often deliver projects not on an efficiency or needs basis, but catering to as many localities and constituents as possible. This is not prevented by the Congressional funds allocations system, as it does not guarantee any consultations between Congress and implementing agencies. The fact that project costs of road concreting projects tend to be very small illustrates this point. As a result, instead of fully paved roads in strategic locations, Congressionally- funded projects often deliver partially concreted roads disperse throughout electoral districts. 38. Differences in Congressional funds per capita across districts are staggering. Because Congressional funds are equally distributed among legislators regardless of the population of their districts, heavily populated districts receive far fewer funds per capita than those which are less sparsely populated. For instance, PDAF fund releases per capita by congressional district over 2003-05 ranged from PHP15 to PHP1307. Similarly, DPWH Congressional Allocation per capita releases ranged from PHP38 to PHP3267 over 2003-2004 and from P6 to P1307 in 2005.30 This variance does not bear any correlation to need. On the contrary, it could be argued that more sparsely populated districts are in greater need of expenditures on local connective infrastructure, have higher costs in public service provision, and tend to be poorer. Therefore, they would warrant greater rather than smaller amounts of public expenditure. 39. Transparency guarantees of Congressional funds are low and accusations of corruption abound. Little information is produced, published and disseminated on those funds and their use. DBM used to publish information on the fund releases, but that information was not sufficient to support proper project monitoring and it is no longer being published. Moreover, because the Commission on Audit does not distinguish between sources of funding of various departmental or LGU expenditures, no specific evaluation has been made of Congressionally-funded projects. Interviews of government officials, legislators, suppliers and contractors by the Philippine Center for Investigative Journalism (1998) point to large scale corruption in the utilization of congressional allocations. In 1998, former DBM Secretary 28 Studies on earlier years (2003 to 2005) found that political party affiliation did not play a significant factor in releases of PDAF and DPWH funds. (Kawanaka, 2007) 29 From 2001 to 2007, the disbursement rate for PDAF and DPWH averaged 74 percent and 75 percent, respectively, while the equivalent rate for the primary national budget (net of PDAF and DPWH agency budget) averaged 87 percent. The fact that infrastructure expenditures tend to have lower execution rates would help explain the lower-than-average DPWH execution rate, but not that of PDAF. 30 Among the districts with the lowest per capita congressional funds because of their big population were the congressional districts of the provinces of Cavite, Rizal, Laguna and the cities of Quezon and Zamboanga. The districts with the highest per capita congressional funds because of their small population were the provinces of Batanes, Siquijor and Camiguin. 45 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Salvador Enriquez estimated that up to 45 percent of congressional funds were lost to graft and corruption. Similarly, PDAF Watch estimated that 7 percent of funds released for road projects in 2004 may have been lost to non-existent road projects while 17 percent may have been spent on defective projects. 40. In addition, the mainstream technical planning process of central government agencies is often politically overridden from above. As a result, a large proportion of infrastructure projects are not identified according to technical efficiency and effectiveness in achieving development objectives. In 2005, for instance, only about 38 percent of the Motor Vehicle User Charge-funded preventive maintenance projects were drawn from the original needs-based list generated by the Highway Development and Management Model 4 (HDM-4). (World Bank 2009). Even the core public investment management process of NEDA and DPWH could usefully be strengthened. This strengthening, however, is likely to yield limited fruits unless the politicization and overall poor governance in the transport sector are resolutely tackled. 41. Therefore, both the low levels of public expenditure and the poor public investment management system need to be addressed if the quality of infrastructure is to improve. The Transport PER concluded that higher public spending on transport infrastructure will be effective only if accompanied by a program to confront institutional and policy distortions. (World Bank 2009). An in-depth reform of the institutions and policies governing the infrastructure sector, jointly with a broader program to improve overall governance in the country, should be a priority of the incoming government. Agriculture31 42. Notwithstanding its high growth and poverty reduction potential, Philippine agriculture continues to under-perform. The agriculture and fisheries sector contributed an average of 15 percent of GDP over the past decade. The Agriculture PER (World Bank 2007) found that a weak policy environment and public expenditure support do not encourage growth and competitiveness in the sector. Farm incomes have lagged non-farm incomes mainly because of low agricultural productivity and slow out-migration from the sector. Despite the potential for higher growth through crop diversification, traditional low-value commodities have continued to dominate production and land use. At the same time, dynamic high-value products with high export potential and income growth, such as tropical fruits, have not been provided sufficient government support. 43. The choice of producing traditional (typically import-competing) agricultural products has been supported through high trade barriers and budget expenditures.32 This has artificially raised the profitability of these products, distorting resource allocation in agriculture and between the agriculture and the non-agriculture sectors. 44. Distortions are the largest in the rice sector. The welfare costs of rice self-sufficiency were estimated by the Agriculture PER to have reached PHP68 billion per year or 1.6 percent of GDP during 2000-2005. Since the food and fuel crisis of 2008-2009, the distortions and costs to the national budget have become even greater. In 2008, NFA’s fiscal costs alone amounted to 1 percent of GDP and the leakage rate of the program is estimated at 50 percent (See Philippines Quarterly Update, November 2009, Box 3). In 2009, the government introduced Family Access Cards in order to better target NFA’s rice subsidies to the poor. The effect of this improved targeting mechanism has not yet been evaluated. This section draws heavily from the Agriculture Public Expenditure Review (World Bank 2007). 31 The major import-competing products are rice, corn, sugar and poultry. 32 46 Trends in Public Expenditures, Key Outputs and Outcomes 45. These costs are disproportionately borne by the poorest Filipinos. Rice self-sufficiency increases income disparity within the sector since only 40 percent of rice farm households obtain two thirds of the benefits from price support. Meanwhile, the poorest members of the rural community, such as small rice farmers, non-rice farmers, marginal fishers, upland dwellers, and landless laborers, as well as the urban poor, bear the costs of the high rice prices that result from the current rice policy. The rice policy has also retarded crop diversification and, together with other protected import-competing wage goods, raised the costs of living in the Philippines, putting upward pressure on wages and undermining the competitiveness of unskilled labor. 46. The allocation of public expenditures has reflected a bias toward supporting traditional commodities over the provision of public goods. During 1998-2005, a more than twofold increase in public spending on agriculture was largely directed to production subsidies, large-scale irrigation systems and the NFA’s operations on rice importation, stock keeping, and distribution. The production subsidies (largely to the rice sector) and irrigation investments (designed to produce mainly rice) accounted for 60 percent of the expenditures of the Department of Agriculture while the share of NFA’s stabilization and tax subsidy in the total agricultural budget (excluding DAR) increased from 4.6 percent in 2001 to 30.4 percent in 2005. Moreover, the contingent liabilities associated with NFA’s borrowing have been rising rapidly. In 2006, they surged to PHP16.4 billion compared to already high PHP10 billion in 2005. 47. On the other hand, NG spending on agriculture as a percent of GDP (excluding the NFA) has declined over the past decade. (Figure 12). NG agriculture spending includes the spending of the Department of Agriculture, Department of Agrarian Reform and the special purpose fund Agriculture and Fisheries Modernization Program (AFMA). Annual expenditures over 1997-2008 averaged 0.6 percent of GDP. They declined between 1997 and 2005 and have increased again since then. Figure 12. National Government Expenditures on Agriculture as a percent of GDP (1997-2008) Source: DBM 47 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 48. Agricultural productivity in the Philippines has grown slowly relative to other countries in the region. Its growth pace has been about half the level of China and Vietnam (Table 9). In addition, while in 1997 the Philippines had the highest crop production index among comparator countries in the region, by 2004 it had been overtaken by Indonesia, Malaysia, China and Vietnam. Table 9: Agricultural productivity in Selected East Asian Countries Source: World Development Indicators (WDI). 49. The best way of increasing the impact of public expenditure on pro-poor agricultural growth in the Philippines is to improve the composition of expenditure. This was the main conclusion of the 2007 Agriculture PER. It recommended the phasing out of rice subsidies, which could be compensated by the expansion of the conditional cash transfers programs. In addition, the rice self-sufficiency policy could be substituted by a policy of food security at the household level. Moreover, the agricultural budget could helpfully be realigned away from commodity-specific support and toward public goods provision and market development support. This would include increasing the sustainability of investments in irrigation and stepping up expenditures on rural roads, wholesale markets, market information, research and development, and food safety and quality. The Judiciary33 50. The resources devoted to the judiciary declined from 1998 to 2005. Between 1998 and 2004, public expenditures on the judiciary as a percentage of GDP were almost cut in half, going from 0.22 to 0.15 percent. Lower revenues and overall expenditures are partly responsible for this decline. In addition, however, the share of the judiciary in the budget declined from 1.2 percent in 1998 to 0.9 percent in 2004. Constrained resource availability has skewed the composition of sector expenditures toward personnel services, squeezing out maintenance and other operating expenses as well as capital outlays. As a result, the condition of physical facilities in lower courts is poor, impairing the efficiency of judicial functioning and the access to justice. The access of the poor and marginalized is particularly restricted. Source: Discussion Note: Strengthening Judicial Performance (World Bank, 2009) 33 48 Trends in Public Expenditures, Key Outputs and Outcomes 51. Inefficient human resource management contributes to weak sector performance. There is a shortage in judges and prosecutors, but overstaffing by administrative personnel. The shortage of prosecutors and public attorneys has led to excessive caseloads, delays in the delivery of justice, poorly prepared cases and low conviction rates. On the other hand, the high level of support staff in courts would provide considerable scope for redeployment: the 2006 Judicial Review Index reported that the average system- wide staff-to-judge ratio was 14.2 in 2005 (12.8 for lower courts and 158.3 for the Supreme Court). 52. The judiciary’s information and communications technology systems need upgrading. Currently, they are characterized by fragmentation, weak interoperability, poor scalability and insufficient resourcing and planning. Due in part to resource constraints, most improvements in IT applications have been stand- alone attempts to improve the performance of individual parts of the system driven by donor priorities. As a result of the absence of a holistic approach, these efforts have not resulted in significant efficiency improvements. In addition, numerous lower courts do not have sufficient computers and IT facilities. 53. Resource availability improved over 2005-2008. Between 2004 and 2009, the real judicial budget increased by 34 percent and the projected budget for 2010 shows a real increase of 69 percent over the 2004 level. Nevertheless, the Supreme Court views this increase as insufficient to address years of under- resourcing. 54. Further increasing the budget of the judiciary and adopting a set of judicial performance indicators could support improved service delivery. Increasing the judiciary’s budget back to the 1997 level by 2016 would require additional resources amounting to 0.1 percent of GDP. This increase, jointly with the adoption of a set of performance indicators and benchmarks pertaining to case backlogs, case disposition rates and case processing, should improve the judicial function and improve access to justice and the overall business climate. Government Owned and Controlled Corporations (GOCCs) 55. GOCCs pose a burden on the government budget. Transfers to the country’s 767 GOCCs have significantly increased over the past 10 years, reaching almost 1.3 percent of GDP in 2008 (Figure 13). Moreover, actual budget transfers to GOCCs are consistently higher than budgeted amounts. In some cases, subsidies are warranted by the GOCC’s performance of quasi-fiscal activities of public utility on behalf of the government. In other cases, however, the welfare rationale for government intervention is less clear. If some of the GOCCs could be privatized or budget transfers to them withdrawn, this could potentially yield fiscal savings of 0.3 to 0.7 percent of GDP, in addition to any one-time privatization revenues.34 34 This estimate is calculated by netting out the cost of rice subsidization through NFA (estimated to be between 0.6 percent and 1 percent of GDP) from the total budget transfers to GOCCs in 2008. In addition, however, it is important to stress that budgetary transfers to the GOCCs only represent a small part of the public support they receive, which includes periodic recapitalizations and the financing of current deficits by issuing debts that are assumed by the national government. The GOCCs also generally operate in a much more favorable environment than private sector companies (e.g., most have tax exemptions and easy access to credit), which has a fiscal cost that usually does not appear in the government budget. See World Bank (2010), “Philippines Development Report.” 49 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Figure 13. Philippines: Government Budget Support to GOCCs Source: Government of the Philippines, Department of Budget and Management 56. Direct budgetary support to GOCCs is only a small part of their fiscal cost. Significantly more public resources are used by GOCCs in the form of off-budget support (the NFA is a prime example), but also through government guarantees, including loans in foreign currencies to companies with no or limited foreign earnings at cheaper interest rates than domestically available. For instance, most of the support that the NFA is receiving from the public purse is off-budget and is mostly financed through government- guaranteed debt that would ultimately require a large recapitalization since the NFA has negative net worth. Overall, therefore, GOCCs are a significant source of non-transparent contingent liabilities. (See the Fiscal Risk analysis of the 2009 Philippines Development Report) 57. Information currently available to the government on the operation of GOCCs is insufficient. The government monitors the 16 largest GOCCs. However, there is a case for revising the overall framework for monitoring the GOCC sector. A strengthened framework could improve the quantity, quality and timeliness of GOCC information available to the executive and enhance its ability to plan and budget. With this enhanced information, it would be better able to monitor the sector’s contingent liabilities, which are a significant source of fiscal risk for the government. (For a review of fiscal risks associated with GOCCs, see Philippines Development Report 2009.) Conclusions 58. Public expenditures in human and physical capital in the Philippines are well below those of other East Asian countries and have been declining over the past decade. This is due both to the low level of public expenditures in the Philippines and to declining shares of national and local government resources devoted to these areas. Education, health and infrastructure expenditures declined over 1997-2005 and began to recover until 2009. Capital expenditures have been particularly affected. Though they began to pick up in 2006, they are expected to decline again as of 2010 due to declining revenues and an increasing salary bill. Over 2001-2008, LGU expenditures on education, health, and economic services declined while expenditures on personnel services and “other purposes” increased. Data to examine key LGU expenditures, outputs and outcomes is weak. 50 Trends in Public Expenditures, Key Outputs and Outcomes 59. As public expenditures declined, key outputs and outcomes in some priority sectors deteriorated. As public expenditures declined, key outputs and outcomes in the education and transport sectors deteriorated and they began to recover with expenditures after 2006. In the health sector, as public expenditures were curtailed, the rate of improvement in key outcomes decelerated despite the marked increase in private and out-of-pocket expenditures. The link between public expenditures and outcomes is less clear in the agriculture sector. This is largely due to the misallocation of agriculture expenditures as well as the importance of other factors in determining sector outcomes. 60. Debt payments have crowded out priority expenditures and GOCCs are a source of fiscal risk. Debt payments as a share of public expenditures rose between 1997 and 2006, crowding out investment in human and physical capital. The share of debt payments declined between 2006 and 2009, but it is expected to begin rising again in 2010. The deterioration in the fiscal position and the attendant accumulation of debt is largely due to the introduction of revenue-eroding measures and the inability to sufficiently adjust expenditures. 61. Closing public spending gaps with other East Asian countries would improve the Philippines’ chances of catching up with their development outcomes. The public spending gap is in the order of 3 to 4 percent of GDP in investment and between 2 to 3 percent of GDP in education and health.35 This yields a total public spending gap in these three priority sectors of between 5 and 7 percent of GDP. Closing this public spending gap will not be enough as the Philippines also exhibits a gap in private investment spending of 7 to 8 percent of GDP with comparator countries in East Asia. Closing the total investment gap, therefore, will also depend on significant increases in private investment that will in turn require substantial improvements in the investment and governance climate. 62. Rolling back non-priority expenditures would reduce the adjustment burden on the tax system. There are a few large expenditure items that merit attention as potential candidates for generating fiscal savings. One of these is the rice subsidy provided by the government through the NFA. Replacing NFA rice subsidies with a well-targeted social protection system and eliminating budget transfers and other non-budget support to the GOCCs could yield potential savings of around 1 percent of GDP. That would reduce the additional fiscal revenue effort required to close the public spending gap in priority areas to about 4 to 6 percent of GDP. Another source of potential fiscal savings is in the budget transfers to the other GOCCs, other than 35 The gap in public investment identified here is similar in magnitude to the shortfall in public infrastructure investment identified in the World Bank’s (2005) infrastructure report on the Philippines. That report had called for a gradual increase in infrastructure investments from 2.8 percent of GDP (in 2008) to at least 5 percent of GDP. Similarly, the gap of 1.6 percent of GDP in public education spending shown in Table 5 is close to the 2% of GDP shortfall in education spending identified in the forthcoming World Bank (2010) public expenditure review on basic education. The World Bank is currently also working on a health sector assessment that seeks to provide a detailed actuarial analysis of the benefits package that should be provided by the Government to achieve better outcomes as well as the costs of mitigating supply side constraints to achieve these goals. Unfortunately, the findings from that analysis are not yet available for comparative purposes. 51 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth the NFA.36 Special Purpose Funds (SPFs), which have grown from under 20 percent of total budget in the late 1990s to 32 percent in 2009, also warrant review. These funds allow for significant discretion in public resource allocation, impair the comparability between budgets and executed expenditures, and are not easily monitored, which seriously impairs the transparency of public finances. 63. The required revenue effort can also be reduced by improving the efficiency of public expenditures. Chapter 2 reviews the efficiency of public expenditures in basic education, health and transport. The chapter points out that significant efficiency gains could be made, among others, by rolling back Congressional allocations, which do not undergo the same degree of scrutiny as expenditures included in the regular planning and budgeting process. 64. Effectively increasing public spending will require improved public expenditure management. This is particularly important in view of the public sector’s chronic under-execution of the budget and lack of budget transparency. The average budget execution rate over the past 10 years has been around 90 percent, with the lowest level being in DPWH. Here, correcting the shortcomings in public financial management identified in the Philippines Discussion Note on Public Financial Management and the recently-concluded Public Expenditure and Financial Accountability assessment would be of great help. Finally, as mentioned in the infrastructure section of this chapter as well as in the 2009 Transport PER, stepped-up expenditures in the infrastructure sector are only likely to bear fruit if preceded by a serious overhaul of the public investment management system. Recommendations 65. Consider increasing expenditures in priority sectors by around 5-7 percent of GDP. For basic education, filling the gap in school infrastructure, hiring additional teachers and extending the cycle to 12 years would cost slightly over 2 percent of GDP. Bringing up public expenditures on health to regional levels requires an additional 0.7-0.8 percent of GDP. Finally, overall infrastructure expenditure should eventually be stepped up by 3-4 percent of GDP. A 0.1 percent increase will be needed by the justice sector. 66. Increase the revenue effort by the same order of magnitude minus any fiscal savings achieved by reducing non-priority expenditures and improving efficiency (See Chapters 1, 3 and 4 for potential efficiency savings in education, health and infrastructure expenditures). 67. Two other potential sources of fiscal savings are the rice subsidies granted through the NFA and the budget transfers made to other GOCCs. The government could helpfully consider reviewing the management and monitoring system of GOCCs as well as their subsidies. (Top 10 Policy Recommendations Number 7). Replacing the rice subsidy with better targeted transfers (such as conditional cash transfers) and eliminating the transfers to GOCCs could reduce the total revenue gap to between 4 and 6 percent of GDP. (Top Recommendation Number 4). The fiscal savings obtainable by eliminating these subsidies and transfers would not accrue to the national government, but to the 36 nonfinancial public sector as a whole. 52 Trends in Public Expenditures, Key Outputs and Outcomes 68. Above all, prioritize investment in human capital. The comparative advantage of the Philippines in the service sector points to the criticality of improvements in the quality of education. The Philippines exhibits promising development prospects in the service sectors, building on its English-speaking population, rapid growth of Business Process Outsourcing (BPO) activities and untapped tourism potential. Fully exploiting this potential will require significantly stepped-up investment in education and telecommunications capacity. In this regard, enhancing the quality and access of basic education is needed to achieve universal completion of primary education, greater participation in secondary education and learning outcomes on par with top performers among middle-income countries. To that end, it will be important to plug the shortage in classroom infrastructure, the inequality across schools in access to books and other learning materials, and to introduce preschool as part of formal education and extend basic education to a 12 year-cycle. Public expenditures on health should also be stepped up to match the regional average with a focus on enhancing the access to health care of the poor and supporting overall progress toward the achievement of the MDGs. This in turn will require increased MDG-related health interventions, upgrading and expanding rural health infrastructure and increasing national government funding of the sponsored program for indigents. (Top Recommendation Number 2). 69. Overhaul the public investment management system. The overhaul should aim to make project selection more technical, less political, and more strategically aligned with the country’s development objectives as defined in its new Medium-Term Philippine Development Plan (MTPDP). In particular, the government should consider reducing and, if possible, eliminating Congressional allocations (DPWH and PDAF). It should review the regulatory framework for any remaining Congressional allocations and for PPPs to foster their alignment with national objectives. To this effect, it should systematically scrutinize all new projects by subjecting them to the regular planning and budgeting system and enhance their transparency, monitoring and evaluation. Public investment management reform is critical to enhancing the efficiency of public investment. A detailed review of public investment management and attendant recommendations for reform could be one of the areas of focus of the next Public Expenditure Review. (Top Recommendation Number 3). 70. Investments in public infrastructure and communications should be increased. Once the public investment management system has been overhauled and revenues sufficiently increased, expenditures on infrastructure and communications should be strategically stepped up. In the short run, increases in total public infrastructure expenditures are unlikely. This is due to the absence of fiscal space owing to the impact of revenue-eroding measures enacted in 2009, the implementation of the SSL III, and the need to rein in the ballooning budget deficit. This period will provide a time window to overhaul the public investment management system. 71. Improved public expenditure management and low execution rates in priority departments need to be urgently addressed. If public expenditures are to be stepped up in priority sectors, the absorptive capacity of the related departments needs to be enhanced. Over the past ten years, the NG budget execution rate has averaged 90 percent, with the lowest-executing department being DPWH. Analyzing and tackling bottlenecks in public expenditure management in priority sectors is thus urgently needed. A recent review of budget execution in basic education found that the key bottlenecks were the late enactment of the budget, procurement delays (particularly the initiation of the bidding process) and weaknesses in the planning capacity of government agencies and spending units. (World Bank 2010) 72. Strengthen overall monitoring and evaluation of government policies, outputs and outcomes. The next MTPDP, sector strategies and the budget’s Organizational Performance Indicators Framework could serve as the basis for a strengthened system. (Top 10 Policy Recommendations Number 9). 53 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth References Herrera, S. and Pang, G. (2005), “Efficiency of Public Spending in Developing Countries: An Efficiency Frontier Approach,” World Bank Working Paper, May. World Bank (2005), “Philippines: Meeting Infrastructure Challenges,” Washington, D.C. World Bank (2007) “Philippines: Agriculture Public Expenditure Review.” World Bank (2009), “Philippines: Transport for Growth,” (Report No. 47281-PH), February 24. World Bank (2009), “Public Expenditure and Financial Accountability Assessment.” World Bank (2009), Philippines Quarterly Update, November 2009, Box 3. World Bank (2010), Philippines Development Report, forthcoming. World Bank (2010), “Philippines: Challenges and Options for 2010 and Beyond,” Discussion Notes, 2010, Draft. World Bank (2010), “Philippines: Basic Education Public Expenditure Review,” forthcoming. 54 Chapter 2. Efficiency of Public Spending 55 Efficiency of Public Spending 2. Efficiency of Public Spending Introduction 73. More effective public expenditure on human capital and infrastructure is crucial to promote growth and equity in the Philippines. The country faces significant challenges in development outcomes: growth performance has been poor; inequality has been on the rise; and progress toward meeting the education and health Millennium Development Goals (MDGs) is under threat. There is evidence that shortcomings in the provision of basic education, health and infrastructure have adversely affected education and health outcomes as well as investment and competitiveness.37 Moreover, efficient and effective provision of human capital and infrastructure is needed to respond to the current global financial crisis, protect the poor, and lay the groundwork for a return to a robust growth path. 74. This chapter provides an analysis of the efficiency of education, health, and transport spending at the aggregate level. Following the literature, efficiency refers to achievement of sector outcomes per unit of sector spending.38 The analysis is focused on basic education, health, and transport, three sectors for which there is reasonable data available in the Philippines and in which public expenditure is particularly critical. The three sectors produce essential public goods and services with significant positive externalities, are critical to generating economic growth and reaching the poor, and have outcomes with robust links to public spending.39 The analysis focuses on benchmarking the country’s overall performance in these sectors in relation to comparator countries to find out where sectoral performance is particularly weak and where it is roughly on par with other countries. The analysis will serve to identify sectoral reforms that are most urgently needed to enhance the efficiency of public funds. 75. The chapter presents three levels of efficiency analysis: a) between peer countries at similar levels of development of the Philippines; b) among the 16 regions within the Philippines;40 and c) between efficiency and intermediate input/output indicators. The findings of this chapter should be complemented with the more in-depth analysis of operational efficiency provided by specific sector reports.41 The analysis is mainly guided by Data Envelopment Analysis (See Annex 1 for methodology and technical description and data), but also draws on evidence from country studies in the relevant sectors. A. Education Spending Efficiency Cross-Country Comparison 76. The Philippines used to have one of the best-educated populations in the region, but this achievement is now under serious threat. As shown in Table 10, its indicators on student achievement have been eclipsed 37 ADB, JBIC and World Bank, 2005; and World Bank, 2009 a; b and 2010 a; b. 38 See, for instance, Herrera and Pang (2005) and Verhoeven and others (2007). 39 In a cross-country study of developing countries, Baldacci and others (2008) document that public spending has a significant impact on primary and secondary education enrollments and on economic growth. 40 Depending on the sector, data are often missing for a few regions in the analysis 41 Basic Education Public Expenditure Review (World Bank, 2010); Health Sector Review (World Bank 2010); Transport Sector Public Expenditure Review (World Bank 2009); Agricultural sector Public Expenditure Review (2007). 57 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth by those in other countries in the Asia Pacific region and there are signs that the situation is worsening. In addition, the Philippines face tremendous challenges in reaching the Millennium Development Goal (MDG) on universal primary education by 2015, with a startling decline in participation rates in primary and secondary schools since the 1998-99 Asian financial crisis.42 Table 10: Student Performance on International Tests on Math and Science Source: Trends in International Mathematics and Science Study 2003 and 1999.. 77. This section examines whether it is the level or the efficiency of spending in basic education that is the main bottleneck to achieving better sector outcomes. We start by comparing the Philippines with its regional peers in spending and education sector performance indicators. Because scores from international tests are not available for many countries, our Data Envelopment Analysis (DEA) focuses on enrollment rates and public education spending.43 The relationship between public spending and enrollment rates in basic education is also supported by the dominant role of public provision in the sector.44 Findings show that public schools account for 93 percent of the primary and 80 percent of secondary enrollment in 2003- 04.45 In addition, public spending has an immediate impact on primary and secondary enrollment rates in developing countries.46 Therefore, there is a sound basis for linking public education spending and outcome indicators in basic education. For cross-country comparisons, national government expenditures are used as inputs due to the absence of total government education expenditures for comparator countries. Since the national government accounts for 93 percent of total basic education expenditure, it should be a good indicator of overall sector efficiency. 42 World Bank 2010. 43 This is consistent with the literature (e.g., Gupta and Verhoeven, 2001 and Herrera and Pang, 2004). 44 It would be desirable to analyze the role of private schools in education sector efficiency, but the required data are not available. While the FIES 1999 provided information on enrollments in private school, such information is no longer collected in subsequent surveys. 45 Manasan and others (2009) 46 As shown in Baldacci et al (2008). The same result is found in the most recent World Bank Public Expenditure Review on Basic Education for the Philippines (2010a). 58 Efficiency of Public Spending Table 11: Average Public Education Spending and Efficiency Estimates 1/ Source: World Bank EduStat database and staff calculation. 1/ Data refer to averages in 1998-09. Also, output VRS DEA scores are presented. Input refers to average national/central government education spending in percent of GDP, while output refers to gross primary and secondary enrollment rates. 78. In recent years, the level of public expenditure on basic education in the Philippines has been well below that of regional peers while its efficiency is largely on par with them (Table 11). The efficiency score is actually above the group average and in between those of China and Indonesia, countries that also have a low level of education spending.48 This result is consistent with Herrera and Pang (2005), who find that the Philippines ranked 4th in education sector efficiency among the same peer group using data for 1996-2002.49 The fact that overall efficiency is in line with peers, combined with the fact that all education outcomes deteriorated as public expenditure was curtailed over 1997-2005 and started to recover once expenditures started to increase again (2006-08), supports the view that the main bottleneck in attaining basic education outcomes is the level of public expenditure.50 In-Country Cross-Regional Comparison 79. The administration of basic education was highly concentrated at the national level in the Philippines until 2001, when the Governance of Education Act was passed to provide an overall framework for decentralization of education management and service delivery to schools. However, the actual practice to date has largely been a de-concentration of authority.51 Therefore, national government spending plays an important role in service delivery in education in the regions. In this analysis of spending efficiency in basic education, the input variable includes both national and local government spending so as to capture the full effect of the role of the public sector. 47 The education spending data used here refer to the cross-country World Bank database to maximize cross-country comparability, which may differ from national data reported by the Department of Education of the Philippines. 48 One needs to be mindful of the limitation of such cross-country comparison given the heterogeneity of these countries and the small number of peers at similar levels of spending. The ranking is generally more reliable than the level of the efficiency scores. 49 The output efficiency score for the education sector of the Philippines was 84%, same as the average of the peer group. 50 Further evidence on the dominant impact of the level of spending has also been found in a recent education sector study (World Bank 2010a). 51 World Bank, 2010a 59 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 80. The level of spending on basic education varies considerably across regions. In 2004-07, Cordillera Administrative Region (CAR) the region with the highest per student expenditure, spent about 9800 pesos per student52, 1.5 times as much as the region with the lowest per capita spending (Central Visayas). In recent years, basic education spending per student has risen in all regions. Comparing the spending from 2004-05 to 2006-07, the average rate of increase is about 8 percent per year and Mimaropa has the highest average increase (23 percent per year). (Annex 1). 81. Efficiency scores also vary significantly across regions and exhibit an inverse U-shape (Annex 6). The spending-efficiency relationship takes an inverse U-shaped pattern, with the highest efficiency being achieved at the medium range of spending. For example, the region with the highest efficiency—Region 4B-Mimaropa—spends about 6600 pesos per student, slightly lower than the median of the regions. The regions with the lowest efficiency (NCR and CAR53) achieve only about 60 percent of the highest efficiency in the country, attained in Mimaropa). The inverse U-shaped pattern suggests that there are some fixed costs that need to be covered before efficiency improvements can take place, and that expenditure starts to show diminishing returns beyond a certain level.54 82. Enrollment and completion efficiency rates are positively correlated, but dispersion across regions in enrollment efficiency is greater than in completion efficiency. Two additional component efficiency scores have been examined through DEA to detect the bottlenecks in raising overall efficiency. The first is enrollment efficiency, which measures the efficiency with which students have been enrolled in primary and secondary schools at a given level of spending. The second is completion efficiency, which measures the efficiency with which the survival and completion rates at primary and secondary schools have been achieved at a given level of spending. The results indicate that these two efficiency scores are positively correlated, but have different dispersion levels across regions. The dispersion of enrollment efficiency scores is about 20 percentage points higher than that of completion efficiency scores. This means that the level of efficiency of regions differs more in raising enrollments than in ensuring completion. In turn, this finding suggests that the disparity in physical capacity to support enrollment across regions is likely to be more of a bottleneck than quality differences in improving education sector efficiency. 83. The variation in efficiency rates across regions points to potential for efficiency gains. For example, NCR has a high level of per student spending, but ranks low in both completion and enrollment efficiency. Since high efficiency tends to be achieved at the medium range of spending, education expenditure could be rationalized among regions to better match their needs for education resources and revenue capacities so as to achieve more balanced levels of spending per student and higher efficiency. Using the average group efficiency as a benchmark, the magnitude of the potential efficiency improvement is about 7 percent.55 84. Physical capacity constraints appear to be a bottleneck for efficiency improvements. An analysis of the correlations between the efficiency scores and education sector output indicators point to physical capacity 52 Data refer to nominal annual spending by the national government and LGUs. 53 CAR has mountainous terrains and thus may have high unit costs, which may contribute to the low efficiency in the region. The difference in efficiency levels, however, is unlikely to be explained fully by differences in unit costs. 54 Due to limitations of the DEA methodology, we cannot provide accurate point estimates on the optimal level. 55 The estimate of this magnitude is based on the average efficiency improvements achievable in regions with relative efficiency lower than the group average. 60 Efficiency of Public Spending constraints as the main efficiency bottleneck. As shown in Table 12, the ratio of students per classroom in primary schools is the only variable that is negatively correlated with the efficiency scores. This finding is consistent with sector studies. For example, the Philippines Institute of Development Studies (PIDS) (2009) found that in school year 2004-07, newly created teacher positions amounted to 80 percent of the required figure while newly built classrooms only covered 49 percent of the required figure. Thus, on the margin, lowering the number of students per classroom would contribute more to improving efficiency than lowering the student to teacher ratio. This finding is also consistent with the large gap in enrollment efficiency. The composition of spending could be adjusted to support more balanced capacity for basic education provision across regions within the same spending envelope. Table 12: Education Sector Efficiency and Output Indicators Source: 2010 Education PER and staff calculations. 85. While greater inter-regional equality of spending could increase average efficiency, the scope for expenditure re-allocation across regions is limited. A key source of funding for basic education is the Special Education Fund (SEF), which depends on collections from real estate taxes. As those are highly unequal across regions, so are SEF expenditures. A revision of the formula for the internal revenue allotment to include a strong equity factor would be of great help in this regard (see Chapter 3 on regional equity). Failing that, the national government will need to play a critical role in diminishing inter-regional inequality in basic education expenditures in order to enhance their efficiency in achieving sector outcomes. 86. Moreover, regions with higher levels of per student spending tend to deliver better attendance and enrollment rates56. Out of the top five performing regions (NCR, Ilocos, CAR, Central Luzon, and CALABARZON), four also rank among the top five in per student spending (NCR, ranked 4; Ilocos, 3; CAR, 1; and CALABARZON, 5). In addition, the region with the highest efficiency —Mimaropa—also has the highest growth in spending per student from 2004-05 to 2006-07. 87. Overall, delivering better basic education outcomes in the Philippines mostly depends on raising the level of public education spending per student. This will require not only higher budget allocations, but also improved absorption capacity in the Department of Education (see Chapter 1). Improvements in the level of sector efficiency will require greater attention to education spending across regions, in particular infrastructure spending. Finally, conditional cash transfers (CCT) to the poor will also be needed to step up demand for education services to match improved supply. This is particularly true for secondary education in which poverty gaps are especially acute (see Chapter 4). Data are based on APIS 2004. (World Bank 2010a) 56 61 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth B. Health Spending Efficiency Cross-Country Comparison 88. In this section, we benchmark the efficiency of public health spending in the Philippines with peer countries in the region. The Philippines underperforms the group average in three key health indicators: life expectancy, child mortality, and maternal mortality. In particular, WHO (2009) finds that only 68 percent of children under 5 have the normal weight-for-age using the National Center for Health Statistics/ WHO standard; “progress to curtail neonatal deaths is dismal, with death rates among this age group showing only the barest decline over the past 20 years;” and the Philippines is “statistically off-track” for achieving the MDG on reproductive health by 2015. 89. Spending for both public and private health services in the Philippines is significantly below the regional average. Public spending on health is over a fourth lower than the regional average. Private health spending is also lower, further constraining the improvement of health outcomes in the country. Because expenditure on public health is progressive and the use of private health care facilities is very limited among lower income groups, stepping up and improving the efficiency of public health expenditures is key to improving the health status of the poor (Chapter 4). For cross-country comparisons, we are limited to using national/central government expenditures due to the lack of availability of total public expenditures on the health sector in comparator countries. This is an important caveat since the health sector in the Philippines is decentralized and Local Government Units (LGUs) play an important role in health sector delivery. Nevertheless, our results are fully consistent with findings of detailed sector studies. 90. Efficiency in utilizing health spending to achieve sector outcomes is roughly on par with peers. While private health spending in the Philippines is significant57, empirical literature supports a strong link between public spending and some health indicators, particularly those related to MDGs. For example, Baldacci et al. (2008) find that public spending on health has a significant link to child mortality. This link guides our application of DEA to infant and maternal mortality indicators as outputs and per capita national government (NG) and LGU spending on health as inputs while controlling for the level of private health spending.58 Because DEA estimates tend to be less accurate in estimating the efficiency of particularly low values of inputs in cross-country comparisons, one should be cautious in drawing definitive conclusions. However, Herrera and Pang (2005) have similar results and rank the Philippines 4th in health sector efficiency59 among the same peer group using data for 1996-2002. Overall, therefore, evidence does not point to the presence of substantially higher inefficiency in health spending in the Philippines compared to other countries. 57 The median estimate of private medical expense per capita based on FIES 2006 is about 530 pesos among the regions within the Philippines. 58 Private health spending is included as a separate input. 59 The output efficiency score of the Philippines was 91%, which is on par with average of 93% based on 4 peer countries for which data were available: China, Indonesia, Malaysia, and Thailand. 62 Efficiency of Public Spending Table 13: Health Spending, Indicators and Spending Efficiency Estimates Source: Department of Finance and NCSB of the Philippines; and staff estimates. 1/ Data refer to central government spending. 2/ Assuming public spending as the only input. 3/ Controlling for both public and private spending. While the level of efficiency estimates change, the ranking remains consistent. In-Country Cross-Regional Comparison 91. The health sector has undergone substantial decentralization. LGUs now play a significant role in financing and providing health services. In 2006, median LGU health spending per capita was 110 pesos and accounted for 80 percent of general government spending on health (NG spending by region and LGU spending, excluding hospitals).60 92. The disparity in public health spending across regions is striking. In 2004-06 CAR, as the region with the highest level of per capita public health expenditure, spent about 340 pesos, or 5 times as much as the regions with the lowest levels (Northern Mindanao, Zamboanga Peninsula, and Davao). In the three regions with lowest spending, the majority of the spending gap comes from differences in LGU spending. On average, LGU spending accounts for 95 percent of the gap between these regions and the top-spending region. Private health expenditures compensate for the disparity somewhat, but the gap between top and bottom spenders remains at about 2:1 after private expenditures are included. LGU health spending refers to spending on health, nutrition & population control as compiled by the Department of Finance; NG 60 spending refers to DOH obligation by region. To match the latest health indicators in 2006 to conduct DEA, NG spending data in 2006 are used, but these data do not cover hospitals. 63 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 93. The efficiency gap in the health sector is also more pronounced than in education. For example, CAR is found to have only half the efficiency level achieved in the Davao region. Therefore, the efficiency gap is 5 percentage points higher than in the education sector. While the health spending efficiency score has a strong negative correlation with the level of per capita health spending, the relationship between efficiency and spending does not form a clear inverse U-shaped pattern as in education.61 94. The disparity in spending and efficiency levels carries into health outcomes. While regions such as CAR that enjoy a high level of per capita spending record good indicators, those with low spending exhibit poor results despite having relatively high efficiency. For example, the infant and maternal mortality rates in the Davao region are 30 and 65 percent higher than the national average respectively. Addressing such disparities is critical to achieving the primary health sector’s goal of “better health for the entire population.” 95. Reallocating health spending across regions could contribute to achieving better health outcomes. Since low-spending regions achieve higher efficiency than high-spending regions, diverting some health spending to the low-spending and high-efficiency regions can improve overall health sector outcomes within the same spending envelope. Using the average efficiency of the regions as a benchmark, the potential magnitude of the improvement is about 11 percent.62 This magnitude is significantly above the 7 percent reported for the education sector. As in the education sector, the reallocation could be achieved either through a reform in the IRA or through the reallocation of national government expenditures. 96. The different performance of the health and the education sectors holds lessons on the impact of decentralization. The higher devolution and higher spread of inefficiency in the health sector compared to the less-decentralized education sector point to decentralization as a source of dispersion in performance. Moreover, the dispersion of physical infrastructure-dependent enrollment rates in the education sector similarly points to the variation in local level funding through the Special Education Fund as a key driver. The World Bank Basic Education PER also found that more efficient schools relied less on LGU-funded teachers. 97. Improved coordination across various levels of government in the health sector is critical. A great degree of responsibility for the health sector has been devolved from the DOH to about 1600 LGUs. Currently, provincial governments oversee provincial and district hospitals, municipal governments manage rural health units and barangay health stations and the DOH maintains specialty and regional hospitals and medical centers. Coordination between DOH and LGUs in service provision is key to ensuring satisfactory 61 The absence of a clear uniform relationship could reflect the fact that health outcomes are affected by more complex factors than education outcomes, including demographics, epidemiology, education, income and environmental factors. DEA analysis only provides a broad-brush analysis to detect whether there appears to be significant efficiency issues between the level of public spending and health sector outcomes. 62 The estimate of this magnitude is based on the potential efficiency improvements that can be achieved in regions with relative efficiency scores lower than the group average. 64 Efficiency of Public Spending sector outcomes. However, the World Health Organization (WHO) (2009) reports that decentralization and other health sector reforms led to lack of coordination and excessive fragmentation of service delivery.63 98. Poorer LGUs do not have the means to deliver on their responsibilities. Decentralization of responsibilities needs to be supported by sufficient financing for the decentralized system to respond effectively to local needs. This has been a particular challenge for LGUs with lower levels of local revenue. For instance, PIDS (2009) reports that while the Vitamin A program supported by NG purchase achieved considerable success, some LGUs were not able to fulfill the delegated obligation to provide other essential supplements to children due to financial constraints and the high cost of small scale procurements. Table 14: Health Spending Efficiency and Intermediate Input Indicators Source: Department of Finance and NCSB of the Philippines; and staff estimates. 99. Increasing the share of expenditures on public health would boost efficiency. An analysis of health spending efficiency with intermediate input indicators points to several directions for improvement (Table 14). The correlation between the efficiency score and the immunization rate among children, which can serve as a proxy of spending on public health, is positive. The potential for improving the efficiency of health spending through higher vaccination rates is corroborated by sector studies. For instance, WHO (2009) reports that 8 out of the 10 leading causes of morbidity in the Philippines are caused by communicable diseases. Similarly, PIDS (2009) finds that child and maternal mortality rates are high relative to peer countries at similar income levels and that these mortality rates can be significantly reduced by public health measures. Therefore, a higher share of spending devoted to public health is likely to boost the overall efficiency of public spending on the sector. 100. A greater focus of human resources on front-line primary health care delivery is likely to generate high efficiency gains. For example, an increase in trained birth attendants working on front-line services is likely to be more effective in raising efficiency than an across-the-board increase in health workers. This could be particularly useful given the country’s relatively high maternal mortality rate. It is also consistent with the finding by the Center for Legislative Development (CLD 2008) that 61 percent of births are home-delivered. The negative correlation between health sector efficiency and the number of doctors per capita is consistent with research findings on the unequal distribution of doctors among the regions. CLD (2008) reports that only 63 Several Government initiatives are aimed at improving coordination across levels of Government. Under the Health Financing Strategy of the Fourmula One (F1) for Health program, there is a DOH initiative to improve coordination of national and local spending for health. Similarly, the Annual Operations Plans of the Province-wide Investment Plan for Health (PIPH) articulate the investment requirements of provinces and their component municipalities to deliver on nation-wide priorities. In addition, major programs such as the Health Facilities Enhancement Program (HFEP) are being prioritized among LGUs based on their PIPHs. Finally, the DOH’s MTEF outlines the logistic coordination requirements between NG and LGUs and their budgetary implications in order for each level of Government to be able to deliver on their respective responsibilities. 65 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 10 percent of the doctors, dentists, and pharmacists practice in rural areas while the vast majority of doctors and nurses operate in NCR and the high growth region of Southern Luzon. Therefore, the unequal distribution of doctors between rural and urban areas could be a key source of inefficiency in the health system. 101. Bridging the physical capacity gap in public health facilities can raise efficiency. Efficiency estimates are strongly correlated with the number of barangay health stations per person. This is consistent with the finding in CLD (2008) that the poor and rural populations use public health facilities more frequently than the rich, who predominantly use private facilities.64 Therefore, the relatively high maternal and infant mortality rates of the poor and rural households can be more effectively reduced by building more barangay health stations in their neighborhoods. Currently, however, CLD (2008) finds that the distribution of health care facilities is skewed towards the NCR and other economically well-off cities and provinces. Hence, bridging the physical capacity gap by rationalizing the development of primary health care facilities is important not only for equity reasons, but also for boosting efficiency. 102. Improving health outcomes will require stepped up public expenditures. Given the comparatively low levels of public and private health spending in the Philippines, reaching the health MDGs will also require higher levels of per capita spending. The adequate provision of health care, particularly of public health and basic health services to the poor, is also important for building human capital and supporting the inclusive growth that the Philippines is pursuing. C. Transport Sector Efficiency 103. Transport infrastructure is critical in a globalized economy (World Bank 2009a). Infrastructure significantly contributed to the integration of export-oriented East Asian countries with global markets.65 Transport accounts for about a quarter of the needs for investment and maintenance of infrastructure expenditures in the region. Therefore, examining the efficiency of the transport sector can provide valuable information about a significant portion of the infrastructure sector that is crucial to the country’s future growth path. This section examines the efficiency of transport expenditure in the Philippines in achieving this sector’s outcomes. Because of the absence of data disaggregated by region, no analysis of within-country cross-regional efficiency is carried out. Cross-Country Comparison 104. Insufficient transport infrastructure appears to be a constraint for growth and development in the Philippines. In surveys of business executives on transport infrastructure quality, the Philippines is ranked at the bottom among peer countries, even underperforming Cambodia, a low-income country, in the perceived quality of road, port, and overall infrastructure (Table 15). The Philippines also lags behind its peers in several other indicators of transport spending and infrastructure. Moreover, the density of paved road and railway in the peer group is 1km and 3.7 km/km2 arable land respectively while for the Philippines it is only 0.7 km/km2and 0.9 km/km2 respectively.66 64 The report cites a survey conducted by the Social Weather Station in 2001, which finds that the richest 40% is found to predominately use private clinics or private hospitals. 65 ADB, JBIC and the World Bank (2005) 66 It is difficult to find consistent cross-country comparators of transport infrastructure given complicating factors such as geography, population density, and trade patterns. However, to maximize comparability, the road density measures used here have been scaled by the area of arable land, consistent with World Bank (2005). 66 Efficiency of Public Spending Table 15: Ranking of Transport Infrastructure 1/ Source: WEF (2010) Global Competitiveness Report and staff calculations. 1/ Data are based on surveys of business executives in the countries except for the availability of airline seats, which are base on scheduled available seat kilometers per week originating in country. 105. The Philippines exhibits both low levels and low efficiency of spending in the transport sector.67 In 2003, the peer group68 spent an average of 2.1 percent of GDP on transport infrastructure while the Philippines spent 0.5 of GDP less. At the same time, there is some evidence of comparatively low efficiency in public spending in transport. The Philippines underperforms other middle-income countries in the peer group as its efficiency score falls below the peer group averages. Our relative ranking across countries is also consistent with an institutional analysis of the public investment management in the Philippines in comparative perspective. For example, ADB, JBIC, and World Bank (2005) find that congressional allocations undermined the efficiency of infrastructure investment in the Philippines. For example, the long-term planning role of the National Economic and Development Authority (NEDA) tends to be undermined by multiple pressures to finance highly politicized infrastructure projects earmarked by congressional allocations. About a fourth of the Department of Public Works and Highways (DPWH) budget is earmarked for projects financed by congressional allocation (See Chapter 1). This results in significant fragmentation of scarce fiscal resources and undermines the strategic planning and budgeting process in infrastructure development. It also seems to contribute to low budget execution rates. 106. Low efficiency is likely due to fragmentation, politicization and poor governance and accountability. Transport spending is particularly fragmented in the Philippines. Despite the existence of a Medium-Term Philippine Development Plan and a National Investment Plan, actual decision-making on infrastructure expenditure is highly fragmented and implementation is poor. Only a third of all transport projects are selected 67 The Government is enacting a number of measures to improve public investment management. These include amendments to the Build-Operate-and-Transfer (BOT) Law and its Implementing Rules and Regulations (IRR) as well as the Joint Venture (JV) Guidelines to streamline the review and evaluation process of PPP infrastructure projects. Moreover, A National Transport Policy Framework (NTPF) and a draft National Transport Plan have been drafted with the goal of improving the institutional and policy framework for the transport sector. The NTPF focuses on the following priority areas for the transport sector: (a) Resource generation and allocation; (b) Criteria for the preparation of agency plans, programs, and projects; (c) Cost recovery and subsidies; (d) Regulation of passenger transport service; (e) Urban transport; (f) Transport logistics; and (g) Governance. 68 There is a lack of comparable cross-country data on transport indicators that are closely linked with national transport spending to support the DEA application. This section aims to keep a consistent set of regional peers as in previous analyses, but data availability prevents the inclusion of Rep. of Korea, Malaysia, or Thailand in this analysis. (See Table 17 for peer countries used). 67 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth through the NEDA Public Investment Management System (World Bank 2009a). Similarly, as much as a third of projects implemented by the DPWH are selected by Congressmen bypassing the national planning and priority- setting system. Moreover, since the advent of decentralization, a significant amount of local infrastructure such as local roads has become the responsibility of the LGUs, and their level of coordination with the national government has been poor. Moreover, due to their low capacity, transparency and accountability, there is hardly any data to assess LGU performance and evidence indicates that voters are unable to hold their local governments accountable. Finally, PPPs have been highly disappointing and infrastructure investments have been hampered by perceived corruption and poor functioning of the overall public infrastructure management system (ADB, 2009; ADB, JBIC and World Bank 2005; and World Bank 2009b). Table 16: Transport Outcome and Spending Indicators and Efficiency Estimates 1/ Source: World Bank, 2005, EAP Infrastructure at a Glance; and World Development Indicators Database. 1/ DEA efficiency score based on one input of national transport spending and two outputs of paved road and railway scaled by the area of arable land. 2/ Average of China, Indonesia, and Vietnam. 107. Addressing the country’s transport infrastructure shortage requires an overhaul of the public investment management system. The particularly low efficiency rates of public transport spending and the high levels of corruption in the sector justify addressing these shortcomings before significantly stepping up public expenditures on the sector. Our macro-economic framework consequently envisages the prioritization of an increase in education and health spending over 2011-2013 while delaying an increase in transport infrastructure until the public investment management system has been overhauled. Conclusions 108. The efficiency of education and health spending in the Philippines is on par with peer countries in the Asia Pacific region. Low levels of sectoral spending seem to be the key driving factor for below- average sector outcomes. Therefore, increasing the levels of spending is likely key to boosting the sectoral performance in education and health. 109. In transport, evidence suggests that low levels of expenditure are accompanied by high inefficiency. As discussed in Chapter 1 and in the Transport PER, such inefficiencies are attributable to a fragmented and politicized project selection and budgeting system. This in turn is due to a large share of Congressional 68 Efficiency of Public Spending allocations in infrastructure expenditures, weak transparency and accountability systems and poorly functioning PPPs (World Bank 2009a). Therefore, improvements in project planning, budgeting, monitoring and evaluation would likely yield considerable gains in sector outcomes. Recommendations 110. There is some scope for efficiency improvements by re-allocating expenditures across regions, particularly in health. In education and health, higher per capita spending in some regions has not reaped results commensurate with average efficiency. Allocating more resources from less efficient to more efficient regions could help achieve better sectoral outcomes within the same spending envelope. Using the group’s average as the benchmark, the potential for efficiency improvements is about 7 percent for education spending and about 11 percent for health spending. Pending a revision of the IRA, only national government expenditures are available for re-deployment. 111. Intermediate output analysis points to some potential areas for efficiency improvements. In education, physical capacity constraints, as reflected in a low classroom per student ratio in primary schools, appears to be a bottleneck. In health, increasing the share of front-line health workers and rationalizing the regional distribution of health care professionals, raising spending on public health, and addressing unequal access to physical facilities appear to be the bottlenecks. 112. The higher level of dispersion of inefficiency rates in the decentralized health sector points to the need for improved inter-governmental relations. (See Chapter 3 for specific recommendations). Potential efficiency gains can also be generated by strengthening the links between sector financing and improvements in sector performance through performance-based grants (See Health Sector Review, World Bank 2010b). 113. Finally, data production and reporting should be improved to facilitate the monitoring and evaluation of the efficiency of public spending. In particular, the following data could helpfully be published: a. Household survey data on private spending on basic education and health by province. b. Key health sector indicators (such as infant and maternal mortality) by province in the annual statistics. c. Some core infrastructure spending and outcome indicators, such as DPWH spending on investment and maintenance of national roads by province; kilometers of roads built and maintained by province; LGU spending on roads and roads built by province; kilometers of national, secondary and local roads by province. d. Strategic infrastructure development plans and PDAF and DPWH Congressional allocations funds, projects, releases and outputs, including maps congressional allocation details by province. References Afonso, A. and Aubyn M., (2004), “Non-parametric Approaches to Education and Health Expenditure Efficiency in OECD Countries.” Manuscript. Asian Development Bank (2009), “Poverty in the Philippines: Causes, Constraints and Opportunities.” Asian Development Bank, Japan Bank for International Cooperation, and The World Bank. (2005), “Connecting East Asia: A New Framework for Infrastructure.” 69 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Baldacci, E., Gupta. S. Clements, B., and Cui, Q. (2008), “Social Spending, Human Capital and Growth in Developing Countries,” World Development. Vol. 36(8), pp. 1317-1341. Banker, R.D., Chares, A. and Cooper, W. (1984), “Some Models for Estimating Technical and Scale Inefficiencies in Data Envelopment Analysis,” Management Science, 30, pp.1078-1092. Center for Legislative Development (CLD), the Philippines (2008), “Issues and Problems on Access to Health Services: A Policy Perspective,” http://www.cld.org/admin/images_products/9169study.pdf Coelli T. J. (1996), “A Guide to DEAP 2.1: A Data Envelopment Analysis Program,” Department of Econometrics Working Paper, University of New England, Armidales, Australia. Charnes, A. Cooper, W., and Rhodes, E. (1978), “Measuring the Efficiency of Decision Making Units,” European Journal of Operational Research, 2 (6), pp.429–444. Farrell, M. (1957), “The Measurement of Productive Efficiency,” Journal of the Royal Statistical Society, Series A, 120, Part 3, pp. 253-290. Gupta, S. and Verhoeven, M. (2001), “The Efficiency of Government Expenditure – Experiences from Africa,” Journal of Policy Modelling, 23, pp. 433-467. Herrera S. and Pang, G. (2005), “Efficiency of Public Spending in Developing Countries: An Efficiency Frontier Approach,” Policy Research Working Paper 3645, World Bank, Washington, D.C. Ilagan, K. (2009), “Big Infra Spending Fails to Life Plight of the Poorest,” Philippine Center for Investigative Journalism, May 7. Lachler, U. (2005), “The Efficiency of Public Spending on Education and Health in Guatemala,” Guatemala Public Expenditure Review. Manasan R., Cuenca J., and Villanueva, E. (2007), “Benefit Incidence of Public Spending on Education in the Philippines,” Philippines Institute for Development Studies Discussion Paper Series No. 2007-09, Makati City. Mandl, U., Dierx, A., and Ilzkovitz, F. (2008), “The Effectiveness and Efficiency of Public Spending,” Economic Papers of the Directorate-General for Economic and Financial Affairs, European Commission. National Statistics Office of the Philippines. Country in Figures, Various issues. Philippine Institute for Development Studies (2009), “Improving Local Service Delivery for the MDGs in Asia: the Case of the Philippines,” Discussion Paper Series No. 2009-34. Verhoeven, M., Gunnarsson, V., and Carcillo S. (2007), “Education and Health in G7 Countries: Achieving Better Outcome with Less Spending,” IMF Working Paper 07/263, International Monetary Fund, Washington D.C. World Economic Forum (2010), Global Competitiveness Report 2009-10. World Bank (2010a), “Philippines: Basic Education Public Expenditure Review,” forthcoming. World Bank (2010b), “Philippines: Health Sector Review.” World Bank (2005), EAP Infrastructure at a Glance. World Bank (2009a), “Reshaping Economic Geography,” World Development Report, Washington, DC: The World Bank. World Bank (2009b), “Transport for Growth: An Institutional Assessment of Transport Infrastructure.” World Health Organization (2009), Country Health Information Profiles-the Philippines. 70 Chapter 3. Decentralization and Regional Inequality Decentralization and Regional Inequality 3. Decentralization and Regional Inequality Introduction 114. With a Gini coefficient of 52.2, the Philippines exhibits the highest degree of income inequality in the East Asia region.69 This inequality across income groups is also reflected in high levels of inequality in incomes and social conditions across regions within the country.70 While some parts of the Philippines resemble those of mid-to-high income countries, others share more similarities with low-income countries. The quality of services in state universities and colleges and in private health facilities in Manila compares favorably with the most advanced facilities in developing countries. At the same time, education and health standards in some parts of the country are more reminiscent of those found in IDA countries. Regional GDP per capita differs widely. For instance, in the National Capital Region (NCR), where per capita GDP is over 11 times that of the Autonomous Region of Muslim Mindanao (ARMM), the poverty rate in 2006 was 10.4 percent whereas it was 62 percent in the ARMM (Table 17). Table 17: Philippines: Poverty by Administrative Region, 2000-2006 Source: World Bank estimates based on FIES data. 69 The Philippines is a diverse and dispersed archipelago. It is composed of 7,107 islands, only about a third of which are inhabited. Most of the islands are mountainous and covered in tropical rainforest, though the largest islands include large agricultural plains. With an estimated population of 92 million people, the country is the world’s 12th most populous country, home to a large number of ethnicities and cultures and 171 spoken languages. The Philippines is divided into three island groups: Luzon, the Visayas, and Mindanao. Administratively, the country is divided into 17 regions, 80 provinces, 120 cities, 1,511 municipalities, and 42,008 barangays or district/village units for a total of 43,719 LGUs. LGUs consist of provinces, cities and municipalities. 70 Differences across provinces are even higher, as regions mask inequality across provinces. However, since no data is available at the provincial level, this chapter uses the region as the main level of analysis. 73 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 115. Inequality is of particular concern because overall growth rates have been low. In a country like China, with its high growth rates, the fact that some regions are growing faster than others is of less concern. For instance, over 2001-2008, the slowest-growing region in China grew almost twice as rapidly as the fastest growing region in the Philippines. In Vietnam, over 1991-2000, the growth rate of the median region was twice as high as its equivalent in the Philippines (Figure 14). Therefore, the analysis of convergence and divergence that this chapter presents needs to be placed in the context of sub-par growth compared to other countries in the region. Figure 14. Average Growth Rate by Region in Selected East Asian Countries (1991-2000) and (2001-2008) Source: Philippines Statistical Yearbook, NSCB, Indonesia-Central Bureau of Statistics 116. Starting in 1991, the Philippines undertook an extensive decentralization of fiscal responsibilities. LGUs are now responsible for providing basic public services and are expected to fund them with transfers from the central government as well as by raising their own sources. The shift in responsibilities or authority between the central government and LGUs was established in the 1991 Local Government Code (Republic Act 7160). This code gives LGUs autonomy to decide the composition of their expenditures and revenues consistent with their local development objectives. To date, LGUs are largely responsible for most basic public services—primary health care, social welfare services and hospital care. Education, on the other hand, remains mostly centralized, although LGUs account for roughly half of capital expenditures in basic education through the SEF financed through real estate taxes.71 Contrary to other decentralized countries, there are neither sector expenditure norms nor minimum service standards with which LGUs are expected and supported to comply, and there are no centrally mandated data they are expected to produce. Capital spending on education (e.g., school buildings) is a shared responsibility between the central and local governments. 71 74 Decentralization and Regional Inequality 117. The 1991 Local Government Code increased fiscal transfers to LGUs, in support of their new responsibilities. Transfers from the National Government (NG) budget to LGUs are formula-based, and are a function of tax revenue collected by the NG during the three years prior to the allocation year. Over time, this formula has resulted in about 20 percent of central government (contemporaneous) tax revenues being allocated to LGUs through the IRA scheme. Transfers rose sharply following the passage of the Local Government Code and have been averaging 3.0 percent of GDP from 1994 to 2009 (Figures 15 and 16). Regulations require that 20 percent of the general IRA transfer be used for development expenditures. Figure 15. Transfers to LGUs rose markedly after Figure 16. IRA transfers account for a growing the 1991 Decentralization Law share of NG tax effort Source: Department of Finance (DOF) Source: Department of Finance (DOF) 118. This chapter assesses regional convergence/divergence of living standards and the extent to decentralization may have been a contributing factor. As the 2009 World Development Report (WDR 2009) on Reshaping Economic Geography points out, government is responsible for providing at least spatially-blind institutions and public services across its territory. When, on the contrary, public expenditures favor the better-off regions, government is contributing to enlarging the natural gap between better- endowed and less well-endowed regions. By doing so, it not only increases inequality, but also dampens the growth prospects of the country as a whole. This chapter documents the evolution of living standards indicators across regions in the Philippines and tracks the geographic distribution of LGU resources and some related key outputs and intermediate indicators. In part due to the absence of comprehensive data on the regional distribution of national government expenditures, the chapter focuses on the role of LGU expenditures. Overall, therefore, the chapter examines whether fiscal decentralization has contributed to reducing interregional equity or whether it may have exacerbated it. It also assesses what current levels of distribution of human capital across regions tell us about the likelihood of convergence in the future. 119. The convergence of regional incomes that was underway prior to decentralization, stopped once it started.72 Although there was some convergence73 in regional incomes over the entire 1975 to 2000 period, the relationship is weak, and there is no measurable statistical relationship in the latter half of this See technical annex (Annex 2) for references, definition of convergence and methodology. 72 σ-convergence refers to the trend in convergence whereas β –convergence measures the speed of convergence (See Annex 2). 73 75 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth period (Manasan and Chaterjee, 2003). Between 1975 and 1986 (before decentralization), the speed of convergence or β-coefficient is estimated at 0.0137 (significant at the 5 percent level). In the 1987 and 2000 (decentralization) period, on the other hand, the coefficient stands at a statistically insignificant 0.0067.74 120. Since 2000, per capita GDP has grown only in the National Capital Region (NCR) and there has hardly been any convergence of regional income (Figure 19).75 The results of our analysis of Gross Regional Domestic Product (GRDP) per capita through 2008 suggest a speed of convergence consistent with the trends in previous estimates, i.e., very slow (β = 0.0029 and statistically insignificant). A simple plot of GRDP per capita over time—whether in logs or in levels—suggests that: 1) regional incomes have been generally flat over the last few years, except generally in the NCR; and 2) there is little discernible movement toward a narrowing of disparities in regional incomes. 121. Like regional GDP, other income indicators exhibit hardly any convergence. Data on GRDP per capita may overstate the extent of disparities in regional living standards. Therefore, to present a more comprehensive picture of well-being across regions, we tested convergence in other indicators such as wages, poverty levels, expenditures and social indicators. These also show little evidence of convergence, as shown below. 122. There is little evidence of convergence in average non-agricultural wage rates across regions since 2000 and some evidence of divergence since 2006 (Figure 17). The σ-convergence result shows some convergence from 2001 to 2005, but a significant increase in regional disparity in 2006. Since then, a very slow pace of convergence resumed. The β -convergence analysis shows that, over 2001-2008, there was little or no evidence of convergence of regional average daily basic pay. The regression result shows a statistically insignificant β-coefficient = 0.00535. 76 This trend is consistent with the absence of convergence in regional incomes. 74 The rate of convergence between 1975 and 1986, while statistically significant, is relatively slow at about half to one third of the rate of convergence observed within the U.S. states since 1880, the prefectures of Japan since 1930, and the regions of eight European countries since 1950 (Barro and Sala-i-Martin, 1995). With β = 0.0137, then regions are converging (closing the gap) at a rate of 1.37 percent per year. This yields a half-life of convergence (i.e., the years necessary to close half of the initial gap) of 51 years (against 23 to 35 among US, EU, and Japanese states and regions). 75 The analysis is hampered by the break in the regional indicators series, when Region IV (Southern Tagalog) information is disaggregated into the “CALZABARZON” (IV-A) and “MIMAROPA” (IV-B) sub-regional information beginning around 2003. In this section and in subsequent ones, we conduct two sets of analyses. The first set includes either data on Region IV or data on Region IV-A and IV-B, depending on which series covers the longest period possible. The second set excludes data on Region IV. The results presented below are invariant to the exclusion/inclusion of Region IV data. 76 The analysis follows the classic, cross-sectional approach to regional convergence analysis. Some caution is thus warranted in the interpretation of results because they are necessarily based on small samples. 76 Decentralization and Regional Inequality Figure 17. Non-agricultural wages have tended to Figure 18. Poverty incidence has tended to diverge diverge since 2006 Source: Bureau of Labor and Employment Services Source: Official Poverty Statistics of the Philippines (1997, 2000, 2006) 123. There is evidence of divergence in poverty rates.77 (Figure 18). The β-convergence analysis of the average growth of poor incidence of the population from 1988 to 1997 shows a positive β-coefficient (i.e. 0.371) strongly significant at 1 percent level, indicating strong evidence of regional divergence within the period. The β-convergence analysis of non-poor incidence of the population for the period after 1997 shows no evidence of regional convergence.78 Further, the β-coefficient is positive and statistically insignificant, indicating probable regional divergence. The σ-convergence results also show strong regional divergence between 1988 and 1997 while the β-convergence analysis for this period shows no evidence of convergence. Since 2000, the trend is less clear, with sub-periods of convergence and divergence. 124. There has been some modest narrowing in household expenditures across regions in recent years.79 (Figure 20).80 Household expenditures suggest some tendency toward equalization in the period after 2000. Regional disparity unequivocally increased from 1988 to 2000, driven in large part by the growth of family expenditures and family income in the NCR outstripping those of other regions. Since 2000, however, the household expenditures of most regions exhibit a modest rate of convergence towards the NCR level, as the NCR stabilized (in logs). Over this period, the lowest quintile’s share of total income has been more or less unchanged.81 77 It is important to note that the β-convergence analysis transforms the relevant social indicators accordingly into their positive form. In this case, we used the non-poor proportion of the population. 78 As previously stated, this is a convergence analysis of the non-poor proportion of the population but Figure 18 shows the poverty incidence (rather than the population share of non-poor) for ease of interpretation. 79 Summary FIES data for 2009 will not be made publicly available until late 2010 or early 2011. Strictly speaking, we should be comparing family expenditures per person and GRDP per person. However, summary data on the FIES over the last twenty years from the National Statistical Coordination Board (NSCB) are readily available only in terms of average families, not on a per-person basis. Nonetheless, for 1997 and 2000 where both summary figures on family expenditures and family expenditure per person are publicly available, the data suggest that that they are very closely correlated (correlation coefficient=0.99). Not surprisingly, the average household size is about the same across regions. 80 See for example World Bank (2004). This, however, may also reflect the same type of discrepancy seen between measures of private consumption drawn from National Accounts data and measures of consumption drawn from surveys. See Ravallion (2001) for a discussion of key issues. 81 This expenditure convergence may be due to the trends seen in recent household surveys of rising household expenditures on education and health as government expenditures declined, higher remittances to poorer regions as emigration increases just as poverty continues to stagnate, and stepped up borrowing rates of cash-strapped households. 77 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Figure 19. Regional incomes have diverged since Figure 20. But there has been some modest 2001 convergence in regional household spending Source: NSCB data and World Bank staff calculations Source: NSCB data and World Bank staff calculations 125. Can some aspects of well-being converge across geographic units, despite diverging income levels? The WDR 2009 concludes that unbalanced economic growth can coexist with the convergence of other indicators of well-being. Namely, if governments provide spatially-blind institutions with equal access to social services across the country, they can promote the convergence of living standards while laying the groundwork for the eventual convergence of growth. The next three sections examine whether in the Philippines there has been convergence in basic indicators of well-being such as education, health, and access to basic infrastructure. 126. Education indicators, which are critical to future income trends, have diverged. Until 2003, literacy rates across regions had converged, with especially rapid convergence in the 1990s (Figure 21). On the other hand, net primary enrollment rates converged through the 1990s, but diverged thereafter (Figure 22). Consistent with this result, primary completion rates exhibit little evidence of convergence up to 2003 and have been diverging since then (Figure 23). Even when excluding the ARMM region, there is little or no convergence. Figure 21. Strong β- and σ- convergence across Figure 22. but recent divergence in net primary regions literacy rates... enrollment rates Source: NSCB data and World Bank staff calculations Note: Excludes ARMM and Caraga. For these regions, data are available only for the most recent years. Source: NSCB data 78 Decentralization and Regional Inequality Figure 23. Primary completion rates have diverged Figure 24. Immunization rates also show some divergence Source: NSCB data Note: Excludes ARMM and Caraga. For these regions, data available only Source: NSCB data for the most recent years. 127. Health indicators that can be directly attributed to the role of the health sector, such as immunization rates, have not been converging (Figure 24). For instance, measles immunizations rates have tended to diverge since 1997, though there have been some sub-periods of convergence over the past ten years. The poor performance of this indicator may be due to poor coordination between national and local government as the responsibility for immunizations is shared, with NG being in charge of procuring vaccines and LG in charge of distributing them and administering them. 128. The most impressive case of convergence can be seen in infant and child mortality rates (i.e., evidence of β- and σ-convergence). (Figures 25 and 26). Between 1990 and 2006, infant mortality rates for the Philippines as a whole fell from about 57 (per 1,000 live births) to 23. These improvements were widespread, with some regions experiencing very steep declines from about 76 in 1990 down to 31 in 2006. Similarly, overall infant mortality rates fell from about 80 in 1990 to 31 in 2006, with Eastern Visayas and the Bicol Region experiencing some of the most rapid decreases. More importantly, the fall has been more pronounced for regions with low initial conditions (β-convergence in infant mortality) and the dispersion in infant mortality rates between regions has narrowed over time (σ-convergence in infant mortality, see Figure 25). β- and σ-convergence between regions is also noticeable for child mortality rates (Figure 26), the prevalence of underweight children (not shown) and maternal mortality rates (not shown), though the analysis of the latter is hampered by the lack of data for more recent years. Figure 25 . Strong β- and σ- convergence across Figure 26 . as well as in child mortality rates regions in infant mortality rates Source: NSCB data Source: NSCB data 79 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Figure 27. International Comparison: Infant Mortality Rate Philippines (2002-2006) and Indonesia (2000-2008) Source: Philippines Statistical Yearbook, NSCB, Indonesia-Central Bureau of Statistics 129. Convergence in infant and child mortality rates needs to be interpreted within a context of an overall slow pace of decline, higher overall mortality levels and greater regional dispersion compared to other East Asian countries. The relatively high level of convergence in infant and child mortality indicates some catch-up effect of the regions with the highest mortality levels. However, under-5 mortality in the Philippines is still 26 per 1000 compared to an average of 18 per 1000 in East Asian comparator countries. The speed of decline in overall infant and child mortality rates in the Philippines has also been much slower than in other countries in the region. Moreover, regional performance is much more unequal in the Philippines than in comparable countries in the region (See Figure 27 for a comparison with Indonesia). 130. The degree of convergence witnessed in infant and child mortality rates is not paralleled in any other indicator of well-being. As described above, all these improvements have taken place without 1) commensurate increases in GRDP per person, or 2) β- or σ-convergence in GRDP per person over this same period. This is a trend that could be seen across the world prior to the advent of AIDS. This seems to be part of a global trend. Over the last several decades, there appears to have been a gradual convergence in some indicators of well-being across countries despite lack of evidence on income convergence. The results of recent cross-country studies by Sab and Smith (2002), Neumayer (2003), and Kenny (2005) show that, despite lack of convergence in incomes per capita, there has been convergence in some education and health outcomes, including life expectancy and literacy.82 This recent literature runs counter to the findings of some earlier studies that suggested divergence, rather than convergence, in social indicators.83 82 The analysis conducted by Sab and Smith (2002), using three-stage least squares, explicitly recognizes the interaction between social indicators, e.g., education and health indicators, in contrast to Neumayer (2003) and Kenny (2005) who use variants of the more typical single-variable σ- or β-convergence. 83 See, for example, Hobijn, and Franses (2001). The sources of discrepancy between the earlier literature and the more recent literature—including measurement errors--are reviewed by Neumayer (2003). 80 Decentralization and Regional Inequality 131. The combination of convergence in some indicators of well-being and non-convergence in income is not fully understood in the literature. For Sab and Smith (2002), the occurrence of human capital convergence in the absence of income convergence suggests the existence of substantial lags between human capital accumulation and the period of measurable increases in productivity. In the health sector, the convergence could be due in large part to the spillover benefits of technological and medical advancements rather than to the impact of government policies. It could also be due, in part, to the role of social services whether provided by government or non-governmental organizations, to the extent that they compensate for income gaps. In addition, transfers may play an important role in equalizing health and education outcomes across regions for reasons similar to those underpinning the less unequal distribution of regional household expenditures (compared to regional incomes). Finally and perhaps most critically, Neumayer (2003) argues that many of the social indicators we are concerned with —such as literacy rates, net enrollment rates, and infant survival rates-- have natural upper bounds, which means that convergence is the natural path of development. 132. Strong caution is warranted in the analysis of non-income dimensions of well-being. As explained by Kenny (2005), there are some measurement issues specific to social indicators, especially when analyzing them using the existing framework for assessing convergence. Because some social indicators have a complement (for example, infant mortality as opposed to infant survival), they may indicate convergence, when measured one way, and divergence, when measured through its complement.84 He also argues that measuring convergence toward zero (as in the case of infant mortality rates) tends to “favor” or place disproportionately more weight on small absolute changes close to zero over much larger absolute changes further away from zero. For example, an infant mortality rate falling from two to one (per 1,000 live births) weighs more (in percentage terms) than an infant mortality rate falling from 100 to 60. 133. The convergence in infant mortality rates is robust to a test of absolute convergence. Taking into account the measurement issues noted in the preceding paragraph, our β-convergence analysis transforms the relevant health indicators accordingly into their positive form (e.g., infant survival), for closer comparability with the previous β-convergence analysis of income. In the case of infant mortality rates (or infant survival), the β-coefficient is 0.035 and strongly significant (at the 1 percent level of significance). This indicates a rapid speed of convergence (half-life of convergence of 20 years), especially when compared with regional incomes (half-life of convergence of 51 years). The results of the analysis of σ-convergence also suggest that the dispersion of infant and child mortality rates across regions has fallen rapidly over time. 134. Access to a safe water supply and sanitary toilets across regions seemed to converge until 2003- 2004 and has diverged since then (Figures 28 and 29). Another indicator of well-being is the population’s access to basic infrastructure services, such as safe water and sanitary toilets. It is also a good indicator of the ability of government to provide basic infrastructure across the country. Our data indicate that, as was the case in most social indicators, convergence took place until 2000, but has been in reversal since then. See also Neumayer (2003). 84 81 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Figure 28 . Access to a safe water supply Figure 29. The same seems to be the case for access to sanitary toilets Source: Philippine Countryside in Figures, NSCB Source: Philippine Countryside in Figures, NSCB A. Contribution of Fiscal Decentralization to Regional Divergence 135. Fiscal resources can play a critical role in curbing geographic disparities in living standards and well-being. Decentralized governments can allocate fiscal revenues so as to alleviate geographic imbalances in resources and ensure equal access to basic services across regions (See Box 1). They can also finance infrastructure investments in lagging regions or, at least, connect these regions and their populations to other regions with better economic prospects and employment opportunities (World Bank, 2009). Moreover, in the right circumstances, fiscal decentralization can enhance government efficiency, encourage innovation and accountability in the provision of services, and foster closer links between revenue and expenditure decisions. On the other hand, in countries with poor capacity and accountability at the local level combined with uneven resource availability across regions, decentralization can widen income and development gaps and undermine transparency and accountability. 136. Decentralization in the Philippines has not played an important role in reducing geographic disparities and instead may be exacerbating them. Devarajan and Hossain, (1998) found that the overall regional allocation of public expenditure on health, education, and infrastructure was progressive and that the findings were invariant to sensitivity tests. Manasan and Chaterjee (2003), using data on LGU expenditures for the 1990s, found that fiscal decentralization has not helped promote interregional equity.85 Analyzing LGU budgets from 1985 through 2000, they found that, in large part, this has been due to the highly divergent levels of fiscal capacity across regions and the overall level of resources in poor regions being incommensurate with newly-devolved responsibilities (see also Bird and Rodriguez, 1999). Among other factors, such mismatches have in turn been due to vertical and horizontal imbalances (namely those across regions as well as the share of national taxes being inadequate to cover the cost of devolved functions). There are substantial discrepancies across regions in the ability to raise revenues. Manasan and Chaterjee reported that, as of 2000, over two-thirds of all local taxes were collected by only four regions. They also noted rising “unfunded mandates,” such as increases in IRA that are not proportional to salary adjustments mandated by the Salary Standardization Law. 85 Part of the discrepancy in some of the findings, in addition to differences in the time-period examined, could be differences in data and methodology, including the scope of the analysis. For examples, Devarajan and Hossain (1998) look at health, education, and infrastructure expenditures, while Manasan and Chaterjee (2003) and Manasan (2007) examine aggregate LGU budgets. By their own admission, Devarajan and Hossain (1998) use strong assumptions to arrive at an integrated estimate of fiscal benefits (accounting for the combined incidence of taxes and spending), while the other studies simply report separate summary statistics for revenues and expenditures. 82 Decentralization and Regional Inequality BoxBox 1. Vietnam’s 1. Vietnam’s Fiscal Fiscal System—Supporting Effective DecentralizationSystem—Supporting Decentralization Effective Service and Infrastructure Delivery Nationwide Service and Infrastructure Delivery Nationwide Vietnam’s fiscal decentralization is a key part of the Government’s economic and political reform program. It began in 1996 and it continues to this day. The focus of Vietnam’s decentralization program has been to ensure that, through the effective delivery of public services and infrastructure nationwide, progress on the achievement of critical development outcomes takes place in a balanced manner throughout the country. To achieve this balanced development, the decentralization program has three central platforms: 1. Delivering appropriate resources to all local governments --especially the poorer and more disadvantaged ones; 2. Ensuring resource allocation is supportive of the achievement of critical development outcomes; and 3. Experimentation with measures of downward and upward accountability of local governments. Delivering appropriate resources to all local governments through a solidly pro-poor intergovernmental fiscal transfer system. • The 2002 State Budget Law required the development of a strongly pro-poor system of equalization transfers from the central to the provincial governments based on criteria including population, poverty rate and whether the province is in a disadvantaged location; • Since 2006, capital expenditure transfers are allocated based in part on the above and other pro-poor criteria, such as the presence of ethnic minorities, the levels of local revenue and the number of districts determined to be in disadvantaged locations; • National targeted programs provide exemptions from education fees to poor households, distribute health insurance cards to disadvantaged individuals, and support communes facing extreme difficulties • Provinces with revenues below minimum expenditure needs are exempted from sharing with central Government the proceeds from shareable taxes; • Program 135 entitles poor and mountainous communes to receive central level funds for infrastructure projects. • Ensuring local government resource allocation is supportive of the achievement of critical development outcomes through expenditure norms. Provincial governments are subject to expenditure norms on their vbudget decisions. In particular, there are certain quotas and standards determined by the central level that must be applied by provincial governments in their expenditure allocations to ensure that certain development outcomes (e.g. in education and health) are met. • New measures of upward and downward accountability of local governments, including through data production and publication. As part of the decentralization process and in order to ensure accountability for the use of public funds, some important forms of downward accountability have been introduced. Since 2004, regulations on public finance transparency require Governments to publicize the estimates and uses of annual state budgets, inter-governmental transfers, and the collection and use of “people’s contributions.” While much of the power is still at the province level, experimentation with downward accountability has been strong at the commune level: Governments are required to involve their citizens in the development of local budgets and people’s contributions, socio-economic development plans, public investment projects and plans, land use management, legal issues and administrative procedures among others. In practice, as is typically the case, progress has been uneven and much progress yet remains to be made in ensure effective access to information and ability to participate in decision making by all. “Local Government” in Modern Institutions. Vietnam Development Report 2010. 83 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 137. A more recent paper using data through 2003 finds that the LGU share of general government expenditures has risen more rapidly than its share of general government revenues (Manasan 2007). Further imbalances are evident when the data are disaggregated into various LGU levels (provinces, municipalities, and cities). The current IRA distribution formula, however, does not allow for equalizing transfers to reduce some of these gaps in resources and responsibilities. This distribution formula proceeds in two stages. At the first stage, IRA resources are allocated by level of government and are distributed as follows: 23 percent shared by all provinces, 23 percent shared by cities, 34 percent shared by municipalities, and 20 percent by barangays. At the second stage, the LGU share depends on the following formula: population (50 percent), land area (25 percent), and equal sharing (25 percent). As such, the formula does not offset disparities and revenue-raising capacities across LGUs; in fact, IRA resources tend to flow toward advanced regions. Not surprisingly, summary household-level data from the Family Income and Expenditure Survey (FIES) through 2000 also suggest that per capita transfers and per capita household incomes were positive correlated (thus suggesting a regressive allocation of resources). 138. Our analysis reveals that regional inequality in total LGU revenue has increased. The gap has grown, in part, as NG transfers have become less progressive. We undertake an analysis of the various sources of LGU income at the regional level in 2001 and in 2008 and draw concentration curves (Figures 30 and 31). These are akin to Lorenz curves and reveal the extent of inequality among units (regions in this case) along several dimensions of interest (total LGUs revenues and its breakdown in this case). The horizontal axis ranks the 17 regions of the Philippines from the poorest86 (zero) to the richest (one). From 2001 to 2008, total transfers from the NG to LGUs (ntotshare?) went from being slightly progressive Figure 30 shows that the IRA was broadly neutral in 2001.87 139. LGU own-revenue sources are starkly inequitable and have become more so, with LGUSs in the poorest regions being able to collect far less revenue than richer ones. In fact, LGUs in NCR alone collected about half of total LGU tax revenues across the country’s 17 regions; i.e., the LGUs of the remaining 16 regions collected about the same amount as NCR. Inequality in both own-revenue mobilization and NG transfers has increased noticeably between 2001 and 2008, as can be seen by comparing Figures 30 and 31. 86 Measured by the real GRDP per capita. 87 An analysis of the extent of progressivity of the IRA is not possible for 2008 as data are missing for a large number of LGUs (only total NG transfers received are recorded, not the breakdown in the sources). 84 Decentralization and Regional Inequality Figure 30. Concentration curves for LGU revenue, Figure 31. Concentration curves for LGU revenue, 2001 2008 Source: World Bank staff calculations. Source: World Bank staff calculations. 140. LGU spending is strongly regressive and has become more so, especially in education. Two regions— NCR and CALABARZON—account for over 60 percent of total LGU spending from 2001 to 2008 even though they only account for 44 percent of GDP and 26 percent of the total population (Figure 34). Of great concern is the fact that the most regressive LGU expenditure category is education. In 2001, LGU spending on education was highly regressive while total LGU spending by region was slightly unequal. Other components of total spending exhibited broadly the same degree of inequality. Expenditure on social security, social services and welfare spending (e_soc) and labor and employment (e_labor), however, showed some slight degree of progressivity (Figure 32). By 2008, total spending had become slightly more unequal, but there were significant changes in sectoral allocations. For example, spending on labor and employment had some degree of progressivity while spending on education had become even more unequal and spending on housing and community development (e_house) had become as unequally distributed across regions as LGU education spending (tote_edu) (Figure 33). Figure 32. Concentration curves for LGU sectoral, Figure 33. Concentration curves for LGU sectoral, spending, 2001 spending, 2008 Source: World Bank staff calculations. Source: World Bank staff calculations. 85 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Figure 34. Education and Housing Spending Shares by Regions, 2001-2008 Source: Statement of Income and Expenditures of BLGF 141. The NCR alone accounts for over half of total LGU spending on both education and housing (Figure 34). The stark inequality in education expenditures across regions is largely due to the (highly regressive) real property tax funding of the SEF and its ring-fencing for education.88 The inequality in the housing sector, on the other hand, is likely due to the large and lumpy outlays necessary in this area.89 Only relatively well-off LGUs have the own-tax and non-tax resources to undertake such large projects, either through their accumulated savings or by borrowing (mostly through LandBank and the Development Bank of the Philippines, both government finance institutions). Poorer LGUs have poor credit-worthiness and therefore limited borrowing capacity. This is of concern, in particular since our chapter on the efficiency of public expenditures finds capital outlays as a key bottleneck in achieving sector outcomes. 142. LGU expenditures account for almost half of total general government non-wage expenditures on basic education. In the aggregate, LGUs contribute less than 8 percent of total general government expenditure on basic education. However, LGU expenditures account for a large share of government non-wage expenditures on basic education. Over 2000-2007, total LGU expenditures on basic education were equal to 73 percent of the Department of Education’s non-personnel services budget and 116 percent of its allocation for current non-wage expenditures. This means that access to learning materials, for instance, is highly unequal across regions. The Department of Education estimates that expenditures on school materials in the Manila district of Makati are equal to equivalent expenditures in the rest of the country. These inequities are likely to translate into large gaps in education quality across regions. 88 Having the additional real property tax levy ring-fenced for education through the Special Education Fund forces LGUs with larger own real property tax bases to spend on education, reducing their choices and increasing inequality in education expenditures across regions 89 Recall that LGUs in education are mostly responsible for capital spending, such as for building schools 86 Decentralization and Regional Inequality 143. NG spending per pupil on basic education across regions is less unequal, but does not systematically or fully compensate for differences across LGUs (Table 18). The variation in LGU per pupil spending appears to be co-variant with average regional household income. Furthermore, the positive relationship between LGU per student SEF spending and per capita household income across regions is found to be statistically significant. In other words, the pattern of LGU spending appears to favor the better- off regions. NG per pupil spending across regions varies less widely, but does not systematically or sufficiently compensate for differences across LGU spending. Table 18: Government Spending on Elementary and Secondary Education in 2007 Source: Family Income and Expenditure Survey (FIES) and the SAAOBs of Department of Education, 2006. 144. Richer LGUs have higher levels of both LG and NG health expenditures. There is a positive and statistically significant relationship between per capita household income and per capita LGU spending on both public health and hospital services (Table 19). This is due to the fact that per capita LGU spending on health is largely driven by the resource envelope of each LGU (Manasan 2008). Per capita DOH spending on hospital services is also found to be positively correlated with per capita household income. This is surprising in light of the implementation of the income retention policy for DOH-retained hospitals starting in 2003.90 90 In the analysis above, the observations for NCR are treated as outliers and excluded from the analysis. If observations for NCR are included, the relationship between per capita household income and per capita LGU spending on hospital care is negative, but not significant. However, a positive and statistically significant relationship still holds for per capita household income and per capita LGU spending on public health and per capita DOH spending on hospital services. 87 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Table 19: Government Spending on Health in 2007 Source: SAAOBs of Department of Health. B. Two decades into decentralization, regional inequality remains stark in incomes, living standards, and access to public services. 145. Two decades into decentralization, regional inequality remains stark in the Philippines. Rapid β and σ convergence in infant and child mortality among regions mask significant remaining differences among regions. Infant mortality remains much higher among poorer regions. This difference is likely due, at least in part, to wide disparities in immunization coverage and access to preventive health care. For example, the level of immunization coverage for DPT3 in NCR is more than twice that of ARMM (89 percent versus 41 percent, respectively in 2003). Similarly, access to basic public health services, such as facility-based deliveries, still differs widely among regions. In NCR 68 percent of deliveries are facility deliveries and 87 percent were attended by a health professional, compared to 15 percent and 19 percent respectively in the ARMM (Annex 2). Moreover, the effects of inadequate access to water and sanitation services and the impacts of air and water pollution vary widely among regions and LGUs and affect predominantly the poor—see World Bank (2010) for more details. 146. Unequal access to public health facilities at the regional level has not been compensated by private sector supply. Evidence in this regard includes the facts that: 1) out-of-pocket spending per household is strongly regressive at the regional level (Annex 2); 2) the share of households that make exclusive use of private health facilities is higher in richer regions (Figure 35); and 3) the share of households that did not use any health facility when sick (either public or private) is sharply higher in poorer regions (Figure 36).91 Regional differences in poverty rates remain wide (Annex 2). 91 Not surprisingly, a recent study finds that household saving has been falling between 1994 and 2006 though it argues that the change in saving behavior is in part explained by the growth of the social security system over this period (Terada-Hagiwara 2009). Another recent study speculates that the rising demand for health services—brought on by the increasing share of young dependents in the total population—is leading to lower saving (Mapa and Bersales 2008). 88 Decentralization and Regional Inequality Figure 35. The richer the region, the higher the Figure 36. The poorer the region, the more likely it is exclusive use of private health facilities that households do not visit any health facility when sick Source: World Bank staff calculations using APIS 2005 data for health Source: World Bank staff calculations using APIS 2005 data for health and NSCB for real GRDP data. and NSCB for real GRDP data. 147. Education deficits reflect geographic (island) differences (World Bank, 2010). For example, age- specific education deficits are smaller in regions of the island of Luzon and larger in regions of the island of Mindanao. Similarly, the distribution of regional education deficits is smaller in Luzon than it is in Mindanao (Annex 2). 148. Local roads, an LGU responsibility, are also highly unequally distributed across regions and inversely correlated with the poverty rate. (Annex 2) This inequality reflects highly uneven levels of own resources and access to finance across regions, especially for lumpy investments. It also reflects the fact that congressional allocations (PDAF and DPWH-CA) tend to (slightly) favor regions with higher per capita incomes. 149. Devolution has not improved service delivery.92 The growing gap in performance across regions is of particular concern because overall service delivery standards seem to have deteriorated since decentralization started. While positive examples of good local governance can be found, they appear to be exceptions rather than the norm. A World Bank discussion note on decentralization presents a simple Human Development Index (HDI) transition matrix which indicates whether provinces, grouped by their HDI scores in 1990, progressed, regressed, or stagnated over the following 10 years. Of the 74 provinces analyzed, only eight managed to improve their scores enough to move up to the next higher cohort while three provinces actually registered lower HDI scores. Moreover, a recent study found that growth in provincial income after decentralization has had a weak impact on poverty reduction, suggesting that LGUs have proven unable to translate growth at the provincial level into poverty reduction through more effective service delivery (Balisacan, 2007). See Discussion note on Decentralization. The World Bank, 2009. 92 89 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 150. Mechanisms for holding local chief executives accountable for performance are weak and frequently ineffective. Aside from unclear assignment of responsibilities across levels of government, there is weak social and electoral accountability. The Local Government Code introduced mechanisms to institutionalize citizen participation in the LGU planning and budgeting processes, primarily through special bodies such as the local development council and the local health and school boards. However, the functionality of these special bodies varies significantly across LGUs and the voice of non-government stakeholders in local governance is generally very weak. 151. No systematic information on LGU performance is available to allow voters to assess their local governments’ performance. Therefore, citizens do not have the information to hold their governments accountable through elections to reward good performers and punish laggards. Not surprisingly, recent research suggests that local elections have been ineffective instruments for holding local chief executives accountable for their performance in service delivery. In fact, the re-election rates of the governors who presided over significant increases in HDI scores (i.e., over 10 percent) were not higher than those of governors who presided over negligible or decreasing changes in HDI scores. (See Discussion Note on Decentralization; World Bank, 2009) Conclusions 152. Convergence in income and living standards among regions in the Philippines provides a mixed picture. Evidence of convergence was found in a few areas, but important areas are converging at a very slow pace, not converging, or even diverging. Rapid convergence over a sustained period of time appears to be a challenge as regional convergence in some areas did take place over a limited time period, but was not sustained. To summarize, the following key findings emerge from the chapter: a) Regional income convergence has been very slow (along both the β and σ dimensions) and wages and poverty rates seem to be diverging since 2000; b) There is some recent evidence of modest convergence in household expenditures; c) Some key health outcomes such as infant mortality rates are converging much faster; d) Education outcomes converged until 2000 and have diverged since then; e) The evidence on the convergence of other social indicators is mixed, but there seems to be modest divergence in access to water and sanitary toilets; f) LGU revenues and expenditures are very unequal across regions and are becoming more so, particularly for education and housing expenditures; g) NG expenditures on basic education and health are not bridging the gap, and neither is private provision (at least in health); h) Inequality in revenues and expenditures translates into unequal access to public services and infrastructure across regions; i) Two decades into decentralization, regional inequality in income, living standards, and access to public services remains stark and in critical aspects, is growing. 153. The non-equalizing character of inter-governmental transfers and high levels of inequality in own resources across regions contribute to regional divergence. While revenue transfers to LGUs are 90 Decentralization and Regional Inequality significant at around 40 percent of overall government resources, their distribution across LGUs has been broadly neutral; that is, the richer LGUs receive a per capita share of NG revenue roughly equal to their poorer counterparts despite their higher capacity to raise own revenue. This fundamental horizontal inequity stems from the fact that the IRA, by far the largest source of transfer from the central government to LGUs, does not include a specific equalization objective. This inequity on the revenue side explains the large inequality in expenditures across regions. Because this inequality affects key areas such as education and local roads, it lays the groundwork for continued divergence in the future. Recommendations 154. The following policy recommendations would address the key shortcomings highlighted in this chapter: a) Including a strong revenue equalization component in the IRA formula would help address growing inequality. This would ensure that NG transfers to LGUs are allocated in a strongly progressive manner. There are two key options for such an IRA reform. The first option is to include a measure that captures differences in revenue-generating capacity in the allocation formula, such as poverty incidence. The second option utilizes a set of minimum expenditure levels that ensure the provision of a basic level of services to the population. A balancing or equalizing transfer would be given to those local governments that are not able to meet that minimum expenditure level. Experiences in other countries may provide a good benchmark for revising the IRA formula (see Annex 2 on Indonesia). In elaborating this reform, considerations other than equity should also be taken into account. Such a reform should be a priority in the next Medium-Term Philippines Development Plan (MTPDP) to be developed by the incoming administration. (Top 10 Recommendations Number 5). b) Strengthen the LGU incentive system, transparency and accountability. Consider establishing a transparent and publicly accessible system for systematically monitoring and evaluating LGU performance, which may be used as the basis for performance incentive systems. (Top 10 Recommendations Number 6).93 c) The NG should at least partially compensate for the disparities in the tax base of LGUs. In the education sector, it should compensate for unequal distribution of funds across schools, as proposed by Atkins and Manasan (2001) and, more recently, by Boncodin et al. (2008). Similarly, the equity- based school maintenance and other operating expenses (MOOE) formula developed by Boncodin et al. (2009) should be implemented so that schools have sufficient funds to implement their school improvement plans. A similar solution to compensate for differences in health spending across LGUs should be devised for the health sector. d) Significant improvement in public finance and other data collected and disclosed is urgent. Enhanced data production and dissemination is critical to improving LGU accountability as well as to providing the information basis to assess government effectiveness. We recommend that the following data be produced on a regular basis: • Macro-economic indicators disaggregated to the provincial level (e.g., provincial domestic product, income, and savings levels). • Poverty and household (FIES) information disaggregated to the city/municipality level. 93 In this regard, it could be useful to use as guidance the system of report cards for LGUS, donors, and Councils for Health Development that the Department of Health is developing for monitoring and evaluating the equity and effectiveness of health services. 91 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth • Annual report on a set of key local government outputs and outcomes in core areas of LGU responsibility, such as health, local roads, waste management and pollution control, agriculture and fisheries development, and access to basic utilities disaggregated to the regional, provincial, city, and municipal level. • Fiscal data disaggregated by region and province, sector and expense class (economic classification): ◊ NG expenditures should be disaggregated by region (and if possible, by province). ◊ NG and LGU budgeted and executed expenditures should be disaggregated by GFS-compliant sectoral and economic classifications. ◊ The inconsistency between audit (COA) and budget (BLGF) data should be bridged. Currently, the two sets of reports (COA and BLGF) are inconsistent with each other.94 Policy analysis requires having consistent data between the two disaggregated sources. The team understands that both COA and BLGF are working to such an outcome. ◊ The BLGF should also revise its expenditure accounts in order to reduce the size of the ‘other purposes’ category, which takes up too large a proportion of LGU expenditures and compromises the quality of any analysis of expenditures. Reforms should also ensure that BLGF’s reporting of expenditures by function is consistent with the IMF-GFS (1986)/UN’s Classification of Functions of Government (COFOG) categories. References Acemoglu, D. (2009), Introduction to Modern Economic Growth, Princeton University Press, Princeton. Barro, R. and Sala-i-Martin, X. (1995), Economic Growth, McGraw-Hill, New York. Balisacan, A., Hill, H., and Piza, S. F. (2008), “Regional Development Dynamics and Decentralization in the Philippines,” ASEAN Economic Bulletin, December, Vol. 25 Issue 3, pp. 293-315. Bird, R. and Rodriguez, E. (1999), “Decentralization and Poverty Alleviation: International Experience and the Case of the Philippines,” Public Administration and Development, Vol. 19, Issue 3, pp. 299-319. Devarajan, S. and Hossain, S. (1998), “The Combined Incidence of Taxes and Public Expenditures in the Philippines,” World Development, Vol. 26 (6), pp. 963-977. Hobijn, B. and Franses, P. H. (2001), “Are living standards converging?” Structural Change and Economic Dynamics Vol. 12, pp. 171-200. Kenny, C. (2005), “Why are we worried about income? Nearly everything that matters is converging,” World Development, Vol. 33(1), pp. 1-19. Llanto, G. (2009), “Fiscal Decentralization and Local Finance Reforms in the Philippines,” Philippines Institute for Development Studies Discussion Paper Series No. 2009-10, Makati City. Manasan, R. (2007), “Decentralization and the Financing of Regional Development,” in Balisacan, A. and Hill, H., (eds.), The Dynamics of Regional Development: The Philippines in East Asia, Ateneo de Manila University Press, Quezon City. 94 COA reports expenditures (only) by economic class (wage bill, capital spending, etc.) while BLGF reports expenditures (only) by functional categories (general public services, health, education, etc.). 92 Decentralization and Regional Inequality Manasan, R. and Chaterjee, S. (2003), “Regional Development” in Balisacan, A. and Hill, H. (eds.), The Philippine Economy: Development, Policies, and Challenges, Ateneo de Manila University Press, Quezon City. Mapa, D. and Bersales, L.G. (2008), “The Link between Population Dynamics and Household Saving: An Econometric Approach,” Working Paper 2008-01, University of the Philippines School of Statistics, Quezon City. Neumayer, E. (2003), “Beyond Income: Convergence in Living Standards, Big Time,” Structural Change and Economic Dynamics, 14, pp. 275-296. Ravallion, M. (2001), “Measuring Aggregate Welfare in Developing Countries: How Well Do National Accounts and Surveys Agree?” Policy Research Working Paper No. 2665, World Bank, Washington, DC. Sab, R. and Smith, S. (2002), “Human Capital Convergence: A Joint Estimation Approach,” IMF Staff Papers, Vol. 49 (2), pp. 200-211. Sala-i-Martin, X. (1996), “The Classical Approach to Convergence Analysis,” Economic Journal, Vol. 106, pp. 1019- 26. Silva, J. (2005), “Devolution and Regional Disparities in the Philippines: Is here a connection?” Environment and Planning C: Government and Policy 23(3), pp. 399–417. Terada-Hagiwara, A. (2009), “Explaining Filipino Households’ Declining Saving Rate,” Asian Development Bank Economics Working Paper No. No. 178, Asian Development Bank, Manila. World Bank (2010), “Philippines: Fostering More Inclusive Growth,” Washington, D.C.: The World Bank. World Bank (2009), “Reshaping Economic Geography,” 2009 World Development Report, Washington, D.C.: The World Bank. World Bank (2004), World Bank EU-8 Quarterly Economic Report, July, Washington, D.C.: The World Bank. World Bank (2001), “Philippines Poverty Assessment,” Report No. 20498, Washington, D.C.: The World Bank. 93 Chapter 4. Equity of Government Spending on Education and Health Equity of Government Spending on Education and Health 4. Equity of Government Spending on Education and Health Introduction 155. Education and health are major determinants of an individual’s human capital, earnings stream, and overall well-being. As such, they are crucial contributors to an individual’s ability to break the inter- generational cycle of poverty. 156. At the national level, the distribution of human capital is one of the most important determinants of the distribution of income as well as a key indicator of government equity and responsiveness to those who most need it. The distribution of benefits from social expenditures is particularly important in a country like the Philippines, in which the overall level of such expenditures is very low. Ultimately, the access of the poor to quality public services will determine the human capital of the “bottom half” of the population, a critical ingredient to the economic success of any country as well as to its social and political stability. Finally, inequality in income and human capital is a root cause of poor governance across the world. 157. In this light, assessing to which extent government expenditures on education and health are benefiting the lower income groups is important for equity reasons, and for evaluating the overall efficiency and effectiveness of public expenditures. 158. This chapter assesses the extent to which various income groups in the Philippines have benefited from publicly provided education and health services. It does so through the use of benefit incidence analysis (BIA), a tool used to assess how government spending or tax policy affects the distribution of welfare in the population by evaluating the distribution of government subsidies among different income groups. (For further methodological information, see Annex 3.) A. Equity in Access to Education Participation Rates 159. High participation rates tend to mask unequal access to education across gender and income groups. The Annual Poverty Indicators Surveys (APISs) show that net participation rates vary widely with per capita household income and gender (Table 20). In particular, children from poorer households are significantly less likely to attend school than children from higher income households, and the school attendance rate of males is lower than that of their female counterparts for all age groups. The poverty gap is found to be wider than the gender gap. 97 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Table 20: Net Participation Rates, by level, by Gender and Poverty Status, 1999, 2004 & 2007 (in Percent) Source: 1999, 2004 and 2007 APIS. 160. Participation gaps between poor and non-poor children rise with age and level of education and are economically determined. Overall, participation rates in the Philippines deteriorated between 1999 and 2004, with the greatest decline in the tertiary level, where the net participation rate went down from 24.3 in 1999 to 20.5 in 2004. Although the participation rate at all levels improved in 2004-2007, the gains were not enough to bring them back to their 1999 levels. The gap between the participation rates of non-poor and poor children rises with age and the attendant opportunity cost of staying in school. In 2007, the gap was 6 percentage points at the elementary level and 25 percentage points at the secondary level. The profile of school leavers indicates that the primary reason for the gap is economic, including the direct costs of education as well as the need to seek employment. 161. Females exhibit higher participation rates than males and the gap rises with age/ level of education. The gender gap in participation rates rises with age. In the elementary and secondary level, the gap widened between 1999 and 2004, but recovered somewhat in 2004-2007. In 2007, it stood at 2 percentage points at the elementary level and at 11 percentage points at the secondary level. The difference is more marked in poorer households, where 58.3 percent of girls aged 12-15 attend secondary school compared with only 44.7 percent of the boys in the same age bracket (Table 20). 162. There are economic reasons for lower male participation. The profile of school leavers suggests that lack of personal interest is the primary reason why boys aged 12-15 years leave schools at this age; however, the profile also shows that “high cost of education” and “employment” rank second and third among their reasons cited, indicating that another important reason for leaving school is economic. 98 Equity of Government Spending on Education and Health Completion Rates 163. The proportion of Filipinos aged 17-24 who completed grade school between 1999 and 2007 remained stable at around 90 percent. On the other hand, the proportion of the same cohort who completed high school increased from 62 percent in 1999 to 66 percent in 2007. 164. Poverty and gender gaps in educational attainment have persisted. Children from poor households are less likely to remain in school than children from non-poor households, while women tend to receive more education than men, regardless of poverty status. Moreover, the poverty gap in the educational attainment of Filipinos aged 17-24 is larger than the gender gap. No significant reduction in the size of either the poverty or gender gap in the educational attainment of Filipinos in the same age cohort is observable between 1999 and 2007. 165. The poverty and gender gaps in secondary school completion are large and have persisted between 1999 and 2007. In 2007, close to 77 percent of Filipinos from non-poor households aged 17-24 graduated from high school while less than 43 percent of counterparts from poor households were able to do so (Figure 37).95 Similarly, about 74 percent of females aged 17-24 years in 2007 completed high school while less than 59 percent of their male counterparts were able to do so. The poverty gap in the proportion of children who complete high school is more than twice that of the gender gap and both gaps have remained largely unchanged between 1999 and 2007 despite overall improvement in completion rates. Figure 37. Educational Attainment of Population Aged 17-24 Source: 1999, 2004 and 2007 APIS 166. Figure 38 shows that the poverty gap in the educational attainment of males is slightly larger than that of females. In 2007, 83 percent of females from non-poor households completed their secondary education compared with 51 percent of females from poor households in 2007. In contrast, 70 percent of males from non-poor households completed high school while only 36 percent of males from poor The educational attainment profile shows in the vertical axis the percentage of the population in the given age cohort who have 95 completed any given grade or higher (as shown in the horizontal axis). The gap between 100 percent and y-coordinate shows the proportion that did not complete grade 1 while the slope indicates the percentage of children who drop out across the years (Filmer and Pritchett 1998). 99 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth households did so (Figure 38). The most commonly cited reason for not being in school by boys aged 12-15 years was “lack of personal interest.” In the case of girls, “lack of personal interest” and “high cost of education” tied as the most commonly cited reason for not being in school. Figure 38. Educational Attainment of Population Aged 17-24 Source: 1999, 2004 and 2007 APIS School Leavers 167. The proportion of children who are not in school increases with the age bracket. The percentage of children who were not in school in 2007 was 5.7 percent for those aged 6-11, 11.5 percent for those aged 12-15 years and 66 percent for those aged 16-24 years (Table 21). Between 1999 and 2007, the proportion of school leavers declined for all age groups. For instance, the proportion of school leavers among children aged 12-15 decreased from 17.5 percent in 1999 to 11.5 percent in 2007. However, the contraction in the proportion of school leavers is hardly significant for the 6-11 years age bracket (from 5.7 percent in 1999 to 5.6 percent in 2007) and the 17-24 years age bracket (from 66.4 percent in 1999 to 65.8 percent in 2007). 168. The proportion of school leavers is higher among males than among females for all age groups (Table 21). Consequently, most school leavers at all levels are males and, on average, males account for 54 percent of all school leavers. The proportion of male school leavers is 63 percent in the 12-15 age group, higher than in the 57 percent in the 6-11 age group. 169. About three-quarters of school leavers in the 6-11 and in the 12-15 age groups are poor (Table 21). In 2007, the incidence of school leavers from poor households was 9 percent in the 6-11 age group and 18 percent in the 12-15 age group, in both cases around three times the incidence level in non-poor households. 100 Equity of Government Spending on Education and Health Table 21: Distribution of School Leavers by Sex and Poverty Status, 1999, 2004 and 2007 (in Percent) Source: 1999, 2004 and 2007 APIS. 170. High dropout rates among children aged 6-11 point to the need to strengthen early childhood education. In 2007, 34 percent of school leavers in the 6-11 years age group said they are not in school because “they are too young to go to school.” On the other hand, 24 percent attributed their situation to “lack of personal interest,” 13 percent to “high cost of education,” 8 percent to “illness/ disability” and 6 percent to the fact that “schools are very far.” The large percentage of school leavers among children aged 6-11 who appear to lack the readiness for school (e.g., those who report they are “too young to go to school” and “lack of personal interest”) indicates the need to strengthen early childhood education.96 “Lack of personal interest” is cited as the reason for leaving school almost twice as often by boys than by girls across all age groups. 171. A high percentage of children aged 6-11, especially those from poor households, do not attend school due to financial constraints. A high percentage of children cite the “high cost of education” and “distance from school” as reasons for not attending school. The higher percentage of children from poor households citing that reason for non-attendance reflects the poverty gap. This finding underscores the importance of conditional cash transfers in reducing the poverty gap. 96 In school year 2008-2009, only 50 percent of the children of the prescribed age (6 years) entered grade 1. Data for earlier years suggest that the remaining half of these children enter grade 1 at age 7 or even later. What we propose is strengthening early childhood education (targeting children aged 3-5 years) so that by the time they reach age 6 they are better prepared to enter grade 1 and have a better chance of staying in school. The important effects of early childhood education are well-established in the literature (See Discussion Note on Early Childhood Education). 101 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 172. Among children aged 12-15, forty-five percent of school leavers cite “lack of personal interest” as the reason for not being in school. Moreover, a significantly larger proportion of male school leavers (53 percent) than female school leavers (31 percent) in this age group cite this reason. Many education experts suggest that “lack of personal interest” in school among older children tends to be associated with poor quality of education. The gender gap can also reflect a better fit of education methods with the learning styles of female students than male students. 173. The “high cost of education” ranks second among the reasons cited by school leavers aged 12-15. Moreover, this reason is more common among female school leavers (31 percent) than male school leavers (21 percent) and among school leavers from poor households (27 percent) compared with those from non-poor households (20 percent). Also, a higher percentage of school leavers in the 12-15 years age group (25 percent) cited this reason for not being in school relative to those in the 6-11 years age bracket (13 percent). This profile of school leavers suggests the need for larger educational cash grants for poor children in the 12-15 years age bracket. 174. “Employment” is the third most cited reason given for leaving school by children aged 12-15 (8 percent). Surprisingly, this reason is slightly more common among children from non-poor households (8 percent) than among those from poor households (7 percent). Also, the proportion of boys in this age group who report that they left school to join the labor market (7 percent) is not much different from the proportion of girls (8 percent). Service Utilization, Government Spending and Benefit Incidence in the Education Sector 175. Public education utilization by the poor declines with age and becomes highly regressive for tertiary education. The distribution of public school enrolment becomes increasingly skewed in favor of richer households as the level of education rises. Thus, while 49 percent of public elementary school pupils come from the three poorest deciles, the comparative figure in public secondary schools is 35 percent and only 17 percent in state universities and colleges (SUCs) (Table 22). Overall, while the distribution of public school enrollment is progressive at the elementary level, it is less so at the secondary level, and in SUCs it is regressive. 176. Benefit incidence of public education varies by region. The distribution of enrollment in public elementary schools is progressive in all regions, while the distribution of enrollment in public secondary schools is progressive in all regions except NCR. Furthermore, while the distribution of enrollment in public secondary schools in Central Luzon and Southern Luzon is marginally progressive, it is almost neutral. On the other hand, while the distribution of enrollment in SUCs is regressive on the whole, it is mildly progressive in the CAR, Bicol, Western Visayas, Davao, and Socskargen and moderately progressive in Caraga. 102 Equity of Government Spending on Education and Health Table 22: Distribution of Public School Enrollment Source: 1999 APIS. 177. Overall, general government expenditure on education in the Philippines was moderately progressive in 2007 (Figure 39). Therefore, poorer households receive a higher fraction of benefits relative to wealthier households. However, the overall average hides important differences in the incidence of public education spending according to the level of education as well as the level of government. In 2007, general government spending on both elementary and secondary education was progressive, although spending on secondary education was less progressive. In contrast, general government spending on SUCs was regressive. Figure 39. Incidence of General Government Spending on Education 178. General government spending on education was marginally more progressive in 2007 than in 1999. This is largely due to the less regressive distribution of government spending on SUCs in 2007 relative to 1999. In turn, this development may indicate that some of the budget reforms in higher education (normative financing and an increasing emphasis on self-financing) may have made government spending in the sector more pro-poor. Since the NG accounts for 92 percent of total general government spending on basic education, the distribution of general government spending on basic education across income deciles closely mirrors that of the NG (Figure 40). On the other hand, while LGU spending on 103 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth elementary education is marginally progressive, that on secondary education is mildly regressive. This may be explained by the fact that the distribution of enrollment in public secondary schools is regressive in NCR and almost neutral in Central Luzon and Southern Luzon, the regions with the highest SEF income nationwide. On average, SEF expenditures accounted for 84 percent of total LGU education spending in 2001-2008. 179. LGU spending on basic education is becoming more regressive while national government spending has become more progressive. A comparison of the estimates of the Suits indices for national and local government spending on basic education in 2007 with those for 1999 (Manasan and Cuenca 2007) indicates that LGU spending at elementary level has become less progressive and LGU spending at the secondary level has become more regressive. This may be attributed to the widening disparity in the distribution of SEF income per student across regions and points to the urgent need for NG spending on basic education to be allocated across regions in such a way as to compensate for differences in SEF spending.97 On the other hand, national government spending on basic education in 2007 was more progressive at both the elementary and secondary levels than in 1999 though the movement is more pronounced at the secondary level. Overall, these changes roughly cancel each other out and have resulted in almost imperceptible changes in the progressivity of general government spending on elementary and secondary education between 1999 and 2007. 180. The distribution of NG spending on both elementary and secondary education does not appear to be biased in favor of either males or females. The same is true of LGU spending on elementary education. However, while LGU secondary education spending for males is regressive, spending for females is marginally progressive.98 B. Equity in Access to Health Services 181. Overall, health outcomes for the poor are worse than those for their better-off counterparts and the gap is widening. For instance, children from households in the lower wealth quintiles have higher rates of infant and under-five mortality than children from higher wealth quintiles (Table 23). In addition, the poverty gap in the infant and under-five mortality rates appears to have widened between 2003 and 2008. In particular, the ratio of the infant mortality rate of the poorest quintile to that of the richest quintile rose from 2.2 in 2003 to 2.7 in 2008. In like manner, the ratio of the under-five mortality rate of the poorest quintile to that of the richest quintile increased from 3.1 in 2003 to 3.5 in 2008. 97 To wit, LGU spending at the elementary level is clearly less progressive in 2007 (-0.0233) than in 1999 (-0.0653). On the other hand, LGU spending at the secondary level is more regressive in 2007 (0.1115) than in 1999 (0.0840). 98 Although the enrollment rate of females is higher than that of males, the number of boys aged 6-11 years who are actually in school is just about the same as that of girls due to the fact that the sex ratio at this age group still favor boys (102 boys for every 100 females). 104 Equity of Government Spending on Education and Health Figure 40. Incidence of Government Spending in Primary Education ... and Secondary Education Table 23: Early Childhood Mortality Rates by Wealth Quintiles, 2003-2008 Source: 2003 and 2008 NDHS. 105 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth  Households’ Access to Health Services 182. Between 2003 and 2008, households’ access to basic health services improved for all income groups although the access gap across income groups has widened. Lower income households continue to have less access to basic health services than higher income households. The poverty gap in households’ access to some basic services such as antenatal care and vaccinations has widened over time. There are a number of reasons for this, including physical inaccessibility of facilities, financial constraints, and weak demand for health care. 183. The proportion of women who did not receive prenatal care has declined, but the poverty gap in access to this service has widened. The share of women who had a live birth five years preceding the National Demographic and Health Survey (NDHS) and did not get prenatal care from medical professionals declined from 7 percent in 2003 to 5 percent in 2008. At the same time, the proportion of pregnant women who did not receive any antenatal care also went down from 6 percent in 2003 to 4 percent in 2008 (Annex 11). However, the proportion of women who did not get antenatal care from medical professionals plus the proportion of those who did not get any antenatal care at all is higher for women from less wealthy households than for women from the higher wealth quintiles. Moreover, the poverty gap (as measured by the ratio of the proportion of women from the poorest quintile who did not receive antenatal care from a medical professional to a corresponding proportion for the wealthiest quintile) widened between 2003 (8.8) and 2008 (14.3). 184. The proportion of live births that took place in a health facility increased and the poverty gap declined, but the overall ratio remains low (Table 24). In addition, the proportion of live births delivered in a health facility is lower for women from poorer households than for those from better-off quintiles. In like manner, the proportion of births which were assisted by skilled health providers went up slightly, and it is higher for mothers from higher income groups. On a positive note, the poverty gap in delivery in a health facility narrowed between 2003 and 2008 while the poverty gap in access to a medical professional during delivery remained unchanged. Table 24: Place of Delivery, 2003-2008 Source: 2003 and 2004 NDHS 106 Equity of Government Spending on Education and Health 185. The percentage of fully immunized children increased from 70 percent in 2003 to 80 percent in 2008, but the poverty gap widened (Table 25). Like access to other basic health services, immunization coverage increases with the wealth status of households. Conversely, the proportion of children in each wealth quintile who were not vaccinated even once declines with the wealth status of households. Moreover, the poverty gap in immunization coverage deteriorated between 2003 and 2008. Table 25: Percentage of children vaccinated, 2003-2008 Source: 2003 and 2008 NDHS. 1/ BCG, measles, and three doses of DPT and polio vaccine 186. Seventy-five percent of mothers with children under five years of age reported having problems in accessing health care in 2008 and the main problem is financial. This is a slightly lower percentage than the corresponding proportion in 2003. Lack of financial protection from the costs associated with illness figured prominently in 2003 as well as 2008. For example, “getting money for treatment” is the most oft-cited problem accessing health care in 2003 (67 percent) and 2008 (55 percent). In addition, 47 percent of mothers interviewed in 2008 cited “concern that no drugs are available” as a problem in accessing health care. As might be expected, these concerns appear to be more important for mothers from poorer households than those from better-off households. As such, reforms are recommended to: expand the social health insurance program coverage, particularly the Sponsored Program for Indigents; improve targeting of premium subsidy beneficiaries under the Sponsored Program; increase support value; and eliminate balance billing and introduce other improvements in the payment mechanisms. 187. Geographic access to health facilities remained problematic in 2003 and 2008 (Table 26). Again, poorer households appear to have less access to health facilities than better-off households. This indicates that improving financial protection without improving geographic access may not be enough to enhance overall equity in access to health care. This is consistent with the finding of Chapter 2 that efficiency 107 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth estimates of public health spending are strongly correlated with the number of barangay health stations per person.99 Table 26: Problems in accessing health care, 2003-2008 Source: 2003 and 2008 NDHS. Service Utilization, Government Spending and Benefit Incidence in the Health Sector 188. Households in the lower wealth deciles tend to use public health facilities more heavily than their wealthier counterparts (Table 27).100 The utilization of LGU-operated provincial/district hospitals is also progressive although less than that of Barangay Health Services (BHSs) and Rural Health Units (RHUs). In contrast, the utilization of DOH-retained hospitals is mildly regressive. This may be explained by the fact that when individuals from poorer households get sick they tend to go to LGU-operated provincial/ district hospitals rather than DOH-retained hospitals because they are geographically closer.101 189. Some regional variation is evident in the utilization of both provincial/district hospitals and DOH- retained hospitals. For instance, while the utilization of provincial/district hospitals in the aggregate 99 The DOH program on Geographically-Isolated and Disadvantaged Areas (GIDA) is being consolidated in the implementation of Health Facilities Enhancement Program so that upgraded health facilities can be provided in GIDA. 100 BHSs and RHUs. 101 DOH-retained hospitals are located in regional centers. 108 Equity of Government Spending on Education and Health is mildly progressive, the utilization of these types of facilities in the Ilocos Region, Central Luzon, Southern Luzon and NCR is regressive. In like manner, while the utilization of DOH retained hospitals is regressive on the whole, household utilization of these health facilities is progressive in the Bicol Region, all the regions in the Visayas and all the regions in Mindanao with the exception of Caraga. Table 27. Distribution of Households Utilizing Various Health Facilities Source: 2003 National Demographic and Health Survey. 190. In 2007 NG spending on public health was progressive while expenditure on hospital services was regressive.102 In the aggregate, NG health spending is found to be regressive since hospital services account for over three-quarters of total NG health expenditures (Figure 41). 191. Between 2003 and 2007, total national government spending on the health sector has become less regressive in the aggregate.103 This occurred partly because the share of public health in the DOH budget expanded from 15 percent to 24 percent. At the same time, the DOH budget reallocation for public health across regions in 2007 (relative to 2003) tended to favor regions where public health service utilization is more progressive, just as the DOH budget allocation for hospital services across regions in the same year shifted towards regions where utilization of hospital services is less regressive. As a result, DOH spending on public health became more progressive in 2007 (with a Suits index of -0.1491) relative to 2003 (with a Suits index of -0.1307). Meanwhile, DOH spending on hospital services became less regressive in 2007 (with a Suits index of 0.0908) relative to 2003 (with a Suits index of 0.1007). The Suits Index for NG spending on public health and on hospital services was (-0.1491) and (.0908) respectively. 102 Comparing estimates of the Suits index for national government health expenditures in 2003 (Manasan and Cuenca 2009) with 103 the estimates for 2007. 109 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Figure 41. Incidence of Health Expenditure by Type of Service by the National Government ... and Local Government Units 192. LGU spending on public health is much less progressive and hospital care is less regressive than the equivalent NG spending. Although, like in the NG, LGU spending on hospital services is regressive, it is only marginally so. In the aggregate, the overall distribution of benefits from LGU health spending is mildly progressive. 193. In the aggregate, LGU health spending has become less progressive between 2003 and 2007. While the change in the distribution of benefits from LGU spending on public health is almost imperceptible (Suits index of -0.0862 in 2003 vs. a Suits index of -0.0833 in 2007), the distribution of benefits on hospital care changed from marginally progressive in 2003 (with a Suits index of -0.0045) to marginally regressive in 2007 (with a Suits index of 0.0168). This change is indicative of a widening disparity in LGU spending on hospital services in favor of LGUs in regions with a relatively more regressive distribution of benefits from hospital services (i.e, the better-off regions like NCR, Central Luzon and Southern Luzon). This development may be linked to the inequality in resource bases across LGUs, including the lack of an equalization factor in the IRA distribution formula. 110 Equity of Government Spending on Education and Health 194. In the aggregate, general government spending on health is marginally regressive. (Figure 42). On the whole, general government spending on health exhibits an almost identical pattern as that of the national government, even if the share of the national government in general government spending on health is less than that of LGUs. The progressivity in general government spending on public health is swamped by the regressivity of general government spending on hospital services given the larger share of hospital services in total health expenditures of the general government. Figure 42. Incidence of General Government Spending on Health Conclusions 195. Public education utilization by the poor declines with age. Expenditure on elementary education is clearly progressive, on secondary education it is less so, and on tertiary education utilization it is highly regressive. Only 17 percent of tertiary education students come from the three poorest deciles. There are no large differences in the distribution of general government spending between males and females. 196. National Government per pupil spending on basic education is fairly uniform across regions while LGU spending varies widely. The variation in LGU spending is correlated with the level of average household income in the region and the level of real-property tax funded SEF expenditures. This is important because LGU expenditures account for half of general government non-wage expenditures. As Chapter 3 points out, infrastructure spending seems to be a bottleneck in improving the efficiency of government expenditures. Therefore, LGU spending-driven inequalities in non-wage expenditures aggravate this inefficiency. 197. Between 1999 and 2007, NG spending on basic education has become more progressive and LGU more regressive. NG spending became more progressive over that time period, especially for secondary education. The increase in the regressivity of LGU spending may be attributed to the widening disparity in the distribution of SEF spending. These changes have roughly cancelled each other out. 198. Poorer households tend to use public health facilities more heavily than their wealthier counterparts. The utilization of LGU-operated provincial/district hospitals is also progressive 111 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth although less so than that of barangay/rural health units. The utilization of DOH-retained hospitals, on the other hand, is mildly regressive. 199. NG spending on public health is progressive while expenditure on hospital services is regressive. In the aggregate, NG spending on health is regressive since hospital services account for over three-quarters of total NG expenditures. 200. Between 2003 and 2007, NG spending on health has become less regressive and LGU spending less progressive. The increased progressivity of NG spending was due to a higher share of public health spending in the budget. Overall LGU spending on health is mildly progressive, but became less so over this time period because of the distribution of benefits from hospital care going from marginally progressive to marginally regressive. 201. In the aggregate, general government spending on health is marginally regressive. This is due to the fact that the progressivity of public health spending is swamped by the regressivity of spending on hospital services. Recommendations 202. Public education expenditure should be reallocated toward basic education. Greater priority should be given to basic education (relative to the tertiary level) in the allocation of total government spending not only because basic education spending is progressive and tertiary regressive, but more importantly because basic literacy affords society at large important additional benefits by facilitating social cohesion and nation-building. Moreover, a mother’s educational attainment is positively linked with child health and nutrition and with declines in fertility. 203. Higher subsidies to State Universities and Colleges are difficult to justify. This is so not only because enrollment in SUCs is regressive, but also because returns to higher education are internalized by students themselves. SUCs also have greater scope for cost recovery by charging tuition at appropriate levels. Such a policy change would also impose some fiscal discipline on the part of SUCs, as paying clients tend to demand better services. At the same time, there is a need to improve scholarship schemes for needy students as well as introduce student loan schemes in order to ensure equal opportunity access to the services of SUCs. 204. Public health should receive greater priority in health sector allocation of public spending. In order for this to be feasible, government hospitals need to operate on a sustainable basis with greater use of cost-recovery through user fees. In this regard, the introduction of income retention policy for DOH hospitals is a step in right direction. A similar reform for LGU hospitals is needed. 205. The social health insurance system needs to be reformed. The reform should seek to: (i) expand coverage, especially of the Sponsored Program for Indigents; (ii) improve the targeting of the beneficiaries of premium subsidies under the Sponsored Program, (iii) increase support value, and (iv) eliminate balance billing and introduce other improvements in the payment mechanisms (Manasan 2009). There is 112 Equity of Government Spending on Education and Health also a need to upgrade government hospitals to enable them to meet PhilHealth accreditation requirements so as to enhance their sustainability. 206. The conditional cash transfers program (4Ps) needs to be expanded to enhance the access of the poor to basic education and health services. In particular, the provision of a higher level of grants for attendance in secondary schools would be an important contribution to preventing the high dropout rates for financial reasons at that level of schooling. 207. Early childhood education should be strengthened and consideration should be given to introduction of pre-school as part of the formal education cycle. Stepped-up early childhood education could help reach children who start elementary school late and/or unprepared. Early childhood education reaching the poorest could be an important contributor to enhancing their chances of success and attainment in basic education and beyond. It would also help improve the preparedness and educational attainment of children in all income brackets. 208. Improvements in the quality of basic education are much needed. Improved quality would enhance the ability of schools to retain their students and provide them with a better education. This would in turn help close the attainment gap between the Philippines and other countries in the region, building the basis for a stronger economy, polity and society. It would also enhance the Philippines’ comparative advantage in its well-educated English-speaking labor force. References Davoodi, H., Tiongson, E., and Asawanuchit, E. (2003), “How Useful are Benefit Incidence Analysis of Public Education and Health Spending,” IMF Working Paper 03/227, International Monetary Fund, Washington, D.C. Demery, L. (2000), “Benefit Incidence: A Practitioner’s Guide,” World Bank, Washington, D.C. Manasan, R., Cuenca, J., and Villanueva-Ruiz, E. (2008), “Benefit Incidence of Public Spending on Education in the Philippines.” Philippines Institute for Development Studies Discussion Paper No. 2008-08, Makati City. Manasan, R., Cuenca, J. (2009), “Benefit Incidence of Public Spending on Health in the Philippines,” Unpublished. van de Walle, D. (1996), “Assessing the Welfare Impacts of Public Spending,” Policy Research Working Paper No. 1670, World Bank, Washington, D.C. van de Walle, D. (1995), “Public Spending and the Poor: What We Know, What We Need to Know,” Policy Research Working Paper No. 1476, June, World Bank. Washington, D.C. 113 Chapter 5. Sufficiency, Efficiency and Equity of Public Revenues Sufficiency, Efficiency and Equity of Public Revenues 5. Sufficiency, Efficiency and Equity of Public Revenues Introduction 209. This chapter will assess the sufficiency, collection efficiency, and equity of tax collections in the Philippines. As substantiated in the previous chapters, the sub-par socio-economic performance of the Philippines is caused in large part by low levels of public expenditure, due to a low and declining tax effort. This chapter will trace the evolution of tax collections over the past 12 years, identify the main sources for their decline and assess the level, sufficiency, and efficiency of revenue collection. Collection efficiency is not only a key determinant of collection rates, but also a good indicator of the state of governance in a country. As such, it affects the investment and governance climate. Finally, the chapter studies the degree of horizontal and vertical equity of the tax system. This analysis can support empirically grounded recommendations on how to enhance tax collections while preserving equity and reducing income inequality. 210. The chapter ends with a series of policy recommendations for the incoming government. At the core of these recommendations is the elaboration of a “fiscal pact” between the Executive and Congress to be enacted during the first few months of the new administration. This pact would include a commitment to raise tax revenues and increase expenditures in priority areas while keeping to a specified deficit path. It would also include a moratorium on any revenue-eroding measures and lay out the basis for an in-depth reform of tax administration. (Top 10 Recommendations Number 10). A. Sufficiency 211. Between 1997 and 2009, tax revenues declined by over 4 percentage points of GDP and are insufficient to cover the country’s expenditure needs.104 The levels of efficiency of the tax collection system are below regional averages and show a compliance gap of around 4 percentage points of GDP. Moreover, the quality of tax administration has tended to decline, contributing to the decline of tax collections. Horizontal equity is particularly poor due to the large number of tax exemptions and high evasion levels among the self-employed and professionals. This leads to large distortions in the neutrality of the tax system, discrimination among taxpayers and windfall revenues to exempted sectors. The tax system is only mildly progressive. Vertical equity can be significantly enhanced by improving the tax compliance of high income groups. Horizontal equity can be enhanced by improving tax compliance of self-employed and professionals. 212. Low revenues have severely constrained expenditures in health, social services, and infrastructure. Between 1997 and 2008, NG expenditure as a percentage of GDP declined by 50 percent in the health sector, 36 percent in education, and 34 percent in infrastructure and other capital outlays. Moreover, in all three sectors, total public expenditure levels as a percentage of GDP are significantly below those of other East Asian economies (see Chapter 1). 213. Low revenues also leave public finances vulnerable to destabilizing shocks. In the past, episodes of particularly sharp declines in revenue levels have resulted in unsustainable deficit levels that the All fiscal data are presented using the government’s definition which is different from the IMF’s GFS definition. For example, the 104 Government of the Philippines includes privatization receipts as revenue rather than as a financing item. 117 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth government eventually addressed by compressing expenditures, increasing public debt and leading to macroeconomic instability (e.g. 2002-2005). The resulting increase in fiscal risk has raised public sector borrowing costs and rendered the country less attractive for private investors. Investors cite an unsustainable macroeconomic stance as one of the main obstacles to a healthy investment climate.105 Trends in Tax Collections 214. Revenue collections have been on the decline since 1997, with a temporary recovery between 2005 and 2006 (Figure 43). BIR tax effort declined from its peak of 13 percent of GDP in 1997 to 9.7 percent in 2004 and, after a temporary recovery in 2005-06, the decline resumed in 2007. The main contributors to this trend are policy changes to excise taxes and tax incentives and weakening tax administration. The BOC tax effort exhibits a similar trend (Figure 43 and 49). In 1993, collections peaked at 5.6 percent of GDP before falling to 3.9 percent of GDP in 1997 and 2.5 percent of GDP in 2002. Between 1993 and 1997, trade liberalization was primarily responsible for the decline in collections, as weighted average tariff rates fell from 30 percent to 20 percent. However, after the East Asian crisis, physical and technical smuggling and weak customs administration were the main contributors to the decline in BOC collections (see the below discussion on the BOC tax gap). Improved administration, notably through the “Run After the Smugglers” (RATS) Program, value added tax (VAT) reforms, and higher commodity prices increased collections to 3.3 percent of GDP in 2006. Lower commodity prices in 2007 and lower imports in 2009 contributed to the lower BOC collections. Figure 43. Tax Effort (percent of GDP) Figure 44. Share of Taxes in BIR (average for 2006-08) Source: BIR Source: BIR 215. Two thirds of the decline in tax collections since 1997 is due to erosion of the value in excise taxes. Of the 3 percentage points of GDP decline in tax revenues since 1997, almost 2 points is attributable to excise taxes, with an additional 0.5 percent attributable to the corporate income tax and 0.5 percent to the personal income tax (Figures 45-48). The loss in excise tax revenue is due to lack of indexation, which has eroded the real value of excise rates and the use of a multi-tier pricing scheme that has distorted Philippines investment climate report. The World Bank, 2005. 105 118 Sufficiency, Efficiency and Equity of Public Revenues production.106 Most of the revenue loss is due to reduced petroleum excises, which fell from 1.2 to 0.2 percent of GDP. Figure 45. Excise Tax (percent of GDP) Figure 46. Corporate Income Tax (percent of GDP) Source: BIR Source: BIR 216. Tobacco and alcohol excise taxes also show large declines relative to their levels in 1997, going from 1.2 to 0.7 percent of GDP between 1997 and 2008. Administrative leakages in these products are low, indicating that tax policy is the main culprit for the revenue erosion.107 Figure 47. Personal Income Tax (percent of GDP) Figure 48. VAT Collection (percent of GDP) Source: BIR Source: BIR and BOC 106 The multi-tier pricing scheme classifies cigarettes into one of four price tiers: low, medium, high, and premium. Each tier has its own specific tax rate. In effect, this adds an ad valorem component to the taxation of cigarettes. The continued use of the tiering scheme based on net retail price has contributed to an erosion of the tax base. Manufacturers have avoided or evaded higher taxes by shifting production from high priced brands to low price brands, mis-declaring high priced brands as low price brands, or pricing cigarettes a few cents below the threshold of the next higher price band (in the case of a new brands). Between 1997 and 2006, the share of low priced brands in total cigarette production increased by 15 percent to reach over 48 percent, while both mid and high priced brands fell by over 9 and 6 percent respectively. 107 For instance, it is estimated that leakages in tobacco excise tax collection is less than 10 percent (World Bank 2008). 119 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 217. The decline in personal income tax (PIT) and VAT is primarily driven by increasingly inefficient tax administration, whereas the decline in the corporate income tax (CIT) is mainly due to lower rates and higher tax exemptions. Despite a 40 percent share of the labor force and a 60 percent share of total non-corporate sector net operating surplus, self-employed and professionals (SEPs) contribute less than 20 percent of total PIT collection. This low collection level is due to weak PIT administration reflected in insufficient audits and lack of control over SEP income tax declarations. VAT collection fell after 2006 because power transmission was exempted and the 70 percent cap on input VAT crediting was lifted as well as an overall weakening in tax administration. CIT collections grew in 2007 and 2008 and fell in 2009 by about 0.5 percentage points due to a cut in the rate from 35 to 30 percent and a proliferation of tax incentives, notably income tax holidays, creating lower than potential CIT collections. 108 Figure 49. BOC collection (percent of GDP) Source: BOC Revenue Performance vis-à-vis Targets 218. Tax performance has been insufficient vis-à-vis the government’s own targets. The 2004-2010 Medium-Term Philippine Development Plan (MTPDP) originally proposed raising public revenues, increasing expenditures and reducing the consolidated public sector deficit (CPSD) to one percent of GDP by 2010. The MTPDP outlined a proposed list of revenue enhancing measures for congressional approval to help attain the targets.109 Of the proposed measures, only the Attrition Act, VAT reform, and an increase in excise taxes were passed. As result of insufficient tax policy measures and weak tax administration, performance has consistently fallen short of targets. Total revenue and BOC revenue under-performed in most years while tax revenue fell short almost every year (Table 28). The concept of potential collection is discussed in the section on tax gap analysis. 108 The MTPDP is a document of the executive branch. The legislative branch is not bound by the MTPDP. As such, there is no 109 guarantee that revenue measures proposed in the MTPDP will be passed. 120 Sufficiency, Efficiency and Equity of Public Revenues 219. The tax targets in the last three Medium-Term Philippine Development Plans110 were consistently not met. Total revenue targets were not met either except in years when revenues are high from items that the government treats as non-tax revenues, such as income from treasury operations and privatization receipts.111 Since 2002, BIR did not meet its assigned collection target. BOC attained its collection target roughly half of the time, mostly due to more favorable macroeconomic variables than anticipated, such as higher oil prices in 2006 and 2008. On average, tax effort was below target by 1.8 percentage points of GDP in the last 12 years. A key reason why these targets have not been met is the progressive enactment of revenue-eroding measures by Congress throughout the MTPDP period. 220. Tax and revenue performance gaps are smaller when compared with budget targets. Performance gaps are smaller compared to the targets established by the Development Budget Coordination Committee (DBCC) due to the fact that the DBCC takes into consideration the latest macroeconomic stance and policy measures before submission of the budget. On average, tax performance was 0.8 percentage points of GDP and total revenue 0.2 percentage points of GDP below target over the past ten years (Table 28). The underperformance of BIR, however, has progressively widened from close to zero in 2003 to about one percent of GDP in 2007-08. Table 28: Revenue Performance vis-a-vis Targets (percent of GDP) Source: MTPDP, DBCC, BTr. 221. All of BIR’s 19 revenue regions and its Large Taxpayer Service (LTS) generally underperform vis- à-vis their assigned collection targets. In six out of the nine years from 2000 to 2008, more than half of the revenue regions failed to meet their respective collection targets without any discernible pattern in regional performance. Similarly, the LTS has failed to meet its collection target in all years since its creation in 2000. The fact that all regions as well as the LTS systematically underperform vis-à-vis their targets indicates that there is a systemic problem. 222. Underperformance with respect to MTPDP targets reflects the political gap between Congress and the executive. The MTPDP has traditionally been a document of the executive, with little involvement and ownership by Congress. Typically, Congress has not shared the desire for revenue increases projected by the executive in various MTPDPs, but instead has tended to enact measures that have precluded 110 We use three MTPDPs, the 1999-2004, 2001-2004, and 2004-2010, each one representing each of the last three administrations. When there is more than one unique revenue program for a year, as in the case when MTPDP coverage overlaps, we use the more recent revenue program. 111 The revenue shortfall would be even larger under the GFS definition where privatization is not treated as a revenue item. Indeed, since 2007, large tax revenue shortfalls have been offset by large privatization receipts. 121 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth achievement of the revenue, expenditure, deficit targets, and development outcomes of the Plans. For illustration, Annex 4 presents a summary of the revenue measures enacted by the 14th Congress, a significant number of which were revenue-eroding measures. They have also severely eroded the neutrality and equity of the tax system. Political economy constraints to tax policy reform are plentiful. However, there are also lessons from experience on what tends to “work” in designing and ensuring the implementation of tax reforms. (See Box 2 below on the Political Economy of Comprehensive Tax Policy and Administration Reforms as well as Annex 4 on how political constraints were handled in Guatemala and Colombia to ensure successful tax reform). Box 2: Political Economy of Comprehensive Tax Policy and Administration Reform Despite the political economy and capacity constraints they face, a number of developing countries have enacted successful tax reform programs in recent years. Comprehensive tax reforms have most often occurred in the midst of a fiscal crisis when non-sustainable fiscal deficits clearly signal the need to mobilize additional resources, or if there has been a change in political regime or after elections which bring to office a new reform-oriented government. Successful in-depth tax reforms have addressed both tax policy and tax administration. Alongside revenue-raising objectives, sweeping tax reforms have been accompanied by improved governance, decreased corruption, and an improved investment climate. The following recommendations are based on lessons learned from successful tax reform programs: 1. Tax reform packages should be presented as a critical component of a comprehensive governance reform program. The reform program should highlight the government’s efforts to improve public sector efficiency, transparency and accountability. 2. Tax reform programs tend to be adopted at the beginning of a new administration. The early days of a new administration provide a window of opportunity to garner political capital to enact a comprehensive tax reform. 3. These programs often reflect an agreement between the executive and the legislature. In some cases, these programs have been underpinned by an agreement between government and the legislature through a “fiscal pact” (see examples in Annex 15). 4. The most successful tax reforms have benefited from detailed planning and preparation as well as close monitoring during and after their implementation. 5. Influential tax policy experts and advisors play a critical role in setting the broad agenda for reform, guiding discussions, and helping bring about satisfactory tax reforms. 6. Tax reform is best seen as an ongoing process in which there is continual refinement of the tax system. While tax reform packages should be quickly passed at the start of a new administration, it should be understood that implementation is a long process, especially in tax administration. 7. Good data and a reliable empirical analysis of the impact of pre- and post-reform tax systems are important to garnering public and political support for proposed tax changes. Successful tax reform measures require detailed knowledge of the weaknesses of the pre-reform system, as well as an understanding of how the distribution of tax burdens will be affected by the reforms. 8. Improvements in simplicity and horizontal equity are strong selling points in favor of tax reform. Simplifying tax policy by broadening the tax base, eliminating unnecessary tax breaks, and making the cost of exemptions and tax breaks explicit in the budget help improve efficiency and transparency while also improving equity. 9. Political support can be garnered when expenditure priorities of the potential revenue gains are made explicit. Putting forth the expenditure areas that additional revenues would be devoted to (e.g. social expenditure, infrastructure) can help galvanize political support for the tax reform program in the legislature and the public at large. This can be done without outright earmarking. (See Annex 4). Sources: Diokno, Benjamin E., Reforming the Philippine Tax System: Lessons from Two Tax Reform Programs. University of the Philippines School of Economics Discussion Paper No. 0502. Philippines: University of the Philippines, 2005; and Thirsk, Wayne, “Lessons from Tax Reform: An Overview,” Policy, Research, and External Affairs Working Paper Series No. 576, 1991. 122 Sufficiency, Efficiency and Equity of Public Revenues 223. The performance gap with budget targets mainly reflects over-optimistic projections of improvements in tax administration. Moreover, in some years, mid-year adjustments in the macroeconomic framework do not lead to a commensurate adjustment in revenue targets. On average, BIR and BOC are expected to collect at least 0.3 percentage points of GDP more in revenues annually owing purely to improvements in tax administration. In 2010, DBCC expects the tax agencies to collect an additional one percentage point of GDP in revenues due to improvements in tax administration.112 International experience suggests that improvements in tax administration are difficult to implement and take time to bear fruit.113 Given this experience and the absence of an ambitious tax administration reform plan, the expected increase in revenue collection for 2010 seems highly optimistic. 224. This over-estimation also reflects information weaknesses, such as the absence of GDP data by revenue district and insufficiently disaggregated information on tax collections. The BIR should estimate its overall collections target and allocation to revenue regions and the LTS based on solid information about the tax base and previous collection history. This data should be sourced from BIR’s own database as well as complementary third-party information. Currently, however, BIR lacks disaggregated collections information by tax type at the regional level. In addition, in most cases there is no one-to-one correspondence between revenue regions and administrative regions114 and no updated listing of enterprises and other business establishments by region. Therefore, although regional income accounts are produced annually, they are not directly useful to BIR. As a result, BIR resorts to using historical collection data plus some estimation of growth rates across regions to allocate its collection targets. This method, aside from its technical weakness, raises moral hazard problems because it tends to reward under-collection. B. Collection Efficiency 225. This section examines the collection efficiency of the Philippine tax system by presenting: i) the country’s tax collection efficiency compared with selected middle income countries; ii) a decomposition analysis showing the evolution of the contribution of tax policy, macroeconomic changes, and tax administration to tax revenue growth; and iii) the tax gap for the major taxes in BIR and for BOC. Comparative Tax Collection Efficiency 226. The tax efficiency performance115 of the Philippines is well below the regional average, with higher statutory rates and lower collections. The Philippine tax effort, at under 13 percent of GDP, is significantly below the regional average of 15.5 percent of GDP. The statutory rates of major taxes (CIT, PIT and VAT) tend to be somewhat above average rates in the region and in other lower-middle 112 BIR and BOC is assigned 0.5 percentage points of GDP each in additional revenue to collect. 113 Countries implementing comprehensive tax administration reforms have typically introduced both tax policy and tax administration reforms and have sustained them over the medium-term. Even in those cases, the additional collections from better tax administration are small relative to the gains from policy reforms. For example, Indonesia’s eight year reform collected only one percent of GDP in additional revenues that is directly attributed to improved tax administration while Bolivia collected about two percent of GDP over 12 years (Tax policy and administration discussion note 2010). 114 A recommendation stemming out of this problem would be for the government to determine with finality the boundaries of administrative regions and order all national government agencies to organize their regional offices based on the new administrative regions. If this cannot be done, then the only recourse is to improve estimation with the available data and make adjustments as needed. 115 Tax efficiency is measured by tax collections as a share of the respective tax base holding tax rate constant. 123 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth income countries (Table 29). However, the revenues generated by these taxes as a share of GDP are not proportionately higher. In other words, the Philippines exhibits comparatively low efficiency of tax collections. VAT efficiency in the Philippines, at around 34, is comparable to Indonesia’s, but below the levels of Thailand and Vietnam.116 It is also well below the average for the region and for lower- middle income countries (Tables 29) respectively. The CIT efficiency of 10.3 is also low compared to Thailand’s and the regional average, although is it well above Indonesia’s.117 Finally, the PIT efficiency of 6.6, though comparable to Thailand’s, is low compared to the regional average. These low collection efficiency rates suggest weaknesses in tax policy and administration. Important features of the tax system that contribute to low efficiency and revenue yield are discussed next. Table 29: Efficiency of tax collection across selected countries Source: USAID, Fiscal Reform and Economic Governance database 2008-09, www.collectingtaxes.net. Note: *Figures for Indonesia, Philippines, Thailand, Vietnam, and Malaysia are for 2007, while the regional averages are based on figures for 2008 or the most recent year available. ** Tax efficiency is calculated as the ration of tax revenue as a share of GDP divided by the tax rate. 227. Between 1997 and 2008, the average efficiency ratio of the Philippine VAT118 was at 31 percent, compared with a regional average of above 40 percent. Figure 50 shows that, while the collection efficiency of the VAT is not low and tended to increase in years following an expansion of the base, these efficiency gains were eroded over time.119 For example, 70 percent of the efficiency gains achieved from the 2005 VAT reforms were lost by 2009 as tax eroding measures were progressively enacted. The figure also shows that the bulk of collections comes from imports, while domestic VAT collection accounts for 116 One reason for Vietnam’s higher VAT efficiency stems from the fact that its VAT system has no threshold and, therefore, all taxpayers are subject to VAT. 117 In Indonesia, income tax is generally reported without disaggregating into corporate and personal income taxes. This could explain the relatively low corporate income tax efficiency rate and the higher efficiency rate for personal income taxes. 118 The VAT efficiency ratio measures VAT collection over GDP as a percentage of the VAT rate while the VAT c-efficiency ratio measures VAT collection over total consumption as a percentage of the VAT rate. 119 A VAT efficiency of 30 percent is not particularly low by international standards. 124 Sufficiency, Efficiency and Equity of Public Revenues less than a third of the overall efficiency and is quite low. When VAT refunds, which are limited by the government, are fully accounted for, this could be even lower. Figure 50. VAT Efficiency Ratio Source: BIR and BOC Decomposition Analysis 228. For most of the 1997-2008 period, tax administration has been weak and has not contributed much to tax revenue growth. We decompose the growth in tax revenues into the contribution from the natural growth of their respective tax bases, the growth from policy changes, and the growth resulting from tax administration improvements. This analysis reveals that the bulk of the growth in revenues in 2000-08 has come from the natural growth of the tax base (Figure 51 and 52).121 The contribution of tax administration has ranged from very weak to moderate. It has been the highest in withholding taxes from compensation earners—where improvements in tax administration accounted for 3.3 percentage points of the 9.5 percent growth, and lowest for the VAT (-7.5 points of the 12.3 percent average growth in VAT revenues). The negative contribution of tax administration to value-added taxes is alarming given its importance in total tax collection. 229. Between 2000 and 2005, tax administration contributed negatively to tax revenue growth. Only in personal income taxes was the contribution of tax administration positive (Figure 51). In the case of compensation earners, it was positive because of the strengthening of the tax withholding system. For SEPs, the positive contribution of the PIT was singularly driven by the strong tax compliance campaign of the BIR against tax evading individuals (i.e., the RATE Program). In 2005, owing to the introduction of this program, collections from SEPs grew by more than 70 percent. In 2006-08, as the tax evasion campaign subsided, this contribution again became negative, as it had been before 2005. For all other taxes, the contribution of tax administration over the period was negative. 120 Appendix A provides the methodology for the estimation. Weaknesses in the source data, in particular the national accounts and households surveys, may have led to inaccurate estimates. Estimates presented here are conservative and focus on period averages rather than annual estimates with the attendant caveat on data limitations. 121 A similar but independent exercise by Manasan (2008) gives broadly the same results. 125 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 230. Between 2006 and 2008, the overall contribution of tax administration to tax collections was positive. This was primarily due to stepped-up efforts to improve tax administration, albeit in a less strategic and non-holistic manner than in 2005. The positive impact of tax administration on the CIT can be explained by the strengthening of the Large Taxpayers’ Office in 2006-07. The contribution of tax administration of the VAT continued to be negative, but less so than in 2000-05 (-4.9 percent compared to -8.7 percent). On the other hand, SEP tax administration worsened significantly as gains made under the RATE program subsided (growth contribution of -9.7 percent from an average of 4.5 percent in earlier years). Figure 51. Tax decomposition average growth in Figure 52. Tax decomposition average growth in percent for 2000-05 percent for 2006-08 Source: BIR, DOF, NSCB, FIES Source: BIR, DOF, NSCB, FIES Note: CIT – corporate income tax, CE – CE, SEP – self-employed and Note: CIT – corporate income tax, CE – CE, SEP – self-employed and professionals professionals Tax Gap Analysis 231. The BIR tax collection gap for income and VAT is conservatively estimated at about 4 percent of GDP.122 This averages about 200 billion Philippine pesos per year for the period 2000-08 in actual income and VAT revenues. Figure 53 shows the tax gap of the four major tax types in the BIR. The severity of leakages is highest among SEP (with a tax gap of 0.8 percent of GDP). It is lowest among CE (0.3 percent of GDP).123 The tax gaps for CIT and VAT amount to about 1.4 percent of GDP each. 232. Between 2000 and 2008, the BOC collection gap averaged 1.4 percent of GDP. A comparison of Philippine BOC import declarations with other countries’ export declarations to the Philippines reveals a substantial gap between what the BOC reports as imports and what other countries report as exports to the Philippines. These leakages are due primarily to technical smuggling (under-declarations or mis- declarations) and physical smuggling of imports. As a result, between 1997 and 2007, the Philippines lost 0.4-2 percent of GDP annually in border tax collections (Figure 54).124 The average revenue gap of 1.4 percent of GDP is close to industry estimates, which put the rate of smuggling at about 50 percent of 122 Appendix C gives the methodology and lays out caveats on data limitations. The annual tax gaps ranges from about 3.5 to 4.5 percent of GDP with no discernible pattern. 123 These figures conform to previous estimates by the Philippines National Tax Research Center (NTRC) and Department of Finance (DOF). The BIR does not conduct tax gap analysis. 124 The smaller gap in 2007 and 2008 is reportedly due to data reconciliation by the BOC with numbers from the rest of the world. If this is accurate, the estimated tax gap between 1997 and 2006 is largely due to data issues rather than technical or physical smuggling. 126 Sufficiency, Efficiency and Equity of Public Revenues potential. A conservative estimate of 0.4 percent of GDP translates into an annual loss of P30 billion in 2009 prices, more than the annual budget for public expenditure on health. Therefore, the combined tax gap of BIR and GOC is around 4.4 percent of GDP. Figure 53. Tax gap for BIR (percent of GDP) Figure 54. Tax gap for BOC (percent of GDP) (Average in 2000-08) Source: BIR, DOF, NSCB, FIES Source: BOT Note: SEP=income tax of self-emplyed and professionals; CE=income tax Note: The decline in the BOC tax gap in 2007 and 2008 suggests both of CE; CIT=corporate income tax; VAT=value added tax. improvement in customs administration and data reporting. Equity 233. This section discusses the de jure and de facto equity of taxes.125 Tax equity involves two important aspects: i) horizontal equity, according to which taxpayers with similar income or profits should pay a similar percentage of their gains in taxes; and ii) vertical equity, according to which lower earning taxpayers should pay a lower proportion of their income in taxes. This section first discusses the horizontal and vertical equity of major taxes and then aggregates the results to present the overall tax burdens of income and gender groups and the country’s overall Gini coefficients before and after taxes. Evolution of the PIT 234. The progressivity of the Philippine PIT has been eroded over time due to over-generous deduction allowances and weaknesses in tax administration. The present PIT system taxes CEs and SEPs using the same tax schedule of seven steps with marginal tax rates ranging from 5 to 32 percent.126 This promotes vertical equity. By taxing CEs and SEPs under the same tax schedule and, hence, taxing taxpayers with the same ability to pay similarly, the system also promotes horizontal equity. However, loopholes, excessive deductions, and weaknesses in tax administration have eroded its progressiveness across income levels and its equity across types of employment. 125 Given the resources available, we only discuss taxes that BIR collects. 126 The present PIT system is a result of a gradual improvement of the PIT design over the last three decades. Prior to 1986, the PIT system was complicated and difficult to administer. CEs and SEPs faced separate tax schedules: CEs followed a modified gross income scheme comprising of nine steps with MRTs from 1 to 35 percent while SEPs were subjected to a net income scheme with five steps and MRTs from 5 to 60 percent. In 1986, the government rationalized the PIT by abolishing the dual tax system and replacing it with a single schedule of 10 steps with MRTs from 0 to 35 percent applicable to both employment types. To prevent excessive deductions by SEPs, ceilings on allowable deductions were introduced but they could not be implemented due to strong lobby from some professional groups (e.g., doctors). Nevertheless, tax buoyancy improved from 1.5 in 1986 to 3.4 by 1997. 127 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 235. The top 10 percent of income earners in the Philippines reportedly pay around 1.7 percent of their income in income taxes.127 Using the reported taxable income in household surveys, the richest decile ought to have paid 5.8 percent of their income in income taxes.128 If income figures are corrected for under-reporting, the tax burden for the top income decile rises to 8.4 percent.129 Due to the relatively generous exemption threshold, the lowest five deciles of the population pay less than one percent of their income as income taxes. Despite the low shares of their income paid as income taxes by the rich, the PIT technically qualifies as progressive because the share paid by poorer income groups is even lower. 236. While the PIT is marginally vertically progressive, it suffers from severe horizontal inequity. Tax collections from SEPs have been very low. They account for only three percent of total BIR collections despite the fact that the share of the self-employed in total employment is 40 percent130 and their share of total non-corporate income is 60 percent.131 Evasion in SEP taxes is the highest among all BIR taxes and is rooted in regulation loopholes and weak BIR capacity to detect under-declarations and enforce tax liabilities.132 As a result, CEs pay around 80 percent of total PIT collection, while SEPs pay only 20 percent—despite the higher ability of SEPs to pay (Annex 4). In terms of tax burden, while a wage earner in the highest decile has an income tax burden of 5.9 percent, the burden of a SEP in the same decile is 2.3 percent. 237. Enhanced tax administration is critical to improving the vertical and horizontal equity of the tax system. Reducing tax evasion by increasing taxes collected from SEPs by 30 percent, and taxes collected from CEs by 10 percent can significantly improve horizontal equity. Under this scenario, the share of total income taxes paid by CEs would fall to about 70 percent while that of SEPs would increase to about 30 percent. This would significantly narrow the gap in income tax burdens across employment types, as the wage earner in the highest decile would have a burden of 6 percent while that of an SEP in the same decile would be 5.2 percent (twice their current rate). 127 Annex 4 Figures show income tax burden by decile (and centile for the top 10 percentiles) computed using reported taxable income and reported income taxes paid. They also show income tax burden by decile (and centile for the top 10 percentiles) computed using taxable income as reported and income taxes as computed from reported taxable income. Therefore, the former uses income taxes paid as reported by households while the latter computes the statutory income taxes that households ought to have paid given their declared income. 128 FIES underreports the income of the highest decile since many of the rich are not captured by the FIES. As such, these average tax rates could be interpreted as lower bounds. 129 Because households tend to underreport their income, the computed potential tax burden is also understated. Therefore, we correct for the under-reporting bias. The result is that the first five deciles of the population would still pay less than one percent of their income as income taxes, but the richest decile would pay 8.4 percent of their income as income taxes. These estimates should be considered as lower bound estimates since our tax gap analysis shows underreporting of income of as much as 20 percent for CE and almost 300 percent for self-employed and professionals. 130 Labor force survey (2000-2008) 131 National income accounts (2000-2008) 132 Foremost, tax collection from SEPs mostly relies on voluntary filing and payment. The withholding tax system applies only when SEPs have business transactions with large taxpayers and the government, which are mandated by law to withhold a portion of the payment to the SEP (usually much less than the statutory tax rate). Second, a generous list of allowable deductions without the benefit of industry benchmarks has encouraged overstating of business expenses. This system, together with under declaration of gross income, has exacerbated the degree of tax evasion. Third, BIR has weak capacity to detect and enforce tax the liabilities of SEPs. The income and expense declarations of SEPs are often not audited properly, if at all. Actual audits of SEPs, especially small taxpayers, are almost non-existent. BIR data shows that, on average, the probability that a small SEP is audited is only two percent. Finally, BIR is unable to implement the provision in the tax code on presumptive income taxation for hard-to-tax groups. 128 Sufficiency, Efficiency and Equity of Public Revenues 238. Male-headed households, which account for over 90 percent of all households, have a higher income tax burden than female-headed households. In particular, male-headed households with entrepreneurial income as the primary source of income have consistently higher income tax burdens across deciles than female-headed households. Moreover, the difference in tax burden between male and female-headed households increases with income level. One reason why female-headed households have lower tax burdens is that many of these households receive significant non-taxable income, such as remittances, from their spouses and other family members working abroad or in the cities. This finding is consistent with anecdotal evidence and survey data on practices of overseas Filipino workers as well as the gender-disaggregated Gini computations that we present later. 239. Non-taxable income reduces the Gini coefficient from 52.2 to 50.0. Based on the 2006 FIES, the income Gini coefficient before income tax and before non-taxable income is 52.2 (Figure 55). When non-taxable income such as remittances and pensions are included, inequality is reduced and the Gini coefficient comes down to 50.0. 240. Improved tax administration can significantly reduce income inequality. If we use the income tax as reported in the Family Income and Expenditure Survey (FIES), the Gini coefficient remains basically unchanged (it goes from 50.0 before income taxes to 49.7 after income taxes). In other words, if taxpayers pay the personal income taxes reported in FIES, the tax system hardly has any impact on income inequality. However, if we compute the income taxes that ought to be paid using declared income in the FIES, the after-income-tax Gini coefficient goes down to 48.8. Finally, if we adjust income for under-declaration as above, the after-income-tax Gini coefficient comes down further to 46.3. Clamping down on the tax evasion of the higher income groups, therefore, can substantially decrease income inequality. 241. Female-headed households exhibit higher inequality than male-headed households. Figure 56 shows that the Gini coefficients of female-headed households are consistently higher than the Gini coefficients of male-headed households. In particular, the before-income-tax and non-taxable income Gini is 60.6 for female-headed households compared to 51.5 for male-headed households. This is consistent with differences in the Gini coefficient on labor income, which is 0.38 for men and 0.43 for women and seems to be explained by a higher dispersion in education levels and returns to education among women.133 However, after other income such as remittances are factored in, the coefficient for female-headed households comes down to 49.8, much closer to the Gini coefficient of male-headed households of 49. A standard regression analysis on wages shows that the returns on education for women are higher for all levels of education 133 group with no education providing a strong incentive for women to achieve higher degrees and their families to finance education 129 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Figure 55. Income Gini Coefficient Before and After Figure 56. Income Gini Coefficient Before and After Income Tax Income Tax (Sex Disaggregated) Source: FIES 2006 Source: FIES 2006 A = before tax and before non-taxable income A = before tax and before non-taxable income B = before tax and after non-taxable income B = before tax and after non-taxable income C = after reported tax C = after reported tax D = after computed tax using reported income D = after computed tax using reported income E = after computed tax using adjusted income E = after computed tax using adjusted income Value-Added Tax 242. If applied uniformly to all goods and services without exemption, a VAT tends to be regressive. This is the case in most Latin American countries, which rely heavily on the VAT and where the Gini coefficient after taxes is often higher than before taxes. On the other hand, discriminate application of the VAT (i.e., exempting goods which are consumed by the poor) can make the tax less regressive. 243. In the Philippines, the design of the VAT took explicit account of distributional concerns and substantially limited its negative impact on the consumption basket of the poor.134 The law exempts all unprocessed food (i.e., food purchased from the wet market), as well as purchases from stores with gross income below P1.5 million135 (e.g., sari-sari stores and ambulant sellers). This has the effect of exempting almost all goods bought from informal sector establishments, where most of the poor buy many of their necessities. Moreover, the application of the VAT to professional services, such as legal and medical services, was designed to counter the income tax evasion of upper income self-employed and professional groups. 244. As a result, the VAT in the Philippines is mildly progressive. A VAT burden analysis suggests that the effective VAT rate (the actual VAT paid as a percent of total income) of the poor is slightly lower than 134 In the 2005 VAT reform, distributional concerns were also addressed by targeting 30 percent of incremental revenue to finance higher spending on social services and infrastructure; this proportion was targeted to increase steadily to 50 percent by 2010. 135 This threshold is adjusted for inflation annually. 136 In computing the VAT burden, we have the following caveats. First, part of the burden may be shifted to producers depending on the elasticity of the product. Second, the calculation assumes 100 percent compliance but this is not always the case in some sectors/industries. A third limitation, which is due to the effect of the exemption on marginal establishments, is resolved by computing Case 2. 130 Sufficiency, Efficiency and Equity of Public Revenues that of the rich.136 Without considering where households buy their goods and services, the lowest decile has a VAT burden of 2.4 percent while the highest decile has a VAT burden of 3.3 percent (Figure 57 Case 1).137 Moreover, if we assume that the poor buy disproportionately more goods and services from informal sector establishments, the VAT burden for the poorest decile falls from 2.4 to 0.9 (Figure 56 Case 2).138 This improves the vertical equity of the VAT. Because the top two deciles account for half of the country’s income, they also pay more than half of the total VAT collected. Figure 57. VAT Tax Burden by Percentile (in percent) Source: FIES 2006 Note: VAT burden is computed using income as base. 245. On average, female-headed households have a higher VAT burden than male-headed households. Annex 4 Figure 2 and 3 show the VAT burden for male-headed and female-headed households respectively while Figures 4 shows the difference in VAT burden between male and female-headed households. One reason why female-headed households have a higher VAT burden is that their incomes are lower than those of male-headed households. The difference is particularly stark in lower income deciles. 246. The present VAT system marginally reduces inequality. Based on the 2006 FIES, the before-VAT Gini coefficient of 50.0 comes down to 48.1 after VAT. However, if we assume that poorer households are more likely to buy from informal sector establishments, the Gini coefficient declines to 47.9 (Figure 58). Figure 59 shows the Gini coefficient for both male and female-headed households. Female-headed households exhibit higher degrees of inequality compared to male-headed households. 137 As we have noted, income is underreported, especially by the rich, so the computed VAT burden may be higher than what it should be because of the denominator. 138 We assume that the poorest decile buys 90 percent of their goods and services from informal sector establishments while the richest decile buys 40 percent of their goods and services from informal sector establishments (increments are use for other deciles). We adjust this ratio for other deciles. The ratios are consistent with purchasing practices of Filipino households. 131 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Figure 58. Income Gini Coefficient Before and After Figure 59. Income Gini Coefficient Before and After VAT VAT (sex disaggregated) Source: FIES 2006 Source: FIES 2006 A = before VAT A = before VAT B = after VAT case 1 (does not consider where households buy their B = after VAT case 1 (does not consider where households buy their goods/services) goods/services) C = after VAT case2 (considers where households buy their goods/services) C = after VAT case2 (considers where households buy their goods/services) Excise Taxes 247. Alcohol and tobacco excise taxes in the Philippines are regressive.139 This is so because the consumption of these goods by poorer households constitutes a higher share of their respective income (Figure 60). Moreover, under the current multi-tiered excise tax system, the rate on low-priced cigarettes is higher than that on medium-priced cigarettes.140 248. In contrast, excise taxes on gasoline and diesel are progressive. Gasoline and diesel are consumed disproportionately more by the better off (Figure 61). As a result, given a fixed excise rate for gasoline and diesel (i.e., there is no multi-tier pricing scheme), the tax burden of the higher deciles is higher than that of the lower deciles. An increase in excise taxes on gasoline and diesel, therefore, would also enhance the equity of the tax system. Overall, total excise taxes are slightly regressive (Figure 62). 139 Excise taxes are applied to goods such as petroleum, tobacco, and alcohol primarily to raise revenues, correct negative externalities, and discourage consumption. Unlike income or other consumption taxes, excise taxes are not typically designed to correct inequality. 140 In computing the tax burden for tobacco, we assume that the first five deciles consume cigarettes falling under the lowest tier, the next three deciles consume cigarettes falling under the lowest tier and the second lowest tier (we get the simple average) while the top two deciles buy from the second lowest tier. We disregard consumption from the highest and second highest tiers (mostly consumed by the very rich). Admittedly, this method uses heroic assumptions but given the complexity of the excise tax system for tobacco and alcohol, especially with regards to the use of multi-tier pricing scheme, this is as far an analysis we can go. For alcohol, we broadly follow the same method. Again, the complexity of the alcohol excise tax system prevents us from achieving higher accuracy. 132 Sufficiency, Efficiency and Equity of Public Revenues Figure 60. Excise Tax Burden of Tobacco and Alcohol Products by Decile (in percent) Source: FIES 2006 Figure 61. Excise Tax Burden of Gasoline/Diesel by Figure 62. Total Excise Tax Burden by Decile Decile (in percent) (in percent) Source: FIES 2006 Source: FIES 2006 249. Male-headed households have higher excise tax burdens across all excisable goods than female- headed households (Figure 64). Given higher consumption rates of alcohol and tobacco by males, the corresponding excise tax burden is higher in male-headed households. 250. The Gini coefficient declines from 50.0 before excise taxes to 48.1 after excise taxes. The trend holds for both male and female-headed households (Figure 63). This decline in inequality is driven by the strong progressivity of the excise tax on gasoline. 133 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Figure 63. Income Gini Coefficient Before and After Figure 64. Income Gini Coefficient Before and After Total Excise Taxes Total Excise Taxes (Sex Disaggregated) Source: FIES 2006 Source: FIES 2006 A = before excise A = before excise B = after excise B = after excise C. Overall Tax Burden and Income Inequality 251. Vertical equity can be improved and income inequality reduced by strengthening tax administration. The total tax burden of the upper deciles under Scenario 1 (Figure 65) would significantly increase if statutory income taxes were collected (Figure 66 Scenario 2) and under-reporting of income was prevented (Figure 66 Scenario 3). The total tax burden of the highest decile, for instance, would increase Figure 65. Total Tax Burden by Percentile (in percent) Figure 66. Total Tax Burden by Percentile (in percent) Scenario 1 Scenario 2 and 3 Source: FIES 2006 Scenario 2 uses income tax as computed from reported income and VAT Source: FIES 2006 case 2 Scenario 1 uses income tax as reported in FIES Scenario uses income tax as computed from adjusted income and VAT case 2 134 Sufficiency, Efficiency and Equity of Public Revenues from 4.9 percent under Scenario 1 to 7.2 percent under Scenario 2, and further to 11.1 percent under Scenario 3. Based on the 2006 FIES, the before-tax (and after non-taxable income) Gini coefficient of 50.0 falls to 47.7 if statutory taxes were paid and further down to 46 if income tax under-declaration was prevented (Figure 67). Gini coefficients among male and female-headed households follow the same pattern (Figure 68). Figure 67. Income Gini Before and After Income Tax Figure 68. Income Gini Before and After Income Tax (Sex Disaggregated) Source: FIES 2006 Source: FIES 2006 A = before tax and before non-taxable income A = before tax and before non-taxable income B = before tax and after non-taxable income B = before tax and after non-taxable income C = after reported tax C = after reported tax D = after computed tax using reported income D = after computed tax using reported income E = after computed tax using adjusted income E = after computed tax using adjusted income 252. The role of the tax system in the Philippines is more equalizing than in Latin America, but less than in OECD countries. In most Latin American countries, the VAT is regressive and accounts for the largest share of revenues while collections from personal income taxation as a percentage of GDP are the lowest of any region in the world. As a result, the tax system tends to weigh more heavily on the poor than the rich.141 In most OECD countries, on the other hand, the tax system is progressive and tends to weigh more heavily on upper income groups. The Philippine tax system stands between both ends and plays a very modest role in income re-distribution. (Table 30) 141 The VAT system does not discriminate much between the goods consumed by the poor and those consumed by the rich. For instance, Engel et al (1998) found that, in Chile, income distribution after tax is slightly more unequal than the before tax distribution (Gini coefficients of 49.6 and 48.8 respectively). This is the case in most other Latin American countries (World Bank 2006) 135 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Table 30: Income Inequality Before and After Taxes and Transfers Source: OECD as reported in Burniaux et al. (1998) for OECD; staff computation for Philippines using 2006 HH survey, and Engels et al (1998 for Chile. 1/ before and after tax only 2/ Using income tax as reported in HH survey Corporate Income Tax 253. The Philippines CIT has a single rate, but a relatively complex design. Since 1986, the tax code prescribes a single corporate income tax rate.142 In the 2005 tax reform, this rate was temporarily increased from 32 to 35 percent for 2006-08. Beginning in 2009, the rate was reduced to 30 percent. Notwithstanding this single rate, the definition of income and expenses which are allowed to be deducted is particularly complex. 254. One percentage point of GDP per year is lost to redundant tax incentives and horizontal inequity is high. (Reside 2006). A large number of corporations are covered by redundant tax incentives (i.e., income tax holidays, reduced rates and VAT and duty-free imports).143 The proliferation of incentives Prior to 1986, dual rates are imposed on corporations depending on size. 142 These exemptions are considered redundant in that the corporations would have invested even without the incentives because 143 the project is sufficiently profitable to start with. Reside (2006) documented that many of the incentives granted by the Board of Investments were for non-exporting domestic corporations with ex ante rates of returns in excess of 15 percent. 136 Sufficiency, Efficiency and Equity of Public Revenues and exemptions to favored industries and firms has led to large horizontal and vertical inequities in the CIT system.144 These pervasive tax incentives generate profound horizontal inequities among taxpayers and erode vertical equity, as the large majority of the corporations receiving tax incentives are among the largest and most profitable in the country (Reside 2006). These inequities, together with a complex and non-transparent administration system, undermine the perceived “fairness” of the tax system, reduce voluntary compliance, and increase incentives for corruption. 255. The wedge between the taxation of Philippine companies benefiting from the maximum tax incentives and those that do not--at 19.7 percentage points-- is one of the largest in the world. (Figure 69). The average statutory tax rate of corporations not benefiting from tax incentives is estimated at 27.9 percent, while that of corporations receiving the maximum tax holiday is 8.2 percent (Botman et al 2008).145 As a result of the relatively high statutory rates and high exemptions, firms not benefitting from tax incentives are at a great disadvantage compared to other firms in the Philippines and the rest of the world. Figure 69. Corporate Income Tax Rates (in percent) Source: Botman et al. (2008) /1 Effective tax rates for companies not receiving tax incentives; p = 20 /2 Effective tax rates for companies receiving the maximum tax holidays; p = 20 144 Every year, the Board of Investments issues a list of industries that are covered by the incentives regime. This list, however, includes many industries, including non-exporters, that would have invested anyway even without the incentives. 145 This refers to the average effective tax rate when the investment consists of plant and machinery, the investment is equity- financed, and has a pre-tax net profit of 20 percent. See Botman et al (2008) for computation of average effective tax rate for debt financed and for other investments. 137 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Conclusions 256. Public revenues in the Philippines are low and declining. They are low compared to average levels in the region and in other middle-income countries. They are also insufficient compared to the goals the country sets itself in its Medium-Term Development Plans and its annual budgets. As substantiated in previous chapters, they are also insufficient to cover expenditure needs and are contributing to the gap in human and physical capital compared to similar countries in the region. 257. Two thirds of the decline in tax collections since 1997 is due to the erosion of the value in excise taxes. Of the three percentage points of GDP decline in tax revenues since 1997, 2 points are attributable to excise taxes, 0.5 percentage points to the CIT and 0.5 percentage points to the PIT. The loss in excise tax revenue is mostly due to the erosion in the real value of excise tax rates because of lack of indexation and adjustment. The decline in CIT collections is the result of a lowering of the rate and higher exemptions while lower income tax collections are in part due to declining collections from the self-employed. Weakening tax administration has also affected collection levels throughout the tax system. 258. The tax collection efficiency of the Philippines is well below the regional average, with higher statutory rates and lower collections. For example, between 1997 and 2008, the average efficiency ratio of the VAT was at 31 percent compared with a regional average of above 40 percent. In 2005, significant efficiency gains were made owing to an expansion in the base and improvements in tax administration. However, those aims were lost by 2009. 259. The contribution of tax administration to tax revenue growth has varied over the past 10 years. Between 2000 and 2005, the contribution of tax administration to revenue growth was negative. The only tax in which it was positive was the PIT. This improvement was driven by the strengthening of the tax withholding system for CEs and a strong tax compliance campaign against tax evading individuals, including the SEPs. Between 2006 and 2008, the overall contribution of tax administration to tax collections was positive. This was primarily due to some improvements in tax administration and, in particular, the strengthening of the Large Taxpayers’ Office. The contribution of tax administration of the VAT continued to be negative, but less so than in 2000-2005. 260. The BIR tax collection gap for income and value-added taxes is conservatively estimated at about 4 percent of GDP. The leakages are highest among SEPs and lowest among CEs. Tax gaps for CIT and VAT are about 1.4 percent of GDP each. The tax gap for the BOC is conservatively estimated at 0.4 percent of GDP. 261. The progressivity of the personal income tax has been eroded over time due to over-generous deduction allowances and weaknesses in tax administration. While it is marginally vertically progressive, it suffers from severe horizontally inequity between self-employed and CEs. 262. The VAT is mildly progressive while excise taxes are regressive. Because the VAT exempts all unprocessed foods as well as purchases from marginal stores, it exempts most of the establishments from which the poor buy their necessities. As a result, VAT payments tend to fall on the middle and higher income groups. Excise taxes on alcohol and tobacco are regressive while those on gasoline and diesel are progressive. 138 Sufficiency, Efficiency and Equity of Public Revenues 263. The CIT is highly horizontally inequitable. The wedge between the taxation of companies benefiting from the maximum tax incentives and those that do not (at 19.7 percentage points) is one of the largest in the world. As a result of relatively high statutory taxes and high exemptions, firms not benefitting from tax incentives are at a great disadvantage compared to other firms in the Philippines and the rest of the world. Because it is the larger firms that tend to get exemptions, the system is also vertically inequitable despite a progressive statutory rate. This inequality also distorts the price system and increases administration and compliance costs as well as incentives for corruption. 264. The overall tax system is mildly vertically progressive, but could be more so with improved tax administration. The Gini coefficient is slightly lower after taxes (47.7) than before taxes (50.0). If individuals paid the taxes they owe based on estimates adjusted income for under-reporting in household surveys, the Gini would fall further to 46.0. 265. Male-headed households have higher income and excise tax burdens, but lower VAT burdens. Male-headed households, especially those with entrepreneurial income as the primary source of income, have consistently higher income tax burdens than female-headed households. The difference in excise tax burden is due to differing consumption patterns across male and female-headed households. The greater VAT burden on female-headed households can be attributed to their lower income levels. Recommendations 266. There is great scope for increasing revenues while improving the neutrality and the equity of the tax system. Increasing revenues would require the establishment of a sound tax policy framework, complemented by an efficient and well-governed tax administration system. This section lays out our recommendations in both areas. A. Tax Policy 267. Enact a comprehensive tax reform package and enter into a “fiscal pact” with Congress to preserve revenue gains. (Top 10 Recommendations Numbers 1 and 10). The package should include both tax policy and tax administration measures and be enacted during the first six months of the new government’s tenure. The fiscal pact with Congress would bring the legislature into a political pact with the executive to ensure ownership by Congress of the overall package and prevent its undermining. This pact should include: a) Measures to be implemented in 2010 • Repeal the revenue eroding measures enacted in 2009 and instituting a moratorium on enacting any others. Reversing the 2009 measures would produce a gain of at least 0.3 percent of GDP over the medium-term. Table 31 summarizes the revenue impact of these proposed measures. 139 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth • Simplify excise taxes, increasing excise tax rates and indexing them to annual inflation:146 ◊ Tobacco and alcohol excise taxes: Collapse the multi-tiered pricing system into a single rate, increasing the rate to at least their real value in 1996, and index that rate to annual inflation thereafter. This reform would generate 0.5 percent of GDP in the first two years of implementation. ◊ Gasoline and diesel excise taxes: increase their rate by about 50 percent and index it to inflation annually. This reform would generate 0.8 percent of GDP over the medium-term. • Implement the Simplified Net Income Taxation Scheme (SNITS).147 This reform would generate 0.3 percent of GDP over the medium-term. In the future, the reform could be expanded to further simplifying the personal income tax system for self-employed and professionals. This involves keeping only a limited amount of deductions and credits and simplifying the rates, while keeping a high personal exemption threshold to at least the level that will fully cover minimum wage earners with a family of six (or about P200,000) to ensure adequate progressivity and easier administration. • Broaden the VAT base by reversing the exemptions for senior citizens and power transmission. This reform would generate 0.1 percent of GDP over the medium-term and, more importantly, would improve VAT administration. b) Medium-term recommendations (next 18 months of the new administration) • Broaden the tax base of the CIT, eliminate exemptions, and simplify the tax code. Such a reform would improve tax efficiency by lowering administration and compliance costs, eliminate tax redundancies, sharply improve horizontal (and even vertical) equity among firms, remove distortions in resource allocation, and improve tax administration and overall governance. Revenue targets permitting, the government could also consider lowering the CIT rate to bring it in line with that of competitor countries. This reform should generate at least one percent of GDP over the medium-term. • Consider increasing the VAT rate from 12 to 15 percent. This measure would generate 0.9 percent of GDP from the first year of its implementation. Alternatively, the government could consider further increases in excise taxes. These reforms are staggered over several years to lessen the increase in specific excise taxes. 146 The simplified net income taxation scheme (SNITS) aims to reduce opportunities for over-declaring expenses by limiting allowable 147 expenses to direct costs incurred in the operation of a business, such as the cost of goods, rent, and utilities. Other expenses such as representation and transportation are either removed or capped. 140 Sufficiency, Efficiency and Equity of Public Revenues Table 31: Reform Scenario Revenue Options Source: DOF, World Bank staff calculation Possible additional item: VAT increase from 12 to 15 percent; 0.9 percent of GDP. B. Tax Administration 268. Sustainable reform of tax administration will require a firm commitment at the highest political level, complemented by sound strategic management at the bureaucratic level. It will also require facing corruption head-on. Given the depth of the governance problems currently affecting BIR and BOC, an effective institutional reform needs to be accompanied by a broader anti-corruption campaign; one that goes beyond BIR to encompass other public sector institutions, including the justice system. Congressional support may also be necessary. A continued focus by civil society, the private sector and the media on good governance in tax administration is also likely to be critical for sustaining the reform process. Furthermore, reform commitment needs to be complemented by sustained leadership in implementation. A major problem that has bedeviled previous reform efforts has been the frequent turnover of BIR commissioners. This needs to be avoided. 269. Strategic performance management will be key to success. In addition to being properly focused, prioritized and sequenced, the reform program needs to be complemented by a detailed implementation plan and monitored through a set of agency-level key performance indicators. The BIR is currently working on such a set of agency-level indicators. These indicators should be stable through time and serve as a management tool within the agency. Also, individual performance indicators and BIR programs should be closely aligned with these key performance indicators. To maximize the likelihood of successful implementation, the key performance indicators could be made public and monitored by congress, the private sector, civil society and the media. 270. The following tax administration reforms are recommended to enhance collections and improve efficiency and overall governance. We estimate that improved tax administration can yield 1.4 percentage points of GDP in additional collection over the medium-term. 141 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 271. Immediately after the new government’s coming into office, the following key measures should be adopted: • Appoint a BIR commissioner for a minimum period of time (e.g., 3 years and renewable), with performance evaluation based on the agency’s key performance indicators. • Adopt a set of agency-level key performance indicators, to be agreed between BIR and DOF, and present them to congress to amend the Attrition Law. • Require public disclosure of all assets, liabilities, and net worth of BIR personnel; commencing lifestyle checks of all BIR managers and disciplining corrupt officials. • Approve the BIR’s rationalization plan so that additional personnel can be hired 272. In the short-term (during the first six months of the new administration): • Strengthen the Large Taxpayers Service (LTS), ◊ Introduce a system of automatic listing, de-listing, and notification of firms meeting LTS eligibility criteria based on a threshold according to which around 1,500 firms would qualify. ◊ Increase the number of staff at the LTS to comparable international levels, particularly in the audit function. ◊ Ensure proper functioning of the e-filing system and remove the requirement for dual accounting (manual and electronic). ◊ Diagnose the poorly-functioning VAT refund system and prepare an action plan to improve its efficiency. • Strengthen tax enforcement: ◊ Implement a comprehensive compliance strategy, including strengthened capacity in the audit, investigations and VAT compliance functions and better preparation of RATE148 and non-RATE cases sent to the DOJ. ◊ Set clear targets for number of cases filed, processing time, and other relevant indicators of effectiveness (such as conviction rates). ◊ Relax the bank deposit secrecy law to allow for proper investigation of fraud and other criminal tax cases. 273. In the medium-term (next 18 months of the new administration): • Clean-up and expand the taxpayer database (registration and transactions): ◊ Improve access to third-party information by promulgating an executive order mandating all government agencies and selected private sector organizations to provide timely and accurate information to BIR free of charge. ◊ Accelerate efforts to computerize/automate and link all BIR offices and LGUS via network to facilitate BIR’s registration and third party information cross checking. Run After Tax Evaders (RATE). 148 142 Sufficiency, Efficiency and Equity of Public Revenues • Strengthen the audit function: ◊ Introduce a risk management approach linked to a national risk-based audit plan. ◊ Increase the number of auditors, strengthen the capacity of the internal audit unit and improve overall audit supervision. ◊ Enforce existing rules (e.g., EO 38 series of 1998) that allow the Commission on Audit to audit BIR’s assessment function and adherence to its own audit regulations. • Increase tax compliance: ◊ Develop a strategy to improve taxpayer services so as to enhance voluntary compliance. This could include simplification of forms, streamlining of certain procedures required of taxpayers, increased access to paying in banks or via internet, overall improved customer service monitored through an independently managed “report card” system etc. This would help increase collections from self- employed and professionals. ◊ Develop an improved arrears management system, including an agreed definition of arrears and reporting all presently unreported arrears. ◊ Introduce a risk-based collection strategy, accompanied by a collections manual to guide implementation. ◊ Improve the VAT refund system over time to reduce the cost of doing business and improve the fairness of the VAT system. • Improve fiscal statistics: • Enhance tax data generation to allow for more accurate revenue analysis, target setting and allocation of revenues to the regions and districts. Required data include detailed tax collection by tax type, by taxpayer, and by districts, and other tax administration information, such as data on registration, audit, and arrears. This data should be reported to oversight agencies (e.g., DOF, Congress) in a timely manner (e.g., at most one month after the reporting period). DBCC could helpfully set reporting standards and penalties for non-compliance to ensure that data is reported properly.149 • Record tax expenditures and reflect them in the budget. Require all corporate taxpayers receiving tax incentives to file their tax returns for proper recording of tax expenditures and reflect them in the government’s budgets. • Strengthen the analytical capacity of the BIR, BOC, and DOF, including the NTRC, in revenue programming and tax gap estimation analysis using available data from taxpayer returns, import declaration, and third party information. • Improve the capacity building of other government agencies that supply data to BIR, such as the NSO and NSCB on household surveys and national accounts. • Enhance tax computerization: ◊ BIR needs to upgrade and expand its information technology systems in order to accommodate the expected increase in registrants and have the capacity to appropriately handle third party information. This will require upgrading the Integrated Tax System, increasing the capacity of data A related reform to improve revenue forecasting and fiscal planning would be for the government to determine with finality the 149 boundaries of administrative regions and order all national government agencies to organize their regional offices based on the new administrative regions. If this cannot be done, then the only recourse is to improve estimation with the available data and make adjustments as needed. 143 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth storage facilities, expanding and upgrading network infrastructure and data exchange facilities, and matching tools among its regional and district offices as well as in government agencies and private sector organizations that supply third party information. In addition, compliance can be improved by exploring how taxpayers can file and pay taxes by using their mobile phones given the very high mobile phone subscription rates. • Consider creating a “Tax Watch” non-governmental organization aimed at monitoring and assessing developments in Philippine tax policy and administration. The organization could be formed by representatives from the private sector, civil society and academia and work closely with government and Congress. Its activities could include monitoring tax collections and key performance indicators in tax administration, carrying out surveys of customer satisfaction, elaborating projections of the costs of potential revenue-eroding measures and studies on tax evasion, compliance costs and other key areas in tax policy and administration. References Botman, D, Klemm, A., and Baqir, R. (2008), “Investment Incentives and Effective Tax Rates in the Philippines: A Comparison with Neighboring Countries,” IMF Working Paper 08/207, International Monetary Fund, Washington, D.C. Engel, E., A. Galetovic, A., Raddatz, C. (1998), “Taxes and Income Distribution in Chile: Some Unpleasant Redistributive Arithmetic,” NBER Working Paper Series 6828. Government of the Philippines. “Medium-Term Philippine Development Plan,” Various years (1998-2004, 2001-2004, and 2004-2010). Manasan, R. (2008), “Sustainability and Improvement in BIR Tax Effort in 2005-2007,” Working paper. Reside, R. (2006), “Fiscal Incentives and Investment in the Philippines,” Discussion Paper. World Bank (2005), “The Philippines: Towards a Better Investment Climate for Growth and Productivity.” World Bank (200x), “Economic Mechanisms for the Persistence of High Inequality in Latin America,” and “Taxation, Public Expenditures, and Transfers,” in “Inequality in Latin America.” World Bank (2006), “Pro-poor Growth in Latin America,” in “Poverty Reduction and Growth: Virtuous and Vicious Circles.” World Bank (2007), “The Political Economy of Reforms in the Philippines: The State of Knowledge and Implications for the CAS,” Annex II, Political institutions and tax policymaking in the Philippines: A case study, Working Paper. World Bank (2008), “Adjusting Tobacco Excise Taxes,” Policy Note. World Bank (2008), “Strengthening Tax Administration,” Policy Note. World Bank (2010), “Tax Policy and Administration,” Discussion Note No. 3. 144 Chapter 6. Medium-Term Macroeconomic Outlook Medium-Term Macroeconomic Outlook 6. Medium-Term Macroeconomic Outlook Introduction 274. In 2010, the Philippines is at a crossroads. The May general elections and a post-global crisis environment offer a unique window of opportunity for undertaking structural reforms. The growth outlook for the Philippines is critically contingent on whether the incoming government is able to successfully implement such reforms. The key areas to address will be investment climate, governance and corruption, and fiscal reforms. Accordingly, two outlooks for growth and other macroeconomic variables are considered here: • A baseline scenario: in the absence of in-depth reforms, medium-term economic growth is projected to be lower than the average during the boom period (2003-2007), given reduced prospects for investment and net exports. Spare capacity is expected to persist while total factor productivity growth would slow down. Some sectors of comparative advantage, however, would nonetheless benefit from global restructuring in the aftermath of the crisis. • Reform scenario: growth would accelerate as structural bottlenecks to inclusive growth are tackled and fiscal reforms are implemented. The fiscal reforms assumed in this case scenario are a central subject of this Public Expenditure Review. They would need to be complemented with improvements to the investment climate and overall governance. Jointly, they have the promise to take the Philippines out of its current low-growth, high inequality trap and propel it into a high inclusive growth path to re-enter the ranks of the region’s leading nations. A. Baseline Scenario (absence of in-depth reforms) 275. In the baseline scenario, growth is projected to be lower than pre-crisis levels, declining from an average of 5.4 percent in 2003-2008 to 4.1 percent in 2012-2016 (Table 32). While growth in 2010 is comparable to immediate recovery after past crises, the pace of medium-term recovery is projected to be less robust than in the past (World Bank, 2010). Unlike in previous recoveries in which the global environment was strong, a soft recovery is now projected for the global economy. The modest pace of the global economic recovery and the new post-global recession equilibrium will impact domestic growth via remittances, exports, capital inflows and access to capital. Some of these factors will structurally benefit the Philippine economy, such as the sea-based Overseas Foreign Workers and the Business Process Outsourcing (BPO) sectors which are expected to benefit from the new wave of global restructuring and cost consolidation induced by the crisis (Gereffi and Fernandez-Stark, 2010).150 At the same time, structural impediments to growth like poor investment and governance climate and insufficient human capital and infrastructure will continue to hold back medium-term growth. 276. Private consumption would be negatively affected by the worsening fiscal stance, weak labor market and investment climate. Private consumption is projected to average around 0.3 point less than in 2003-2008. As growth is projected to be moderate, and given the rapidly growing labor force, the labor market is estimated to remain weak, dampening private consumption notwithstanding good remittance prospects. Inclusive growth would remain a challenge. The sea-faring industry has been sharply affected by the global recession and the collapse in global trade and cruise ship tourism. 150 This has created pressure to drastically reduce costs and has led some companies to accelerate their staff sourcing from countries such as the Philippines, which has a large pool of comparatively cheap, English-speaking, and well qualified sea-farers. 147 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Table 32: Philippines: Baseline Growth Projection, 2006-2016 Source: National Statistical Coordination Board, World Bank staff estimates. 277. Overseas deployments and remittances would continue to perform well. Global remittance flows are projected to grow at a slower pace since the global recovery is expected to have lower job intensity and tighter immigration controls are being introduced. As a result, remittances to the Philippines are projected to grow by 9.6 percent annually in 2012-2016, or 6.1 percent below the level of the boom period. This compares favorably with global remittance growth prospects of 7.9 percent over the same period (Figure 70). This is largely explained by the wide geographic and skill distribution of deployed Filipinos and the continually rising number of deployed workers. Robust deployment of workers overseas partly reflects recession-induced global staffing restructuring and the attractiveness of Filipino workers in the global labor market. 278. Structural bottlenecks and weaknesses in the Government’s fiscal performance are projected to result in a persistently low investment-to-GDP ratio. Private sector incentives to invest would be limited in the absence of structural reforms, including stepped-up expenditures to improve the country’s human and physical capital and hence its competitiveness, and with macroeconomic stability endangered by weakening public finances. Public expenditures on human capital and infrastructure investment would also be constrained by the state of public finances. Total investment as a ratio of GDP is projected to decline from 15.6 percent in 2003-2008 to 15.1 percent in 2016, following its declining trend in the boom period. Net foreign direct investment (FDI) inflows are expected to be broadly similar to the pre-crisis period. Once global fiscal and monetary stimuli are withdrawn, however, portfolio investments held by foreign investors would be lower than during the boom period due to tighter global liquidity. The electricity sector, which has been deprived of sufficient investment for expanding generational capacity, runs the risk of a supply shortage. This would affect the economy from 2013 onwards. 148 Medium-Term Macroeconomic Outlook Figure 70. Remittance are Projected to Perform Well as Figure 71. Debt Dynamics Would Worsen as a Result of Deployment Increase and the Peso Weakens High Deficits, Weak Growth, and Rising Spreads. Source: BSP, World Bank Source: World Bank staff estimate Figure 72. Spreads are Projected to Rise in the No Figure 73. Deficit Financing Could Crowd Out Reform-No Fiscal Consolidation Scenario Domestic Invesment Source: JP Morgan, World Bank Source: BTr, World Bank 149 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Figure 74. Domestic Interest Rates are Projected to Rise Figure 75. Capital Spending* is Set to be Negatively Due to Increased Risk Aversion Affected by the Weaker Macro Environment Source: DBM, World Bank staff estimate Source: BSP, World Bank *Includes net lending 279. Net exports are not expected to significantly contribute to GDP growth. Export growth is projected to continue tracking developments in the market for electronics and semi-conductors as these sectors account for about 60 percent of the country’s total exports (Figure 76). With the recovery of high income countries projected to be soft, the growth rate of exports of goods and services is projected to average 6.1 percent annually, slightly below the boom period average. Imports of goods and services are also projected to expand broadly at the same rate as exports. The terms of trade are projected to remain at the pre-crisis average level (Figure 78).151 Figure 76. Share to Exports, Average 2008-2009 Figure 77. Share to Imports, Average 2008-2009 Source: NSO Source: NSO Development in electronics and semi-conductor prices have limited impact on the Philippines’ term of trade as exports in that 151 sector are part of a global supply chain and domestic value added is relatively small. The country’s terms of trade are more sensitive to the prices of imported oil and consumer goods, especially rice. As international prices of these commodities are projected to be broadly stable, the terms of trade over the medium term is projected to remain at the pre-crisis average level. 150 Medium-Term Macroeconomic Outlook 280. Trade in services is expected to strengthen, in part owing to the global recession. The BPO sector is expected to continue to strive over the medium-term as cost consolidation among global firms accelerates due to the global recession. The country’s competitive edge—especially English proficiency, competitive telecommunications infrastructure and relatively low labor costs—will continue to lure foreign investment in this sector. However, the industry may face growing difficulties in recruiting quality staff in sufficient numbers given its projected growth and declining quality of the education sector. 281. In the no-reform scenario, potential growth would remain below pre-crisis boom levels over the medium-term. Potential growth is estimated to have averaged 4.6 percent in 2002-2008. Based on a growth accounting framework, annual potential growth over the medium-term is estimated to slow down to 4.2 percent from 2009 to 2012 and to stay broadly stable through 2016. As a result, the output gap that opened up with the food and fuel price shock and widened with the global recession to about 3.2 percent of actual GDP is projected to close in 2013 (Figure 79). The slow recovery is due to a slower increase in capital and a decline in productivity growth. Figure 78. Terms of Trade Projected to be Stable Figure 79. Growth Path, Transition and Recovery Source: NSCB, World Bank Source: NSCB, NSO, World Bank staff calculation 282. Public finances would increasingly become a source of risk to macroeconomic stability. Growing deficits, along with low growth and rising spreads, would cause public debt to continue rising as a share of GDP over the forecast horizon. Combined with the public sector’s large gross financing requirements, currently at about 19 percent of GDP, this increased public debt would have the potential for creating macroeconomic instability. 283. Without policy actions to reduce the fiscal deficit, risks to debt dynamics are tilted to the upside. The baseline scenario projects a debt path that is broadly stable with significant uncertainty and risks on the upside.152 Based on the historical shocks that have hit the Philippines, there is a 90 percent probability that, by 2013, NG debt would be within the range of 55-65 percent of GDP. Were the sovereign debt crisis to deepen, risk premia could rise further, driving up the cost of servicing the government’s debt. This projection is made based on the “fan chart” approach (Celasun et al., 2006) which summarizes risks to debt dynamics by 152 representing the frequency distribution of a large sample of debt paths generated by means of stochastic simulations. 151 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 284. Any minimal gains from improved tax administration would be lost to tax incentives and the continued erosion in the real value of excise tax rates. Low tax yields would likely result in further expenditure compression. Primary expenditure is projected to fall from 14.9 percent of GDP in 2009 to 13.3 percent of GDP in 2016 while interest payments would continue to rise. The wage bill will drive overall spending but at the expense of MOOE and capital outlays. MOOE and capital outlays are not expected to grow, as a ratio of GDP, given the tight fiscal space. Internal revenue allocation would also fall given the lower taxes collected by BIR. In the absence of local tax reforms, this would squeeze the budget of most local governments. Low revenues and subsequent expenditure compression would lower the deficit (by government definition) from 3.9 percent of GDP in 2009 to 2.3 percent of GDP by 2016. 285. Lack of improvement in the welfare and human capital of the poor would continue to constrain the country’s overall growth prospects. With weak economic growth and labor markets, progress in fighting poverty and inequality will remain disappointing. Using micro-macro simulations, the global recession is projected to throw 1.9 million Filipinos into poverty by 2010, compared to what would have been without it (Habib et al, 2010). The long-term effects of the crisis in a no-reform scenario could have a more severe impact on the poor as even a small reduction in their income can have a lasting impact on their overall welfare as households withdraw children from school and further reduce their access to health services. Moreover, tightening fiscal space could also lead to cuts in investment in public education and health, further endangering the country’s human capital base for the future. B. Reform Scenario 286. In the reform scenario, comprehensive tax reform, increased investment in human and physical capital and reduced fiscal deficits would support inclusive growth. A comprehensive tax policy and administration reform program would be enacted in 2010. The program would be geared at increasing revenue collections, widening fiscal space, preserving macroeconomic stability, and improving expenditure efficiency and equity as well as overall governance. The envisaged revenue measures would include a broadening of the tax base of the VAT, a simplified net income taxation scheme, and increases and indexation of excise taxes on tobacco, alcohol and petroleum products. In the outer years, revenue gains would also be derived from the phase-out of tax incentives and the impact of tax administration reforms (both assumed to start in 2010). In addition to generating revenues, the phase-out of tax incentives would improve the horizontal and vertical equity of the tax system and reduce distortions to resource allocation. These reforms are projected to result in an increase in the tax effort from 14.2 percent in 2008 to 17.2 percent in 2016. If the government wished to increase expenditures further, an increase in the VAT rate from 12 to 15 percent would yield around 1 percent of GDP. 287. Strengthened tax collections could support greater investment in human capital and infrastructure, fostering inclusive growth. With increased tax collections, the government could step up spending on priority sectors. It would enable an increase in primary spending from 13.5 percent of GDP in 2008 to 17.7 percent of GDP by 2016, benefiting mostly the social and infrastructure sectors. Starting in 2011, the social sectors, especially health and education, would benefit from these expenditure increases. The increase in primary spending would be driven by increases in capital outlay and MOOE by 2.0 and 1.6 percentage points of GDP, respectively, relative to their level in 2008. While the wage bill would increase 152 Medium-Term Macroeconomic Outlook as a result of the Salary Standardization Law (SSL III) and better remuneration for key civil servants such as teachers, health workers, and technical staff, it is assumed that these would be part of a broader civil service reform that would not only limit overall wage growth but also improve overall public sector efficiency. In terms of sectoral spending, education spending would increase by 1.7 percent of GDP (through a combination of personnel services, MOOE, and capital outlay), health by 0.7 percentage points of GDP and social protection by 0.3 percentage points of GDP (with agriculture subsidies decreasing by a similar amount). The justice and security sectors would also see a total budget increase of about 1 percent of GDP relative to 2008. 288. Enhanced macroeconomic and growth prospects would lower interest payments, with the savings being reallocated to priority sectors. At the same time, government would rationalize spending on low- priority and inefficient areas (such as subsidies to the NFA) and reallocate the savings to increase conditional cash transfers (by about 0.3 percent of GDP) to enhance the demand of the poor for education and health services. Strengthened public sector management would enhance the transparency and accountability of government expenditures and the efficiency and effectiveness of government institutions. This would improve the delivery of public goods and services. The deficit would be reduced to about 1.5 percent of GDP by 2016, the debt stock would fall and the debt stance would become sustainable. Reforms are also assumed to include key sources of fiscal risks such as NFA, thereby further widening of the fiscal space and limiting fiscal risks that could derail the overall budgetary process. 289. A strengthened public investment management system would significantly enhance the efficiency of public investment. For example, a reduction in the share of Congressional allocations in infrastructure projects or a strengthening of the selection and implementation process for projects funded by these allocations could result in significant gains in the efficiency of infrastructure expenditures. 290. In addition to fiscal reforms, robust and sustained improvements to the investment and governance climate are needed to enhance the country’s overall competitiveness. Priority reforms should focus on forcefully tackling corruption and the weak contestability of markets, which currently lead to a high cost of doing business (World Bank, 2007 and 2009). In this regard, clarity, simplicity and stability of overall investment regulations (including in the tax system) are critical. Improving the access of micro, small and medium enterprises (MSMEs) to finance would also be instrumental in generating more inclusive growth, as MSMEs are essential players in job creation (World Bank, 2009). The potential power crisis of the baseline scenario would be averted as improvements in the business climate and the institutional and regulatory quality of the power sector attract investments. 291. In the medium term, a small fiscal deficit and a return to a sustainable debt path are projected. By 2016, the fiscal deficit (based on government definition) is projected to come down to 1.5 percent of GDP (from 3.9 percent in 2009) and the NG debt is projected to reach 44 percent of GDP (from 57.3 percent in 2009). Not only are the debt dynamics sustainable but, with the reduction in fiscal risks and improved debt management, the resiliency to shocks would be significantly strengthened. 292. The reform scenario would generate substantially higher growth rates and a rapid decrease in poverty incidence over the medium term. Enhanced investment in human and physical capital and higher exports would be the catalysts of growth (Figure 80). Improvements in the investment climate, jointly with stepped-up public investment in human and physical capital, would trigger private investment, 153 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Figure 80. Growth Path Under the Reform Scenario Figure 81. Reform Scenario Growth Path, Transition and Recovery Source: NSCB, NSO, WB staff calculation Source: NSCB, NSO, WB staff calculation including large foreign direct and portfolio investment. Private consumption, especially in the latter part of the forecast horizon, is expected to rise as fiscal consolidation and improved employment prospects boost confidence and incomes. Poverty incidence is projected to decline strongly over the medium term as growth is more widely shared. 293. Exports would benefit from increased investment in human and physical capital and a more enabling business environment. Total exports growth is projected to rise under the reform scenario. This is in part due to a surge in commodity exports (agricultural, metals and minerals), which are currently negatively impacted by inadequate infrastructure, a high cost of doing business and a weak investment climate. Trade in services, such as tourism, is projected to expand significantly with increased competitiveness and greater social stability. Imports are also projected to accelerate as the investment surge would likely be import-dependent. Overall, the trade deficit is expected to increase, but remain sustainable over the medium term. 294. As the domestic economy strengthens and investment rises, remittance growth is expected to slow down and, along with a higher trade deficit, drive the current account into a moderate deficit. The economic upturn of the reform scenario would encourage more Filipinos to stay in the country leading to a decrease in remittances as a percentage of GDP. The peso is expected to appreciate along with the strengthening of the economy and rising capital inflows (this could also limit incentives to migrate). Coupled with a strong increase in domestic investment, the current account is projected to turn slightly negative in 2015-2016. 295. As pointed out at the beginning of this chapter, the incoming government is at a crossroads. Lack of reform of the government’s finances and the investment and governance climate would continue to lead the country down a path of low investment, low productivity and GDP growth and poor progress on poverty reduction. Forceful reform, on the other hand, holds the promise of returning the Philippines to a high- investment, high-productivity growth path, resulting in fast economic growth and poverty reduction. 154 Medium-Term Macroeconomic Outlook Table 33: Growth Projections in the Reform Scenario, 2006-2016 Source: World Bank. References Gereffi, G. and Fernandez-Stark, K. (2010), “The Offshore Services Value Chain: Developing Countries and the Crisis,” Policy Research Working Paper No. 5262, World Bank, Washington, D.C. Habib, B., Narayan, A., Olivieri, S., and Sanchez-Paramo, C. (2010), “Assessing ex ante the employment, poverty and distributional impact of the global crisis in Philippines: A micro-simulation approach,” forthcoming Policy Research Working Paper, World Bank, Washington, D.C. World Bank (2007), “Invigorating Growth, Enhancing its Impact,” 2007 Philippines Development Report. World Bank (2009), “Battling the Global Recession: Managing the Global Recession and Preparing for the Recovery,” 2009 Philippines Development Report. World Bank (2010), Philippines Quarterly Update, February 2010. World Bank (2010), “Philippines: Fostering More Inclusive Growth,” Washington, D.C.: The World Bank. 155 ANNEXES Annex 1 Annex 1 (For Chapter 2: Efficiency of Public Spending) Efficiency Analysis Methodology (Chapter 2) 1. The efficiency analysis in Chapter 2 is mainly guided by Data Envelopment Analysis (DEA). DEA is a non- parametric programming method for estimating the efficiency frontier and distances to the frontier in order to evaluate relative efficiencies. Pioneering work is generally attributed to Farrell (1957) and popularization to Charnes, Cooper and Rhodes (1978). With assumptions of common technology, it allows for the estimation of an efficiency frontier without the need to fully specify the production function and thus helps provide quantitative estimates with parsimonious assumptions. Banker, Charnes and Cooper (1984) further extended DEA from Constant Returns to Scale (CRS) to Variable Returns to Scale (VRS) situations. 2. The principle of DEA can be readily adapted to assess the efficiency of public spending between and within countries. In this context, DEA can first define the efficiency frontier by countries or regions that achieve the same or better outcomes with spending lower than other countries/regions in the sample. The relative efficiencies of other countries can then be estimated by their distances to this frontier. DEA also has potential drawbacks, such as its sensitivity to measurement errors and to sample selection. In small samples, the estimates tend to be less accurate for observations with the lowest input values. 3. In cross-country studies, DEA results may compound inefficiencies in sectoral spending with extra-sectoral institutional inefficiencies that also affect the sector’s input-output relationships. These drawbacks can be mitigated by limiting the analysis to countries at similar levels of development and by using period averaging. Applying DEA to regions with a shared institutional environment within a country can further address potential drawbacks. VRS is generally considered more suitable for a cross-country setting while CRS is more suitable for within-country comparisons because the former better accommodates the heterogeneity among countries while the latter fits better with a relatively uniform institutional environment. Annex 1 provides the technical details on the implementation of DEA for efficiency analysis 4. DEA has been widely used in applied analysis of the efficiency of public spending in both developed and developing countries. In developed countries, Verhoeven et al. (2007) use the technique to assess the efficiency of education and health spending in G-7 countries and analyze how institutional factors, such as staffing and wage spending, class sizes and immunizations, can explain relative efficiencies. Afonso and Aubyn (2004) apply DEA to compare the efficiency of education and health spending among OECD countries and argue that this tool is most useful in assessing the efficiency of public spending in areas dominated by public provision. Mandle et al. (2008) use DEA to analyze the efficiency of education and R&D spending in EU member countries and argue that efficiency analysis can provide the most useful results when focused on individual functional spending areas. 5. In developing countries, Gupta and Verhoeven (2001) compare the efficiencies of education and health spending in Africa to similar developing countries in Asia and Latin America and conclude that there is significant scope for efficiency improvements in many African countries and that improving education and health outcome requires more than higher budget allocations. Herrera and Pang (2005) use the method to estimate the efficiency of education and health spending in a broad set of developing countries and find that the efficiencies are negatively associated with the size of the wage bill, income inequality, and aid dependency. 159 Philippines Annex 1 Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 6. In Chapter 2, DEA is applied to education, health, and transport spending in the Philippines. The analysis begins with cross-country benchmarking and then delves into cross-regional analysis within the Philippines. In cross-country comparisons, VRS output efficiency scores153 are presented while for cross- regional comparisons, CRS results are presented.154 7. DEA is a non-parametric programming method for estimating the efficiency frontier and distances to the frontier in order to evaluate relative efficiencies in business applications. It was generally attributed to Farrell (1957) for pioneering the work and to Charnes, Cooper and Rhodes (1978) for popularizing it. With assumptions of common technology, it allows estimation of efficiency frontier without the need to fully specify the production function and thus helps provide quantitative estimates with parsimonious assumptions. Banker, Charnes and Cooper (1984) further extend DEA from Constant Returns to Scale (CRS) to Variable Returns to Scale (VRS) situations to support the analysis of production models that may have different optimal scales. 8. The principle of DEA can be readily adapted to assess the efficiency of public spending between and within countries. In this context, DEA can first define the efficiency frontier by countries or regions that achieve the same or better outcomes with spending lower than other countries/regions in the sample. As shown in The Efficiency Frontier graph, countries are represented by observations A, B, C, D, E, F, and G with their respective inputs and outputs. The straight line OA illustrates the CRS efficiency frontier while the piece-linear curve illustrates the VRS frontier. Country A is the most efficiency country under CRS, while countries A, B, C, and D are all on the efficiency frontier under VRS. 9. Once the efficiency frontier is determined, the relative efficiency scores of other countries can be estimated by their distances to this frontier. There are two types of efficiency scores, input efficiency score and output efficiency score. Take observation E for example, it is located below the efficiency frontier, and its output efficiency score is OY/OY’<1, E could have achieved higher output at Y’ instead of Y; its input efficiency score is OX’/OX<1, E could have been achieved with lower input of X’ instead of X. Under CRS, given its linear efficiency frontier, the output and input efficiency scores are equal to each other. Under VRS, however, the value of the two efficiency scores may differ, but the rankings determined by the two are generally consistent. 153 Given the focus on analyzing efficiency to help inform policies to reach the MDGs, output scores are presented. Input scores are generally consistent qualitatively with output scores. As shown in Annex 1, results based on input efficiency score largely provide the same ranking for the Philippines in comparison with the peer group. In addition, as explained in Annex 1, the input and output efficiency scores are equivalent in within-country comparisons of CRS efficiency. 154 The two types of estimates generally show qualitatively consistent results, although VRS tends to recognize more observations as located on the frontier. 160 Annex 1 The Efficiency Frontier Source: Adapted from U. Lachler (2005) 10. The linear programming used in calculating the efficiency scores (1/θ) is described in (1) to (3) below. 11. In the efficiency analyses conducted in Chapter 2, within country comparisons are based on CRS efficiency score and thus the input and output efficiency scores are equivalent. Cross-country comparisons are based on VRS. Given the focus of the analysis on what sector indicators can be achieved under the current level of spending, output efficiency scores are presented. The corresponding input efficiency scores are also analyzed and found to provide consistent ranking with the output efficiency score. Annex 4. Data for Efficiency Analysis155 (Chapter 3) 161 Annex 1 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Data for Efficiency Analysis155 (Chapter 2) 12. Data used in the analysis are mainly drawn from two sources. The cross-country data used for peer comparison comes from cross-country datasets to ensure consistency in comparison. These include the World Bank’s World Development Indicators database, the Edustat database, other Bank cross-country studies, and the World Health Organization’s cross-country database. The within country (regional) data are mostly obtained from the authorities of the Philippines during the PER mission. These include education and health indicators, spending by region by national government (NG) and local government units (LGUs) on education and health, and NG spending on transport in 2004-2006.156 The input and output indicators used in the DEA for each sector are summarized in the Table below. Summary of Input and Output Indicators used in DEA (Chapter 2) 13. Data availability limits the scope of the analysis. For example, actual spending data are not readily available and obligation data have to be used instead; functional and economic classifications are not consistently followed in reporting NG and LGU spending; the Commission on Audit (COA) follows different reporting standard from the Department of Budget Management and publishes actual spending The annex is adapted from Coelli (1996) and Lachler (2005). 155 However, data are not available for all indicators for all years. Also, some earlier data are used in the analysis of the basic 156 education and transport sectors. 162 Annexes data in a format that is not easily usable. Also, while it would be desirable to analyze the role of private schools and private health facilities in assessing sector efficiency, the required data are not available and good data collected in the past stopped being collected. 14. In particular, FIES 1999 included information on enrollment levels in private schools, but such useful information is no longer collected in subsequent surveys. Although many education and health sector intermediate output indicators are published in Country in Figures by the National Coordination Board for Statistics, the relevant spending input indicators and some output and outcome indicators (such as infant and maternal mortality rates by region157) are not. Similarly, infrastructure spending and outcome indicators by province or region are not available. Furthermore, there is a lack of coordinated reporting of public spending in similar areas (e.g., education, health, and infrastructure) by NG and LGUs. The absence of these data hampers timely efficiency analysis to inform planning and budgeting of public spending as well as transparent monitoring and evaluation. The team was able to obtain such data through the Department of Health. 157 163 Philippines Annex 1 Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Efficiency Scores for Public Spending on Education and Health in East Asia & Pacific (Chapter 2) Source: World Bank staff calculations, based on data for 1996-2002. Note: These figures represent single input/single output measures based on the Data Envelopment Analysis model with variable returns to scale, and range between 0 (least efficient) and 1 (most efficient). 164 Annex 1 Cross-Country Efficiency Comparison Based on Input Efficiency Scores 1/ (Chapter 2) Education Sector Efficiency (Chapter 2) Health Sector Efficiency (Chapter 2) 165 Annex 1 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Transport Sector Efficiency (Chapter 2) 1 VRS input efficiency score based on the same data used in calculating the VRS output efficiency score. Note the same ranking of the Philippines has been preserved as in the comparison using output efficiency scores. 166 Annex 1 Annex 6. Additional Tables and Figures for Chapter 3 Education Spending Education by Region Estimates Spending Efficiency Efficiency Estimates Region 1/ by2) 1/ (Chapter Source: 2010 Education PER and staff calculations. 167 Annex 1 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Health Spending, Indicators, and Efficiency Estimates by Region 1/2/ (Chapter 2) Source:Department of Health Field Health Information System 2006; Department of Finance; and staff calculations. 1/ DEA results based on two outputs of infant and maternal mortality rates and one input of health spending. Results are presented for 2006, but the inclusion of lagged spending in 2004 and 2005 leads to similar results. 2/ Output data on infant mortality rate and maternal mortality rate by region are available only for 2006, while input data on health spending data analyzed covered 2004-06 and refer to the sum of DOH spending on health by region and LGU spending on health, nutrition & population control as compiled by the Department of Finance. Data are not available for National Capital Region or for ARMM to allow DEA analysis. 168 Annex 1 Average Education Sector Spending, Outcome Indicators, and Efficiency Estimates by Region, 2004-07 1/ (Chapter 2) 1/Public spending per student at primary and secondary level are standardized as a percent of the highest spending to simply the presentation; 2/ The efficiency score of achieving primary and secondary enrollment at a given level of spending; 3/ The efficiency score of achieving primary and secondary completion at a given level of spending. 169 Annex 1 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Average Health Sector Efficiency by Region, 2004-06 (Chapter 2) Sources: DOH, FIES 2006, and staff estimates. 1/ Standardized as percent of the highest per capital spending region of CAR to simplify the presentation. 2/ CRS output efficiency scores using per capita health spending in 2005-06 as input and infant mortality and maternal mortality in 2006 as outputs and controlling for private health spending using estimates based on FIES 2006. 170 Annex 2 Annex 2 (For Chapter 3: Decentralization and Regional Inequality) Intergovernmental Fiscal Transfers in Indonesia158 (Chapter 3) 1. Indonesia’s fiscal decentralization started 2001; from the beginning it included an important revenue equalization component, the DAU. The DAU, which accounts for about 45 percent of sub- national revenues, employs a formula-based allocation mechanism. The overall DAU pool at the national level is calculated as a share (currently 26 percent) of net national revenues (i.e., net of shared revenues). The DAU formula159 has two components, the ‘basic allocation’ (BA) component and the ‘fiscal gap’ (FG) component. The BA component consists of a lump sum and a civil service wage bill part. The fiscal gap is calculated as the difference between fiscal capacity (FC) and expenditure needs (EN). The FG component of DAU is allocated to the districts pro rata of their fiscal gaps. It is the main driver of equalization. However, the importance of the fiscal gap formula in the distribution mechanism is only partial as only 50 percent of the total DAU pool is distributed using the fiscal gap formula. 2. In transitioning towards increased revenue equalization, Indonesia introduced a “hold harmless” provision which prevents LGUs from losing out in the short-term. The “hold harmless” provision limits the extent of fiscal equalization through the DAU allocation as it stipulates that the regions will not receive fewer transfers than in the previous year. It was introduced in the first year of decentralization when the FC component accounted for only 18.5 percent of DAU and did not include the natural resource revenue as part of the fiscal capacity component. Today, this provision favors resource-rich districts, but by law it was phased out in FY2008. 3. The second-largest transfers to sub-national governments are shared revenues, including shared taxes and revenues from natural resources. In 2004, shared revenue amounted to some 20 percent of sub-national budgets. While shared taxes represent about two-thirds of these transfers, revenues from natural resources are highly concentrated in a small number of oil producing regions, making them one of the main beneficiaries of decentralization. In 2006, only 62 out of 440 kabupaten/kota and only five out of 33 provinces are oil and gas producing regions and therefore receive oil and gas shared revenues. Most of these kabupaten/kota are located in the Rio archipelago and East Kalimantan. Thanks to a provision in Law 33/2004, in 2009, the regions received an even higher share (an additional 0.5 percent) of oil and gas revenues; the increase allocation is earmarked for the basic education budget of these LGUs. Oil and gas producing regions have benefited significantly from decentralization, as they now receive 15.5 percent of oil and gas revenues. 4. The poorer and politically more unstable provinces, particularly Aceh and Papua, have been among the main beneficiaries of fiscal transfers. Both provinces have also been granted “special autonomy status” which also conferred additional resources. Starting in 2002, Aceh and Papua received a higher share of oil and gas revenues. In addition, Papua received a substantial Special Autonomy Fund (SPF) representing two percent of the national DAU pool. Since 2008, Aceh also received a SPF. 158 Adapted from Chapter 7 of World Bank (1997) Spending for Development: Making the Most of Indonesia’s New Opportunities, Indonesia Public Expenditure Review 2007, Washington DC. 159 The DAU is allocated according to the following formula (law 33/2004): (1) DAUi = BAi + FGi, where (2) FGi = ENi – FCi; (3) ENi = [0.3*Population Indexi +0.1* 1/HDIi + 0.15*Area Indexi + 0.3*Cost Indexi + 0.15*GRDP per capita Indexi] * Average expenditure of sub-national government; and (4) FCi = OSRi + STXi + SDAi. STX = Shared Tax Revenue, SDA = Shared Natural Resource Revenue, HDI = Human Development Index, OSR = Own Source Revenue). The subscript i refers to LGU. 171 Annex 2 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Additional Figures and Data on Regional Inequality (Chapter 3) Birth Deliveries in a Health Facility by Region Poverty and Regional Contribution to National GDP in the Philippines Source: World Bank (2010); estimates based on FIES 2006; Regional GDP Source: World Bank (2010). Data from NDHS 2008. data are from NSCB Density of Local Roads per Region Source: Countryside in Figures and Philippine Statistical Yearbook (various years). 172 Annex 2 Access to Basic Services, 2006 Per Household Spending Source: World Bank (2010) and 2006 FIES; NTC; DPWH. Source: FIES 2006 for out-pocket household spending; NSCB for real 1/ Number of telephone lines installed per 100 persons. GRDP data Age-specific Education Deficits, by Regions in Luzon Age-specific Education Deficits, by Regions in Mindanao Source: World Bank (2010) Source: World Bank (2010) 173 Annex 2 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Convergence Methodology (Chapter 3) 5. The convergence methodology applied in the analysis of regional equity across the Philippines relies on two measures of convergence. The first measure, which is the σ-convergence, makes use of a given measure of dispersion to analyze the trend of convergence among the regions. In this case, we use the standard deviation of the log of the variable measured (in this case: average daily basic pay, poverty incidence, access to safe water and access to sanitary toilet). We say that there is σ-convergence if the resulting value declines systematically over a certain period and conversely, that there is σ-divergence if the dispersion increases. 6. The second measure is the conditional β-convergence which is based on a regression analysis of growth of the variable being measured as a function of its initial level. We base our estimation on the regression form below: log (Yi, t+T/Yi, t)/T = α – β log (Yi, t) + ε i, t where, i refers to a region observed between period t and t+T and β refers to the coefficient estimate of the independent variable. If β is negative and statistically significant, then there is evidence that the indicator is converging across regions. Average Daily Basic Pay 7. The data from 2001 to 2008 is directly taken from the Bureau of Labor and Employment Statistics. The figures were adjusted to real levels using 2001-2008 Consumer Price Index for all income households by majority commodity group. In calculating for the σ-convergence, we first take the log of the average daily basic pay each year starting from 2001 to 2008. After this, the standard deviation across regions for each year is then calculated. In estimating the β-convergence, we compute for the log of the initial average daily basic pay from the data in 2001 and the log of the average growth per year from 2001 to 2008. We then use Ordinary Least Squares to run a regression analysis. Poverty Incidence 8. The data from 1988 to 2006 is taken from Official Poverty Statistics of the Philippines. For social indicators such as poverty incidence, the β-convergence analysis transforms the relevant social indicators accordingly into their positive form first. In this case, we use the non-poor proportion of the population. Thus, instead of using the poverty incidence ratio, we calculate for the non-poor incidence ratio (1-poverty incidence ratio). We then proceed using the same protocol performed for average daily basic pay. Access to Infrastructure 9. The data from 1998 to 2006 is directly taken from the Philippine Countryside in Figures. In calculating for the σ-convergence, we first take the log of the number of households with access to infrastructure (i.e. safe water supply and sanitary toilet) each year starting from 1998 to 2006. After this, the standard deviation across regions for each year is then calculated. 10. In calculating for the β-convergence, we compute the log of the initial number of households from the data in 1998 and the log of the average growth of the number of households with infrastructure. We also use Ordinary Least Squares to run the regression analysis. 174 Annex 3 Annex 3 (For Chapter 4: Equity of Government Spending on Education and Health) Methodology and Data Sources for Benefit Incidence Analysis (Chapter 4) 1. Benefit incidence analysis combines information on the household utilization of government services with information on the cost of providing said services to assess the incidence of the benefits from government spending across income groups. BIA basically involves three steps: (i) array individuals or households by per capita income (or expenditures) and group by deciles or percentiles; (ii) estimate the unit subsidy of providing a particular type of government service as derived from official data on government spending; (iii) identify users of the government service (based on data of individual/ household service utilization) and impute unit subsidy to said households or individuals (Demery 2000). 2. Benefit incidence thus depends on the household/ individual behavior on the use of the government service and the composition of government spending. Benefit incidence studies generally assume that the value to consumers of a public service is equal to the cost of providing it. They then assign benefits to the users of the service ranked in accordance to their per capita income or per capita consumption. 3. In doing benefit incidence analysis, the redistributive impact of government spending is assessed with reference to the benefit concentration curve, the graphical representation of the distribution of benefits. The benefit concentration curve is plots the cumulative distribution of the “benefits” of public spending on the y-axis against the cumulative distribution of population sorted by their per capita income on the x-axis. The progressivity or regressivity of a public spending is then assessed by comparing the benefit concentration curve with the 45-degree diagonal. The 45-degree line or the diagonal indicates neutrality in the distribution of benefits. If the distribution of benefits lies along this line, the poorest 10 percent of the population gets 10 percent of the government spending; poorest 20 percent account for 20 percent of the subsidy; and so on. Thus, the 45-degree line is also referred to as the line of perfect equality (PE) line. Lorenz and Concentration Curves 175 Annex 3 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth 4. The distribution of benefits of public spending is said to be progressive if the benefit concentration curve lies above the diagonal. In this case, then the poorest 10 percent of the population receives more than 10 percent of the benefits, the poorest 20 percent of the population receive more than 20 percent of the benefits, and so on. Conversely, if the benefit concentration curve lies below the diagonal then the distribution of benefits is said to be regressive. In this case, the poorest 10 percent of the population captures less than 10 percent of the benefits, the poorest 20 percent of the population gets less than 20 percent of the benefits, and so on. 5. The concentration coefficient, also called the Suits index, is the most common summary measure of benefit incidence. It is estimated in like manner as the Gini coefficient160 but it is based on concentration curve instead of the Lorenz curve. While the Gini coefficient is computed as the ratio of the area between the diagonal and the Lorenz curve (represented by A in figure below) to the total area below the diagonal (i.e., triangle cde or Area B), the concentration coefficient is the ratio of the area bounded by the diagonal and the concentration curve to the total area below the diagonal. Suits Index and Gini Measure of Inequality Gini Coefficient (Suits index) = Area of A/ Area of Triangle cde, but area of triangle cde = 0.5. 160 Thus, Gini Coefficient (Suits index) = 2A ( ) 1 where Area of A = — – 1 Ci + (1/N) CN , 2 — N S CN = 1, and N is the number of equal divisions 176 Annex 3 6. If the distribution of benefits is progressive, the Suits index is negative. Conversely, if the distribution of benefits is regressive, then the Suits index is positive. Thus, the Suits index varies from -1 to +1. 7. It should be emphasized that the Suits index is only sensitive to the relative magnitude of subsidies across income groups and not to the absolute amount of the subsidy. However, if the unit subsidy varies across regions and if the regions are not homogeneous, it is important that the regional level analysis is done first and then aggregated to arrive at the national level distribution and estimates of the Suits index. 8. The expenditure of both the central governments and LGUs are included in the following analysis. Because of the large variation in LGU fiscal capacity, LGU spending on health and education tends to exhibit large disparities across regions. This situation tends to have an impact on the equity of public expenditures on education and health. Data sources 9. National government spending data on an obligation basis was obtained from the Statement of Appropriations, Allotments, Obligations and Balances (SAAOBs) of the Department of Education (DepEd) and the Department of Health and from the Budget of Expenditures and Sources of Financing (DBM various years). On the other hand, the expenditures of local government units (LGUs) came from the Annual Financial Report (AFR) for local government units of the Commission on Audit (COA) and the Statement of Income and Expenditures (SIE) of the Bureau of Local Government Finance (BLGF). 10. Enrolment data at the elementary and secondary level is available from the Department of Education’s Basic Education Information System (BEIS). Aggregate enrolment for the different regions from BEIS was then distributed across income deciles in accordance with the distribution of number of students in public schools in the 1999 Annual Poverty Indicators Survey (APIS).161 On the other hand, the household utilization of the various types of public health services/ facilities is based on the 2003 National Demographic and Health Survey (NDHS).162 More recent updates of the APIS do not distinguish between public and private school enrollment. 161 Although the 2008 NDHS has recently become available, service utilization information from this survey is not enough to do 162 BIA. 177 Annex 3 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Additional Tables and Figures for Benefit Incidence Analysis (Chapter 4) Reasons for Leaving School 178 Annexe 3 Reasons for Leaving School ... continued Provider of Antenatal Care, 2003-2008 Source: 2003 and 2004 NDHS 179 Annex 3 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Assistance During Delivery, 2003-2008 (Chapter 4) Source: 2003 and 2004 NDHS 180 Annex 3 Incidence of NG Spending on Elementary Education (Chapter 4) ... and Secondary Education 181 Annex 3 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Incidence of LGU Spending on Elementary Education (Chapter 4) ... and Secondary Education 182 Annex 4 Annex 4 (For Chapter 5: Sufficiency, Efficiency and Equity of Public Revenues) Tax Growth Decomposition Methodology (Chapter 5) 1. The decomposition of the growth of tax revenues into the contribution of policy changes, growth in natural base, and improvements or changes in tax administration is straightforward. The method involves determining the growth of the natural base and the incremental revenues due to policy changes. These two are then subtracted from the growth in tax collection to determine residually the contribution of tax administration. 2. To derive the natural growth of the tax base, we compute tax elasticity of the major tax types. The tax base for personal income taxes of self-employed and professionals, and CE is the taxable income of households and is derived from the Family Income and Expenditure Survey. The tax base for corporate income taxes is equal to the net operating surplus of public and private corporations and is taken from the national income accounts. The tax base for the value-added tax is equal to the VATable sectors of the economy and is taken from the national accounts. The input-output table is used to determine the VATable inputs and outputs per sector. Growth rates for each of the four tax types are computed on an annual basis for 2000-08. 3. The incremental collections due to policy changes for PIT, CIT, and VAT are derived from Department of Finance estimates of the impact of policy changes on tax revenues. 4. Caveat: The accuracy of these estimates is heavily depended on the availability and quality of data. We recognize that weaknesses in the source data, in particular the national income accounts and the households surveys, may bias our estimates. 183 Annex 4 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Tax Gap Estimation Methodology (Chapter 5) Personal income tax calculation of self-employed and professionals 5. The basic data is the Family Income and Expenditure Survey (FIES) for the years 2000, 2003, and 2006. We impute income for years in between surveys by using historical averages. Households that reported ‘own account’ as their main employment types are considered.163 Weights are used to arrive at population income. 6. To compute potential taxable income, we use reported income from “own accounts”, and subtract deductibles such as 13th month pay not exceeding P30,000, contributions to pension and other mandatory government institutions. Personal and additional exemptions are then applied to each household based on their family characteristics. 7. Since approximating deductions for cost of doing business or practice of profession is very difficult, an average presumptive tax rate of 5 percent on gross income is used instead to arrive at potential tax liability per household. This is equivalent to an effective tax rate of 10 percent if the cost-revenue ratio is 50 percent (applicable mostly to professionals) and 25 percent if the cost-revenue ratio is 80 percent (applicable mostly to businesses). A presumptive tax of 2 percent was considered but proved to be too generous for most self-employed and professionals except the retail trade sector. The potential tax liability per household is summed up to arrive at total potential tax liability. 8. To arrive at the tax gap, actual tax collection is subtracted from potential tax liability. Personal income tax calculation of compensation earners 9. The basic data is the Family Income and Expenditure Survey (FIES) for the years 2000, 2003, and 2006. We impute income for years in between surveys by using historical averages. Households that reported ‘wages and salaries” as their main employment types are considered.164 Weights are used to arrive at population income. 10. To compute potential taxable income, we use reported income from “wages and salaries”, and subtract deductibles such as 13th month pay not exceeding P30,000, contributions to pension and other mandatory government institutions. Personal and additional exemptions are then applied to each household based on their family characteristics. 11. To arrive at potential tax liability per household, we apply the statutory schedule of rates to the potential taxable income. The potential tax liability per household is then summed up vertically to arrive at total potential tax liability. 12. To arrive at the tax gap, actual tax collection is subtracted from potential tax liability. 163 The FIES is believed by experts to underreport income in all deciles and much more so in the highest decile. This further adds to the conservatism of the estimates. 164 The FIES is believed by experts to underreport income in all deciles and much more so in the highest decile. This further adds to the conservatism of the estimates. 184 Annex 4 Corporate income tax calculation 13. The basic data is the national accounts by factor share of institutions for the years 2000 to 2008. There is a change in methodology in 2005 so the series 2000-04 and 2005-08 are not perfectly comparable but should not pose a problem given the conservative method. Both private and public corporations are considered. 14. The computation of potential tax liability is straightforward. Net operating surplus of corporations is used as the potential taxable income. This is net of indirect taxes, compensation, and depreciation. The corporate income tax rate of 32 percent (35 percent in 2006) is applied to the net operating surplus. 15. To arrive at the tax gap, actual tax collection is subtracted from potential tax liability. Value-added tax calculation 16. The basic data is the gross value added of the lowest level sector/industry in the national accounts. The 2000 input output table was used to compute VATable input/output ratios and VAT-exempt ratios of each sector/industry. The ratio of marginal value-added is estimated at about 20 percent of VATable GVA. The VAT rate of 10 or 12 percent is then applied to the net VATable GVA is then. 17. To arrive at the tax gap, actual tax collection is subtracted from potential tax liability due to BIR. 185 Annex 4 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Guatemala Fiscal Pact (Chapter 5) Political System Guatemala is a constitutional democratic republic with a presidential system. The president, who is elected for a four-year term without the possibility of consecutive re-election, appoints his executive branch. The Guatemalan Legislative Branch is made up of a unicameral congress with 158 seats (increased from 113 seats in 2003) where members are elected by popular vote. Objectives The Fiscal Pact proposal, which was launched by President Arzu in 1998, aimed at a multi-sectoral, long-term consensus between the principal political, economic, and social bodies over the future tax regime and fiscal policies in Guatemala. Specifically, the Fiscal Pact had the following objectives: • Raise the tax-to-GDP ratio from under 8% in 1998 to 12% by 2002 • Narrow the fiscal deficit from 2.8% in 1995 to 1.8% in 2000 • Formulate a tax policy based on the taxpayer’s capacity to pay to promote equitable development • Keep public investment above 4% of the GDP per year • Set targets that emphasize the implementation of projects in justice, security, education, and health. • Create a stabilization fund, made up of 20 credit unions to cover revenue shortfalls • Create a semi-autonomous tax and customs administration agency, the Superintendencia de Administracion Tributaria, (SAT)165 • Strengthen the judiciary’s capacity to process cases of tax evasion Process In 1997, the Guatemalan government established a semi-autonomous revenue administration agency, the Superintendencia de Administracion Tributaria (SAT), which became fully operational in 1998. At the same time, President Arzu launched the Fiscal Pact proposal in an effort to raise the tax-to-GDP ratio to 12% by 2000. The Fiscal Pact Preparatory Committee (CPPF) was established to organize a national debate on tax policy166. The UNDP funded technical assistance in the preparation of background studies, and a public information program to publicize the contents of the negotiations and subsequent agreements. A variety of actors were involved in the discussions, including Government, Congress, civil society, the Church and the private sector. In 2000, President Alfonso Portillo of the Guatemalan Republican Front (FRG) took office, while the FRG maintained its majority (63 seats) in Congress. Portillo, who was committed to supporting the liberalization of the economy and increasing revenue by stricter enforcement of tax collections rather than increasing taxation, declared the pact a priority. In May 2000 the pact was reviewed and approved by both the legislative and executive branches of government. Outcomes The Pact was the end product of discussions that aimed to address Guatemala’s historically weak revenue performance. The pact also recognized the large socio-economic gaps existent in Guatemalan society, and the urgent need to reduce them by increasing public spending on education, health, and infrastructure. It declared that “the state will ensure the consolidation of a socioeconomic order founded on principles of social justice.”167 It represented a consensus viewpoint of leaders from all sectors of Guatemalan society that “fiscal balance will be primarily the result of an increase in tax collection, not a decrease in social spending.”168. The achievements of the Fiscal Pact include the following outcomes: • A reduction in VAT tax evasion from 40% in 2000 to 32% in 2006 • An increase of social expenditures to 5.7% of GDP by 2006, almost double the level of 1996. • The fiscal deficit declined from 2.8% of GDP in 1995 to 1.6% of GDP in 2008 • Increase of tax revenues as a percent of GDP from 8% in 1990-1998, to 10% in 1999-2003 • Tax revenue per employee increased from $8.4 million in 2003 to $9.6 million in 2006. • The creation of the Fiscal Pact Tracking Commission for monitoring and evaluation • The incorporation of civil society into the decision making processes of government 165 Regional Surveys of the World: South America, Central America, and the Caribbean, Volume 2004 By Europa Publications Limited 166 South America, Central America and the Caribbean 2003 By Europa Publications, Eur 167 “Fiscal Pact, A Major Achievement. Tax Reform, Next in Line.” www.envio.org. retrieved May 16, 2010 168 “Fiscal Pact, A Major Achievement. Tax Reform, Next in Line.” www.envio.org. retrieved May 16, 2010 186 Annex 4 Colombia Medium-Term Fiscal Framework (Chapter 5) Political System Colombia is a democratic country with a presidential system and a bicameral legislature. The president was elected for a four-year term without the possibility of re-election until 2005 when the constitution was amended and (s)he was permitted to run for consecutive re-elections. The Legislative Branch is made up of a bicameral congress with 161 members of the House of Representatives and 102 members of the senate. Colombia is a multi-party system with 3 major parties: the Social Party of National Unity, the Colombian Conservative Party, and the Colombian Liberal Party. Process & Objectives In 2002, President Uribe was elected as an independent liberal, while the liberal party took the majority of the legislative branch. The Uribe administration began a national development campaign that included correcting misalignments in public finances, including the modernization of the National Directorate of Taxes and Customs (DIAN) in order to increase administrative efficiency and reduce administrative costs. In collaboration with the World Bank and other development agencies, an integrated management and services platform called Unified Single Model for Revenues, Services and Control (MUISCA) was designed, and a new Fiscal Responsibility Law (FRL) was drafted and submitted to Congress. The law, which aimed at setting fiscal targets linked to debt sustainability, promoting transparency between the executive and legislative branches, and civil society, as well as including fiscal impact and source of financing within any law that creates new tax expenditures, was approved and enacted in 2003. The law promoted transparency by mandating that: 1) Any law relating to expenditure must have a technical analysis 2) the SIIF (Sistema de Administracion Financiera) is open to the public 3) SUIPF (Public Investment System) is open to the public, including pre evaluation and follow up of any public investment. Furthermore, the FRL allowed the establishment of the Medium-Term Fiscal Framework (MTFF). The MTFF is prepared jointly by the MoF and the National Planning Directorate (NDP), which represents the sectors. In 2005, a subcomponent of the MTFF, the Medium Term Expenditure Framework (MTEF) was established. The MTEF provides expenditure targets for all the ministries for each of the next four years. Ministries are held accountable and must inform progress in investment, programs and policies every year. In addition, there is a Government Plan that establishes sectoral goals for the presidential term. The preparation of the sectoral plan is the result of a collective process in which different sectors, such as education, unions, universities, NGOs, and political parties, participate. While the NPD is in charge of investment expenditure and the MoF is in charge of only current expenditures, the annual budget must be agreed by both entities before approval by Congress. Furthermore, The Superior Council of Fiscal Policy (CONFIS), an inter-ministerial council chaired by the finance minister, coordinates and manages the fiscal process between the central government and the public sector. The main objective of the MTFF is to reach a net debt level of 25 % of GDP by 2018 from 52% of GDP in 2003. In order to achieve this, a surplus of 2.5 percent is considered necessary, with GDP growth of 4.5 percent. Outcomes • Tax revenues increased from 13.5% of GDP in 2002 to 17% of GDP in 2007 • The administrative cost of DIAN was reduced from COL$10.3 per each COL$1000 collected in 2003, to COL$ 9.3 in 2005 • Fiscal deficit declined from 3.7% of GDP in 2002 to 2.5% of GDP in 2004 • Public expenditure on education rose from 2.4% of GDP in 1991 to 4.7% of GDP in 2006 • Sub-national debt dropped from 5.5% (of GDP) in 2000 to 4.1% in 2004, and the external debt from 3.5% to 1.9%. • The VAT base was expanded from 51% of GDP in 2003 to 54.3% • The publication of the MTFF established the basis for a more coordinated, efficient, transparent, and responsible fiscal performance by the national and sub-national governments. • The MTFF has been widely distributed and used by civil society, and international organizations. 187 Annex 4 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Additional Tables and Figures for Chapter 5 Summary of revenue measures of the 14th Congress Source: DOF 188 Annex 4 Income tax burden by percentile using reported Income tax burden by percentile of male-headed income and reported income taxes paid (in percent) households using reported income (in percent) Source: FIES 2006 CE = compensation earner Source: FIES 2006 SEP = self-employed and professional Income tax burden by percentile of female-headed Income tax burden differential in percent households using reported income (in percent) (male-headed minus female-headed) Source: FIES 2006 Source: FIES 2006 CE = compensation earner CE = compensation earner SEP = self-employed and professional SEP = self-employed and professional 189 Annex 4 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth VAT burden by percentile of male-headed households VAT burden by percentile of female-headed households (in percent) (in percent) Source: FIES 2006 Source: FIES 2006 Case 1 does not distinguish where households buy their goods/services. Case 1 does not distinguish where households buy their goods/services. Case 2 distinguishes where households buy their goods/services Case 2 distinguishes where households buy their goods/services VAT burden differential in percent (male-headed minus female-headed) Source: FIES 2006 190 Annex 4 Excise tax burden of tobacco and alcohol products by Excise tax burden of tobacco and alcohol products by decile (in percent), male-headed household decile (in percent), female-headed household Source: FIES 2006 Source: FIES 2006 Excise tax burden of gasoline by decile (in percent), Excise tax burden of gasoline by decile (in percent), male-headed household female-headed household Source: FIES 2006 Source: FIES 2006 191 Annex 4 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Total excise tax button by decile (in percent), Total excise tax button by decile (in percent), male headed households female headed households Source: FIES 2006 Source: FIES 2006 192 Annex 5 Annex 5 (For Chapter 6: Medium Macroeconomic Outlook) Macroeconomic and Fiscal Framework: Baseline Scenario (Chapter 6) Source: GOP for historical, World Bank for projections. 193 Annex 5 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Baseline Scenario Fiscal Framework 1/ (Chapter 6) Source: World Bank staff calculations 1/ Government definition Baseline Scenario Expenditure Framework 1/ (Chapter 6) Source: World Bank staff calculations 1/ Government definition 194 Annex 5 2. Macroeconomic Annex Macroeconomic and and Fiscal Fiscal Framework: Framework: Reform Reform Scenario Scenario (Chapter 6) Source: GOP for historical, World Bank for projections. 195 Annex 5 Philippines Public Expenditure Review: Strengthening Public Finance for More Inclusive Growth Reform Scenario Fiscal Framework 1/ (Chapter 6) 1/ Government definition Reform Scenario Expenditure Framework 1/ (Chapter 6) 1/ Government definition 196 1818 H Street, NW Washington, DC20043 USA Internet: www.worldbank.org World Bank Ofce Manila 23rd Floor, The Taipan Place F. Ortigas Jr. Road, Ortigas Center Pasig City, Philippines Telephone: (63-2) 637-5855 Internet: www.worldbank.org/ph/cas International Finance Corporation 11th Floor, Tower One Ayala Triangle, Ayala Avenue Makati City, Philippines Telephone: (63-2) 848-7333 Internet: www.ifc.org