Report No. 36630-LK Sri Lanka Selected Public Expenditure Issues, 2003/2004 June 29, 2006 Poverty Reduction and Economic Management Sector Unit South Asia Region Document of the World Bank PREFACE This report is a summary o f the series o fbackgroundpolicy notes prepared under a process-based public expenditure review (PER) program managed by the World Bank and the Ministry of Finance from September 2003 to March 2004. The PER process and the preparation ofthese backgroundnotes involved close collaboration among staff from the World Bank, Ministry o f Finance; Ministry of Health, Nutrition and Welfare; Ministry o f School Education; Ministry o f Samurdhi; Ministry o f Social Welfare; Ministry o f Irrigation and Water Management; Ministry of Public Administration, Management and Reforms; the Institute o f Policy Studies (IPS); and the Center for Poverty Analysis (CEPA). The PER program was intempted due to the transition that followed the change o f Government inApril 2004 and later the tsunami, but has resumed beginninginlate 2005. A key motivation for the publication o fthisreport is to provide a written record of the analytical work carried out under the PER program during 2003/2004, and thereby preserving the institutional memory on this work. As such, the report is intended to provide a `snapshot' o f the state o fknowledge on the issues covered at that time. TABLE OF CONTENTS IIntroduction . ........................................................................................................ 1 PART 1 .CROSS-CUTTING EXPENDITURE ISSUES .............................................. 3 I1 Fiscal policy and debt sustainability . 3 Fiscal sustainability.................................................................................................... . . . ........................................................................... 3 K e y challenges ........................................................................................................... 6 8 I11.Pension Reform priorities........................................................................................................ schemes for publicsector employees ........................................................ 10 Overview o f system ................................................................................................. 10 K e y challenges ......................................................................................................... 11 12 I V Publicsector employmentand pay issues . Emergingpolicy options .......................................................................................... .............................................................. 14 Overview and challenges ......................................................................................... 14 Policy options........................................................................................................... 16 PART 2 SECTOR EXPENDITURE REVIEWS . ........................................................ 19 V Health . .......................................................................................................................... 19 Sector performance .................................................................................................. 19 Health expenditure ................................................................................................... 22 K e y challenges ......................................................................................................... 22 24 VI Education . Policy recommendations .......................................................................................... ................................................................................................................... 26 Education achievements and challenges.................................................................. 26 Public Spending in education................................................................................... . . 29 Challenges in governance and service delivery....................................................... 31 Policy development for the medium-term budget framework ................................. 31 34 VI1 Social welfare . Increasingthe........................................................................................................... quality o f service delivery ................................................................ 36 Overview.................................................................................................................. 36 Policy framework o f the ministries and related issues............................................. 36 Expenditure patterns and trends............................................................................... 38 Reforms inprogress ................................................................................................. 39 VI11 Irrigation .Policy recommendations to support and complement the reform process ..............40 ................................................................................................................ 44 Sector performance.................................................................................................. 44 Irrigation sector expenditure trends ......................................................................... 45 Sector policy and planning....................................................................................... 48 Policy recommendations .......................................................................................... 49 ANNEX ............................................................................................................................ 53 Tables Table 1. Sri Lanka: Summary o f Government Fiscal Operations....................................... 3 Table 2.Baseline Macroeconomic Assumptions ................................................................ 5 Table 3. Sri Lanka: Sources o f Government Financing...................................................... 7 Table 4. Sri Lanka: Pension Expenditures, PSPS and Widows and Orphans Scheme,.... 10 Table 5. Sri Lanka: Evolution o fPensioners. 1998-2003 ................................................. 11 Table 6. Sri Lanka: Number o fAutonomous Agencies and Their Employees ................ 14 16 Table 8. Sri Lanka: Growth inEmployment, Population. Labor Force. and Real GDP. Table 7.Wages and Salaries o f Consolidated Central Government ................................. 1998-2002 .......................................................................................................... 17 Table 9. Education Expenditure as a Share o fNational Income and Government Expenditures. Sri Lanka and Selected Other Countries..................................... 30 Table 10. Sri Lanka: Net Enrolment Rates inMajor Grade Cycles by Economic Groups31 Table 11. Samurdhi Coverage and Benefits by Income Quintile ..................................... 38 Table 12. Total expenditures o f the Ministryo f Samurdhi and Social Welfare ............... 39 Table 13.Irrigation investmentsin Sri Lanka by type o f investmentand their shares. Table 14.1950-97 1/.......................................................................................................... 46 Sri Lanka: Percentage Share o f Major Subsectors inTotal Agricultural Expenditure. 1981-2000.................................................................................................... 46 Figures Figure 1. Sri Lanka: Components o f Government Expenditure. 2000.2004 ...................... 4 Figure 2.Primary Balance and Interest Payments Consistent with Most FMRA Targets .6 Figure3. Sri Lanka: Financing o fthe Fiscal Deficit. 1998-2005 ....................................... 7 Figure4.PSPS expenditures. % o f GDP. 2003-2075 inthe baseline scenario ................ 12 Figure 5. Sri Lanka: Shares o f Total Public Employment. 2002 ...................................... 15 Figure6.Per CapitaExpenditure on HealthinSri Lanka by Province (1997) ................20 Figure7. Sri Lanka: Percentage o f children aged 3 to 59 months who are stuntedbyplace o fresidence ........................................................................................................ 21 Figure 8. Sri Lanka: Health Outcomes by Level o fMothers Schooling .......................... 21 Figure9. Sri Lanka: Poverty Incidenceby Education Level o fHousehold Head. 1995/627 Figure 10.Sri Lanka: Unemployment Rates and Job SearchDurationsby Education Level. Menand Women. 2000........................................................................... 29 Figure 11. Investmentsin Sri Lanka by type o f investment and their shares. 1950-97.... 45 Figure 12.Major Irrigation: Spread of Investments in2003 ............................................ 47 Acknowledgements The report was prepared by Emily Sinnott under the guidance of Rocio Castro and Ijaz Nabi. The task team leader o f the 2003/2004 PER program was Alma Kanani. The original background policy notes were prepared by Mario A. Cuevas (Fiscal and Debt Sustainability), Ismail Radwan (Contingent Liabilities o f State Owned Enterprises), Esperanza Lasagabaster (Pensions), Nick Manning, Robert Beschel and Princess Ventura (Public Sector Employment and Pay Issues), Institute of Policy Studies, Michele Gragnolati (Health), Harsha Aturupane (Education), Center o f Poverty Analysis (Social Welfare) and Nihal Fernando, Ranjith Ratnayake (Irrigation). An annex is included with the Draft National Budget Circular prepared by Malcolm Holmes (advisor on the budget formulation process). The PER process benefited from advice o f peer reviewers: Shanta Devarajan, Allister Moon. I.INTRODUCTION 1. This note contains a summary o f the background policy notes prepared in the context o f the public expenditure review (PER) work carried out for the Government o f Sri Lanka in2003/2004. The PERwork involved World Bank assistance at two levels: (i) the preparation o f background policy notes on selected cross-cutting issues (i.e. fiscal sustainability, pay and employment, and pensions) as well as sector-specific public expenditure issues (education, health, social welfare, and irrigation); and (ii)the design o f revised budget guidelines aimed at enhancing the strategic focus o f the budget formulation process and the elaboration o f the Medium Term Budget Framework for 2005-2007. 2. The main objective o f the background policy notes was to provide the Ministry o f Finance and selected line ministries with background policy analysis as an input to assist the government in improving the budget formulation to better link policies with resource allocation. Policy notes were prepared in a few selected sectors that were deemed most significant for the Government, either as major cost drivers or perceived as being inefficient. The summary given in this paper is based on the final draft policy notes prepared by March 2004. The draft budget circular drawn up by the PER team in March 2004 is also included as an annex. 3. The analysis inthis report concentrates on the major expenditure issues facing the Government of Sri Lanka. The topic o f government revenues i s also, albeit briefly, touched on. The policy notes commence with a look at three overarching issues o f concern for fiscal policy. The first focus i s on the sustainability o f the rapidly rising public debt and persistent fiscal deficits. Next, a potential contingent liability, the public service pension scheme, i s examined. The note looks at the budgetary impact o f the pension scheme and its medium- to long-term financial viability. The final cross-cutting expenditure issue investigated-public sector pay and employment-is a critical subject for Sri Lanka, a country with one o f the highest per capita staffing rates inthe developing world. The analysis reviews the ability o f the administration to deliver services effectively to its citizens, particularly in a setting marked by the proliferation o f various institutions with unclear mandates and overlapping functions. 4. The next part o f the policy notes deals with expenditure issues in four specific sectors. An assessment is given o f the appropriateness o f expenditure patterns and policies to meet remaining and emerging challenges in the health and education sectors. For health, there is a need to consolidate the impressive gains made in health outcomes and respond to nascent trends associated with an aging population, such as a growth in non-communicable diseases, through greater focus on preventive. Ineducation, the issues to be confronted are access beyond the primary level and the achievement o f higher quality educational outcomes. Another area investigated is social welfare spendingby the Ministries o f Samurdhi and Social Welfare. There are some key issues o f benefit incidence and emerging social protection that need to be addressed. The fourth sector turned to is irrigation and water management. In light o f the diminishing amount o f 1 public spending going towards this sector, the budget allocation and water resource management policy is examined. 5. We follow a similar pattern in summarizing the seven issues covered in these policy notes viz. fiscal sustainability, pensions, public sector employment, health, education, social welfare and irrigation. To begin, there i s a brief review o f performance, followed by an investigation o f the key issues and challenges, including any relevant equityhegional considerations. Next, expenditure trends in the major program areas are covered, including analysis on the consistency o f spending allocations with the stated objectives o f the sector. Finally, a range o fpolicy proposals are presented for each sector. 2 PART 1. CROSS-CUTTINGEXPENDITUREISSUES 11. FISCAL POLICYAND DEBT SUSTAINABILITY The targets set out in the 2002 Fiscal Management Responsibility Act (FMRA) remain elusive, with the overall fiscal deficit and public debt in 2004 equal to 8.2 percent and 1OS percent of GDP, respectively. Sri Lanka 'spublic debt resultingfrom persistent and large deficits is unsustainable. I n particular, the heavy reliance on domestic (non- concessional) borrowing for deficit financing has been detrimental to growth by crowding-out productive public spending andprivate credit. Given the heavy public debt burden-which absorbs over half of tax revenue-the public debt needs to be brought to manageable levels by eliminating the primary defcit and by limiting non-concessional borrowing. FISCAL SUSTAINABILITY 6. Despite recent fiscal consolidation efforts, including the enactment of the FMRA in 2002, the fiscal situationremains strained. A period of fiscal consolidation occurred during2002-03 when the primary fiscal deficit was cut from 4.1percent o f GDP to 1.2 percent. However, the fiscal situation worsened thereafter as the primary fiscal deficit increased to 2.2 percent o f GDP in 2004. The overall fiscal deficit in 2004 remained at 8.2 percent o f GDP, a similar level to 2003 (Table 1). Table 1. Sri Lanka: Summary of Government FiscalOperations (as a percentage of GDP) 1985 1990 1995 1998 1999 2000 2001 2002 2003 2004* Total Revenue 22.3 21.1 20.4 17.2 17.7 16.8 16.6 16.5 15.7 15.3 Tax Revenue 18.7 19.0 17.8 14.5 15.0 14.5 14.6 14.0 13.2 13.9 Total ExpenditureandNet Lending 34.0 31.0 30.5 26.3 25 2 26.1 27.5 25.4 24.0 23.5 Current Expenditure 20.1 22.3 23.1 19.6 18.7 20.2 21.6 20.9 19.0 19.2 Interest Payments 4.6 6.4 5.7 5.4 5.6 5.7 6.7 7.4 7.1 5.9 Foreign 0.8 0.8 0.7 0.7 0.7 0.7 Domestic .... 4.8 4.9 6.0 6.7 6.4 5.2 Subsidies and Transfers 5.5 6.5 6.1 4.6 4.2 4.2 4.6 4.7 4.0 5.2 Wages and Salaries 4.2 4.9 5.2 5.3 5.3 5.5 5.5 5.6 5.1 5.2 Civilianwages and salaries 3.6 3.9 3.4 3.0 3.0 3.2 3.4 3.4 3.2 3.2 Security .. 4.1 6.5 5.0 4.4 5.6 4.9 4.1 3.5 3.6 CapitalExpenditureandNet Lending 14.0 8.7 7.4 6.7 6.5 6.5 5.9 4.6 S.0 4.3 Overall Fiscal Balance 1/ -11.7 -9.9 -10.1 -9.2 -7.5 -9.9 -10.8 -8.9 -8.3 -8.2 PrimaryFiscal Balance 1/ -7.1 -3.5 -4.4 -3.8 -1.9 -4.2 -4.1 -1.6 -1.2 -2.2 Memorandum Items: Total public debt 80.2 96.6 95.2 90.8 95.1 96.9 103.2 105.6 105.8 105.4 Foreign 49.6 55.0 51.9 45.3 46.0 43.1 45.2 45.6 47.9 49.1 Domestic 30.6 41.6 43.3 45.5 49.1 53.8 58.0 60.0 57.9 56.3 Educationand Health .. 4.5 4.5 4.0 4.0 4.1 3.3 3.9 3.9 Source: CentralBank of Sri LankaAnnual Report (various issues), IMF and World Bank Staff estimates. Notes: *Estimated. I/Excludinggrants 7. Just three components of public sector spending-security related expenditures, civil service wages and salaries, and interest payments on public 3 debt-are equal to total tax receipts (Figure 1). The average combined expenditure on security related outlays, civil service wages and salaries, and interest payments was 14 percent o f GDP over 2000-2004, while taxation proceeds averaged 14.2 percent o f GDP inthis period. Average expenditure on subsidies and transfers accounts for an additional 4.5 percent o f GDP over 2000-2004. There is then little room for other discretionary spending. Ifthe level of spending on these constituents remains unchanged, then the only w a y to reduce the fiscal deficit will be to substantially increase revenues. 40 As a PercentageOfGDp 0 Security relatedeqmditures 35 - Civil service wages and salaries 30 Totalexpenditure and net lending Interest payments + 25 - Current expenditure 20 ............ - ....-...............-_.............._................ 15 10 5 0 2000 2001 2002 2003 2004* Source: Central Bank of Sri Lanka Annual Report(various issues), IMF and World Bank Staff estimates 8. Large fiscal imbalances and high public debt have beset the Sri Lankan economy for the past three decades. Despite repeated attempts at reform, fiscal imbalances persisted and the public debt burden increased. The average fiscal deficit (before grants) in the 1991-2001 period exceeded 9 percent of GDP. While high public debt levels contribute to fiscal difficulties, persistent primary deficits exacerbate the situation. After a decade of running high primary fiscal deficits (around 3-4 percent o f GDP), the primary account moved closer to balance in2002 and 2003, only for a further deterioration in2004. 4 Box 1. The Fiscal Management (Responsibility) Act I In view o f the importance o f strengthening fiscal management, Parliament approved the Fiscal (Management) Responsibility Act (FMRA) on December 10, 2002, effective from fiscal year 2003. The purpose o f the Act is to provide a formal framework for fiscal discipline, and to increase transparency and accountability ingovernment fiscal operations. A key element o fthe FMRA is the introduction o fmedium-termfiscal targets. Specifically, the fiscal deficit has to be reduced to 5 percent o f GDP by 2006 and maintained under that level thereafter. The FMRA also sets ceilings on total government debt, which should not exceed 85 percent o f GDP by 2006, and 60 percent of GDP by 2013. Another important characteristic o f the FMRA i s that it sets a limit on explicit government guarantees o f 4.5 percent o f GDP. As the FMRA is also designed to improve the credibility o f fiscal policy, the Act mandates the government to present a number o f reports, including: (a) a Fiscal Strategy Management report, with the Budget Speech; (b) Mid-Year Fiscal Position Reports; (c) Final Budget Position Report; (d) Pre-Election Budgetary Position Reports, within three weeks o f the announcement o f a general election; and (e) Statements o f Responsibility by the Finance Minister and Secretary, together with a Pre-Election Budgetary Position Report. Several of these reports have been presented by the government. 9. The fiscaVdebt sustainability analysis for Sri Lanka shows that achieving the FMRA mediumand long term targets will be very challenging, especially given the recent increase in fiscal imbalances. The FMRA targets are: (a) fiscal deficit not to exceed 5 percent o f GDP by 2006; (b) total government debt at 85 percent o f GDP by 2006; and (c) government debt not to exceed 60 percent o f GDP by 2013 (see Box 1). The fiscal/debt analysis was conducted using as a basis the estimated fiscal position o f 2004 and projecting onwards the path for the primary deficit that would be required to meet the FMRA targets. The baseline macroeconomic assumptions underlying the analysis are presented in Table 2. 10. Meeting the F M R A targets would require a zero primary balance between 2005-2013, resulting in a public debt to GDP ratio of 60 percent by 2013 (see Figure 2).* However, these projections are based on an estimated primary deficit of 1.3 percent o f GDP in 2004, as envisaged in the pre-election report, as opposed to the currently estimated deficit o f 3.1 percent. This implies that a further adjustment would be needed to remain within the FMRA targets. Table 2. BaselineMacroeconomic Assumptions 2004 2005 2006 2007 2008-2013 GDP growth ("YO) ' 6.0 6.3 6.5 6.3 6.0 Change in GDP deflator (%) 7.0 6.0 5.0 5.0 5.0 Nominal interest rate o n public debt* ("A) 5.9 6.0 6.0 6.0 6.0 External financing (as % o f GDP)** 0.4 2.8 1.8 1.8 1.8 Notes: The real exchange rate i s assumed to appreciate mildly during2004-2006 and to remain constant thereafter. * Average interest rate that applies to total government debt, expressed innominal terms. **Net External financing, including grants. L Under this scenario, public debt is projected at 88 percent o f GDP in2006, somewhat above the FMRA target. 5 11. An early convergence towards FMRA targets would bring substantial fiscal savings due to sharply reduced debt servicing costs. The baseline projections show that meeting the primary balance and debt ratios FMRA targets would result in annual fiscal savings equivalent to 1percent o f GDP by 2006, and 2.6 percent o f GDP by 2012- 2013. Total interest payments on public debt would stabilize at around 3.5 percent o f GDPby2012. Figure2. Primary BalanceandInterest Pa! ientsConsistentwith Most FMRA Targets 2% ,. ..... .-_................... ... 2004 2035 2006 2007 2M8 2009 2010 2011 2012 2013 Primary Balance PublicDebt , 7 f f ~ ~ ........... . . . . . . I0% 00% I 2 m 2W5 2 M 2037 2038 2009 2010 2011 2012 2013 Interest Payments Overall deficit Note: Inthis scenario, the government slightly misses e 2006 FMRA debt/GDP target o f 85 percent. However, the 2006 fiscal deficit target of 5 percent of GDP is met. 12. However, the magnitude of the fiscal adjustment is highly sensitive to changes in key macroeconomic assumptions, such as real GDP growth, the nominal interest rate, and the exchange rate. For instance, the baseline scenario assumes that annual real GDP growth stabilizes at 6 percent in the long-run. But, if real GDP growth remained at its historical level o f 5 percent, ceteris paribus, a primary surplus closer to 0.5 percent of GDPwould be required to reach a debt ratio o f 60 percent in2013. KEY CHALLENGES 13. Reversing the decline in tax revenue. Government revenues have declined over time in Sri Lanka, falling as a share o f GDP from 22.3 percent in 1985 to 20.4 percent in 1995 to 15.3 percent in 2004. At the same time, while there has been a decline in total expenditures since 1995, government spending has remained highrelative to revenues at 23.5 percent in2004. Inparticular, tax revenue losses would seem to be driving the recent 6 deficit growth, with tax revenues having fallen from 19 percent o f GDP in 1990 to 13.9 percent o f GDP in2004. 14. The rapid rise in public debt resulting from persistent and large deficits is unsustainable.Aggregate public debt increased from about 95 percent o f GDP in 1999 to over 105.4 percent in 2004-a rise in the total public debt-GDP ratio o f more than 10 percentage points of GDP in just two years. The government's financing needs were primarily met by issuing domestic debt, with the result that the domestic debt ratio increased by 7 percent of GDP, going from 49 in 1999 to almost 56.3 in2004. While the burden o f external interest payments remained low between 2000 and 2004 (under 1 percent o f GDP), domestic interest payments rose by 2 percent o f GDP between 1999 and 2004 (to 5.2 percent o f GDP). Table 3. Sri Lanka: Sources of Government Financing As a Percentage of GDP Source o f Financing 1985 1990 1995 1998 1999 2000 2001 2002 2003 2004* Foreign Project Loans 4.4 3.8 2.1 1.9 1.0 0.9 Non-Project Loans 0.0 -0.2 1.0 -0.9 -0.8 -0.9 Grants 2.0 2.1 1.4 0.7 0.6 0.4 0.4 0.4 0.5 0.4 Total Foreign 6.4 5.7 4.5 1.7 0.7 0.4 1.0 0.1 2.4 1.8 Domestic Market Bank 4.6 1.5 1.1 1.4 2.4 4.2 3.5 -0.3 -0.9 2.1 Non-Bank 6.5 4.0 3.9 5.2 4.5 4.9 5.3 8.3 5.7 3.6 Total 7.9 5.6 5.1 6.6 6.9 9.2 8.7 8.0 4.5 5.5 Non-Market -1.2 0.1 0.1 -0.1 -0.1 0.3 0.1 -0.1 0.0 0.2 UseofCashBalances -1.5 -1.5 -0.1 0.5 0.0 0.0 Total Domestic 5.2 4.2 5.1 7.0 6.8 9.4 8.8 8.0 4.8 5.8 Privatization 0.0 0.0 0.4 0.4 0.0 0.0 0.6 0.4 0.6 0.1 Total Financing 11.6 9.9 10.1 9.2 7.5 9.9 10.8 8.9 8.3 8.2 Source: Central Bank of Sri Lanka Annual Report (various issues), IMF and World Bank Staffestimates *Estimated. Figure 3. Sri Lanka: Financing of the Fiscal Deficit, 1998-2005 25 10 20 8 15 6 10 4 - 5 2 0 0 1998 1999 2000 2001 2002 2003 2004* 2005** m i Total Domestic Fmancmg(nght scale) Total Foreign Fmancmg(right scale) -Total --- Expenditure andNet Lending(left scale) Total Revenue (left scale) *Estimated, **Projected. 7 15. The heavy reliance on domestic financing of public sector deficits is detrimental to growth. (Table 3 and Figure 3). A significant consequence o f the government's reliance on domestic borrowing to finance the deficit has been the implicit taxation o f public pension funds and the domestic banking system. Sri Lanka finances domestic public debt through two sources. The first are rupee loans issued at below market rates. The use o f this type o f financing has declined from 62 percent (total rupee securities/total domestic debt) in 1997 to 14 percent in 2004. The main supply o f domestic financing i s now pensiodother social security funds, e.g. Employee's Provident Fund (EPF). These funds are required to invest largely in government paper. For example, the EPF is restricted to mostly rupee securities and a small amount o f treasury bills. The result is that an implicit tax is imposed on non-civil service workers. One estimate puts the after tax real rate o f returns over the past three decades for rupee loans at 0.25 percent (IMF,2002).2 REFORM PRIORITIES 16. Reducing public debt to manageable levels. Given the heavy public debt burden,which absorbs over half of tax revenue, it is imperative to bringthe public debt to manageable levels by at least eliminating the primary deficit and by limiting non- concessional borrowing. 17. Reforming wage and recruitment policies. Sri Lanka has one o f the largest bureaucracies in the region, with a ratio o f 3.9 civil servants per 100 people. Although the size o f the (civilian) wage bill i s not unmanageable at present (around 3 percent o f GDP), the trends are worrisome. Between 1990 and 2001, public sector employment grew at 3.6 percent annually, outpacing growth in population and labor force. While keeping the wage bill in check, strong political commitment will be needed to address well-known constraints to public service delivery. These include overstaffing (particularly at the lower grades); excessive salary compression (8: 1); administrative fragmentation, duplication, and wastage (partly exacerbated by the ineffective devolution of functions); and outdated processes and procedures. 18. Rationalizing public spending and linking it to the poverty reduction strategy. The scope for expenditure rationalization is significant, not least because o f the need to address the duplication and overstaffing problems o f the public administration. Inaddition, there is a serious imbalance inpublic spending, with the bulk of the budget being directed to fbnd recurrent costs (i.e., interest payments, wages, and subsidies) and very little to investment. The Ministry o f Finance is developing a medium term budget framework (MTBF) which could potentially combine macro (i.e., attaining fiscal sustainability) and micro objectives (such as enhancing the development impact o f public spending). Following the recent introduction o f budget ceilings and o f a three-year planning horizon, there is a needto strengthen the links between priorities, resources, and outputs/outcomes. This will require a more strategic and consultative budget formulation process and the identification o ftrade-offs within and across sectors. L The implicit tax i s measured as the gap between interest rates o n rupee securities and market-based instruments. 8 19. Improvingthe performance of state-owned enterprises (SOEs). The financial burdeno f SOEsis highand service delivery is poor. Most are overstaffed, incur operating loses, and have large debts (mostly to the state-owned banks). Direct subsidies alone absorb 3 percent of GDP annually, which is greater than the entire education budget. These companies should be substantially restructured (through privatization or other methods), be allowed to operate on a commercial basis with no political interference, and be subject to a hardbudget constraint. 20. Reversing the decline in revenue by strengthening tax administration and expanding the tax base for major taxes (VAT and income tax). The challenge in designing tax policies i s to reverse the massive decline in the tax-to-GDP in a manner that i s supportive o f growth and efficiency. The task i s daunting given the sharp drop in trade taxes (from 6 to 2 percent o f GDP since 1990); the stagnation o f income taxes at about 2 percent o f GDP, which i s very weak by international standards; and the incomplete transition from a complex system o f turnover taxes and special levies to a Value Added Tax (VAT). Increased revenue in the order o f 2-3 percent o f GDP may be achievable through: 0 Strengthening VAT. While the 2004 unification o f the two VAT rates into a single rate o f 15 percent facilitated administration and reduced leakages, its immediate impact was revenue-reducing. Going forward, additional revenue will be possible if the VAT coverage is extended to retail trade and exemptions are reduced. 0 Raisingthe incometax yield. Underlyingreasons for low collection include: the provision o f long term tax holidays for the rapidly growing Board o f Investment (BOI) sector, a long history o f investment allowances, concessionary rates, generous depreciation, and exclusion o f most public sector employees from personal income tax. The scope for substantially raising the income tax yield i s limited inthe short term, but significant gains could be made if a moratorium on tax holidays were to be gradually phased in and existing exclusions and concessions pared down. 0 Improvingtax administration.For several reasons, including the coexistence o f parallel systems (for customs and BO1firms), Sri Lanka's tax administration has limited experience inadministering modem taxes (e.g., VAT and income tax) that require verification, audit and risk monitoring. A well-integrated revenue administration, along the lines o f the originally envisaged Revenue Authority, needs to be established as soon as possible. Complementary measures include the establishment o f a separate tax audit unit and making the large tax payer unit more proactive. 9 111. PENSION SCHEMES FORPUBLICSECTOR EMPLOYEES While the ratio of retirees to active civil servants is high, expenditures on the Public Sewice Pension Scheme (PSPS) are not estimated to grow substantially in future years, following the introduction of a Contributory Pension Fund scheme. A major problem with the PSPS benefits design is that theflow ofpension payments to an individual retiree is heavily skewed towards early years of retirement, in part due to the ad-hoc indexation of benefits at rates below inflation. International experience suggests that, optimally, real benefit value should remain constant over time. However, a full indexation of benejh could be quite costly if other parameters of benefits design are not changed. Another critical issue is the need to increase the investment return on pension assets by diversijjing awayfrom low-return government bonds. OVERVIEW OF SYSTEM 21. There are currently two old-age pension schemes for permanent employees o f government and public enterprises in Sri Lanka. Until 2002, all such employees have been insured through the previous non-contributory pay-as-you-go (PAYG) Public Service Pension Scheme (PSPS). Expenditures o f the PSPS totaled 1.8 percent of the country's GDP (7.1 percent o f the total government expenditure) in 2002. Starting January 1, 2003, a new contributory, funded defined-benefit scheme (Contributory Pension Fund, CPF) was introduced for the new employees o f public sector. The workers whojoined before 2003 remainedinsuredthrough the PSPS. 22. Due to confusion about its legal status, not all new workers have been shifted to the CPF. The law that introduced the CPF, while approved by parliament, was never signed. This creates some uncertainty regarding the pension rights o f new recruits. This note presents model-based analysis o f the PSPS and the CPF-assuming that all entrants into the civil service after November 2003 are transferred to the new scheme-and suggests policy options regarding future development o f the pension schemes. Table 4. Sri Lanka: PensionExpenditures,PSPS and Widows and OrphansScheme, 1998-2002,Millions of Rupees Year Expenditures on Expenditureson Contributionsfor Total Pension civil pensions* widows and widows and expenditures expendituresas orphans orphans percentof GDP 1998 15,151 4,277 1,816 19,428 1.9 1999 15,199 3,857 2,284 19,056 1.7 2000 17,592 4,360 3,045 21,952 1/ 1.7 2001 20,217 5,63 1 2,655 25,847 1.8 2002 25,728 5,094 2,541 30,823 2 Source; Departmento f Pensions, Ministry ofPublic Administration Notes. Includesgratuities * 1/ Basedon the dataprovidedby the PensionsDepartment to the World Bank in September 2003. Accordingto mother source, Departmentof PensionsAnnual PerformanceReportand Accounts, 2002, total expenditures in2000 amountedto SLR20336.5 million. 10 23. Since 1947, the PSPS has constituted the primary pension program for civil servants, armed services, provincial and local govemment employees, and teachers. The PSPS is a non-contributory scheme directly financed from the public budget. Together with the contributory scheme for widows and orphans, its expenditures amounted to about 2 percent of GDP in 2002 (Table 4). 24. The ratio of retirees to active civil servants (dependency ratio) i s high. As o f 2002, this program covered an estimated 506,000 govemment employees and 389,000 pensioners (Table 5). Minimumretirement age i s 55 years, and mandatory retirement age i s 60 years. Even though the effective retirement age i s close to 60 years, it i s low compared to countries with similar life expectancy. The current demographic profile o f the insuredworkers is skewedtowards younger and middle-age cohorts: about 90 percent o f them are younger than age 50. The program provides for a lifetime unreduced pension representing between 85-90 percent o f the last salary at the time o f retirement for 30 years o f service. Alternatively, a retiree could receive a pension equal to 75-85 percent o f the last salary, increased by 5-10 percentage points o f the last salary after 10 years o f retirement and complemented by commuted gratuity-a lump-sum payment at retirement, worth 24 months o f initial pension. Almost all retirees chose the second option. The benefit diminishes by two percentage points for each year short o f 30 years o f service. Table 5. Sri Lanka: Evolution of Pensioners, 1998-2003 Year Number o f Numberof Total Rate of growth civil widows and number of o f total pensioners orphans pensioners pensioners, % 1998 269,645 88,583 358,228 1.7 1999 274,345 90,127 364,472 2.0 2000 279,802 91,920 371,722 3.0 2001 288,276 94,704 382,980 1.8 2002 297,041* 97,584* 394,625 1.3 Source: Department of Pensions,Ministry of Public Administration * Estimate KEY CHALLENGES 25. The expenditures of the PSPS are estimated to grow as a percentage of GDP over the next two decades, peaking at about 2.5 percent around 2025. The driving force behind this growth will be the increasing size o f the cohorts retiring from Sri Lanka's public sector. This will occur because o f rapid aging o f the public sector current and former labor force. After reaching the peak, the liabilities will decline as the scheme is being gradually phased out. Figure4 gives the estimated stream o f PSPS expenditures over 2003-2075 in baseline simulations done in a background paper for this note. The baseline scenario assumes 3 percent annual growth in wages, 3 percent annual inflation and indexationby 20 percent of the inflation rate. 26. Liabilities of the PSPS could be reduced if the retirement age is increased. However, the retirement age increase should be gradual, because o f another government's objective-elimination o f surplus labor in the public sector through the 11 Voluntary Retrenchment Scheme (VRS). Since the increase inretirement age could act in the opposite direction, the timeframe for implementation o f this measure should allow for the VRS to take hold. Figure 4. PSPS expenditures, % of GDP, 2003-2075 in the baseline scenario 3.0% - 2003 2013 2023 2033 2043 2053 2063 2073 27. The major problemwith the PSPS benefitsdesign is that the flow of pension payments to an individual retiree is heavily skewed towards early years of retirement. This is due to two factors. The first one is commuted gratuities paid at retirement. The second one is an ad-hoc indexation of benefits, on average below the inflation level, leading to a substantial fall in the real value o f an individual's pension over time. 28. Indexation of benefits could be quite costly if other parameters of benefits design are not changed. For full indexation by inflation to be feasible without increase inthe scheme's outlays, the average replacement rate at retirement for the retirees opting for commuted gratuity should be reduced from about 80 percent to 65-70 percent if the commuted gratuities are to be paid at their current levels, to 70-75 percent ifthey are cut inhalf, andto 75-79 percent ifthey are abolished. EMERGING POLICYOPTIONS 29. Establishment of the CPF has been a positive development for at least two reasons: introduction of employee contributionsand some reductionof the benefits levels comparedto the PSPS. These allow for gradual reduction o f the financial burden imposedby the pension system on the government budget. 30. There are three areas for improvementinthe CPF design: (a) like any fully funded defined-benefit arrangement, it is exposed to a large number o f risks, and its financial sustainability is highly sensitive to the values o f economic, financial, and demographic parameters, which could be quite volatile in the long term. The value o f the scheme's reserves in the two most distant scenarios o f our simulations differs by over 30 percent o f GDP after 2050. (b) in the long run, when the scheme matures, the still generous benefits it promises may not be covered by its revenues. For the first decades o f its 12 operation, the CPF will have few or n o retirees and will rapidly accumulate reserves, which will peak around 2040-2050. After this point, however, the financial status of the scheme will rapidly deteriorate and value of its reserves i s likely to become negative. (c) benefits design (largely the same issues as for the PSPS). 31. Depending o n the political and other constraints, three options o f the CPF reform could b e suggested. Option 1: change the parameters of the scheme without changing its structure. Increase inthe retirement age (synchronically with PSPS) and decrease in the effective replacement rate and/or amount of gratuity paid at retirement would enhance financial position of the scheme. However, it would not eliminate inherent volatility o f the funded defined-benefit arrangements. Option 2: introduce a defined-contributioncomponent into the CPF. Under this strategy, the replacement rate provided by the scheme on the defined-benefit basis should be reduced. In exchange, the scheme's participants would b e offered additional benefits based o n the defined-contributions principle. Shifting to this "multipillar" structure would diversify the risks and improve the long-term financial performance of the scheme. Option 3: unify as much as possible design of the pension schemes covering private and public sectors of the economy. This would allow for portability o f benefits between the two schemes and remove labor market distortions (in particular, obstacles for the labor mobility between private and public sectors) caused by the fragmentation of the country's pension system. 32. Whichever of the three strategies i s chosen, there i s a need to increase the yield rate by diversificationof investmentof pensionassets. Currently, the bulk of funds of all funded pension schemes in Sri Lanka is invested in government bonds. Historically, real returns on government bonds in Sri Lanka were low, which puts financial sustainability o f the scheme at risk. 33. Last, but not least, development of the effective governance structure for the CPF is a necessary condition o f its successful performance. Design of this structure should depend on the choice of reform strategy. Ifpension schemes for public and private sector have similar designs, they could b e managed by the same government agency; otherwise, development of a separate governing body for the CPF is desirable. 13 IV. PUBLICSECTOR EMPLOYMENT AND PAY ISSUES Sri Lanka 3 share of public employees to population is among the highest in the world. Since 1998, the government has added about I O percent more individuals to the public sector workj5orce. Shortcomings in the devolution process, numerous and overlapping ministries and political patronage have contributed to the excess number of public employees. A further problem faced by the public sector is the low compression ratio of wages. Targeting wage increases to higher-level staff would ease dfjculties in recruiting staff with skills highly sought-after by the private sector and increase incentives for advancement in thepublic service. OVERVIEW AND CHALLENGES 34. During the decades immediately after independence, the Government o f Sri Lanka was able to achieve impressive gains in leading social indicators. Even now, Sri Lanka's public administration retains certain elements o f good governance, such as a dedicated professional staff, a transparent regulatory system, and levels o f corruption that are among the lowest o f South Asian states. However, the last three decades have witnessed increasing politicization, patronage and fragmentation in administration, leading to an overall decline in civil service standards, capacity and work ethos. Sri Lanka's public administration suffers from tremendous over-staffing, an excessive number o f institutions, an excessive reliance on administrative procedures, weak institutional control mechanisms and politicization. 35. Within CentralGovernment,autonomous agenciesalso havebeen growingin number-including regulatory bodies, research institutes, and service delivery organizations. Table 6 reveals the growth inagencies and their total number o f employees. As with ministries, there are evident cases o f unnecessary balkanization among statutory boards, as illustratedby the Coconut Cultivation Board, Coconut Development Authority, and Coconut Research Institute. Table 6. Sri Lanka: Number of Autonomous Agencies and Their Employees 1994 1998 2002 No. of Agencies 11 111 130 161 Employees l/ 60,209 65,551 106,654 Central Govt. Current Transfers to Public Institutions (Rs.mn.)21 4,320 12,091* Sources: 11Public employee census documents, various years. 21 Central Bank of Sri Lanka, Annual Report 2002. These transfers are not all dedicated to salaries. * provisional 36. Devolution has led to an expansion of the central government's administrative structure, as seen in the establishment o f ministries for regional development (e.g., Ministry o f Southern Region Development, Ministry o f Western 14 Region Development). The functions o f these central government development ministries clearly overlap with those o f the Provincial Council governments. Coordination is complicated by the fact that the geographical coverage o f the central government's regional development ministries is not coterminous with the boundaries o f the Provincial Councils. It has been suggested, however, that these development ministries o f central government compensate for a shortcoming in Sri Lanka's devolution design. Devolution did not establish effective policy linkages between central government and the provincial councils, but the Secretary o f the relevant development ministry may provide a channel for provincial councils to raise policy concerns with central government. 37. Sri Lanka's share of public employees to population is among the highest in the world. Public employment exceeds 1 million out o f a population o f little more than 19 million. The problem o f excessive employment i s long-standing. Yet, the situation has worsened in recent years. In 1998, there were roughly 930,000 public employees in Sri Lanka-already a worrisome number. Today, there are an additional 100,000. Figure 5 presents the relative share o f Civilian Central Government, Provincial Council, state- owned enterprises (SOEs), and armed forces employment. Three related factors clearly have contributed to the excess number o f public employees: (1) shortcomings in the devolutionprocess, (2) numerous and overlapping ministries, and (3) political patronage. Figure 5. Sri Lanka:Shares of TotalPublicEmployment, 2002 State Civilian Central Govt. 41% Armed Forces 12% 38. Despite having such a large share of public employees to population, Sri Lanka's expenditure on public sector wages and salaries is not higher than other countriesin Asia. Table 7 presents data on the evolution o fthe central government wage billfrom 1995-2001 inSri Lanka and three comparator countries. 39. On average, monetary allowances constituted roughly 40 percent of employees' total pay in 2002. All employees were entitled to two interimcost-of-living allowances, which combined accounted for nearly two-thirds o f the total expenditure on allowances (excluding the armedforces). The remaining third i s difficult to account for in detail, as apart from the two cost-of-living allowances already mentioned, there is considerable variation in the allowances different employees receive. Indeed, the 2000 15 Salaries Commission identified as a significant problem the pay disparities arising from allowances. Table 7. Wages and Salariesof Consolidated Central Government 1995 1997 1999 2001 SriLanka GDP 667,772 890,272 1,108,845 1,397,453 Rs. mn. Current Revenue 136,161 164,779 195,895 231,420 Current Expenditure 195,880 228,732 267,611 367,966 o/w Wages & Salaries 34,909 44,676 58,532 78,056 % current exp. 17.8 19.5 21.9 21.2 % current revenue 25.6 27.1 29.9 33.7 % GDP 5.2 5.0 5.3 5.6 India Wages & Salaries 170 266 322 373 Rs.billion % current exp. 9.7 11.5 10.6 9.4 % current revenue 11.6 14.4 13.8 12.6 % GDP 1.4 1.7 1.7 1.6 Malaysia Wages & Salaries 12804 14494 Ringgit mn. YOcurrent exp. 33.3 33.9 % current revenue 23.6 22.3 % GDP 5.8 5.1 Thailand Wages & Salaries 223168 262316 289511 302269 Bahtmn. % current exp. 51.6 49.4 48.5 38.3 % current revenue 28.6 30.1 39.2 33.7 % GDP 5.3 5.5 6.2 5.9 Source: IMF, GovernmentFinanceStatistics Yearbook, 2002. Note. ExcludesArmed Forces. POLICY OPTIONS 40. Changes to the Administrative Structure. The "Report o f the Salaries Commission 2000" recommends limiting by statute the overall number o f central government ministries. Provincial Councils must comply with such a limit; the 13th amendment restricts the number o f ministries to five. However, a restriction of this sort should not be necessary to reduce the number of central government ministries to a more reasonable number.Moreover, the statute i s unlikely to succeed unless policy makers are convinced of the fiscal dangers created by an expansive administrative apparatus. Rather than devote energy toward agreeing an arbitrary cap on the number of ministries, efforts might be better devoted to studyingwhich o f the current ministries ought to be slated for elimination. 41. Employment Reductionsvia Attrition. Given the enormous number of public employees in Sri Lanka, personnel reductions through attrition alone may not be an adequate solution. In any case, the data show that over the last two years of the current hiringfreeze a significant numberofvacancies havebeen filled, andnewposts havebeen created (Table 8). 16 Table 8. Sri Lanka: Growth in Employment, Population, Labor Force, and Real GDP, 1998-2002 percentagechange 1998 2000 2002 1998-00 2000-02 1998-02 CentralGovt Emdoyment . - 354,160 413,300 436,390 16.7 5.6 23.2 Population 17,935,000 18,467,000 18,9681480 3.0 2.7 5.8 GDP (constant 1995 us $) LabourForce 7,722,811 8,103,320 8,410,623 4.9 3.8 8.9 15,067 16,658 16,909 10.6 1.5 12.2 Sources:Sri Lanka Departmentof Census& Statistics, Census of Public and Semi-Government Sector Employment " 1998" and Census of Publicand Semi-GovernmentSector Employment 2002" (preliminaryreport); Sri LankaMoF `` " BudgetEstimates2002'' (vol. 1-111); andWorld Bank DevelopmentIndicators. 42. Employment Reductions via a Voluntary Retirement Scheme. A carefully crafted, targeted voluntary retirement scheme might be combined with a hiring fi-eezeto generate more rapid reductions in surplus staff, particularly at mid and lower grades. The "Report o f the Salaries Commission 2000" sets an ambitious goal o f a 30 percent reduction instaff, focused on Categories C & D employees. 43. Creating a New Pay Scale. The 2000 Salaries Commission, in accordance with its mandate, has proposed a detailed revision o f Sri Lanka's public pay scales. One goal o f the Commission i s to reduce the number o f pay scales. That would be a welcome change to strengthen the coherence, consistency, and transparency o f pay policy. However, the Commission's proposals in this area are relatively timid. The Commission recommends eliminating 22 scales and creating another seven. If implemented, the reform would still leave government with more than 100 separate pay scales. Greater consolidation i s warranted. 44. Target pay increases to higher-level staff. The 2000 Salaries Commission proposed a more generous pay increase for manual laborers than for any other employee group. However, the Commission's study o f private sector wages reveals that manual laborers in government receive 25 percent to 125 percent more than a similar person in a private company. Clerical workers, too, receive up to one-third more than their private sector equivalents. By offering a more generous percentage increase to the less skilled segment o f the public sector labor force, the effect o f the Salaries Commission proposal would be a further erosion o f the compression ratio, thereby weakening incentives for advancement inthe public service. 45. The compression ratio was already a modest 9.1 under the salary scales fixed in 1992 and fell to 8.1 in 1997. It would fall further to 7.1 under the current proposals o f the Commission. Targeting increases to higher-level staff i s an obvious alternative approach to decompress the wage scale and thereby ease difficulties in recruiting staff with skills highly sought-after by the private sector (a problem identified by the Commission). Targeted increases also would also be fiscally more affordable. 46. PromotingMerit in Appointments.Patronage-style appointments have become commonplace in Sri Lanka. The Ceylon constitution o f 1948 established an independent Public Service Commission with authority over the appointment, promotion, transfer, and disciplinary control o f permanent employees. However, the authority o f the Public Service Commission was severely weakened in 1972 when the new constitution 17 transferred the power of appointment to the Cabinet of Ministers. An important remedy is the recently approved 17th amendment to the constitution. In accordance with that amendment the national Public Service Commission (PSC) is granted the authority to appoint, promote, transfer and dismiss all public officers, with the exception of department heads. However, the PSC must receive greater support from central government inorder to reestablish its former authority. 18 PART2. SECTOREXPENDITUREREVIEWS V. HEALTH Sri Lanka has achieved exceptional improvements in health outcomes despite relatively low levels of spending in the sector. There is however a need to consolidate impressive gains and respond to emerging challenges, such as non-communicable diseases, associated with the rapidly agingpopulation. F'hile unit costs are low, there is evidence of inefficiency, such as an increasing share of resources going to curative rather than preventive care. Malnutrition is a major problem with over 30 percent of pregnant and lactating mothers suffering from anemia. There are also some inequities in theprovision of health services and in health outcomes, with differences in nutritional outcomes particularly notable. Inequalitiespersist in terms offinancial inputs (though much of this is accounted for by differences in private expenditures)-notably in the Western Province. SECTOR PERFORMANCE 47. Sri Lanka has achieved exceptional improvements in its health status despite relatively low level of spending in the sector. Infant and maternal mortality rates o f 16.3 per 1,000 live births and 2.3 per 10,000 live births are already close to those enjoyed inOECD countries. As aresult significant additional improvements inthese indicators, as proposed by the Millennium Development Goals, are not achievable within the Sri Lanka context. 48. Sri Lanka's success reflects factors within and outside the health sector-a highly educated populationmaking full and informed use of available health services and adopting reasonably healthy lifestyles. Geography, climate and a range o f socio-cultural factors also play a major role. Nonetheless, the health sector can also take significant credit for these trends by ensuring that resources have generally been utilized in an equitable and efficient manner. There is almost universal coverage o f essential services. Immunization rates exceed 95 percent even for disadvantaged groups whilst the share o f deliveries which are assisted by trained health personnel-estimated at 93.9 percent in 2000-is exceptionally high. Moreover, such coverage has been provided at extremely low cost. Studies show that unit costs are low in comparison to other countries in the region-largely a reflection o f the high (in some cases, excessive) levels o f utilization- andhave been declining over time. Public services have generally been allocated ina pro poor manner. Though a large share o f resources i s allocated to secondary and tertiary care poorer groups appear to have significant access to services provided by these institutions. As a result, the hospital sector appears to provide an effective, if basic, safety net for those facing the potentially catastrophic effects associated with hospitalization. Benefit incidence studies demonstrate that the poor utilize a far greater share o f public subsidies thaninIndia, Nepal andBangladesh. 19 49. However, there i s an unfinished agenda. Despite low unit costs there is still evidence of inefficiency. Utilization of in-patient services is much higher even than OECD countries indicating that much o f this care could perhaps be delivered more cost effectively at lower levels in the system. Maternity homes, on the other hand, have very low levels o f utilization-partly a reflection o f the declining number o f births but also due to the fact that such facilities are typically bypassed. In practice, the lack o f administrative or financial incentives to encourage more appropriate referral system has encouraged such behavior. 50. Nutrition is a major problemwith over 30 percent of pregnantand lactating mothers suffering from anemia. The poor underlyng health o f mothers has also contributed to persistently high level o f low birth weighed babies.. Vulnerable groups including the economically disadvantaged living in the estates sector and those living in conflict-affected areas are particularly affected. Malnutrition i s on the increase in these areas and the incidence o f malaria i s increasing with over 50 percent o f reported cases in conflict-affected areas. 51. There exist some inequities in the provision of health services and health outcomes. Sri Lanka's achievements inproviding universal and free education andhealth care to the general population are reflected in good social indicators for the poor and the poor fare considerably better in some key non-income dimensions o f poverty than in most other developing countries, a few crucial pockets o f inequity persist. In the health sector, this i s illustrated by lower access for some health services and weaker health indicators for poor and vulnerable groups, especially those living in estates, or remote or conflict-affected areas. Figure6. Per CapitaExpenditureonHealthinSriLankaby Province (1997) 2500 2000 1500 2 1000 500 0 52. Inequalities persist in terms of financial inputs (though much of this is accountedfor by differencesin privateexpenditures)-notably inWesternProvince. As indicatedinFigure 6, there are significant differences inper capita public spendingon health by province although these are partly offset by expenditures through central programs. However, there has been a trend toward greater equity inpublic expenditure on health by province over the 1990s. Yet, there are some substantial differences interms o f infrastructure and human resources across the regions. The number o f hospital beds 20 varies from 4.8 per thousand population in Colombo to 2.0 or less in Jaffna, Kilinochi and Vavuniya whilst the number o f medical officers per 100,000 population varies from around 99 in Colombo to less than 12 in Jaffna, Kilinochi and Mannar. Such differences are mainly due to the presence o f teaching, provincial and specialized hospitals in the major urban centers. 53. Performance against some of the key health service indicators is generally good with little, if any, inequality. The proportion o f children between 12 and 23 months immunized for DPT3/4-a good indicator o f health system performance-is around 98 percent (97.2 percent inestates). The proportion o f children immunized against measles-an indicator closely linked to child mortality rates-averaged 94 percent according to DHS 2000 (86.1 percent in the estates sector though more recent date suggests this gap has reduced significantly since). Overall utilization o f public services does differ widely by district. The number o f inpatient admissions per head in Colombo, for instance, is double that in Nuwara Eliya. Per capita out patient contacts in Anuradapura are almost double those inMatara. Figure 7. Sri Lanka: Percentage of children aged 3 to 59 months who are stunted by place of residence 60 50 40 El 1993 2000 30 20 I O 0 ColomboMetro Other Urban Rural Estate Figure 8. Sri Lanka: Health Outcomes by Level of Mothers Schooling 35 0 Neonatal MR Post NeoNatal MR IMR USMR 21 54. As noted above, significant inequity in outcomes persist. Differences in nutritional outcomes are particularly notable. A child living in the estates sector i s over four times more likely to be stunted than one living in Colombo and almost three times more likely than one living in rural areas (although the differential is narrowing) (see Figure 7). Outcomes are closely linked to socio-economic status as children with mothers with higher levels o f educational attainment having far better outcomes (see Figure 8). HEALTH EXPENDITURE 55. Sri Lanka has achieved extraordinarily good health outcomes with relative modest level of spending on health. Total expenditure on health was SLR 39,177 million in 1999 of which 13 percent was capital investment. This amounts to 3.5 percent of GDP, which i s somewhat lower than other regional countries (Bangladesh 3.8 percent, India 5.0 percent, China 4.5 percent, and Thailand 4.0 percent). Government funding for health mainly comes from central funds (47 percent) with provincial taxes contributing 4 percent. Private expenditure i s predominantly out-of-pocket expenditure (43 percent), with a further 4 percent coming from employers and 3 percent from private insurance. Health expenditure has been increasing, albeit fi-om a low base. Since 1990, total expenditure has been between3-3.5 percent o f GDP o f which the Government share has fluctuated between 1.4 percent and 1.7 percent. The level o f external development assistance to this sector i s also low (currently between 4-6 percent o f the total health expenditure). 56. By regional and international standards the health system is extremely equitable and displays high levels of technical efficiency (although less so for allocative efficiency with the increasing share of resources going to curative care). Nonetheless, there i s room for improvement in a number o f areas. There are still sections of the population who lack access to basic health services. In terms o f technical efficiency whilst unit costs are low, in part because hospitals are overcrowded, many services could still be provided more cost effectively at lower levels o f the system. Perhaps more importantly, the large and rising share o f public spending going to the hospital sector in response to strong public demand for the provision o f such services, threatens future progress in other important areas. Overall spending on public health and preventive services, for example, declined by around 15 percent inreal terms during the 1990s. Sri Lanka also faces new challenges particularly those associated with its rapidly aging population which will require a reorientation o f support. KEY CHALLENGES 57. Sri Lanka needs to consolidate impressive gains, address the remaining shortcomings, and respond to emerging threat such as non-communicable diseases associated with the rapidly aging population. The latter include a rising incidence o f cancers, health disease, diabetes, accidents and suicides. Sri Lanka is unusual in having gone into the demographic transition at a low level o f per capita income. This makes an emphasis on the prevention of NCDs especially pressing, as the financial means to provide effective treatment for such conditions are unlikely to be available. The system 22 also has to face new threats such as HIV/AIDS and the re-emergence o f dengue and malaria. 58. Support needs to be targeted towards meeting the needs of the most vulnerable, the elderly, disable, adolescents, estates workers and those affected by conflict. Steps will need to be taken to make more rational use o f the existing health infrastructure-addressing issues o f overcrowding at higher level facilities and improving utilization at lower levels. 59. In order to improve efficiency it will be essential to develop more flexible management approaches to promote greater delegation o f authority, the development o f more effective financial and information systems, better planning and management o f human resources. Government will need to take a lead in developing and strengthening partnerships and coordination between key stakeholders in the sector including the private sector, civil society andtraditional systems o fmedicine 60. Financing mechanisms need to be strengthened. This will require the strengthening o f incentives for better performance, ensuring resources continue to be allocated in a pro poor manner and the development o f further mechanisms to promote this, ensuring that adequate provision is made for new initiatives, reviewing strategic health financing options and strengtheningthe information base. 61. Continued progress will depend heavily on the performance of the provincial councils who are responsible for the delivery of services. There i s no mechanism, at present, for ensuring that councils devote adequate resources to meet the health needs o f their populations. Currently, they are not even required to report formally on their spending. Provincial councils face major human resource and financial capacity constraints and remain heavily dependant upon central Government for their funding whilst there are general concerns about the overall methods o f allocatingresources to the provinces. Future resourcerequirements 62. Future health care costs over the coming decade will depend on patient behavior, specifically the propensity of the population to use health services when ill. Recent macro costing studies suggest that traditional increases o f 1-3 percent per annum for outpatient care imply an increase requirement o f 0.2 percent o f GDP. The changing demographic structure-and additional costs for meeting the needs o f an increasingly elderly population-is anticipated to require a similar amount o f additional expenditure. Improvements in productivity o f up to 2 percent per annum based on historical trends would be expected to offset some o f this increase. A key area o f uncertainty i s the rate o f price inflation in the private sector which has major implications both for overall health expenditure as well influencing whether patients seek to access care inpublic or private sectors. Just to stand and maintain existing levels o f provision the study estimates that the share o f public spending on health needs to increase to around 1.6-1.7 percent o f GDP. This relates to the base case scenario-the figures could be much higher under less optimistic assumptions. 23 63. Achieving significant improvements in health services would require additional expenditure. Preliminary estimates suggest that significant improvements in the quality o f care could requireup to 0.5 percent o f GDP whilst new programs including the provision o f anti retrovirals to HIV/AIDS patients, integration o f the estate sector, development o f the infrastructure inthe North and North East could add the same again. The establishment o f a publicly funded family practitioner program could add up to 0.7 percent o f GDP. POLICY RECOMMENDATIONS 64. The key reform areas that needto be tackled by the sector are: e Improve curative health care services (through hospitals and other providers) at the Provincial and district levels in order to make these services more accessible inpoor, ruralareas; e expansion o f health care services to meet the needs o f specific groups such as the elderly, victims o f war and conflict and promoting specific areas o f health care such as occupational health problems, mental health care and estate health services; development o f health promotional programs, with special emphasis on outreach through the schools; e improvement o f disease prevention amongst the poor can reduce the health financing burdenand contribute to more rewarding and productive lifestyles; e reform o f health care fimding, with the aim o f concentrating public support on the poor, including a better effort to mobilize and manage resources inboth the public and private sectors; and coordinating financing mechanisms such as social insurance; e rationalization o fhumanresources andhealth infrastructure; e development o f a coherent policy framework for redefiningthe roles o f the public and private sectors inhealth care provision. 65. Interms of resource allocation priorities the changing nature of the health sector inSri Lanka suggests aneedfor the following: e a modest increase in overall spending on health to maintain the current levels o f provision and significant increases if Government is to take serious steps to improve quality and expand access to services; e a general increase inthe share o fpublic expendituregoing to preventive care; e greater focus on specific program areas notably nutrition, NCDs, maternal and child health, mental health and less on program areas where the threat is declining (whilst continuing to provide sufficient support to consolidate previous gains and respond to outbreaks as they occur); e greater focus on effective approaches within program areas, e.g. behavioral changes programs aimed at alcohol and tobacco; 0 the development o fmechanisms to ensure that resources are increasingly focused on those district andprovinces ingreatest need; 24 0 increases inthe allocation o f resources to support measures to increase hospital efficiency and to make more effective use o f lower level facilities. 66. Over the last decade there has been a major reorientation o f resources away from promotive and preventive services towards curative care. This shift has a large constituency-it i s popular with the public, politicians and medical professionals. A shift in the allocation o f resources towards preventive care would signify a significant break with recent trends. Two other areas requiringmajor attention are: 0 the relationship betweenthe centre and provinces and 0 public sector reform efforts to improve the efficiency o f the public sector. 67. Neither o f these can be taken forward at the sector level but both need greater clarity and action if sustained health development i s to be achieved. 25 VI. EDUCATION Universal access to primary education has been largely attained through the establishment of an island-wide network of schools offering tuitionfree public education. However, raising the quality of education constitutes the next major policy challenge. Education continues to enjoy a prominent place in the government budget, accountingfor about 8 percent of public expenditure. Public expenditure in education is currently chiefly absorbed by salaries and wages and administrative costs in the recurrent budget and building construction in the capital budget. Obtaining higher quality educational outcomes will necessitate a greater share of resources being shifted towards capital expenditures. EDUCATIONACHIEVEMENTS AND CHALLENGES 68. Education continues to enjoy a prominent place in the government budget, accountingfor about 8 percentof public expenditure.Sri Lanka is well-known inthe development policy literature as a country that has attained high levels o f human development relative to its per capita income. Historically, the Government o f Sri Lanka (GOSL) awarded high priority to investment in education, devoting the largest share o f the government budgetto education. Inrecent years, as military expenditures increased as a result o f the conflict in the North-Eastern region and crowded out resources for development, the share o fpublic spendingon human capital declined. 69. Educational attainment is strongly associated with poverty in Sri Lanka. Poverty rates decline steeply as education levels increase (see Figure 9). Poverty incidence was 40 percent among households whose head was uneducated and 38 percent among households with a primary educated head in 1995/96. Among households with junior secondary educated heads, poverty incidence was 31 percent. Poverty rates declined continually across school levels to just 1 percent among households with heads who were university graduates or higher. 70. GOSL's education policy framework emphasizes three key elements: (a) providing universal access to public primary and secondary schooling, and controlled access to public university education; (b) enhancing the quality o f education, especially cognitive achievement, and other dimensions of quality such as good team work, a disciplined and productive work ethic, a positive, problem solving approach, and attitudes favorable to a multi-ethnic, multi-religious society; and (c) increasing the economic and social relevance o f the education system. In support o f this overarching policy framework, GOSL under the leadership o f the National Education Commission (NEC) has developed wide-ranging policy initiatives and measures to develop the education system [seeNEC (1997), (2003)l. GOSLhas invited the World Bank (WB) to support the development o f a medium term budgetary framework for the education sector to facilitate investmentplanning and enhance the efficiency and equity o fpublic spending. 26 Figure 9. Sri Lanka: PovertyIncidenceby Education Level of HouseholdHead, 1995/6 45 ~----40 38 - 7 40 35 30 0 6 25 e 20 c. 15 F 10 5 0 No Schooling Primary Lower Upper GCE O/L GCE AIL Graduate and Secondary Secondary Above Source: Gunewardena, 2000. Compulsory basic (Grade 1-9) education: enrolment and completion 71. One of the chief priorities of education policy makers is the provision of universal access to primary and secondary education, especially the compulsory education cycle, grades 1-9. This policy objective has largely been attained at the primary level through the establishment of an island-wide network o f schools offering tuition free public education. Net enrolment at grade 1 i s about 97 percent, and net primary completion around 95 percent. However, further progress is required in the junior secondary (grade 6-9) cycle. Only about 82 percent o f students complete grade 9. GOSL has adopted several initiatives, such as catch-up education programs, special education programs and non-formal education programs, targeted at these groups o f children. These needto be further strengthened to reach the final set o f children inthe age group 6-14 outside the school system. Education quality 72. In contrast to the high levels of education attainment at the primary and junior secondary levels, GOSL policy makers have become increasingly aware that education achievement is unsatisfactory, and raising the quality of education constitutes the next major policy challenge. Cognitive achievement tests among primary school children, for instance, show substantial shortfalls in mastery o f basic language and numeracy skills. In first language (Sinhalese or Tamil), average mastery i s only 37 percent. This implies that two out o f every three children in the primary cycle have not mastered their first language. English language writing skills are virtually non- existent, with just 10 percent o f children exhibiting the required skill level. In mathematics achievement, too, overall mastery is only 38 percent. The low level o f cognitive achievement among primary students i s especially worrying, from a policy perspective, as primary education forms the foundation upon which higher levels o f education and various types of skills training are built. Hence, the quality o f the entire 27 education system i s severely constrained by the low achievement levels in primary grades. 73. The unsatisfactory state of education quality is also evident at the level of senior secondary education. The proportion o f students passing the GCE O/L i s low, only 37 percent, implying that about two out o f every three students taking the examination fails. Further, this low pass rate has been fairly constant over the period 1998-2002. A substantial majority o f students appear to struggle with subjects such as mathematics, Englishlanguage, science and social studies. At the GCE A/L examination, too, pass rates have been around the 50-55 percent level since 1998 over the 1998-2002 period. This implies that only about one out o f every two students actually passes the GCE A/L examination. These are low pass rates at the GCE O/L and GCE A/L examinations, especially since successful completion of these examinations i s necessary for a variety of skills training courses or access to tertiary education programs, or for entrance into a range o f labor market occupations. 74. Inaddition to the problem ofpoor education quality as measured by low learning achievement scores or high examination failure rates, policy makers and employers have argued that other, non-cognitive dimensions of education quality are also unsatisfactory and need urgent improvement [NEC (1997), (2003)l. Education policy makers have stressed the importance o f producing characteristics such as a disciplined work ethic, good team work, creativity and initiative, a problem solving approach and sound civic consciousness among school children and university students. Educated youth unemployment 75. A major problem facing GOSL policy makers is that of unemployed educated individuals, especially women (see Figure 10). Unemployment rates tend to be highest among university graduates and GCE O/L and GCE A/L completers. The high unemployment rates o f educated labor force participants are mainly caused by lengthy job search durations by new entrants into the labor market. The lengthy periods o f job search has led to considerable social unrest in the past, particularly among university graduates. The main cause o f high unemployment rates and long job search durations among educated young people has been the slow growth rate o f the economy, which has prevented the demand for educated labor rising sufficiently fast to absorb the rising supply o f educated young people entering the labor market. Secondary causes of educated youth unemployment have included the production o f graduates with low generic skills in demand within the labor market, such as English language fluency and IT skills, and the moderate quality o f some universitydegree programs. 28 Figure 10. Sri Lanka: UnemploymentRates and Job Search Durationsby EducationLevel, Men andWomen, 2000 Average unemployment rate (in percent) 30 25 20 15 10 5 0 Primary Lower Upper GCEOL GCEAIL Graduate Postgraduates All secondary Secondary Average job search duration (in months) 25 20 15 10 5 0 Primary Lower Upper GCEOIL GCE AIL Graduate Postgraduates All secondary Secondary Source: Department of Censusand Statistics, Labor ForceSurvey 2000. PUBLIC SPENDING INEDUCATION Comparativelylow publicexpenditureon educationby internationallevels 76. Public education expenditure in Sri Lanka currently amounts to about SLR40 billion (US$415 million) annually. In recent years, the education budget has accounted for approximately 3 percent o f GDP and 7-8 percent o f government spending. This level o f spendingis somewhat lower than South Asian countries and low income nations which devote about 3.2 percent o f national income and 11percent o f government expenditures to education. The group o f lower middle income countries, to which Sri Lanka is expected to belong in the near future, spends about 4 percent o f GDP on public education. Countries such as South Korea, Malaysia, Thailand and Singapore, which act as policy models for Sri Lankan policy makers, devote considerably higher resources to education (Table 9). 29 Table 9. EducationExpenditureas a Shareof NationalIncome and Government Expenditures,Sri Lankaand Selected Other Countries Country Education Education Recurrent education Average teacher expenditure as a expenditure as a % expenditure per salaries as a % o f % of GDP o f government student as a % of GDP GDP per capita expenditure per capita Sri Lanka 2.9 6.8 9.7 1.5 India 4.1 12.7 16.3 3 Bangladesh 2.5 15.7 3.3 Pakistan 1.8 7.8 Nepal 3.7 14.1 14.5 Malaysia 6.2 26.7 20.7 2.9 Maldives 3.9 11.2 Bhutan 5.2 12.9 Thailand 5.4 31 25.3 3.1 South Korea 3.8 17.4 4.1 Singapore 3.7 23.6 Phillipines 4.2 20.6 Costa Rica 4.4 22.8 South Asia 3.2 11.2 14.5 2.8 Low Income Countries 3.2 16.3 Lower Middle Income Countries 4.1 18.5 Upper Middle Income Countries 5 23 Sources; Sri Lanka, calculations from Central Bank o f Ceylon Annual Reports, various issues; Other Countries, World Development Indicators (World Bank) and UNESCO Statistics, various issues. The information above for Sri Lanka is computed from 2002 data. Other countries and regions are from the closest available year inthe late 1990's and early 2000's. Equityof publiceducationspending 77. The enrolmentpatternacross economic groups shows a highdegree of equity in education attainment at the lower education levels (Table 10). Net primary enrolment is almost equal across consumption quintiles and among poor and non-poor households. Net primary enrolment in 1995196 rangedonly from 95 percent to 97 percent between the poorest and richest consumption quintiles. At the junior secondary education level net primary enrolment ranged from 61 percent among the poorest consumption quintile to 76 percent among the richest consumption quintile, while at the senior secondary education level net primary enrolment varied between 31 percent among the lowest consumption quintile to 60 percent among the highest consumption quintile.These are progressive enrolment patterns by developing country standards. 78. At the tertiary education level, however, the pattern of enrolment is heavily weighted in favor of the richest quintile, with 13 percent enrolment, as against the poorest two consumption quintiles, where only 2 percent o f students enroll in tertiary education. This i s an unfavorable distribution o f tertiary education enrolment, from an equity perspective, with a large gap between the highest consumption quintile and the two lowest consumption quintiles. Overall, public education spending at primary and secondary education level appears equitable, while public education expenditures at tertiary level appear comparatively inequitable. 30 Table 10. Sri Lanka: Net Enrolment Rates inMajor Grade Cycles by Economic Groups Consumption Net primary Net junior Net enrolment Net tertiary quintiles enrolment secondary enrolment enrolment (grade 1-5) (grade6-9) (grade 10-13) % % % % Lowest 1-20 95 61 31 2 Quintile21-40 96 66 35 2 Quintile41-60 95 67 41 4 Quintile61-80 96 77 47 5 Highest 81-100 97 76 60 13 Poor 95 63 32 2 Non-Poor 96 71 47 6 Sri Lanka 96 68 42 5 Source: Calculatedfromthe HouseholdIncomeand ExpenditureSurvey, Department o f Censusand Statistics, 1995196. Composition of public education expenditure 79. Public investment in education is chiefly absorbed by salaries and wages in the recurrent budget and building construction in the capital budget. In consequence, investment in quality inputs, such as equipment, technology, reading material and tools, and quality processes, such as professional development o f principals, sections heads and teachers in the school system and university academics and administrators in the university system, has been crowded out. As Sri Lanka moves from the first stage of education development, providing universal access to basic education, to the second stage of education development, producing a good quality and high performing education system at all levels, an increasing share o f resources needs to be investedinquality inputs and quality processes. CHALLENGESINGOVERNANCEAND SERVICE DELIVERY 80. While the level and pattern o f public expenditure on education can be important determinants of education attainment, equally important i s the quality and effectiveness of service delivery. Inthis context, the following areas have beenhighlighted as the main issue that need to be addressed: (i)poor teacher deployment and high teacher absenteeism: (ii) inadequate empowerment o f education institutions, such as schools, NCOEs and TCs, especially in staffing, planning and administration; (iii)limited administrative capacity and inadequate use o f technology, which leads to delays inpolicy implementation, operations and procurement; and (iv) the difficulty o f attracting and retaining highly qualified university academic staff, given the enormous difference in academic salaries paid by developed countries and Sri Lanka, and the ability o f well- educated individuals to compete on overseas labor markets. POLICY DEVELOPMENTFOR THE MEDIUM-TERMBUDGET FRAMEWORK 81. The government faces several key challenges to increase the quality o f education, enhance equity o f public spending, strengthen service delivery within the system, and improve the economic and social relevance o f schools, universities and technical 31 education institutions. These challenges exist at a time when public spending in education i s relatively low by international standards and the state experiences a high fiscal deficit which compels it to adopt a conservative fiscal policy. In this context, the government can consider four sets o f measures to increase education spending and enhance the efficiency and equity o f public spending. Increaseprivatesector investmentandparticipationin education 82. There is a need to promote private sector participation in education, especially at tertiary level. Relaxing the legal barriers to private sector participation in education could produce several benefits. First, it would increase the overall volume of resources invested in the education sector. Second, since the students attending private schools and education institutions are likely to be drawn from upper income families, it would release more public resources, on a per student basis, for students from poorer families. Third, it would stimulate economic activity in a sector where investment has been artificially restricted and contribute to higher growth. Fourth, it would provide an alternative mode o f service delivery, with considerable power and responsibility at the level o f the individual education institutions, such as private schools and education institutes. These private education organizations would be compelled to offer highquality services to remain viable inan economic context where they are in competition with free public education institutions. Protectandgraduallyincreasepublicspending in the schoolsystem 83. Government policy to improve education quality envisages several important education policy measures (see NEC (2003)). These policy initiatives require considerable investment o f resources in the education system. Given the low level and decreasing trend o f real public education spending, and the major challenges Sri Lanka faces in improving the quality and economic and social relevance o f the education system, it is important that the government preserve the current level of real expenditurein the short-termandincreasepublic investmentin educationgradually over the medium-term. The real value o f the education budgetneeds to be preserved in 2005, and from 2006 onwards gradually increased, with the real value o f the increase in resources targeted at the school system. Enhancethe equity of public educationspending 84. The pattern of student enrolment over major grade cycles and the unit costs of education by grade level, show that spending in primary and secondary education is relatively progressive and benefit students from low and middle income households, while spending in tertiary education tends to benefit students from upper-income households. Inconsequence, there is a strong case for allocatingthe major share of any increasein public resources for the education sector to primary and secondary schooling, while carefully controlling the expansion o f the public university system and allowing enrolment expansion in tertiary education take place mainly in the private sector. Opportunities for poorer students to access tertiary education could be expanded 32 through schemes such as student vouchers and student loans, rather than through direct public sector provision. Increase cost-effectiveness of the education system 85. The tight resource constraint faced by the government makes it extremely important that the education system generates internal savings to reduce costs. In this context, three important policy measures are available. i. Schoolrationalization.Thecost-effectivenessofoperatingtheschoolsystem can be enhanced by consolidating and amalgamating small, uneconomical schools, where this can be accomplished without reducing access to schooling and damaging enrolment and attendance. The public school network is such that a sizable proportion o f small schools are located close to other, larger government schools. MHRECA and the provincial councils designed and implementeda successful school rationalization program, during 1998-2002. This program, which was temporarily suspended due to intense political pressure, needs to be re-opened, but with careful safeguards to protect access for poor and vulnerable groups. ii. Raise the student-teacher ratio in schools and tertiary education institutions. Currently, the student teacher ratio in schools, at 21:1, i s low. Countries with outstanding education systems and far higher income levels, such as South Korea, Singapore and Hong Kong, have higher student-teacher ratios and more cost-effective systems. The current teacher employment and deployment policy i s based on a target student-teacher ratio o f 22:l for secondary grades and 26:l for primary grades. The government could consider increasing the target student-teacher ratio to 23: 1 for secondary grades and 27:l for primary grades. In addition, the student-teacher ratio in the public university system tends to be very low, at 14:l. The government could consider increasing this student-teacher ratio to the internal academic norm o f 18:1, so that cost savings can be generated within the system. iii. Rationalizeadministrativestaffing.Theroles,functionsandresponsibilities o f staff within the complex, multi-tiered education administrative system is unclear. A careful analysis o f work loads, and de jure and de facto roles, functions and responsibilities o f staff in the various education ministries, provincial councils, zonal education offices, division education offices and tertiary education institutions i s likely to identify both duplication o f roles and functions, and gaps and shortages o f staff. Overall, there may be an excess o f staff, which a carefully designed voluntary retirement scheme could decrease, reducing pressure on the budget and releasingmore resources for investment. 33 Re-investing resources saved within the education system back in the sector 86. In order to increase resources for quality enhancing education policy initiatives through the implementation o f cost saving measures within the education system, it i s vitally important that the Ministry of Finance permits funds saved to be channeled back into the education system for investment. If the funds saved through cost- reducing measures are not re-investedinthe education system the objective o f enhancing education quality would be defeated. Further, the education system would lose the incentive to implement difficult measures such as school rationalization and higher student-teacher ratios. Intra-budget shifts in resource allocation in favor of quality inputs and quality processes 87. Over time, the balance of capital and recurrent spending within the education budget needs to shift to allocate a greater share of resources to capital expenditure. Within the capital budget, a higher proportion o f resources needs to be invested in quality inputs such as IT centers, science laboratories, libraries, activity rooms, multi-purpose rooms, equipment, technology and tools. The construction and expansion o f buildings, which absorbs the highest share o f the capital budget, needs to be rationalized, prioritizing the urban school system which i s experiencing rising demand. Construction work within the university system, beyond the current cycle o f buildings in progress, should be halted until a careful needs assessment has been completed. Within the recurrent budget, too, resources need to be reallocated from salaries and administrative expenses to quality processes, such as teacher education and training, management and academic training o f school principals, and the purchase o f teaching material and learning resources. INCREASING THE QUALITY OF SERVICE DELIVERY Decentralizing education management and establishingschool boards 88. Measures to improve the quality of service delivery within the education sector are extremely important. In this context, the government has identified a key policy proposal to improve service delivery within the school system. This i s the establishment of school boards and decentralization o f education management to schools. The objective of this policy initiative is to empower principals, section heads and teachers, enable schools to forge links with local communities to improve resource mobilization and public accountability, increase the efficiency of decision making by decreasing administrative layers, and improve service delivery. A sound evaluation o f this pilot program would be extremely helphl for future policy making on the structure o f school governance. 89. The government should also consider applying the principle of education decentralization to other education institutions, such as National Colleges of Education (NCOE's) and Teacher Centers (TC's). Currently, the NCOE's have little managerial responsibility and autonomy. Curriculum design, assessment and certification 34 of teacher education courses offered by NCOE's are undertaken by the National Institute of Education (NIE). Recruitment o f academic staff i s to a centralized teacher educators service, and NCOE's rarely have a voice in the choice o f academic staff allocated to them. TC's operate in an inefficient administrative structure. The NIE designs the curricula o f continuing teacher education programs offered by TC's. The NCOE's act as academic advisors to TC's. For administrative purposes, including finances, the TC's come under the zonal education offices. This complex administrative and academic structure has badly hampered the development o f TC's. Devolving a greater degree o f managerial autonomy and responsibility to NCOE's and TC's could increase the operational efficiency of the institutions and enhance the quality o f service delivery, Monitoringandevaluation 90. The development o f a public expenditure tracking system would help identify the quantity o f resources flowing to and within service delivery points, such as schools, NCOE's, TC's, universities, vocational training and technical education institutions, and the NIE, and the transactions cost o f the system. Such a public expenditure tracking mechanismwould be an extremely usefulmanagementtoolto increasethe efficiency andequity of resourceflows andthe quality of service delivery. 35 VII. SOCIALWELFARE A review of expenditures of the Ministry of Samurdhi and Social Welfare reveals a high concentration of spending on direct welfare related beneJits for both Ministries. Of the two, the Samurdhi Ministry is responsiblefor the larger share of direct welfare payments. Yet, thepoverty impact of the Samurdhi consumption transfer is weak: the scheme misses 37-47 percent the two lowest income quintiles while over 40 percent of transfers are made to the wealthiest 60percent. Most of the Ministry of Social Welfare spending goes towards disabled soldiers andfamilies of soldiers killed in action. While this is a needy segment of society, the Ministry of Social Welfare needs to carve out a rolefor itself as a primary agency for the design and delivery of social welfare to all the vulnerable groups in the country. A number of areas require enhanced social protection schemes. One such area is theprovision of social protectionfor the elderly. OVERVIE$ 91. Social welfare spending in Sri Lanka comprises o f a multitude o f overlapping programs administered by a number o f Ministries. Totalwelfare expenditureamounted to an estimated 3.2 percent of GDP and 12.2 percent of total government expenditure in 2003-higher than the overall education budget. Welfare programs include Samurdhi transfer payments, pensions to retired government personnel, expenditures on school uniforms and textbooks, payments to disabled soldiers and dry rations to rehgees. The main agencies directly involved with program delivery at the central level are the Ministries of Samurdhi, Social Welfare, Education and Health, and Triple R (Relief, Rehabilitation & Resettlement). The Provincial Ministries are also involved in the implementation o f social welfare programs, notably the disbursement o f the "Poor Relief',which is financed by the decentralized budgets. 92. The Samurdhi transfers program, run by the Ministry of Samurdhi, is the most significant of welfare programs. Samurdhi constitutes the second highest budget item inwelfare spending (after pension payments to retired public servants). At its peak in 2002, expenditures of the Ministry of Samurdhi-of which the transfer payments comprise a significant share-consumed close to 1 percent o f GDP and 4 percent o f the government budget; the 2003 estimates are 0.75 percent of GDP and about 3 percent o f total government spending. POLICY FRAMEWORK OF THE MINISTRIES AND RELATEDISSUES 93. Broadly defined, social welfare i s the provision o f government assistance to people inneed. The categories o fpeople inneed are defined as the vulnerable sections o f the elderly, disabled, children and mothers, and the poor. Strategies to support these segments o f society typically take the form o f safety nets and income support programs. L The analysis here i s confined to the two main agents at the central government level delivering welfare programs, the Ministry o f Samurdhi and the Ministry o f Social Welfare, as they have the most explicit mandate to serve the neediest and the most vulnerable sections o f society and are the most significant in terms o fbudget allocation. 36 Ifthis understanding is translated into the policy frameworks of the two Ministries that are assessed in this study, the vision and mandate o f the Ministry o f Social Welfare comes closest to that o f a welfare program. It explicitly aims to target the disadvantaged sections o f society and make them partners in the national development process. Samurdhi, on the other hand, presents itself as a poverty alleviation program, which pledges to improve the social and economic conditions o f disadvantaged sections o f society. 94. As the expenditure analysis that follows reveals, the Ministry of Social Welfare dedicates the bulk of its spending on a disadvantagedsection of society-disabled soldiers and families of soldiers killed in action. It could be argued that care for this section o f society should be the priority and responsibility o f the defense institutions. They are different from war victims, which i s rightfully the target group o f the welfare agencies. This dilemma i s all the more important to be addressed since this group receive a predominant share o f the benefits, with little left for all other disadvantaged groups- vulnerable children, or those affected by natural disasters, disability, or old age- targeting whom i s actually the stated objective o f the Ministry. Thus there appears to be a mismatch between the profile o f expenditures by the ministryand its stated objective. 95. Conversely, the Samurdhi program, which describes itself as a development program, i s in terms o f its expenditure profile primarily a welfare program, providing income support to poor families (as the expenditure review below will show). The substantial transfer component o f the program cannot be reconciled with the stated aim o f the Samurdhi program to "empower the rural poor" to mobilize their potentials. Thus there is a need to re-define the objectives o f the program to better match the composition o f its expenditures. This would also help rationalize the objectives and target group o f the Samurdhi, in the context o f the broader Welfare system that includes a multitude o f smaller programs, including those administered by the Ministry o f Social Welfare. Effectiveness o f spendingi s also an issue, due to the fact that expenditures for Samurdhi transfers have not realized their potential interms o f improving welfare. 96. As studies on the Samurdhi program reveal, the poverty impact o f the consumption transfer i s weak. The high degree o f politicization o f the program has resulted in considerable mis-targeting, with the result that Samurdhi covers many o f the non-poor (total coverage o f almost half o f the population), while missing some o f the neediest, and consequently provides too little to have a marked impact on consumption levels o f the poor. The Sri Lanka Integrated Survey (SLIS) supported by the World Bank, found clear evidence o fboth leakage and exclusion inregardto Samurdhi coverage. SLIS found that while 14-42 percent o f households in the three higher income quintiles also receive Samurdhi transfers, 36-47 o f households in the two lowest income quintiles did not receive the transfers (see Table 11). More than 40 percent Samurdhi transfers in the sample were made to the wealthiest 60 percent o fthe population. 37 Table 11. SamurdhiCoverageand Benefitsby Income Quintile Benefit Quintile 1 (Lowest) 2 3 4 5 (highest) Total %ofrecipients 64 53 42 30 14 40 Rs./ HH (gross) 403 390 378 362 303 381 Rs./ capita (gross) 88 96 93 105 107 95 Rs./ HH(net) 322 315 307 301 260 310 Rs./ capita (net) 71 78 77 88 94 78 Source. SLIS quoted in World Bank 2002, p.36. Note. Gross receipt includes compulsory deductions while net receipt excludes them. 97. The lack of clarity on objectives o f the Ministries has also resulted in insufficient coordination across programs, in the context o f targeting. This has led to considerable duplication between the beneficiaries o f different programs, while at the same time some o f the poorest and most vulnerable groups are missed altogether. Effectiveness o f welfare expenditure interms o f its impact can thus be enhanced by better rationalization between the programs of different Ministries, interms o ftheir stated objectives and expenditures. EXPENDITURE PATTERNSAND TRENDS 98. The review of expendituresof the Ministry of Samurdhi and Social Welfare reveals a high concentration of spending on direct welfare related benefitsfor both Ministries-in practice both Ministries are thus strongly oriented towards welfare, although the stated objective o f Samurdhi would suggest otherwise. More than two thirds of the total budgets o f the Ministries are spent in the form o f direct transfers to households. Together, the two Ministries cover about half the population o f the country. Inthe case of the Ministry of Samurdhi, which administers the Samurdhi program, the transfer goes into the payment o f the consumption grant to poor households. Inthe case of the Ministry o f Social Welfare, a large part o f the transfer payments goes to a more specialized group4isabled soldiers and families o f soldiers killed in action; other beneficiaries include vulnerable children, or those whose vulnerability is caused by natural disasters, disability, or old age. At the same time, there i s evidence o f considerable overlap between the beneficiaries o f the programs administered by the two ministries. 99. Since the transfers are inthe form o f regular cash payments, the expenditures are classified under recurrent expenditure, which dominate the spending pattern. The other significant category of recurrent spendingi s administrative costs, including payments on personal emoluments, which consume a substantial portion o f their budgets-especially for the Ministryo f Samurdhi. 100. An annual average of SLR 13.2bn was allocated to Samurdhi between 2000 and 2003, which amounts to 1 percent of annual GDP, over 10 percent of social service expenditures and a third of welfare expenditures (Table 12). Total Samurdhi expenditures (nominal) increased from around SLR 10billion in2000 to SLR 15.2 billion in 2002, and then declined to around SLR 12.2 billion in 2003. These movements have 38 largely occurred due to changes in the size o f the transfer component. Recurrent spending typically accounts for about 98 percent o f total expenditure, and about 97 percent o f that consists o f payments for the consumption grant to households and salaries paid by the Samurdhi Authority to the more than 26,000 cadre o f staff that administers the program. For instance in 2003, recurrent spending accounted for SLR 12.2 billion out o f a total o f SLR 12.5 billion, o fwhich SLR 9.2 billion was spent on consumption grants or transfers, and SLR 2.8 billion on the salary bill o f the Samurdhi Authority (about 23 percent o f the total expenditure o f the program). The large salary bill constitutes a drain on the expenditures o f the Ministry, and reduces the welfare impact o f spending. Table 12. Total expenditures of the Ministry of Samurdhi and Social Welfare Ministry 2000 Actual 2001 Actual 2002 Estimate 2003 Estimate I n SLR million Ministryof Samurdhi 10,682 14,103 15,384 12,484 MinistryofSocial Welfare 2,555 3,779 4,513 4,047 Combined expenditure 13,237 17,882 19,897 16,531 As % of GDP Samurdhi 0.85 1 0.97 0.75 Social Welfare 0.2 0.27 0.28 0.24 As % of totalgovt. spending Samurdhi 3.19 3.66 3.82 2.84 Social Welfare 0.76 0.98 1.12 0.92 101. In contrast to the Ministry of Samurdhi, the Ministry of Social Welfare and its two associated departments, consume only 0.2 percent of GDP, 3 percent of the social services expenditure and 7.5 percent of the spending on welfare. The breakdown o f expenditures o f the three institutions shows that recurrent expenditure accounts for about 98 percent o f the total. Mirroring the pattern seen for Samurdhi, there was a steady increase in total expenditures from 2000 to 2002, from around SLR 2.5 billion to SLR 4.4 billion, followed by a decline to around SLR 4 billion in 2003. The increase i s primarily explained by an increase in the size o f the transfer component; the share o f this component in the total budget has also significantly increased over time, from 79 percent in 2000 to 94 percent in 2003. As much as 92 percent o f transfers in 2003 went for social security for disabled soldiers (including families o f soldiers who have lost their lives), leaving only 8 percent for beneficiaries within other vulnerable groups. REFORMS INPROGRESS 102. The issue o f politicization and mis-targeting o f the Samurdhi program started to be addressed through the introduction o f the Welfare Benefits Act (WBA) (in 2002) and the establishment o f the Welfare Benefit Board (WBB). The WBB has the mandate to select persons eligible to receive welfare benefit payments, on the basis o f transparent and prescribed selection criteria. The selection criteria considered involved a proxy- means test formula, based on a poverty predictor model developed usinghousehold data, to identify households whose "scores" fall below a pre-determined cut-off as eligible for benefits. A computerizeddatabase would be used to monitor the eligibility of households, entry and exit, as well as crosscheck for benefits received from multiple sources by the same household. A pilot survey covering 48,000 households in 114 GN divisions was 39 conducted in 2003 to test a proxy-means tested application process for selection o f beneficiaries of the Samurdhi program under the new Act. After completion o f the enrolment phase o f the pilot, the island-wide application and targeting process would be initiated, along with a widespread publicity campaign to generate awareness about the reformed system, which i s expected to be operational by the early part o f 2005. The role of the WBB would be limited to selection o f beneficiaries and defining the payment schedules, while the actual disbursement o f benefits under the various welfare schemes would be carried out by the relevant agencies. 103. While the Samurdhi program, being the largest transfer program for the poor in the country, represented the first phase o f reforms, the medium-term agenda would be to rationalize the selection o f beneficiaries of all transfer programs in the country, by bringing them under a set of criteria applied by the WBB. This would help minimize duplication and inefficiencies across programs, and enable each program to be better aligned to its stated goals and rationale. 104. Some key implications o f the implementation o f WBA are noted here. First, the WBA provides for transparency and accountability not only on the part o f the applicant who seeks benefits, but also on the part o f officials who sanction those benefits. Second, a number o f easily observable characteristics would be used indetermining a household's eligibility for the Samurdhi transfers rather than the current practice o f relying on subjective reports on income levels. Third, the WBA would explicitly define the criteria for entry and exit o f households in and out o f the program, and the status o f households would be tracked so that overlaps in the coverage o f the different welfare programs can be minimized. 105. The cut-off point for the new selection mechanism is proposed as such that the program would cover a proportion o f the population that i s roughly in accordance with the incidence o f poverty in Sri Lanka (as estimated by the Department o f Census and Statistics). The proposals for cut-off point and payment schemes that are being considered, informed by simulations using data, will likely lead to: (a) limiting the Samurdhi transfers to the bottom 25-27 percent o f the population, which will substantially increase the amounts o f transfers to target households; (b) better accuracy in identifying the poor; (c) making the transfers progressive, i.e. giving larger amounts to the poorer households. As a part o f the reform plan, the possibility o f separating the Samurdhi transfers from the compulsory savings, insurance and lottery schemes, will also be considered. POLICY RECOMMENDATIONSTO SUPPORT AND COMPLEMENT THE REFORMPROCESS 106. On the basis o f the policy issues highlighted previously and the review of expenditures-and inthe light of the ongoing reforms inthe sector-a number o fpolicy implications emerge. These should also be looked upon as steps that could complement, and strengthen, the reforms in the targeting and payment mechanisms initiated by the Welfare Benefits Board. 40 Strategic areas of policy 107. In addition to the issue of mis-targeting, which remains the main focus o f the ongoing reforms, the experience with Samurdhi program suggests that some thought be given to its conceptual focus. According to its declared goals, the Samurdhi program should be regarded more as a poverty alleviation program; this i s however belied by the fact that a very large share o f its budget goes to the transfer program, which in turn "subsidizes" the savings and credit components that are more aligned towards poverty alleviation. There i s a good case for the latter to be de-linked from the welfare component, operationally and financially. This i s likely to improve the efficiency and impact o f both types o f programs, on account that the objectives and the potential clientele are likely to be quite different from each other, and the incentives o f beneficiaries and program staff often work at cross-purposes when the two kinds o f programs are combined. 108. To the extent that such de-linking may lead to financial difficulties for the savings and credit components, alternate budget transfer mechanisms can be thought o f to provide subsidies directly to these programs. Such support would however have to be justified by a close examination o f the potential impact o f these programs, and can be limited to a period o f time deemed necessary for the programs to become financially viable on their own. Moreover, as far as the microfinance component i s concerned, it can be argued that in the long run, the government's role should be replaced by non- government and private sector institutions. Major policy and regulatory reforms envisaged to streamline microfinance in the country creates the potential for working towards a long-term goal o f minimizing the role o f Ministry o f Samurdhi in the delivery o fmicrofinance services. 109. On the budgetary front, the planned reforms in the targeting and payment o f Samurdhi benefits do not have any implication as far as the budget for transfers i s concerned, since all simulations that drive the current proposals hold the current budget of SLR 9 billion for transfers constant in real terms-recognizing that this would be the minimum budget required to provide a meaningful subsidy to the poorest 25-27 percent o f the population. At the same time, with the WBB taking over the selection o f beneficiaries, the administration o f the consumption grant can be handed back to the decentralized bodies to a large extent. This would obviate the need to employ a large cadre o f development officers, and result in considerable savings on the salaries bill and other administrative overheads (that amounted to 22 percent o f the total budget in 2003). The exact impact o f reforms on the staffing needs o f the Samurdhi Ministryis however not known yet, and would require careful examination once the fillreform takes shape. The implementation o f administrative and staffing reforms would also have to be integrated with the broader public sector reforms underway in the country, for instance, the implementation o fthe VRS plan. 110. There is scope for the various other associated activities implemented by various departments of the Samurdhi Ministry to be streamlined and rationalized with regards to functions of other ministries. These include activities in health, nutrition, education, training, and social security, when other ministries are specifically 41 dedicated to these functions. The budgetary implications o f eliminating some o f these functions would save expenditures at the Samurdhi Ministry, including administrative expenses. Although this would not amount to a significant saving, since the budgets on these activities are low to begin with, it could result inbetter efficiency emerging from a focus on the delivery o f services ina few key areas. 111. The Ministry of Social Welfare needs to carve out a role for itself as a primary agency for the design and delivery of social welfare in the country. To start with the payment to soldiers could be reallocated to the budgetso f the Defense Ministry. While they do represent a marginalized segment of society, responsibility for the provision o f care for disabled soldiers and their families rightfully rests with the defense institutions and the required finances allocated from its budgets. In the event o f a re- allocation it is not recommended at this stage that the budget o f the Ministry be reduced, and any such steps should be preceded by careful examination o f the needs o f the programs administered by this Ministry and their target groups. 112. It will be useful for the Ministry to undertake stocktaking of its many programs and activities, many of which overlap with other Ministries (e.g., health) and even involve duplication within its own departments. If the activities o f the Ministry are rationalized, the same budget canbe used for activities that directly address the creation o f safety nets for disadvantaged groups inneed, such as the provision o f care for the elderly. Sri Lanka's demographics point to a serious crisis inthe provisiono f care to its aging population. The provision o f social insurance schemes for the informal sector is another area in need o f attention. The International Labour Organisation is providing significant technical input into this area, and the support for such programs could be another focal area for the Ministry. 113. Further, the administrative costs of the Ministry have room for rationalization. While administrative costs as a share o f total budget are lower than that for the Samurdhi Ministry, the spending on personal emoluments is still highrelative to the total spending minus the transfer (13 percent in 2003). The expenses are partly justified due to the need for a high level o f skilled counselors and trainers (e.g., probation officers), but if this i s the case, the programs themselves should be expanded in size and coverage to justify the personnel cost. 114. The welfare reform's mediudlong-term vision involves the WBB coordinating the beneficiary selection o f all welfare programs involving transfer payments in the country, with the goal to minimize duplication, and achieve better targeting and budget planning. This would however require that the rationale, objectives and potential target group o f all transfer programs be examined carefully before the selection criteria for different programs can be devised and applied in an integrated manner. Inorder for this to occur, the ministries administering these programs would need to look across the entire landscape o f programs, to identify the areas o f duplication and gaps in the coverage o f existing programs, and rationalize their programs accordingly. For instance, to complement the primary transfer program, Samurdhi, which aims to target chronically poor households, the programs o f the Ministry o f Social Welfare can be positioned to 42 focus on more narrowly defined vulnerable groups, including those affected by short- term shocks like natural disasters. Issues of process: budgetingand audit 115. There are a number of areas where the budgetingand auditing process can be improved, to better align the strategic direction of different ministries and departments with budgetplanning.While many o fthese are not limited to the Welfare sector, but rather reflect endemic problems, they have explicit implications for this sector, and thus deserve a mention. 116. First, there is a need for greater coordinationbetween the Treasury and the line Ministriesand Departments.Frequently, the reclassification o f expenditures by the Treasury in the Budget Estimates i s not clear to the planners at the Ministries, which constrains effective budgeting and planning by the Ministries. Second, monitoring o f programs and activities can be enhanced beyond the largely financial and physical monitoring o f targets that presently takes place, to outcome monitoring that involves looking at what has happened once the output i s delivered and utilized by the beneficiary or recipient. With the significant reforms in the welfare sector that are planned, such monitoring o f outcomes assumes special importance, to inform how the programs can be adjusted and fine-tunedover time to achieve better impact. 117. Third, greater efforts are required to capture public expenditure at the decentralized level, especially those of the Provincial Councils. Although Constitutional requirements do not oblige Provincial Councils to report their expenditures in detail to the Treasury, some link to decentralized spending should be made in the Budget Estimates. This will lead to improvements inpublic accountability and facilitate current efforts towards greater devolution; it will also lead to better coordination between the strategic planning on Welfare at the federal and provincial levels and avoid duplication o f efforts between the central line ministries and Provincial Councils. 118. Fourth, both internal and external audit processes suffer from a lack of trained staff and facilities that limit the effective functioning of the audit process. The internal auditing by the audit units o f the Ministries could also be made more independentso that these units play a more stringentrole inoverlooking the expenditures of Ministries and Departments. Instruments such as the Audit and Management Committees could be enhanced by the compulsory attendance o f the Audit Superintendentfrom the Auditor Generals department, and the profile and legal authority of the Internal Audit and Investigation Division o f the Ministry o f Finance could be enhanced to allow more autonomy and legislative clout in the internal auditing process 43 VIII. IRRIGATION Public investment and donor support for the irrigation sector has diminished. Operation and maintenance (O&M) work is given inadequate priority and funding. A review of irrigation expenditure on major systems (> 80 hectares) over 2000-2003 reveals an inadequate provision for recurrent expenditure. A low allocation for recurrent expenditure could be one reason why forward planning/programming of irrigation activities are not possible. Expenditure on rehabilitation is largely driven by donors with insuficient beneficiary involvement. A considerable problem faced by the sector is the political pressure to maintain supply drivenfree services. SECTORPERFORMANCE 119. Irrigation, together with new seed and fertilizer technologies, has been the main driver of agricultural productivity in the post-independence period. Irrigation has been of particular importance for the rice sector, which has moved from producing 36 percent o f the total domestic rice requirement in 1948 to supplying nearly all the country's rice needs today. In addition to maintaining rice self-sufficiency, the agricultural sector aims to further raise productivity and diversify into high value alternative crops. 120. Poverty in Sri Lanka is closely linked with agriculture. A large proportion o f the poor are dependent on the agricultural sector for their incomes, with approximately 43 percent o f principal income earners in poor households employed in agriculture. Less than 2 percent growth in this sector has contributed to keeping poverty at relatively at high levels. More rapid agricultural productivity growth is fundamental for reducing poverty in Sri Lanka since nearly 90 percent o f the poor live in the rural sector. A JBIC study in 2002 highlighted the strong empirical evidence on the role o f irrigation infrastructure on poverty alleviation. The findings suggest that incidence, depth and severity of poverty are highest inareas without irrigation. 121. The rural population has shifted from farming as returns from agriculture have declined. About 80 percent o f the population was still based inrural areas in2000, but only 23 percent o f the mean earnings o f rural households came from agriculture. The share o f the workers employed in agriculture fell from 47 percent o f total employment in 1990 to 36 percent in 1999. Workers employed in irrigated agriculture account for 18-20 percent o f the total employed. At the same time, agriculture productivity per worker stagnated at around SLR 53,000 per annum (in constant 1996 price). Agriculture i s becoming increasingly divided between a small dynamic sector concentrating on higher returnfruits, vegetables and spices anda large, relatively stagnant sector producingpaddy and tree crops. Domestic (especially tourist) and export demand spurs sector growth, while the reduction o f import restrictions and relatively high cost restrict demand for the 44 stagnant sector. Much o f the irrigated land is used for paddy production, which has seen its share o f GDP fall from 28 percent in 1982-85 to 22 percent in 1996-2000.4 Figure 11. InvestmentsinSri Lanka by type of investment and their shares, 1950-97 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 1997 Source:IWMI, 2002. IRRIGATION SECTOREXPENDITURETRENDS 122. Public investment and donor support for the irrigation sector has diminished.The trends and compositiono fpublic irrigation expenditureinconstant 1995 prices i s given in Figure 11 and Table 13. Total public expenditure on imgation has fallen to under 10 percent o f its peak levels in the 1980s. Sri Lanka i s not alone in experiencing a stark reduction in public investment in the irrigation sector. Imgation investment fell from the 1980s onwards in other Asian countries due to fiscal pressures and foreign donor disenchantment with the low economic returns from investments inthe sector. Total investment in agriculture has been declining at a similar rate to irrigation expenditure. Even as spending falls, irrigation continues to dominate total agricultural expenditure. During the late 199Os, irrigation represented a half o f total public expenditure inagriculture (Table 14). L World Bank, Sri Lanka: Promoting Agricultural and Rural Non-farm Sector Growth (Report No. 25387), February26,2003. 45 Table 13. Irrigation investmentsin Sri Lankaby type of investment and their shares, 1950- 97 1/ Investment(USmillion in 1995prices) Investment(US million in 1995 prices) Share ("YO) Public investment Public investment New Rehabiliation O&M Private Total New Rehabiliation O&M Private Total construction investment21 construction investment 1950 48.2 1.8 49.9 96 4 100 1955 46.0 2.1 48.0 96 4 100 1960 30.0 6.2 36.3 83 17 100 1965 31.0 3.1 34.1 91 9 100 1970 49.8 3.9 53.7 93 7 100 1975 55.8 0.2 6.4 0.4 62.8 89 10 1 100 1980 151.4 11.3 6.8 0.6 169.9 89 7 4 100 1985 138.7 22.6 7.8 1.6 170.5 81 13 5 1 100 1990 33.8 10.1 5.3 4.5 53.5 63 19 10 8 100 1995 13.5 11.9 5.5 7.2 38.2 35 31 14 19 100 1997 12.1 18.0 5.1 8.6 43.5 28 41 11 19 100 Source: IWMI, 2002. Notes, I/Five-year averages centering on theyears shown. 2/ Investmentson agro-wells and irrigationpumps by farmers. Table 14. Sri Lanka: Percentage Share of Major Subsectors in Total Agricultural Expenditure, 1981-2000 Average Share (%) Period 1981-85 1986-90 1991-95 1996-00 Subsector Plantation 4.2 11.0 15.3 12.1 Imgation 79.6 62.6 49.9 50.5 Land Related 1.2 2.0 0.6 0.6 Crops 10.0 14.3 21.3 21.7 Others 5.0 10.1 12.9 15.1 Source: World Bank, 2003 (Abeysekera, forthcoming). 123. Expenditure on rehabilitationis largely driven by donors with insufficient beneficiaryinvolvement.New construction has declined substantially both inlevels and as a share o f total irrigation expenditure since the 1980s; a larger share o f the irrigation budget is now devoted to rehabilitation and operation and maintenance (O&M). However, expenditure on rehabilitation has decreased to less than half its maximum level inthe mid-1980s. The lack of a national policy, setting criteria and technical guidelines has by and large made system rehabilitation a donor driven exercise. Beneficiary contribution i s stipulated (at 10 percent to 20 percent o f tertiary improvement costs) in some donor projects, while inother donor and consolidated fund projects it i s not, leading to inconsistency and inequity o f approach. Beneficiary contribution entails consultation, leading to user-oriented improvements, quality control and a sense o f ownership by the beneficiaries. This i s vital ifturnover o f management is a condition o f rehabilitation (e.g. MRRP-SIRP,NIRPetc.). Withpractices not consistent across the sector, the result canbe two adjacent projects operating under different criteria, creating problems and confusion among the respective beneficiaries. 46 124. Operation and maintenance (O&M) work is given inadequate priority and funding. O&M expenditures have been falling since the mid-l980s, despite the slight increase in the irrigated area for this period. Nominal resources provided for O&M have declined by 30 percent since 2000. Inadequate funding has led to a rapid deterioration o f canal systems, which results in a poor quality o f water service to farmers and thereby contributes to low crop yields. Poor maintenance o f the large dams and their major structural components has created potential safety risks and in turn potential threats to public ~ a f e t y The' lack o f O&M resources also bring about the need for repeated and . ~ expensive rehabilitation programs. The fact that donors support rehabilitation rather than O&M has led to the practice whereby O&M investments are substituted by shorter donor funded soft loadgrant rehabilitation cycles in order to overcome system deterioration. In a monsoonal climate where O&M costs remain high and in an environment o f fiscal constraint, this practice has now almost become the norm. Figure 12. MajorIrrigation: Spread of Investmentsin2003 RecurrentExpenditure SLR 527 Million Personal Emoluments Related 91% CapitalExpenditure SLR 893 Million Measurement 8% s, Studies Improvement Rehabilitation /Structures 39% Machinery 5% *Head 310 -Project 02 & Head 853 -Project 1,2,3 & 4 L World Bank: Mahaweli Restructuring and Rehabilitation Project, Implementation Completion Report, May 2004 47 125. A review of irrigation expenditure on major systems (> 80 hectares) over 2000-2003 reveals an inadequate provision for recurrent expenditure (Figure 12). Recurrent expenditures for Imgation System Management and Construction (policy i s excluded) show that a major part o f the provision goes to support personnel emoluments with meager 10-15 percent provision for operational support. The result is invariably that recurrent costs are charged to capital provision. For meaningful performance in relation to investment on high value humanresources in the sector a more performance oriented ratio for operational support needs to be provided. While technical support in construction activities could be subsumed within capital provision, aspects such as institutional development work which could have a major impact on long term savings andspecializedservices vital for planningdesign, investigationetc. suffer. 126. A low allocation for recurrent expenditure could be one reason why forward planning/programming of irrigation activities are not possible and such services can only survive on provision made available through currently implemented activities from capital budgets. It would appear that a minimum provision o f 25 percent needs to be provided inthe recurrent budgetto enable proper utilization o f the humanresources made available. Unless such a development occurs development o f effective programmatic fonvard-looking planning for the sector will not be possible. Adequate provision for recurrent expenditure i s crucial to ensure functional effectiveness o f dedicated staff performing duties that cannot access funding from capital projects, such as monitoring and evaluation (M&E), project preparation, specialized technical services, etc. 127. Presently around 35 percent of capital votes go to support recurrent expenditure. Major costs such as vehicle fuel/ maintenance etc. were charged approximately 80 percent to Capital votes. The mixing o f recurrent and capital expenditure streams leads to misallocation and distortions in use o f resources. While it may be attractive to show large capital budgetsthe fact that really a third o f such budgets go for recurrent costs anyway while gravely affecting programs that do not have support of capital projects leads to a skewed and distorted evaluation o f performance. One reason for the low recurrent-to-capital ratio could be attributed to the lack o f adequate operational provision under recurrent votes. For example, o f the SLR 242 million, SLR 393 million and SLR 434 million in o f recurrent expenditure in 1999, 2000 and 2001 respectively, approximately 91 percent, 89 percent and 92 percent respectively went for personnel emoluments. Adequate levels o f recurrent operational costs need to be provided for balanced program implementation. SECTOR POLICY AND PLANNING 128. Sri Lanka has invested heavily in irrigation, but the results have been disappointing. About 90 percent o f irrigated land i s taken up with paddy as a result o f combined agricultural policies to promote rice production. This ties most farm households to low productivity activities, constraining the public sector from introducing charges needed to fund the maintenance o f the irrigation system. The subsequent lack o f fiscal funds continues to impede water delivery blocking farmers from diversification into higher value crops inmany areas o f the dry zone. The strategy must be to break this 48 vicious circle. The main challenges faced by the public sector in providing irrigation services are outlined inBox 2. 129. A considerable problem faced by the sector is the political pressure to maintain supply driven free services. First, this results an under-funding o f the sector. Second, the lack o f cost sharing o f services and improvements by beneficiaries result in agency driven infrastructure development and rehabilitation. Third, the free public good nature o f irrigation infrastructure attracts political leverage to be exerted to undertake local projects o f doubtful economic viability, sometimes led by local interests and nexus between local political authorities and agency staff. Lack o f proper technical investigation and economic and financial assessments result in cost over-runs and reduction o f benefits with opportunities for attention to more viable projects being reduced. Political patronage precludes a rational transformation o f irrigation agencies to provide services required under a new environment. Internal resistance to change, unless faced with severe financial crisis, i s also a factor. Crucially, the lack o f cost sharing encourages an inefficient usage o f scarce water resources as farmers do not have any incentive to conserve water inline with its marginal cost. Box 2. Main challenges for public sector in irrigation services provision 1) Transformthe public sector role from provider of supply-drivenfree goods and services for the peasantbased irrigated subsistence farming sector to supporter of commercially oriented viable farming enterprise development of small holder agriculturethroughmanagementandcost sharing. 2) Developmento f an institutional and operational framework within which irrigation institutions will regulate through quality control and essential intervention rather than providing resources and services that may be effectively outsourced. 3) Address the political dimension of provision of free public goods and services. The irrigation sector remains the only sector where aunit of economicinfrastructurecostingupto and over Rs. 1millionisprovidedfree of charge, together with subsequent follow-up services and improvements. These are given to persons without proveninvestment or other capabilityresultingin a supply driven expansion. Such a policy does not promote full employment, and invariably holdsthe recipient in a povertytrap andresults in investment with inadequate returnsfor the economy. 4) Through augmentation, improvements, technology adoption and better management transform irrigation systems established earlier to provide only supplementary irrigation to systems being capable of providing higherincomesthroughfull employment (250 mandays/ yr). 130. Water resources need to be managed in an integrated fashion within a river basin context supported by a coherent water sector policy and institutional framework. Competition among different sectors and users for water has increased dramatically resulting in disputes and conflicts. A recent study forecasts that, if the current rates o f highusage and low irrigation efficiency continue growing water scarcity would affect a number o fpredominantly poverty stricken rural districts by the year 2025. Therefore, an integrated water resources management approach i s indispensable to strike a sustainable supply and demand balance o f resources while complementing economic, social, and environmental development needs o f the country and recognizing the inherent political dimensions. This i s particularly true given that some major rivers are shared between the South and North-East areas which are presently inpolitical conflict. 49 POLICY RECOMMENDATIONS 131. Clear policy directions and criteria are required with respect to new development and expansion, especially where it may affect current availability of water to stakeholders in a basin and create greater irrigation and environmental problems. While rehabilitation has been shown to have better economic returns, reduction in O&M provision coupled with take up for rehabilitation o f those systems at high investment, where potential returns are low need re-examination. This calls for longer rehabilitation cycles which are now ever getting shorter with poor O&M. 132. Budgetary allocation criteria and guidelines need to be established. A focus on construction and sector ceilings on funding should not occur at the expense o f adequate levels o f O&M. High investment in "development" at the expense o f O&M i s self defeating. A fixed proportion o f the budget should be allocated for O&M, while some rationalization o f O&M provision between types o f systems needs to be established. Provision for Head Works O&M is inadequate in relation to functions and responsibilities both in Irrigation Department (ID)and Mahaweli Authority o f Sri Lanka (MASL). Similarly for O&M related institutional activities. Greater mileage from funds for rehabilitation & O&M with beneficiary contribution with linkages to farmers' organizations appears possible. In the context o f operations o f field canals through private sector, IDneeds to develop and investment strategy that includes private sector as a player. 133. Policy review and M&E of policy implementation is ancillary to expenditure and physical progress review and should be undertaken by the central planning units in the ministry and agencies. As irrigation is an input to agriculture and a vehicle for poverty reduction, these `efforts need harmonizing and integration. The Central Coordination Committee on Irrigation Management (CCCIM) could be the forum for integrating irrigation and agricultural policies and needs to be reconvened regularly. Policy studies offer opportunities for examination o f policy gaps while reinforcing capacity building. Surveys and case studies help provide the necessary feedback for impact assessment o fpolicy and implementation. 134. There i s a need to transform the agencies involved in irrigation and reassess their missiodfunctions in line with overall economic and structural changes and the future investment strategy for the sector. Both ID and Irrigation Management Division (IMD) can transfer some ground level activities and take up more upstream activities focusing on water resource management through multi-disciplinary approaches. There i s additionally a need for integration o f O & M items, improvements, and rehabilitation with segregation from new development activities to ensure greater clarity with respect to investment. Need to adhere strictly to construction timelines, as take up o f many new items in the context o f sector ceilings, result in most initiated itemized works programmed for 1 to 2 years sometimes taking 5 to 7 years to compete with reduced annual budgets resulting in cost overruns and reduction in benefits. Even anniversary programs, such as Centenary Irrigation Agriculture Development Project, seem to continue far beyond anticipated conclusion o f event. 50 135. While strategic plans, including six-year investment plans and pipe line projects were initiated by the Ministryo f Irrigation and Water Management, there is a mismatch and a lack of coherence in the implementationprocess, due to: (1) Investment ceilings and fimding constraints delay implementation o f on-going projects and reduce opportunities for take-up new initiatives. (2) Increased ground level political pressures promote take up o f "ad-hoc projects" and resources are directed to discrete activities rather than plannedprogrammatic approaches, leading to distortions o f intentions vis-a- vis goals. (3) New items funded through the consolidated fund invariably reflect very localized interests and though they should be channeled through the local authorities (Provincial Councils) by virtue o f falling within their mandate are taken up at national level on the basis that they are within the major system classification (>80 ha). While Provincial Councils are generally unwilling to support these projects financially. Consequently, the investigation, planning, and design processes for such "itemized work" are mostly short circuited to meet political priorities leading to, (a) Take up o f items even without proper preliminary investigation reports resulting in: complications to construction during implementation; escalation o f costs through changes to the Technical Evaluation Committee during construction; reduction o f benefits due to poor planning assumptions; delays due to unanticipatedsocial/ land issues; doubtful financial/ economic viability. (b) Investment budgets are diffused and lack focus leading to: most local fimding channeled to construction activities normally within purview o f local authorities; lack o f priority by other agencies affect realization o f benefits due to problems in settlement, land development, agricultural extension or credit and social infrastructure; long gestation periods due to many competing items within investment and institutional capacity affects costs and benefits; reduced interest and investment to support program areas such as O&M, joint management and turnover, policy studies, specialized services and monitoring, project planning and design. 136. The more feasible approach to reformingthe water sector in Sri Lanka may be a "bottomup" approach for policy and institutionalreforms as indicatedby the Mahaweli Restructuring and Rehabilitation Project (MRRP). A key lesson drawn from MRRP i s that a practical approach to water sector reform is to involve water sector agencies and water users inpilot demonstrations and activities that incorporate the major policy recommendations and instruments (e.g. entitlements, farmer managed systems, bulk water based water management). Such a method is likely to be more fi-uitfulrather than pushing a reform agenda on a purely academic/conceptual basis. Pilot interventions are likely to get more support from water users and agencies, particularly ifthe programs are action oriented and aimed at resolving real time and emerging water management problems and issues inriver basins. The idea would be to get farmers closely involved in designing pilot interventions, and create strong links between bottom up reforms and investmentto create the right incentives. 51 52 ANNEX March 23,2004 Draft National Budget Circular No: Department o f National Budget, Secretariat Colombo To: All Secretaries o fMinistries Chief Secretaries o fthe Provincial Councils Secretary to the Finance Commission Preparationofthe 2005 Budget andMediumTermBudgetEstimates in Consultationwith Ministries 1. IntroductionandPurposeof This Circular To build on budget reform efforts and to assure that the 2005 Budget and the estimates for the subsequent two years reflect: 0 the requirement o f the Fiscal Management (Responsibility) Act No. 3 o f 2003 to achieve a deficit not in excess o f 5 % o f GDPby 2006; 0 the integrationo f policy priorities and expenditure; 0 information on program performance, including the initial results o f performance budgeting and the identification o f performance indicators to measure progress in terms o fresults; and 0 a sectoral perspective to priority setting and elimination o f low priority and duplicative sectoral projects and activities; the following procedures will be adopted for developing the 2005 Budget and the estimates for the two subsequent years. 2. Sectoral ConsultativeCommitteeon Budgets(SCCBs) To strengthen the focus on priorities and performance over the medium term, the Ministry of Finance and the Planning Department will engage spending agencies in a dialogue from the outset o f the Budget process. As a means o f advancing this objective, Sectoral Consultative Committees on Budgets (SCCBs) for each sector (as shown inAnnexure I) are being established. Depending upon their responsibilities, individual Ministries may be represented on more than one SCCB. Please also be informed that Sectoral worlung groups already set up under the Public Expenditure Review Program supported by the World Bank, on Health, Irrigation, Education, and Social Protection will be absorbed into the appropriate SCCB. 53 3. Compositionof the Committees SCCBs will be comprised o f Secretaries to the Ministries and Heads o f Spending Agencies in each sector, officers from the Departments o f National Budget, National Planning, Public Enterprises, External Resources, and Management Services as appropriate. In order to assist with the integration o f policy priorities and the Budget, officers o f the National Planning Department will be appointed to the appropriate SCCBs. Provincial Councils and representatives of the Finance Commission will also be co-opted to the SCCBs that are reviewing devolved subjects. 4. Objectives ofthe SectoralConsultativeCommitteeson Budgets a) To review Sectoral priorities and re-define the role o f each Ministry within the Sector, if necessary; b) To identify issues pertaining to resource allocation within the Sectors, including duplication and overlapping o f activities among Ministries within each sector; c) To work with preliminary Sectoral budgetary limits and to ensure their most appropriate allocation. d) Prepare Policy Notes for each sector on the basis o f findings under (a), (b), and (c) above, including: 1) an analysis o f expenditure trends in the sector in general and sub-sectors in particular; ii) areviewofcadrelevelsanddistributionamongprojectstoassurethatstaffingis aligned with administrative responsibilities and priorities; iii) comments/views onexistingallocations ofresources, policypriorities, andthe current performance of the sector, drawing on a review o f results, as measured by performance indicators, including those required by Budget Circular 111, dated July 31,2003; iv) the identification o f specific policy options and the recommendedre-allocation o f resources to priority areas in2005 and over the medium term; and 4 the identification o f low priorities, including duplicated activities, that can be terminated to permit funding o f high priority sector policies, programs, and projects. Where sectors believe there i s a case for any increase inthe preliminary expenditure limit they must present a rigorous case for any such increase. For those sectors that have not engaged in developing a sector based strategy (strategic plan, master plan, a sector policy note as part o f the public expenditure review process) the SCCB should focus particularly on expenditure trends and identifying options for eliminating duplicated functions. 54 5. Preliminary2005 Expenditureand CadreLimits Preliminary expenditure and cadre limits for each sector are given at Annexure 11. The starting point for these limits was the spending agency estimates for the year 2005 as shown in the printed 2004 estimates. The 2005 expenditure estimates have been increased, and there has been some reallocation between objects, to accommodate the estimated costs o f the 2004 salary adjustments. The challenging macroeconomic environment i s illustrated in Annexure 111. This provides the background against which SCCBs will have to ensure that they work within their limits and recognize that any increase inthose limits will be at the expense o f other sectors. 6. Activitiesto be Performedby SCCBsby mid-June, 2004: SCCB meetings should initially focus on: a) The identification o f key issues confronting the sector; b) A review and re-affirmation or adjustment of sectoralpriorities; and c) The identification o f instances o f duplication and overlapping activities. Regular SCCB meetings to discuss progress and issues emerging from the work to date will be necessary. On the basis of SCCB deliberations, the following should be forwarded to the Treasury by mid-June: d) The Policy Note, containing a realistic assessment of the priority needs of the individual sectors in relation to the preliminary expenditure limits and the analysis referred to in 4 (4; e) Distribution of the 2005 preliminary expenditure and cadre limits. It is expected that SCCBs will fully engage their Ministers in the issues arising from SCCB deliberations. 7. FinalExpenditureCeilings The Treasury will submit a Budget Framework Paper on the macroeconomic and fiscal outlook, cross-cutting issues, such as pay and employment, and proposed expenditure ceilings (after review o f the recommended allocations o f the SCCBs) for the approval o f the Economic Policy Committee o f the Cabinet by the end o f July 2004. Upon approval o f same by the Economic Policy committee, such approved ceilings will be conveyed to the SCCBs for final agreement on ministry/sub-sector allocations. 8. Submissionof Estimatesfor 2005 and Projectionsfor 2006 and 2007 On the basis o f the final 2005 expenditure and cadre ceilings approved by the Economic Policy Committee, and the sub-sector/ministry allocations within the final sectoral ceilings agreed 55 by the SCCBs, Secretaries to Ministries should submit final 2005 estimates and expenditure projections for the years 2006 and 2007 to the Treasury by end o f July 2004. 9. Preliminary SCCB Meetings In order to discuss the logistics of SCCB operations, including arrangements for chairmanship of each SCCB, an initial meeting for each sector will be scheduled by the Director General o f National Budget for the last week o f April, 2004. Additional guidance on the role and functions o f SCCBs and on what might constitute a case for any variation to the preliminary limit to be entertained will also be provided at this meeting. Your active participationinthe Budget preparation process is solicited inorder to ensure the optimal allocation o f scarce resources. Secretary to the Treasury 56