Sri Lanka - Public Expenditure Review

CONFIDENTIAL Report No. 15940-CE SRI LANKA PUBLIC EXPENDITURE REVIEW November 5, 1996 Country Operations Division FILE C O Country Department 1 ' South Asia Region This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. SRI LANKA PUBLIC EXPENDITURE REVIEW Table of Contents Page No. EX E C U TIV E SU M M A R Y ............................................................................................. I: IN T R O D U C T ION........................................................................................................ A. Objectives and Scope of the Review......................................................................... B . M acroeconom ic O verview ....................................................................................... II: FISCAL PERFORMANCE AND SUSTAINABILITY...............................................3 A . B ackground .............................................................................................................3 B . Fiscal D eficit and Financing.....................................................................................3 C . Fiscal Sustainability ................................................................................................4 III: CIVIL SERVICE REFORM........................................................................................5 A . B ackground .............................................................................................................5 B. Developments in Employment and the Wage Bill......................................................7 C . System ic Issues ................................................................................................s..... 9 D. Re-launching Civil Service Reform ........................................................................ 12 IV: REFORMING THE PUBLIC SECTOR PENSION SYSTEM ................................ 17 A . B ackground ........................................................................................................... 18 B . The R eform A genda............................................................................................... 22 C . T ransition Issues ................................................................................................... 26 V: SOCIAL TRANSFER PROGRAMS ......................................................................... 27 A . Food Subsidies ...................................................................................................... 28 B . Sam urdhi Program....................................-...28....................................................... 28 C . O ther Transfer Program s ....................................................................................... 29 VI: PUBLIC ENTERPRISES...............................--.......................................................31 A . Role, Size and Com position ................................................................................... 31 B . Corporations and Com panies ................................................................................. 32 C . P rivatization................................................ ...................................................... .35 D . Statutory B oards .................................................................................................. 36 This report is based on the findings of a World Bank mission which visited Sri Lanka in April-May, 1996. The mission comprised of Eric Bell (Fiscal Devolution and task manager, SAICO), Maurice Gervais, David Steedman (Civil Service Reform, ASTPH), Christina Harder (Social Transfers, Pensions, SAlCO) and Sona Varma (Fiscal Sustainability and Public Enterprise, SAICO). Contributions were received from David Lindeman and David RaJnes (on Pensions, PSP). The ADB and OECF also participated in the mission. Peer reviewers were Barbara Nunberg (PSP), Jeffrey Hammer (PRD) and Sarath Rajapatirana (OPRPG). Mieko Nishimzu is the Director and Lorene Yap is Division Chief. The report was processed by Annisa Cline-Thomas. Page No. V II: FISC A L D E V O L U T IO N ............................................................................................37 A . B ackg rou n d ...........................................................................................................3 7 B. Political Autonomy and Accountability ..................................................................38 C. Expenditure and Revenue Assignments ..................................................................38 D . Intergovernm ental Transfers ..................................................................................40 E . B orrow ing R ights ..................................................................................................4 1 F . O ther Issues ..........................................................................................................4 1 ANNEXES BUDGET DATA Annex 1 Table 1: Summary of Central Government Operations, 1990-96............................43 Table 2: Central Government Revenues by Component, 1990-96...........................44 Table 3: Classification of Expenditures, 1990-96 ..................................................45 A nnex 2 Fiscal Sustainabilty ...............................................................................................46 Annex 3 Civil Service Reform: Background Note.................................................................50 Annex 4 Social Transfer Programs in 1995, 1996 and 1997.................................................59 Annex 5 Sri Lanka: Background Information on Public Enterprises......................................62 LIST OF TABLES 1. K ey Econom ic Indicators........................................... .......... -...............................2 2. Central Government Revenues and Expenditures......................................................4 3. Modifications in the Pension Formula .................................................................... 20 4. Increasing Benefits for an Average Civil Servant....................................................21 5. Cross-country comparison of Civil Servants Pension Systems ............................. 21A 6. Social Transfer Programs in 1995-1998............................................................31 7. Profitability of Public Enterprises..........................................................................33 8. Public Enterprise Contributions to Government Revenues.......................................33 9. Impact of Public Enterprises on the Government's Capital Budget..........................34 LIST OF BOXES 1. Civil Service System s in East Asia......................................................................... 13 2. The Teacher Training and Re-deployment Project (TTRDP)................................... 14 3. Sri Lanka's Public Sector Pension System (PSPS) and its Components .................. 18 4. Country Examples of Successful Reforms of Pension System.................................23 5. The "Three Pillar" Pension System ......................yst.........................................em525 6 Other Existing Pension Schemes in Sri Lanka ........................................................26 7. Revenue-sharing: The Experience of Argentina .....................................................40 8. Borrowing by Local Authorities in Certain Countries .............................................41 LIST OF FIGURES lA. Total Civil Service Employment 1985-96.............................................. IB. Employment by Level of Government 1990-96................. 2A . Total W ages and Pensions, 1987-96..................................9...................................... 9 2B Wages/Pensions Wrt. Other Goods and Services (G&S)...........................................9 3. N ew Retirees PSPS, 1990-95.........................-...9..... -. --.-22.... ...............................22 4. Pension B ill, 1990-95 .........................90-95..-.....-...--......--.--22-..-............................... 22 B IB L IO G R A PHY...................................O..A....64 ......................... ..............................64 ABBREVIATIONS AND A CCRONYMS ARC Administrative Reform Committee ETF Employees Trust Fund EPF Employees Provident Fund GDP Gross Domestic Product PSPF Public Service Provident Fund PSPS Public Sector Pension System PSC Public Service Commission SCC Salary and Cadre Comission PMB Paddy Marketing Board CWE Co-Operative Wholesale Establishment CTB Central Transport Board COLA Cost of I iving allowances EXECUTIVE SUMMARY Introduction 1. Sri Lanka has achieved remarkable success in human development, and the country's level of social development is well ahead of other countries at similar income levels. That success has however not been matched on the macroeconomic management side. Economic policies have been implemented in a stop-and-go mode which has had a dampening effect on growth and the ability to utilize fully the country's human resource base. As a result, Sri Lanka's long-term per capita growth, which has been good compared to other low income countries, has remained well below potential and lagged behind its East-Asian neighbors that it tries to emulate. Tragically, this relative lack of success has been marred by an ethnic conflict that has exacted a high cost in human lives and suffering and indirectly sapped growth prospects. 2. New opportunities have appeared with the 1994 change in the political scene and the Government's current efforts at ending the civil war. Peace would set the stage for a sustained economic boom if complemented by appropriate development policies. But in a first stage, the Sri Lankan government would have to reestablish macroeconomic stability not only to stem the country's economic decline but also in preparation for the major challenges that are emerging regarding rehabilitation of the Northeastern part of the country. 3. The Government of Sri Lanka (GOSL) is currently engaged in internal discussions on how to revitalize the economy and expects that economic program will be supported by the IMF under the ESAF. These discussions have highlighted the need for Sri Lanka to effect a strong upfront fiscal adjustment as part of a credible program of macroeconomic adjustment. This report has been prepared in response to the Government's request for assistance in identifying policies and reforms that could reduce underlying budgetary expenditures quickly while strengthening the budgetary position. Fiscal Sustainability 4. Sri Lanka's fiscal position deteriorated significantly over recent years. Growth is projected to decline sharply in 1996 and inflation has risen to 14 percent. Financial policies, especially fiscal management, failed to respond to the exogenous shocks that have beset the economy, such as the drought and the war escalation. The overall budget deficit exceeded 10 percent of GDP each year during 1994-96 and monetary policy was left to accommodate the excesses. The main reason for the fiscal deterioration was gradual across-the-board slippages in current spending at a time when both debt service payments and defense expenditures were high. 5. The weakening in fiscal policy did not result in an increase in the debt/GDP ratio over recent years mainly because of the high economic growth rates and the high level of external financing that Sri Lanka has been able to get on concessional terms as well as the high level of low- cost domestic financing obtained from large savings institutions. Fiscal policy has, however, had serious adverse effects such as macroeconomic instability, crowding out of the private sector, and delays in implementing structural reforms. As a consequence, the long-term growth trend of the economy will ultimately suffer. The GOSL has established clear economic objectives for the next five years for which fiscal policies have to be formulated. Sustainability analysis based on both debt dynamics and broad macroeconomic considerations indicate that Sri Lanka would need at a 11 minimum to eliminate its primary fiscal deficit to achieve such objectives. In these circumstances, further delays in introducing reforms would have dire consequences on economic growth and also on the poor who are the least able to protect themselves against inflation. 6. Sri Lanka's budgetary revenues, at 20 percent of GDP, are quite high for countries of its level of development and offer little scope for significant enhancements in the short term. As a result, the burden of fiscal adjustment will fall on expenditures, namely, civil service, pensions, transfers to public enterprises and social transfers, given the already low level of capital spending, pressures on defence expenditures and the incompressible nature of interest payments. These are the four areas which have been the focus of the Public Expenditure Review (PER), and various avenues for short-term fiscal savings have been identified in each of these sectors. Civil Service Reform 7. Sri Lanka's civil service has several fundamental weaknesses which repeated reforms have not succeeded in eliminating. It remains overstaffed and fiscally unaffordable, the incentive structure is not conducive to good performance with ample evidence of low efficiency, and it has not adapted to the new liberal economic environment. Since 1990, a considerable amount of job creation has occurred and there have been substantial increases in salaries and pensions. These developments are the result of a number of political and economic factors that are deeply rooted in the Sri Lankan society, namely the long-standing role of Government in the labor market, weaknesses in governance and inadequacies in the management of public resources. 8. The GOSL faces a pressing need to control and contain the growth of the civil service and pension bills. But the institutional capacity to undertake an ambitious reform program has yet to be created. In this context, the priority is to implement fiscal containment measures while tooling up for medium-term reforms. Meanwhile, it is crucial to introduce administrative improvements to the management of positions, staff and costs, including: (a) establishing more stringent budget preparation processes to tighten the wage bill; (b) setting in place a monitoring system for movements of staff; (c) avoiding general salary increases which tend to distort the salary scale; and (d) imposing a moratorium on further institutional proliferation. 9. A more comprehensive reform over the medium term would acquire credibility only if the government implements the above policies. Sustained public administration and civil service reforms require strong political commitment and a high level of planning. In the meantime, it would be best to move quickly to a transition period that would be driven by hard budget constraints, to streamline government functions and reconfigure staff profile. This would involve: (i) redefining the functions of government accompanied by a program of administrative audits; (ii) preparing a program of institutional modernization; Ill (iii) implementing a program of staff "right-sizing". A useful tool in this regard is ratio analysis, in that it permits standards to be applied throughout government. Under this scenario, a more detailed review of ministerial functions would only be done later; (iv) adopting a business plan for the railway and postal departments; and (v) creating a capacity within government to re-deploy redundant staff. 10. It is not possible at this early stage to predict the nature of the plan for civil service renewal, as the program will be influenced by several factors that are still at an evolutionary stage, such as the implementation of the Teacher Training and Re-deployment Project (TTRP) and the new framework of cooperation between the Center and Provinces. It is clear that the second stage of reform would also need to focus on establishing a robust compensation system. It is recommended that Sri Lanka civil service move toward a compensation system linked to productivity and market-determined wages rather than tenure. In moving toward this long-term objective, a step-by-step pay reform could be undertaken that would reestablish relativities among various groups; reduce the number of pay scales and integrate most of the existing allowances in basic pay. Pension Reform 11. Sri Lanka's Public Sector Pension System (PSPS) is a defined benefit system that operates without actuarial analysis and requires a marginal contribution from civil servants. It relies heavily on government budgetary support which has become increasingly burdensome over the years. 12. Modifications to the system since 1985 have resulted in a generous pension scheme for civil servants. This generosity is reflected in the possibility for retirement after only ten years of service; a replacement rate not linked to years of service; a large lumpsum benefit; a pension base linked to the last salary drawn; a surviving spouse eligibility for 100 percent of pension, etc. The latest modification to the system came with Circular 44/90, which was meant as a temporary measure to support early departures in the civil service in 1990. This scheme however remains in force. 13. At this stage, Sri Lanka has two options for the reform of its public sector pension system. The first option consists in proceeding as fast as possible with measures aimed at reducing the deficit on pension operations, namely by increasing the rate of contribution and reducing benefits, leaving a clearly identifiable budgetary cost for this operation. The second option consists of engineering a comprehensive reform focused on long-term sustainability of the system, but which would have a major impact on the budget. However, the design of a detailed reform agenda is a major undertaking that is likely to be lengthy. In addition to in-depth economic and financial studies, the reform would require political consensus on major issues such as grandfathering, harmonization with the private sector pension system, and the liberalization of investment opportunities for pension funds. In the meantime, several short-term measures could be implemented immediately to reduce the fiscal burden without undermining the ground for successful medium-term reforms. These would include: IV (a) withdrawal of Circular 44/90; (b) reducing the lump sum benefit from 24 to 12 months of pensionable base; (c) scaling back the survivor benefit from 100 percent of worker's pension to 50-70 percent; (d) reducing the replacement rate for retirees with less than the minimum number of years of service; and (e) relinking the pensionable base to an average of the salary of the previous 5-7 years. 14. Long-term viability of the pension system would require a major overhaul toward the "three pillar" system recommended by the World Bank. Sri Lanka already has a publicly-finded pay-as-you-go pillar and a mandated capitalized pillar that is supported by a means-tested back-up for those with limited labor participation. The two large provident funds are good schemes that could be extended across the country; and the private pension funds covering farmers and fishermen have some of the features that are being recommended to Sri Lanka. The main recommendations for strengthening the system on a long-term basis are: (a) unifying the treatment of government and non-government employees by merging the PSPS into the Provident Funds; (b) offering pension annuity based on actuarially sound calculations; and (c) enhancing the remuneration of pension funds through the use of private management. 15. Various transition solutions could be envisaged to move the system toward a self- sustaining position. The PSPS could be closed to new entrants who would be enrolled in the EPF; but a more ambitious scheme would be to move all employees below a given age into the EPF. An intermediate strategy would be to cover new entrants and to allow (not require) already employed civil servants to the EPF. Social Transfers 16. Sri Lanka's impressive record in social development reflects the government's long- standing commitment of public expenditures to social sectors. Despite various consolidations in recent years, namely through the launching of the Samurdhi, there remains considerable scope for reducing the cost of the overall social welfare program without jeopardizing the objective of assisting the poor. The main recommendations include: (a) the elimination of the wheat and fertilizer subsidies; (b) strengthening the Samurdhi program, namely through containment of the number of mobilizers, maintaining mobilizers on one-year employment contracts, freezing budgetary allocations in 1997, and introducing time limits for transfers to beneficiaries; (c) eliminating the Janasaviya interest payments; V (d) consolidating the three schemes for the Northeast--food stamps, kerosene stamps, and emergency assistance--under one single program until the Samurdhi can be introduced in this region; and (e) integrating the school uniforms and textbooks schemes into the Samurdhi. Public Enterprises 17. The review focused on the financial operations of some of Sri Lanka's largest and most important corporations and companies. The consolidated balance sheet of 12 of these enterprises indicate that they are in relatively good financial health on their current operations and most of them are net revenue earners for the Treasury. However, several of these companies have major weaknesses because of government's policies and most of them provide a level of service which is quite low. Many of them have large capital expenditure needs that the government is unable to meet. In addition to capital transfers, several companies benefit from a fairly large volume of guarantees from the Treasury on their bank borrowing. In the short term, fiscal savings can be generated by: (a) rationalizing the poorly performing public corporations such as the PMB, the CWE and the CTB; (b) reviewing and closing down statutory boards that have lost their raison d'etre; and (c) intensifying the privatization program. Fiscal Impact 18. It is difficult to quantify the impact of the various measures being proposed. It is clear, however, that implementation of measures like the elinination of the wheat subsidy; the partial embargo on civil service recruitment and the enforcement of more stringent budgetary preparation processes; the closures of certain statutory boards; the reduction in pension benefits; and the containment of expenditures on the Samurdhi, combined could generate fiscal savings equivalent to 1.5 percent of GDP in 1997. Sustained implementation of these measures could bring additional savings each year thereafter, reaching 2.5 percent of GDP by 1999. The impact of the medium- term program for the civil service and the pension fund reform is not quantifiable at this stage because major uncertainties remain regarding their most important sub-components. They are likely to be cost-neutral if introduced simultaneously with the short-term measures recommended. Nonetheless, cross-country experience indicates that financial costs of similar reform programs can be recovered within 3 years. In this context, it is clear that Sri Lanka would also need a reduction in other expenditure items, such as defence, in order to achieve the medium-term fiscal goal. Fiscal Devolution 19. Sri Lanka's first experience with fiscal devolution since 1989 has been inconclusive for lack of real political motivation and technical difficulties at the local government level. A second devolution package is currently being designed for introduction in 1997. This review has examined various issues to be considered in this exercise, focusing primarily on its possible adverse effects on macroeconomic management. The main recommendations include: vi (a) introducing mechanisms to improve accountability and transparency of fiscal operations at the provincial level; (b) strengthening expenditure management across the economy; (c) linking the system of central transfers to revenue collection at the provincial level; (d) establishing a fiscal and financial framework for the evaluation of alternative schemes for decentralization; and (e) not allowing borrowing rights to provincial authorities. 20. The Bank stands ready to provide assistance to the Sri Lankan authorities on the design of these policies. In a first stage, this assistance could aim at building institutional capacity in the Ministry of Finance to formulate devolution strategies within a framework that gives greater recognition to the macroeconomic implications of various alternatives. Introduction A. Objectives and Scope of the Review 1. As indicated in IDA's Country Assistance Strategy', the most urgent task is to assist the Government of Sri Lanka (GOSL) in achieving fiscal adjustment. This is a precondition for a full resumption of the structural reform program where a rationalization and modernization of the public sector stands high on the agenda. This Public Expenditure Review (PER) provides a first round of inputs in pursuit of this objective. It takes particular relevance in the context of the discussions that the GOSL is conducting currently with the IMF on the formulation of a macroeconomic adjustment and structural reform program that could be supported under the ESAF. These discussions emphasize the need to achieve a convincing up front fiscal adjustment. The scope and focus of this report responds to the Government's request and is primarily aimed at identifying measures that could be easily implemented and quickly strengthen the fiscal situation while laying the ground for long-term strengthening of the economy. 2. The last in-depth review of public expenditures took place in December 19932 and annual IDA/government Public Investment Reviews have been carried out to facilitate discussion of the quality and composition of public investments. More recently, a report3 has been prepared jointly by the IDA and the Government elaborating on policies to increase Sri Lanka's growth rate over the medium term. The aim was to help Sri Lanka realize its dream ofjoining its most prosperous neighbors in South East Asia. The study indicated that expenditures need to be reduced by some 4-5 percent of GDP together with a change in their composition towards physical and social infrastructure. Care needs to be taken that expenditure adjustments do not compromise social and growth objectives by protecting capital and key recurrent spending. 3. The review focuses on four areas of budgetary expenditure: civil service, pensions, social transfers, public enterprises and fiscal devolution. It has not addressed the question of expenditures on defence and debt service, which are less manageable in the short term. On the other hand, expenditures on civil service, pensions, social transfers and public enterprises have taken increasing shares of total expenditures over recent years and are good candidates for a permanent strengthening of the budgetary position. This study also reviews briefly the Government's policies and intentions with regard to devolution of authority to Provinces. B. Macroeconomic Overview 4. The Sri Lankan economy did well after 1989, spurred by renewed adjustment efforts and private sector confidence. Efforts were made to reduce macroeconomic imbalances and create incentives for the private sector through privatization, trade and external payments reforms. The rebound was broad-based, with strong growth in industry and an unprecedented increase in domestic and foreign direct private investment. Real GDP rose by an average 5.5 percent per annum during 1990-94, compared to 2.2 percent per annum over the preceding three years (Table 1). Average domestic savings amounted to 14 percent of GDP for the same period, and underlying inflation remained below 10 percent per annum. Gross official reserves rose to more than 5 months of imports in 1993-94, fueled by strong export performance-- especially in garments - an increase in remittances, and large inflows of foreign direct investment. Report No. 15633-CE; May 21, 1996. Report No. 12337-CE, December 22, 1993 Sri Lanka in the Year 2000-An Agenda for Action, Report No. 15455-CE; March 14, 1996. 2 Table 1: Key Economic Indicators (In percent) Prel. Est. \g 1960-94 1990 1991 1992 1993 1994 1995 1996 Real GDP growth rate (per capita) 2.8 4.8 3.3 3.0 5.6 4.4 4.1 2.1 Consumer price index \a 8.2 21.5 12.2 11.4 11.7 8.4 7.7 14.0 Gross domestic investment/GDP 20.1 21.9 22.6 24.6 27.0 26.0 25.1 24.7 Private investment/GDP \b 13.6 \e 13.5 13.0 17.2 17.7 19.5 17.7 17.6 Budget deficit/GDP \c 9.6 9.8 11.7 7.9 8.6 10.5 10.2 10.6 Primary balance/GDP (including grants) - -1.2 -2.0 -0.7 -1.0 -2.5 -3.0 -3.2 External current acc. deficit/GDP \d 7.2 \f 5.4 7.5 6.1 5.3 7.8 6.8 6.7 Gross national savings/GDP 12.9 16.5 15.0 18.5 20.3 19.1 19.7 18.0 Foreign direct investment, net (US$m) - 33.0 63.7 121.3 187.0 158.0 54.0 183.0. Portfolio investment, net (USSm) - 9.0 32.1 25.7 67.0 28.0 -1.7 N.A. a/ Period average Colombo consumer price index. b/ Includes investments by public corporations not financed through the budget c/ Excluding grants and privatization proceeds. d/ Excluding official transfers. e/ Period 1980-94. f/ Period 1970-74. g/ IMFStaff Report on the 1996Article IV Consultations (projections include implementation of adjustment measures but lower defense expenditures than in tables of Annex I). Note: Not available. Source: Central Bank ofSri Lanka, MSF and Bank staff estimates. 1/ Includes investment by public corporation not financed through the budget. 5. This good economic performance slackened starting in 1995 due to both a deterioration in the security situation, which increased defence expenditures to 6.4 percent of GDP, and some degree of reform exhaustion, especially as a result of the 1994 election campaign. The promises of that election campaign translated into budgetary expenditure increases: civil service employment increased and the flour and fertilizer subsidies were reintroduced. These slippages were compounded by the legacy of a large, incompressible interest bill. In the event, the fiscal deficit widened and economic growth slowed down to 5.5 percent. (4.1 percent per capita). 6. A further deterioration is expected in 1996 as Sri Lanka is adversely affected by several factors. Inclement weather conditions reduced agricultural production and water supplies for electricity generation, while the war escalated. Real GDP is now projected to grow by 3.5 percent (2.1 percent per capita) only. Investments are expected to decline to 24.7 percent of GDP as investor confidence is set back by the macroeconomic deterioration, heightened security concerns and strike activity. The country was paralyzed for two and a half days in May by a general strike at the Central Electricity Board in protest against, both, declining purchasing power and the privatization program. Foreign investment inflows plummeted to one quarter of the average level in the previous two years and the Colombo Stock Exchange dropped to historical lows. Inflation has already risen to over 15 percent. The external current account is projected at 6.7 percent of GDP in 1996. Exports have slowed down in real terms, which have been more than offset by a larger slowdown in imports in line with the decline in investments. The balance of payments is projected to register an overall deficit of US$ 12 million in 1996, the first in many years, and foreign reserves are expected to decline to 4.2 months of imports. 7. The economic decline is to a great extent due to the continued weakening in fiscal policies. The overall fiscal deficit (excluding foreign grants and privatization receipts) is likely to exceed 10.5 percent of GDP in 1996, a double digit number three years in a row, again due to expenditure slippages in both defence and other expenditures, while monetary policy continues to accommodate the need to finance the high budget deficit. 8. Considerable progress had been made in reducing structural distortions in Sri Lanka during the early 1990's (namely tariff reforms and a reduction in the size of the public enterprise sector) which has injected great resilience to the economy. Since late 1994, however, progress has stalled: the Goods and Services Tax (GST) was not introduced, further tariffs reductions did not take place, and financial sector reform is yet to come. On the positive side, the privatization program has been intensified and considerable success has been achieved. The danger now is that further deterioration in economic management could compound the difficulties associated with the security situation. 9. The weakening of fiscal policies is a major concern in Sri Lanka, especially in the light of the economic objectives of the Government and future developments. First, the Government has repeatedly expressed the ambition of joining the South East Asian countries in growth performance. Second, some rehabilitation/poverty alleviation programs will ultimately be needed in the Northeastern part of the country in order to establish peace on a permanent basis. These type of programs are known to exacerbate a country's economic difficulties should the macroeconomic framework be unprepared for sudden inflows of external assistance. These factors clearly indicate the critical juncture at which the Sri Lanka economy stands, and the need for urgent action on both the fiscal and structural fronts. IL Fiscal Performance and Sustainability A. Background 10. Sri Lanka has endured persistent high fiscal deficits over the past two decades. These deficits have led to rising public debt which has grown from 50 percent of GDP in 1974 to 95 percent twenty years later. By comparison with other countries, the Government of Sri Lanka is highly indebted. After some policy correction in 1990-93, fiscal performance deteriorated strongly and the overall budget deficit widened to more than 10 percent each year over 1994-1996. This deterioration has taken place in the context of a more difficult economic environment, namely increased civil strife and a drought and power crisis. B. The Fiscal Deficit and Financing 11. The primary budget deficit (after grants) increased from 1.0 percent to 3.5 percent of GDP between 1993 and 1996 (Annex I). Total revenues, which had declined somewhat in 1993 and 1994, due to the granting of waivers and exemptions on various taxes and duties, recovered in 1995 on account of better than expected indirect taxes and dividends from state enterprises; but a new deterioration is expected in 1996. The more substantial fiscal problem is the rapid increase in current expenditures, which rose by almost 3 percent of GDP between 1993 and 1996 (Table 2). Expenditure slippages occurred mainly in defence and welfare payments and subsidies (on wheat and fertilizer). Spending on civil service wages and pensions remained high; together with defence, they consume 56 percent of total revenues. Interest payments also remained high, exceeding 6 percent of GDP over 1991-95, and compounded the difficulties associated with the widening primary budget deficit. (Annex 1, Tables 1-3). 4 Table 2: Central Government Revenues and Expenditures 1985- 1989/a 1990 1991 1992 1993 1994 1995* 1996 /f Total Revenue 20.9 21.1 20.4 20.3 19.7 19.0 20.5 19.8 Total Expenditure and net lending 33.3 31.0 32.3 28.3 28.3 29.5 30.8 30.4 Current Expenditures 20.5 22.3 22.5 21.2 20.5 22.0 23.3 23.3 olw: Wages and salaries/b 3.9 4.1 3.6 3.1 3.3 3.3 3.5 3.3 Pensions 1.7 1.4 2.4 1.9 2.1 2.3 2.3 2.3 Defense 2.4 2.0 2.8 3.7 4.1 4.7 6.4 6.3 Interest payments 5.2 6.4 5.9 6.1 6.0 6.6 5.8 5.9 Social Transfer programs 2.7 2.7 2.1 1.9 1.8 2.2 3.3 5.5 Capital expenditures /d 12.8 8.8 9.8 7.1 7.9 7.5 6.3 5.4 Overall budget deficit (exc. grants)/e -12.4 -9.9 -11.9 -8.0 -8.6 -10.5 -10.3 -10.6 Primary Deficit before grants -7.2 -3.5 -6.0 -1.9 -2.6 -3.9 -4.5 -4.6 Grants 2.4 2.1 2.0 2.0 1.6 1.4 1.3 1.2 Primary Deficit After Grants -4.8 -1.4 -4.0 0.1 -1.0 -2.5 -3.2 -3.2 Financing in domestic borrowing 6.4 6.3 4.1 3.0 4.9 6.5 5.6 7.8 * Estimate a/ Annual average. h/ These nunbers include expenditure on defence until 1992 only. c/ Including wheat and fertilizer subsidy. d/ Including net-kending and excluding privatizaion proceeds cl excluding privatization proceeds. (f) Projected out-turn for the year based on IMF estimates made in July 1996. 12. From the financing point of view, these large deficits took place because of the availability of concessional foreign loans as well as the possibility for the Government to obtain high levels of domestic financing from pension funds and the National Savings Bank (NSB). In 1995, external financing of the budget deficit amounted to 3 percent of GDP and were extended by bilateral donors and international institutions on IDA terms. Nonbank financing was about 4.5 of GDP from three sources essentially. These included: (i) the Employees Provident Fund (EPF) which is the single largest captive source of funds for the government budget, with more than 98 percent of its total assets (Rs 100 billion, 15 percent of GDP) invested in Treasury Bills (TBs) and other government securities; (ii) the Employees Trust Fund (ETF), which because of its strong risk aversion, voluntarily places 65-70 percent of its assets (Rs 13 billion, or 2 percent of GDP) in TBs; and (iii) the NSB, which mobilizes household savings and term deposits across the country. It places more than 80 percent of its assets (Rs 58 billion, or 9 percent of GDP) in government securities. C. Fiscal Sustainability4 13. The intertemporal budget constraint equation (solvency test) has been applied on Sri Lanka's economic aggregates. On the basis of this test, it appears that Sri Lanka's fiscal stance over 1991-95 was sustainable due to the relatively high rates of growth enjoyed by the country and the financing facilities mentioned in the above paragraph. However, these factors cannot be taken for granted in the future (IMF, 1995). In particular, the balance of payments pressures that have emerged in 1996 will ultimately create pressures for a reversal of the liberal policies introduced earlier, thus endangering further growth prospects. 14. The solvency test has also been applied to the Sri Lankan fiscal situation on an ex ante basis, that is in the context of the overall macroeconomic framework in which Sri Lanka is likely to operate in the next few years and the objectives that the Government has been assigned, as described in the 1995 and 1996 Budget speeches. The main aim is to achieve 7-8 percent economic growth with low inflation (5-6 percent). The results are much less favorable. The mam conclusion is that, on the basis of an annual growth A more detailed description offiscal sustainibilty analysis is provided in Annex II. 5 rate of 6-7 percent and a rise in real interest rates on government debt toward the marginal cost of funds in the country, Sri Lanka would need to move quite quickly toward a surplus of at least 1 percent of GDP in the budget primary balance. More importantly the primary budget surplus is needed just to stabilize the debt/GDP ratio which may not be ambitious enough for a heavily indebted country like Sri Lanka. Under the circumstances, Sri Lanka would need to aim at a primary budget surplus higher than 1 percent to reduce its debt/GDP ratio. More detailed calculations made by the IMF (IMF, 1996) and taking into consideration the projected external constraint confirm the strong, positive impact of fiscal consolidation which, in particular, could reduce the debt burden to 70 percent of GDP by the year 2001 (Annex 2). 15. An important corollary of this conclusion is that it is essential that the authorities embark urgently on a durable program of fiscal consolidation. As the evidence in 1995-96 indicates, interest rates and capital inflows into Sri Lanka are highly vulnerable to political and economic developments, especially market sentiments about the state of public finances. Fiscal consolidation would also facilitate implementation of a rehabilitation program for the Northeast later. 16. The brunt of fiscal retrenchment lies on the expenditure side. Tax revenues themselves, at more than 20 percent of GDP, are already reasonably high for the country's per capita income and offer little scope for significant enhancements in the short term. Over the long term, however, revenue buoyancy and efficiency could be improved somewhat (for example, by including civil servants in the tax net and introducing the GST). As indicated in the CAS, there are several opportunities to restructure current expenditures: restraining the wage and pension bill and reducing transfers to households and public entities. These measures would produce medium-term gains in efficiency and considerable fiscal savings could come from containing and reducing entitlements and subsidies. Government has been a growing source of employment and, as a by-product, has reinforced young people's desire to wait for such jobs. 17. The sustainability of fiscal consolidation also depends on the participation of Provinces in the adjustment process. By and large, Provincial financial management have had a mixed record in the past, but a number of underlying problems exist for the future which the ongoing devolution package needs to address in advance (See Chapter VII). III. Civil Service Reform 18. Sri Lanka's civil service has several fundamental weaknesses, which a decade of attempted reforms has not been able to overcome. It remains over-staffed and fiscally unaffordable; the incentive structure is not conducive to good performance, and it has not adapted to the new liberal economic environment. A. Background 19. A persistent decline in the effectiveness of the public sector began in the early 1970's, abetted by the growth in its size and scope since independence and by widespread political interference in established hiring and personnel management procedures. This decline had reached serious proportions by July 1986 when the Administrative Reform Committee (ARC), set up by the Government, began a review of the sector. Its Report of 1987 provided a comprehensive analysis of the deficiencies in the system and recommended a broad strategy to rejuvenate and modernize government administration in a comprehensive manner. The aim was to equip government for its new role as policy formulator, monitor and evaluator as the economy was being liberalized. 20. The Government started implementing some of ARC's recommendations in 1990 in the context of an economic program that was supported by an IDA Economic Restructuring Credit -- ERC (CR 2128- 6 CE).' While the ERC relied primarily on the ARCs reform agenda, specific performance criteria were tied to staff reductions which were driven by the need to reduce budgetary costs. The target of reducing over-staffing 60,000 over 3 years (over an estimated base of about 450,000 employees), was based on ARC's own assessment that there was 20 percent over-employment in government. Actions to adapt the administration to the more liberal economic environment and to improve the efficiency and effectiveness of government services were not well specified. Rather, the government relied on an ill-defined process of organizational reviews to be carried out by Organizational Development Units in the main ministries, which had yet to be created. Efforts were also intended to revitalize program budgeting and to introduce performance measures in development programs. 21. A partial embargo on new recruitment was imposed (education, health and the security forces were excepted). Measures were announced to improve civil service management, including the re-creation of the Public Service Commission (PSC). The Salary and Cadre Committee (SCC), which was created in the wake of the ARC Report, was given the responsibility to investigate the need for new staff and for changes in salary policies based on requests from the Cabinet. The Government also planned to reform public and private sector pension systems with the objective of bringing them into line and improving portability. 22. The 1990-91 Reform Outcome. The program failed to achieve its purposes, especially with respect to staff reductions and the reform of the pension system: * More staff were hired than separated. According to data collected by the SCC, public service personnel who retired under the scheme were: Staff Grade officials (cadres in categories A and B)- 1,123; Clerical grades (category C) - 9,676; Minor grades (category D) - 8,258; teachers - 11,798. This outcome did not reflect any targeting by government. Some of the experienced retirees had to be re-hired as consultants, and the Government found it necessary to replace departing teachers by 14,000 trainees, many of whom were ill-equipped for the task. * As the program unfolded and political constraints impeded the targeting of redundant staff, the government improved incentives to induce staff to opt for early retirement 6. Enhanced separation benefits were given to many who would have retired anyhow: out of the 31,000 staff who left under the incentive scheme (from November 1990 to end-1991), about 18,000 would have reached their normal retirement age any way. * The authorized budget positions were not reduced to reflect staff departures, creating a total of 61,000 vacant positions by mid-1993. Failure to deflate the authorized cadre maintained line ministries' authority to recruit staff once the embargo on hiring was relaxed. * The incentive scheme to induce early departures under Circular 44/90 was not rescinded. 23. The institutional measures to reform personnel management and provide for sustainability were only partly effective: * The monitoring of staff movements and the retrenchment program was undertaken by the Ministry of Policy, Planning and Implementation (MPPI), with the use of quarterly surveys. Reporting compliance weakened after the staff departure program was completed and by mid-1993, only 65 For various reasons, some of the major building blocks of the ARC strategy were omitted. For example, the Government did not follow through on the recommendation to establish the Senior Management Group, re-establish an independent Civil Service Commission with real powers over the whole civil service and equip the government with in-house capacity to carry out a long-term reform program (e.g. organization and administrative audit capacities). Additional incentives were introduced under Circular 44/90 allowing staffto retire much earlier than what was permitted under the existing pension scheme without suffering penalties for early departures. See chapter IV. 7 percent of budgeted positions were being reported. The surveys were discontinued in 1995 and were not replaced by any other instrument to track actual employment and staff movements. * The personnel management system has remained largely decentralized. It regulates each professional group through a number of "services minutes" and gives substantial freedom to line ministries and departments to manage careers. However, it has lost its capacity to regulate and manage the civil service as a whole so that there is no strong central mechanism to ensure that rules and guidelines are adhered to and to minimize politicization of the civil service. The new PSC did not enjoy the same authority as its 1972 predecessor. * The process by which the government would have restructured itself (through professional administrative and program audits) never really took hold. This undermined the sustainability of the administrative reform as evidenced by the renewed recruitment after 1993. In particular, the SCC did not have the strong pro-active mandate that it needed to assume some leadership in investigating cadre needs and salary developments, as they arose. 24. Lessons. The administrative reform program had major design flaws. First, the institutions needed to carry out an ambitious administrative reform program and make it sustainable were not in place when the program started. There was no strong central management of the program and of the civil service, no in-house capacity to carry out administrative audits and staff redundancy programs, no program to facilitate re-insertion of redundant staff, and the program failed to create this capacity '. Second, there was no hard budget constraints imposed by the government on the size of the wage and pension bills. This led the Government to use voluntary departures as the driving force toward staff retrenchment. When the incentives to stimulate departures were deemed insufficient, the generosity of the incentives was increased, without paying due regards to the overall financial consequences.' Third, the retrenchment program was not targeted to identify redundant staff and positions. Fourth, while the reduction in the total wage bill was expected to provide fiscal space for salary adjustments, there was no strategy.to use these savings to reform the incentive structure. B. Developments in Employment and the Wage Bill 9 25. From 1985 to 1996, budgeted positions in the civil service (Central Government and Provincial Councils) increased by 150,000 and, as a share of population, grew from 2.6 to 3.0 percent (See Fig. IA) o. Most of the job creation has occurred since 1990: 81,000 in the Central Government"-- and 57,000 in Provincial Councils (Fig. IB). The Government did benefit from a UNDP-financed program for administrative audits (RMU). However, this project failed to be integrated into the administration and was used to carry out ad hoc administrative reviews rather than to support the core administrative reform effort. 8 In effect, the pension bill doubled from 1990 to 1992, under the double impact of numbers of staff leaving the civil service during that period, the nature of the payoffs accorded to staff as they departed and the suspension of rules to allow departing staff to collect pensions after 10 years of service rather than wait until they reach their regular retirement age to do so. Annex 3 provides additional background information on this subject. o For the Central Government, employment figures correspond to budgeted positions. They include official figures for officers and soldiers in the military of 78,000, each year, since 1985. For the Provincial Councils, the statistics come from the government's quarterly surveys. Over 1991-93, the bulk of newly created positions occurred in both the Central Government - police department, teachers and medical staff transfers of staff from agricultural projects, and the creation of the Divisional Secretariats, and Provincial Councils - 14,000 new teaching positions. Since 1994, however, most new jobs were created in the Central Government: additional teachers and medical staff, transfers, and new positions in the District Secretariats. 8 Rg. IA: Total Civil Service Employment, 1985-96 Rg. 18: Employment by Levei of Government, 1990-96 Thousands and Percentages of Populadon Thousands 70 3.40. Central Prov. 600 Population Thousand 3.4 35.0 GVt. Council employe 3.3 2300 400 C 0 11 1 300 3.04! 1500 200 a ' .t'. 0.0 19919 5 1958 1990 1992 10m 19190 192165 I Sources: Budget Estimates, Quarterly Survey. 26. Actual employment is difficult to quantify since the Government does not have a central data base to keep track of "staff on board". Ministries manage staff independently and information on staff is available only in pay sheets maintained by budget heads. Quarterly surveys were carried out during 1990- 95 at the national level by the MPPI, which monitored approved positions versus actual employment as reported in the pay sheets. The survey covered most staff in the Provincial Councils but only about 65 percent of budgeted positions in the Central Government mainly because of non-reporting by the police department. The results of the survey show that the total number of vacant positions peaked at 61,000 in March 1993. Two years later (March 1995), there were only 18,000 empty positions, mostly in the Provincial Councils. The latest indications are that, by end September 1996, there were about 20,000 vacant positions. 27. The total cost of employment, including the payment of pensions to retired civil service staff, has not been reduced. Initially, economies on wages were offset by increased pension payments. Subsequently, overall costs increased as a result of additional recruitment and salary adjustments. Salaries, which had not been adjusted since 1987, were raised by 30 percent across-the-board in 1993. This was followed by ad hoc adjustments for certain categories of staff (teachers) and discretionary increases in allowances which in turn, created distortions in the wage structure (for instance, teachers get higher pay than school principals) and created pressures for additional adjustments. Under the combined effects of staff increases and pay and pension rate adjustments, expenditures on wages and pensions increased five-fold since 1987, increasing from 5.7 to 7.4 percent of GDP (Figs 2A). Since the Government protects wages versus other categories of current expenditures, the ratio of expenditures on wages/O&M has increased significantly since 1987 (Fig 2B). This crowding out of O&M by personnel expenditures has adversely affected the efficiency and effectiveness of government services. There is ample evidence of this in all sectors, in particular infrastructure (notably routine road maintenance and railways) and health and educational services, documented in various World Bank and other donor reports." 12 See World Bank, "Public Expenditure Review of 1993" and "Teachers Training and Re-Deployrnent Project-, 1996. Fig. 2A: Tota Wages and Pensions, 1987-1996 Rg. 2B: Wages/Pensions wrt Oth. Gds and Serv. Rs Billions and Percentages of GDP Index Numbers 1/ 7 4.0 *1 70 400, -100 120 I 20.0 Source: Central Bank Annual Reports 11 Wages and pensions divided by other goods and services. 28. The Government's wage policy has consisted of: (a) periodic across-the-board salary adjustments to catch up with inflation and, in 1988, to restructure the pay system; (b) providing a variety of allowances to supplement monetary incomes between major adjustments; and (c) providing staff automatic salary increments, generally unrelated to performance because of inadequacies in staff performance monitoring and reviews. On several occasions, allowances were re-integrated into basic salaries, only to be re- introduced later. Since 1994, egregious adjustments in pay and allowance levels have led to a serious imbalance in the pay structure, with consequent dissatisfaction of the staff and tensions among staff groups. As a result of this policy, there are now three major problems in the current incentive system: (a) compensation, including pension benefits, are not linked to market-determined levels and depress incentives at higher grade levels due to the traditional egalitarian bias. Thus, the Government is becoming increasingly non-competitive in attracting and retaining the "best and the brightest" in Sri Lanka and in motivating performance by higher-level staff; (b) increases in staff pay are not linked to performance, as the performance evaluation system has broken down; and (c) there is an excessive number of wage grades (currently 94) and monetary and non-monetary allowances, geared to satisfying the interest of specific staff groups rather than the interests of the Government, which makes the compensation package non- transparent.'" C. Systemic Issues 29. The civil service developments described above are the result of a number of social and political factors that are deeply rooted in Sri Lankan society. These main systemic issues are: * The political economy of the labor markets in Sri Lanka. * Weaknesses in the management of government and public resources. * Inadequacies in the management of positions and staff and of the incentive structure. 30. The Political Economy of Labor Markets. Recent governments, including the present one, have made it clear that employment is their most important concern, as approximately 200,000 young people - of whom 25,000 are skilled - enter the labor market annually. Despite a professed desire to reduce the size 1 For instance, a Supreme Court Judge receives allowances equivalent to about four times the basic salary. 10 and scope of the public sector, they have not been able to restrain growth in public employment. The public sector remains the employer of first resort . 31. Studies have shown that most Sri Lankan workers search for jobs essentially in the public sector because of job security. For qualified university graduates who can work in English, the private sector pays higher wages and offers better career prospects. In addition, shedding staff is difficult and costly due to administrative and legal obstacles, such as anti-layoff legislation and the absence of unemployment insurance and private pension schemes. This inhibits job creation by the modern, regulated private sector. There is also evidence that a deterioration in educational standards has reduced the number of well- qualified entrants in the labor market and Government. These factors are at the source of the high rate of unemployment, which has exceeded 14 percent since 1980. Of the unemployed about half are 0 or A level graduates who are not willing to take up lower-prestige and lower-paying jobs in the unregulated sector. Recent economic difficulties and the low rate of job creation in the private sector have further exacerbated labor market tensions and made public sector employment even more attractive than during the early 1990s. 32. These structural factors have consistently made public sector employment attractive particularly for lower-skilled entrants who constitute the bulk of government employees at the C and D grades. At these levels, wage differences with the private are not as great as they are at middle and senior management levels. Given the uncertainties of employment in the private sector, job security becomes a major concern, considering the virtual guarantee of employment once one has joined the civil service and the government's relatively generous pension scheme which provides income security for life at no cost to the beneficiary. 33. The situation has been complicated by the fact that the present government is closely linked to the labor movement and rests on a fragile parliamentary majority. Sri Lanka's labor movement has traditionally been active in opposing a reduction in the size and scope of the public sector while expecting increased labor protection from the State. Each service and segment of the civil service is organized under their own unions and associations." Pressures to create new employment opportunities within the public sector remain strong, as evidenced by the increase in employment since 1993 and by specific programs like Samurdhi (See Chapter V). The unions and professional associations have also, of late, been quite successful in promoting the interests of specific segments of the civil service, particularly with respect to wage conditions. This is one of the causes of the present distortions in the wage system (see below). 34. Weaknesses of -overnment management. The weak management of the public sector is also due to the absence of a strong framework for decision making at the Cabinet and executive levels. There is little prior scrutiny of the financial impact and opportunity costs of decisions submitted for Cabinet approval by individual ministers, including appointments, new programs and modifications to the pay system, and mass recruitment of large staff cohorts such as teachers." Many critical decisions are approved, without discussions, as "Other Business", e.g. the decision in 1996 to create four new marketing boards. 35. Public Finance Management. Sri Lanka's public finance management system is based on program budgeting and is underpinned by detailed financial regulations. The system is quite comprehensive and, with appropriate improvements, would provide an adequate platform for its modernization toward performance budgeting. Nevertheless, it is currently focused on accounting for inputs rather than on stimulating performance and the production of outputs. In addition, constitutional rules which forces the Government to resign if it loses the appropriations vote, encourages the Government to focus on budget The "Report of the Salaries Commission, 1995", published in August 1996 provides a comprehensive listing of the trade unions and associations in the civil service. '5 In most Parliamentary systems, this function is exercised by a strong Privy Council, which manages and coordinates the business of government and scrutinizes the fiscal implications of proposed legislation or executive actions. 11 preparation, to the relative neglect of implementation. control and evaluation." Although in theory. the Government's financial regulations include provisions for the regular evaluation of programs as a basis for their improvement over time, in practice piogram evaluation is practiced in an ad hoc fashion and many of the recommendations of the Government's evaluation unit in the MPPI are not integrated into the design of new programs. As a result, budget preparation remains basically incremental, with very little scrutiny of programs once they are approved. Thus, many programs continue to linger on for years without review, mobilizing resources that could be used more effectively elsewhere in public service delivery functions. The budget cycle is long, with the approval of the budget outcomes by Parliament occurring two years after the end of the fiscal year. This makes it difficult to effectively enforce hard budget constraints and accountability for the use of resources. 36. Financial management is very decentralized: once the budget has been approved, each ministry is fully in charge of its own budget. As a result, the Treasury has little a priori control over expenditures and lacks the authority to require that changes be made during the year. Decentralization of spending authority is not coupled with a strong centralized management information system: the imprest reporting system does not provide the Treasury with timely and quality information on spending trends during the fiscal year in time to take corrective action. 37. Traditionally, the wage bill was budgeted on the basis of the approved cadre of staff, thus giving line ministries a blanket authority to fill vacant positions through new recruitment. In the Provincial Councils, the authorized cadre is still used as a basis for determining budgets and obtaining block grants from the Central Government. Within the Central Government, the budgeting of staff has been tightened somewhat, but remains weak - the Government recently indicated that there are about 20,000 vacant positions in the budget. Once funds are appropriated for wages, financial regulations forbid the "virement" of resources from the wage bill to other expenditure categories, although such transfers are allowed in the opposite direction. This reduces the fingibility of resources and, in the context of a fiscal tightening, crowds out expenditures on O&M. 38. The Manaement of Staff Careers and Positions. The complexity and fragmentation of the personnel functions constitute a major impediment to effective deployment of civil service staff. The Sri Lankan civil service is organized along specialized services (all-Island Services, various technical services, planning services, etc.). Each service is regulated under "Minutes" by a parent Ministry (for instance, the Ministry of Public Administration regulates the combined services and the clerical services, which account for 20 percent of the staff). Personnel within the various services may be deployed in several ministries (such as the Sri Lankan Administrative Services). Each service has its own scheme of recruitment, so that there are no common entrance requirements for all staff and no centrally regulated exams except for the combined, administrative and clerical services. Thus, staff and career management are fragmented among a number of authorities, with the budget head in the department having the greater authority and influence over civil servant's career progress and prospects, so that the normal checks and balances have weakened over time and have lost their transparency. Devolution has further complicated the situation by transferring the responsibility of the management of about half of the civil service to provincial Civil Service Commissions. Lastly, at the senior level especially, where appointments are a ministerial prerogative, decisions are seldom if ever challenged by the Cabinet. 39. In theory, cadre management is more centralized and control over the cadre is vested in the Department of the Budget. Nevertheless, the budget system fails to adequately distinguish the staff member from the cadre position being occupied although administratively, there is separation between demand for staff (the authorized cadre) versus supply of staff (managed by the various ministries and Budget preparation in Sri Lanka is an arduous task. The budget itself is a detailed document of some 2,800 pages, translated in three languages, before presentation to Parliament This requires that preparation be initiated in March of each year, which does not alow the authorities to take into account budget outcomes of the previous year. 12 departments and regulated by the services). In the absence of a centrally located data base of staff and of positions, the Treasury Department cannot keep track of vacant positions and the Government cannot effectively macro-manage the country's largest single labor market (keeping track of the skills mix and adjusting it to demand, ensuring the re-deployment of qualified staff as administrative reform changes the structure of government, monitoring entry and exit of staff and the Government's position as an employer vis-a-vis the competition). During a period of administrative reform, central management becomes mandatory to ensure success. The only central unit able to provide support in cadre management is the Salary and Cadre Committee (SCC), which is attached to the Ministry of Finance, but receives its mandates from the Cabinet, on a case-by-case basis. The SCC is equipped in-house to carry out routine cadre assessments and recruits local consultants for more substantive tasks. 40. All the systemic issues described above should not, however, distract from the fact the Sri Lankan civil service also has important strengths: the quality of the staff, notably at the mid and higher level, remains quite high; the central government framework is sound and comprehensive although rationalization of overlapping agencies and functions would need to be undertaken; the existing management system, although not operating efficiently, is well established and could quickly be modernized. The main "malaise" seems to be that the civil service is more focused on administration and input controls than on management and output production. Future reforms could build on these strengths. D. Re-launching Administrative and Civil Service Reform. 41. Long term vision. The most basic requirements is for Sri Lanka to articulate a vision on the kind of civil service system it needs to face up to the challenges of the 21st century. The lessons learned from successful reformers point to the need to re-focus the role of government on the critical functions which are needed to overcome gaps in essential service created by market failure. As the market economy develops, the cases of market failure will be progressively reduced. Accordingly, the size and composition of the public administration would shrink over time, with many of the services hitherto provided by the government being performed more efficiently by the private sector. In this respect, Sri Lanka's administrative reform is lagging behind the development of the private sector and many services could be devolved to the market or, at the very least, be "contracted out". As for the civil service itself, Sri Lanka would need to return to basic orthodox management practices. The Bank's East Asia Miracle study7 provides some guidance on the basic underpinnings of an effective civil service system (Box 1). From the models which have emerged in East Asia the main requirements are: political insulation, a competitive working environment, adequate pay and clearly delineated career paths. Important determinants of a new vision will be the new framework of cooperation between the Center and Provincial Councils and commitment by the Government to de-politicize the management of the civil service (e.g. by strengthening the PSC). Both will require constitutional amendments which are now being discussed. See World Bank, "The East Asian Miracle, Economic Growth and Public Policy", A World Bank Policy Research Report, Oxford University Press, 1993. 13 Box 1 Civil Service Systems in East Asia - Getting the Basics Right * Working environment. The ability of economic technocrats to formulate and implement policies depends on their effective insulation from lobbying for special favors by politicians and interest groups. Countries have evolved various systems to insulate technocracies: in Japan and South Korea, administrative and legislative systems give primacy to the bureaucracy in proposing laws. In Japan, the bureaucracy is also protected from political pressure by the National Personnel Authority (NPA), an independent body that sets the bureaucracy's pay scales and promotion policies, administers civil service exams, and makes most appointments. In Indonesia and Thailand, legal limits on the size of the public debt and on the budget deficit afford little margin for introducing politically motivated programs. * Merit-based recnitment and promotion. Highly qualified staff will be attracted to a civil service career if entry and promotion are directly related to their ability to compete. In Japan and Korea, recruitment revolves around highly competitive entrance exams, administered by national personnel authorities. In Japan, promotions are based on a combination of seniority and a host of performance indicators that differ across ministries. Because the number of personnel in the bureaucracy is fixed by law, competition for promotions is intense. Throughout the career of a civil servant, the lack of career progression points to inadequate performance and will be sanctioned by early dismissal. * Incentive-based compensation. In bureaucracies as in all other walks of life, you get what you pay for. Most successful bureaucracies have put into place incentive systems which are geared to make the civil service competitive with jobs offered in the private sector at most ranks. Senior civil servants may not receive compensation equivalent to Chief Executive Officers in the private sector, but at those levels, other intangible incentives are present. Thus, monetary compensation is a very powerful incentive and the divergence between compensation in the civil service and the private sector, at all hierarchical levels, is minimized. * Predictable career paths. An important requirement is a well-defined, competitive career path, with a substantial prize for those who make it to the top. Young entrants into the civil service will be willing to commit to a civil service career if there are reasonable prospects that sustained high performance will allow them to enjoy relatively rapid promotions and to achieve a high rank in the administration. In JaMpn as in other East Asian civil service systems, retirement comes early, and the rewards to a successful bureaucrat are substantial, extending beyond the pay, perks and prestige to include golden parachutes (such as lucrative jobs in the private sector). 42. Best Practices. In preparing an administrative reform, the Government needs to take stock of the successful reform efforts undertaken in a number of countries, notably Commonwealth countries which have been at the forefront of reform in the 1980s and 1990s. The most striking examples are found in New Zealand, Australia, and Great Britain. A common approach has been to use hard budget constraints as a means to drive government departments toward greater efficiency in the production of client-driven services, while relating outputs (efficiency) to outcomes (overall effectiveness of programs). 43. Current programs being undertaken by Sri Lanka may help develop the country's own best practices. In particular, the Teacher Training and Re-deployment project (TTRDP) may point to a home- grown example of structural reform in which all of the basic elements of a reform program are present: (a) restructuring a vital sector in which the role of the state will remain pre-dominant; (b) criteria for establishing a service delivery system which are scientific and objective and are derived from national and international best practices; (c) a regulatory system which provides for the central management of the teaching profession; and (d) a staff adjustment program which caters to skills mix improvements on entry, re-deployment of qualified staff and staff reduction through attrition and some involuntary departures which are regulated by existing rules and legal provisions (such as failure to occupy one's designated position (see Box 2). To a lesser extent, a project under preparation which seeks to re-structure the Mahaweli Authority of Sri Lanka will provide additional lessons in re-structuring semi-autonomous government bodies; whose role has changed over time. 14 Box 2 The Teacher Training and Re-Deployment Project (TTRDP): A Model for a New Beginning? The problems facing primary and secondary education are symptomatic of many of the issues facing the Sri Lankan civil service. The above project, approved by the World Bank in June 1996, focuses on reestablishing basic controls over the recruitment and deployment of teachers. To achieve this, several new institutional mechanisms have been designed which will permit the central Ministry of Education to analyze and project needs, link them to supply capacity and regulate the annual trainee intake and the deployment of trained teachers. The approved cadre will be determined annually on the basis of a formula (Ready Reckoner) linked to budget targets; this will be done by a Cadre Committee consisting of representatives from the Ministry of Finance, other key ministries and the Provincial Councils. Transfer Boards will ensure that requirements for initial appointment and other policies, rules and regulations are enforced. The project supports an attrition policy, whereby only one-third of departing staff are replaced, and those refusing re-deployment will be considered as having abandoned their office, and therefore subject to dismissal. In short, the project is designed to overcome the difficulties of devolution and to introduce radical improvements in resource allocation and personnel management in the largest functional segment of the civil service (173,000). A broader reform I of the education sector will follow if this first effort succeeds. 44. The results of these experiences have yet to be translated into best practices, and Sri Lanka may not be ready to undertake ambitious reforms because of the systematic weaknesses mentioned in Section C above. The use of hard budget constraints for comprehensive reforms needs to be preceded by a strengthening of financial management tools and processes so that, once budget constraints are imposed, they become binding and monitorable. At this stage, Sri Lanka should therefore aim at: (a) conventional reforms in civil service management that would restore the country's pre-1972 strong civil service traditions while positioning it for meeting the challenges of the 21st. Century; and (b) focus reforms on agencies which deliver services where there exists a clear relationship between inputs and outputs, such as the Postal and Railway Departments. As experience is gained and the budget system is strengthened, the coverage could be expanded to all government departments and agencies. 45. Short Term Measures - Getting Basic Controls in Place. The Government of Sri Lanka faces a pressing need to control and contain the growth of the wage and pension bill. The first priority should be to implement fiscal containment measures while simultaneously introducing administrative improvements to the management of positions, staff and costs. These measures would also set the stage for medium-term reforms. They include: * Re-imposing an embargo on recruitment by replacing only part of the 8,000 to 9,000 staff who retire annually. Thus, a hard manpower constraint would be imposed which would contain the size of the staff pending the implementation of functional reviews at a later stage. * Establishing more stringent budget preparation processes to tighten the wane bill. A first step would consist of budgeting the wage bill for FY 1997 on the basis of actual staff levels at end 1996. Additional savings could possibly be achieved by enforcing more stringent budget preparation, i.e., ensure that the Government's salary scales and rules of calculation are adhered to 1s * Setting in place an efficient centralized cadre information system. This could be done, at first, by expanding imprest reports by budget heads and requiring them to report on staff numbers "on board". The budget system allows for this by listing the number of budget positions authorized in the budget by staff groups. Budget heads have in hand detailed information on "staff on board" DTere is some evidence of loose budgeting" of the wage bill among budget heads when comparing the rate of increase of unit wages for the same categories of staff. There are significant differences in wage adjustments from 1995 to 1996 in some departments in spite of the fact that there has been no approved wage adjustment. 15 in pay sheets. The threat of withholding the wage bill would enforce timely and comprehensive reporting. A similar staff reporting system for the Provincial Councils should be set into place under the authority of the Finance Commission. * Ensuring that the salary adjustment in 1996 and 1997 are fiscally affordable and correct the most glaring distortions in wage rates. Given the size of a comprehensive correction proposed by the Salary Commission of 1995 (e.g., Rs 14 billion), the Government should consider a process of correcting egregious distortions over three years during which time it would prepare a major overhaul of the pay structure. * ReviewinR and closing down statutory boards which may have lost their reason for continued existence (see Chapter VI) and declaring a moratorium on new creations. 46. Systemic Improvements. The next 18 months should be a transition during which the pre- requisites of a long-term administrative reform are put into place, namely; (a) mobilizing political suport at the highest level of Government. This is an essential pre-requisite for sustained reforms. The planning and implementation of reforms must be well organized and centrally located in a unit which enjoys a high profile and the full backing of the political leadership. Clear lines of responsibility are needed for implementing the reform and for the overall management of the civil service during the reform period. The implications are that the present fragmented personnel management system will need to be suspended during the implementation; (b) developing the basic tools to manage a sustained reform effort which may extend well into the 21st century. This includes a strengthening of public finance management with the selective introduction of performance budgeting, shoring up establishment controls and setting up a strong administrative/program audit capacity, possibly in the Ministry of Finance to carry out administrative and program audits. Carefully selected pilot audits in a number of government agencies and statutory boards would be carried out during the preparatory phase to build up capacity and provide the basis for a first group of adjustments in the structure and composition of service delivery units. 47. If the Government pursues the above agenda and initiates these reforms over the next 18 months, then it would be realistic to undertake a more comprehensive reform over the medium-term. A key indicator of commitment would be the implementation of measures agreed to in the TTRDP Project and decisions made on the restructuring of the Mahaweli Authority. This should become evident during 1997. 48. Medium Term Program - Restoring Efficiency in the Public Administration - Policy Framework. During the next five years, public administration reform should aim at improving efficiency of government services while setting the stage for longer-term improvements in program effectiveness. This would need to be supported by a strong policy framework that emphasizes: * fixing fiscally affordable limits on the wage bill and on the number of authorized positions. These hard budget constraints will be the driving force behind the reforms once the controls are in place. Setting a target for the size and the cost of the civil service will force the authorities to focus on the affordable and appropriate role for the State and its constituent parts - core government services; provincial councils; statutory boards - and to identify the choices and trade-offs which emerge from the program. The State can no longer afford to maintain a large presence in the economy, nor is it necessary, given the recent liberalization of the economy. Sri Lanka 2000 has already provided a few broad principles for the reform of government role in the future; * determining affordable compensation policies implied by budget and staffing targets, including for the Provincial Councils. This implies that Provincial Civil Service Commissions and the Central Civil Service Commission would coordinate personnel management, establishment control and wage policies in order to provide for a single public sector labor market. This is also imperative to ensure the long-term sustainability of fiscal devolution; 16 * reviewing the structures. organizations and functions of government to obtain an indication of the number and profile of staff needed. A useful tool in this regard is ratio analysis. Under this scenario, the more detailed review of ministerial functions would be done later. The advantage of this approach is that it gives sufficient time to reform the pay and grading system before fine- tuning the structures, organizations and staff positions. 49. Action Program. With the institutional and policy prerequisites in place, a medium-term action program would most likely emphasize: * closing down obsolete government agencies and boards and adjust the public administration to the present economic environment; * preparing a program of institutional modernization, including computerization; * implementing a program of staff "right-sizing", entailing the retrenchment of some staff categories, staff re-training and the recruitment of staff with skills not available in the administration. Downsizing of staff numbers in certain generic categories of staff which are present in all administrations (minor staff of all types and clerical staff with only generalist skills) would also be undertaken. This right-sizing effort would be driven by newly defined staffing parameters through the use of ratio analysis that would permit common standards to be applied throughout government. It would also take into account the possibilities of "contracting a number of services" and reducing the need for staff through computerization. While this program is being undertaken, the hiring embargo would be maintained; * carrying out a program of administrative and program audits to complete the institutional rationalization program and identify means to improve the efficiency and effectiveness of essential government services; * reforming the railways and postal departments to position them for commercialization. Improvements in efficiency together with tariff adjustments would reduce the financial burden of these two large government departments (together they number 50,000 employees- 10 percent of the civil service). Both produce commercial outputs. Rigorous business plan for these two departments would assist in clarifying measures to rationalize their activities, reduce costs and improve cost recovery. * creating a capacity within government to re-deploy and assist the placement of redundant staff in the economy. This should be viewed as an essential component of a medium-term reform program. Such a facility would assist in making the market for public servants more fungible among sectors and levels of government, help to build a consensus on staff retrenchments by associating the various stakeholders (labor unions, government agencies, the various public service commissions, and the staff being retrenched). 50. Salary reform is a major item in the civil service reform agenda. Sri Lanka would need to evolve a compensation system which is linked to productivity and market-determined wages rather than to tenure and inflation. This will imply overcoming the four basic weaknesses mentioned above (paragraph 28): the absence of a link to the market, inadequate linkages of pay increases to performance, excessive number of wage grades and allowances with a resultant loss in transparency. In addition, the absence of a contributory pension scheme does not provide a clear basis for evaluating the real cost of employment. 17 51. To overcome these distortions and establish a market-based incentive system over the medium and long term, a step-by-step pay reform process needs to be undertaken. focusing on: (a) simplifying the pay structure by reducing the number of pay scales from the present 94; (b) integrating all allowances into basic pay; (c) establishing an efficient performance evaluation system to link pay with performance; (d) carrying out job grading to allow valid comparisons of job contents with similar jobs in the private sector; and (e) establishing a pay comparison survey system to provide credible wage comparisons with the private sector. 52. Costs and Benefits. It is difficult at this stage to quantify the impact of the various measures being proposed. The short-term measures listed in paragraph 45 are primarily aimed at stopping the growth of the civil service and dampening the size of the wage bill as a percentage of GDP. These measures could help realize fiscal savings of Rs 600-850 million in 1997, including: Rs 250 million as a result of the partial embargo on recruitment; Rs 300-400 million resulting from a more stringent budgetary preparation process; and an additional Rs 200-300 million resulting from close monitoring of staff movements during the budget year, which would inhibit non-budgeted recruitment. Further short-term gains could be achieved should the Government pursue a more aggressive policy of closing down obsolete agencies and Boards. Although fiscal savings would be modest initially, over time the sustained implementation of these short- term measures would have significant cumulative effects on the size of wage bill provided that the Government refrains from excessive general salary increases. 53. The medium-term action program presented in paragraph 49 mainly consists of a process to be followed to achieve specific objectives. It lays out specific targets for reforms but does not include precise sub-programs because of the major uncertainties surrounding the reform agenda itself. Such uncertainties include, for example, the cost of an incentive program to encourage staff departures and the cost of re- inserting staff in the economy. In the absence of specific reform programs, costs and benefits cannot be accurately quantified. Fiscal gains are likely to be small in terms of savings on wages and pensions, but potentially important when considering the improved efficiency and effectiveness of the public administration over the longer run. From cross-country indications obtained by the Bank, well-conceived programs recover their financial costs over a relatively short period, 3-5 years. IV. Reforming the Public Sector Pension System 54. Modifications in recent years to Sri Lanka's Public Sector Pension System (PSPS) have resulted in civil service pension costs assuming an increasing proportion of GDP, from 1.4 percent in 1990 to 2.3 percent in 1995, and the total pension liability to existing and new pensioners over the next 30 years (the pension debt overhang) is quite large. The PSPS is liability-driven and generous in comparison with other countries. The increasing pension cost has squeezed out high priority government expenditures. The imperative of fiscal adjustment over the next few years now requires an early reform of the system. 55. This chapter first reviews the civil service pension scheme and key modifications in recent years. It also attempts to compare the Sri Lankan system with that of some other countries and discusses the different short-term options available in reducing the fiscal burden. Finally, the medium-term reform agenda, including the transitional alternatives, is presented in broad terms. 18 A. Background 56. Sri Lanka's PSPS is a defined benefit system that is managed as one element of civil service remuneration. Its two major programs - for old-age retirement and for widows and orphans, respectively - - operate without the benefit of actuarial analysis (see Box 3). Only the latter program requires an explicit contribution from civil servants (3-6 percent depending on the salary level), but this is inadequate to fund promised benefits. Therefore, it too has had to rely increasingly on government budgetary support, from 56 percent of pension liabilities in 1989 to 72 percent in 1995. PSPS liabilities are thus largely unfunded and benefits are financed on a pay-as-you-go basis. While benefit generosity has played a key role in driving these public sector pension costs ever higher, factors such as increased life expectancy have also contributed to the future fiscal burden. Box 3: Sri Lanka's Public Sector Pension System and its components. There are four pension plans for civil servants, while the employees in the state-owned enterprise sector and the private sector are covered by two other plans. * Civil service pensions and lumg-sums. This main program, with an estimated 310,854 itretirees at end-1995, covers the core civil servants, armed services, and provincial govemment employees. Payments include the pensions, lump- sums (gratuities), and disability pension for these employees. * Widows and Orphans (W&O) Pensions. The program covers pension payments for widows and orphans (i.e. below the age of 24) of the above-mentioned groups, with an employee contribution of 3-6 percent of salary. Widows and orphans, about 68,000 beneficiaries at end-1995, are eligible for pensions provided under the civil service pension scheme. * Teachers Pensions. This program covers pension payments for the teachers in non-government schools, such as Pirivena (Buddhist temple schools), private schools, and special schools (e.g. deaf and blind schools). Since 1970, these teachers receive pensions provided under the civil service pension scheme. The number of retirees is estimated at 16,250 end-1995. * Local Government Pensions. Following the 13th Amendment to the Constitution of November 1987, the local government employees who retired after September 1993 are covered by the civil service pension scheme. Local government employees retired prior to September 1993, about 65,000 end-1995, receive pension from a separate provident fund. * Public Service Provident Fund. This was established in 1942 for the benefit of public officers who are not entitled to a pension at retirement, e.g. temporary staff required to serve on a temporary capacity before they are appointed to a pensionable position, contractual employees, etc. The contribution rate is similar to the EPF, and there were 170, 847 contributors at end-1995. * Employees Provident Fund (EPF). The national program covers pensions for permanent public servants not eligible for pensions (e.g. statutory boards) and the private sector. It is a provident fund, and investment is restricted to government securities. The employee contributes 8 percent of salary, employer 12 percent and no contribution is made to the W&O plan. Number of beneficiaries about 50,000 end-1995. * Other Funds. Since 1981, the Employees Trust Fund (ETF) also covers all non-government employees and has about 2 million members. It is only about 10 percent of the size of the EPF. The employer contributes 3 percent of salary to the ETF. Although the ETF is not required to place any of its assets in government securities, its portfolio is concentrated in Treasury Bills. Of which civil servant pensioneers - 227 265, W&P beneficiaries - 64,999; teachers - 3,046; teachers' W&P - 3,358; local govemnment - 12.186. 57. Recent changes to the pension system. Major modifications to the pension system occurred in 1985 and in 1990 making the system more generous. Prior to 1985, the pension payable to a retiree (optional age of 55 and mandatory age of 60) was linked to the number of years of service, and there was a clear tradeoff between the lumpsum and annuities, although the system as a whole was quite generous. In 1985, benefits were increased. At that time, severe delays in the payments of pensions, due to poor maintenance of records, had occurred which prompted the authorities to exclude the number of years of service from the formula in order to reduce the delay in payments. From that time, a retiring employee 19 satisfying the double condition of 55 years of age and 10 years of service could retire with a pension of 80- 90 percent of final earnings. In 1990, Circular 44/90, which was introduced as a temporary measure to induce staff to leave the civil service," withdrew the age requirement for retirement, while increasing the benefit for all retirees, in particular those with more than 30 years of service. Under Circular 44/90, a civil servant with 10 years of service could retire with 90 percent of salary.20 As expected, this resulted in a surge in retirements, totaling 43,801 staff in 1991 compared with about 11,000 in 1989, and costing Rs 8.8 billion or 2.4 percent of GDP (compared with Rs 4.7 billion or 1.9 percent GDP in 1989). The Circular 44/90 scheme has now been in place for more than 5 years and has defacto become the regular scheme. The modifications in the pension formula since 1985 are presented in Table 3. 19 This was introduced in the context ofthe IDA assisted Economic Recovery Credit of 1990. 20 Pension payments would, however, not start until 20 years after the first day ofemployment or 55 years of age is reached. 20 Table 3: Sri Lanka - Modifications in Pension Formulall Pre-1985 1985-1990 Sep. 1, 1990 Oct. 18, 1990 (44/90) \2 Optional age of retirement: 55 \3 55 \3 55 \3 Not related to age, but YOS Mandatory age of retirement: 60 60 60 60 Years of service: Min. 10 years \4 Min. 10 years \4 Min. 10 years \4 Min. 10 years \5 Pension base: Last month's salary Last month's salary Last month's Last month's salary salary Replacement rate with 67-97%, depending 80-90/., depending Max. 90%, - 10-20 YOS: 90% after unreduced scheme: on salary (negative on salary grade reduced by 2% for completing of 20 YOS function) and YOS (negative function) each year short of or reaching age 55 (positive function) and closeness to 30 YOS - 20-30 YOS: 90% retirement age of 60 -30 YOS or more: 90% (positive function) + 24 months lump-sum Replacement rate with 30 months lump- 24 months lump-sum 24 months lump- - less than 30 YOS: 24 reduced scheme: \6,7 sum in exchange for in exchange for 10% sum in exchange months lump-sum in 25% reduction of reduction of for about 10% exchange for 10% unreduced pension unreduced pension reduction of reduction of unreduced for the initial 10 for the initial 10 unreduced pension for the initial years after which years after which full pension for 10 10 years full pension pension years after which - 30 YOS or more: 24 full pension months lump-sum with no reduction of unreduced pension Source: Department ofPensions. YOS =Years ofService. 1/ Personnel in the armed services (ay, navy and air force) are entitled to a pension as in the case of civil servants, but on a different basis. Optional age ofreirement is afler 20 YOS in the case ofcommissioned officer, and 22 YPS in the case ofnon-commissioned officer. As of /1/1i985 the pension benefit, unlike civil servants, is linked to YOS with a replacement rate of55-88% oflast month's salary. 2/ Applicable to core civil servants only. Public sector employees outside the core civil service fall under the Pension Minute of September 1, 1990 and see footnote 1/for armed services. 3/ Locally appointed nurse, nursing sister, matron or midwife of the Health Department, female teacher of Education Department, and female employee ofPost and Telegraph Department at age of5O or af?er completing 20 YOS, or member ofpolice force after the age of 50. 4/ Subject to pensionable age (55), abolition ofoffice, or on medical grounds. 5/ Subject to abolition of office, or on medical grounds. If less than 10 but more than 5 YOS, the retiree receives a lump-sum equivalent to 1 month salary for each YOS. 6/ Lump-sum is calculated on the pension base. 7/ In practice, almost all retirees have preferred the reduced scheme to the unreduced scheme. 58. Table 4 compares the benefits received by a civil servant under the pre-1985, 1985-1990, and post- 1990 (Circular 44/90) schemes respectively. The table shows the present value of pension liabilities for a civil servant retiring under each scheme, with the base (100) defined as a civil servant retiring at age 55 with 35 years of service. This analysis leads to four observations. First, benefits increased substantially with each change in the pension formula. Second, for an employee retiring at age 55 or more and having 10-20 years of service, the increase in benefits was larger in 1985 than with Circular 44/90. Specifically, in 1985 benefits for retirees with 10 years of service more than doubled, while those for retirees with 20 years of service increased by only about 50 percent. Third, for employees with a period of service earning them the maximum pension benefit (i.e. 30 or more years), the Circular 44/90 scheme increased pension costs by at least as much as the 1985 pension formula modifications. Finally, it should be emphasized that 21 the cost scenarios presented in the table underestimate the actual impact of Circular 44/90 on pension liabilities. More precisely, the data do not show the effect of the young retirees now present in the post Circular 44/90 retiree population. On average, they add relatively more to pension expenditures by virtue of their receiving benefits over a longer period of time. Table 4: Sri Lanka - Increasing Benefits for an "Average" Civil Servant. Era PV Era PV Index Index Retiring at age Retiring at age 55: pre-1985 41 60: pre-1985 35 10 years of 1985-1990 85 10 years of 1985-1990 74 service post-1990 89 service post-1990 76 20 years of pre-1985 60 20 years of pre-1985 52 service 1985-1990 93 service 1985-1990 82 post-1990 99 post-1990 84 35 years of pre-1985 100 40 years of pre-1985 87 service 1985-1990 108 service 1985-1990 98 post-1990 120 post-1990 108 Note: PV refers to the present value of pension liabilities discounted at the rate of 2 percent real interest. The index uses as the base the present value of maximum benefits received by a typical pre- 1985 em civil servant, who worked 35 years and retired at age 55. For simplfication, a civil servant, earning the average salary, is assumed to work for a fixed number ofyears and immediately retires at the age noted, choosing the lump sum option 59. Cross-country comparison. Sri Lanka's civil service pension scheme is generous compared to other comparable countries. First, there is no employee contribution in Sri Lanka.' Where there is no employee contribution, other factors such as wages and the basic parameters of the pension formula offset the pension benefits to a greater extent than is found in the Sri Lankan system. Second, Sri Lanka's pension payment is linked to the last month's salary drawn, compared with other countries where this is usually based on an average of the previous 5 to 7 years. Thus, Sri Lanka's system creates an incentive for employers to accord pay raises before the employee's departure. Third, Sri Lankan civil servants are allowed to receive a pension after only 10 years of service without any penalty reduction,2 while in most countries a public sector employee is entitled to a pension upon reaching a prescribed age (50-70 years) and 30 to 35 years of service. Other countries strongly reduce the replacement rate when the number of years of service falls short of an expected target, with at least 3 percentage points penalty per year. Allowing a pension entitlement after 10 years of service, while being good for labor market flexibility, could turn out to be very detrimental over the long run by creating excessive turnover in certain segments of government. This may not have appeared yet in Sri Lanka because Circular 44/90 is relatively young. Fourth, granting 100 percent of a worker's pension to a surviving spouse is generous, compared to cross-country standard of 50 to 75 percent, given that the size of the household contracted. Finally, a 24 month lumpsum payment is granted to all retirees in Sri Lanka; most pension systems offer such payments in conjunction with a reduced pension involving an actuarially sound procedure with amounts linked to the retiree's life expectancy at a particular age and a reduction based on the notion that benefits due to be paid in the future are instead paid up-front. The main features of the civil service pension schemes for some countries are presented in Table 5. Except for the 3-6 percent contribution to the widows and orphans plan. 2 Employees with 5-10 years of service are entitled to a lump-sum equivalent to I month salary for each year of service. Table 5: A Cross Country Comparison of Civil Servants pension Systems Country Type Plan Contribution Rate Pension Base Accrual/Replacement Rates Normal Ret. Age Pension or Service Period Indexation Argentina Flat + Optional Employee=11% 10-Year DB Accrual = 0.85% per annum Age: Men=65 Ad hoc increases Chile jDC or DB Employer=16% Salary (DB) Women=60 Chile Defined Contribution Employee=10% Not in DC Plan Not applicable in DC system Age: Men=65 Employer=0% Women=60 Peru Defined Benefit Employee=6% Final Salary Accrual=3.33% per annum YOS: Men=30 Wage Indexed Employer=6% Women=25 Burkina- Defined Benefit Employee=16-22% Last 3 Years Accrual=2% per annum up to a Age: Men=60 Faso Salary Max RR=80% Women=60 Pakistan Defined Benefit Employee=O% Final Salary Accrual=2.33% for 30 YOS; 2% Age: Men=60 Ad hoc increases + Savings Employer=100% after; lump sum=50% pension for Women=60 15 years, reduced pension=50% or 25 YOS Turkey Defined Benefit Employee=15% Final Salary Accrual=3-3.75% per annum YOS: Men=25 Wage Indexation Employer-20% Max RR=85% Women=20 Sri Lanka Defined Benefit Employee: Final Salary Accrual=9% with 10 YOS for max YOS: 10 to 20 Ad hoc increases Retirement=0% unred. pension RR=90% + lump Widow & Orphans sum; =4.5% with 20 YOS for max 20 to 30 = 3.6% unreduced RR=90% with no lump Employer: sum; Retirement=100% =3% for 30 YOS for unred. RR= 30 + Widow & Orphans=rest 90% + lump sum (24 months). Ireland Defined Employee=O% Final/Last 3 Accrual=1.25% per annum Age: Optional=60 Wage indexation 1Benefit Employer=100% Years Salary Lump sum3.75% Mandatory=65 Source: World Bank Staff. N.B.: QC refers to defined contribution plan;,,_i refers to defined benefit; Yos means years of service; RR refers to replacement rate or percentage of specified salary level that is slated for replacement by pension formula. Accrual refers to proportion of the wage base to which a worker acquires pension rights for each year or covered employment. M:\AnnisalPER\Table6.xis 22 60. The pension bill. The fiscal consequences of the modifications in the pension formula were a sharp; immediate increase in the pension bill for retirees who benefited from enhanced eligibility conditions. and permanent increases in pension liabilities (the total payments to existing and new pensioners over the future) through a reduction in the average retirement age and an increase in the yearly pension payments. As shown in figures 3 and 4, total pension expenditures increased from 1.4 percent of GDP in 1990 to 2.4 percent of GDP in 1991, in tandem with the surge in new retirees, and has remained at about that level since then. Fig. 3: New retirees PSPS 1990-95. Fig.4: Pension Bill 1990-95. 2.5- 45,000 40,000 2.0 35.000 30,000 0.5 25.000 20,0001. 15.000 NoG)P 10.0000. 5,000 0.0 1990 1991 1992 1993 1994 1995 1990 1991 1992 1993 1994 1995 Source: Department ofPensions, and Ministry ofFinance. 61. Without any reforms, total expenditures on civil service pensions are projected to increase from an already high level of 2.3 percent of GDP in 1995 to 2.8-3.0 percent of GDP in 2020. In light of the need for fiscal adjustment, these high levels of expenditure on civil service pensions are worrisome. B. The Reform Agenda 62. Sri Lanka's PSPS requires a major overhaul in order to reduce its budgetary cost, move it towards a self-sustaining system, and correct existing deficiencies such as worker disincentive and the absence of an actuarial basis for the calculation of pensions. Several countries have undertaken successful pension reforms (Box 4) and provide examples for engineering the reform of the Sri Lankan system. It should be stressed however that, ultimately, the design of a pension system depends on the country's objectives with respect to the promotion of savings and income redistribution within the retiree population. 63. At this stage, Sri Lanka has two options for the reform of its public sector pension system. The first option consists in proceeding as fast as possible with measures aimed at reducing the deficit on pension operations, namely by increasing the rate of contribution and reducing benefits, leaving a clearly identifiable budgetary cost for this operation. The second option consists of engineering a comprehensive reform focused on long-term sustainability of the system, but which would have a major impact on the budget. However, the design of a detailed reform agenda is a major undertaking that is likely to be lengthy. In addition to in-depth economic and financial reviews, the reform would require political consensus on issues such as (i) grandfathering for workers close to retirement; (ii) the relative importance of private versus public funded schemes; (iii) harmonization with the private sector pension system; (iv) tax treatment of benefits; and (v) the regulatory framework for financial management of pension funds. In the light of 23 these constraints, and given that the fiscal situation cannot wait for this time-consuming operation, a few short-term measures could be introduced to reduce the fiscal burden immediately without undermining the ground for successful medium-term reforms. However, should Sri Lanka decide to undertake long-term reforms along the lines of the systems adopted by Chile and Argentina, for example, it would be prudent to minimize stop-gap measures aimed at reducing the deficit as these could undermine the long-term sustainability of reforms. Box 4: Country Examples of Successful Reforms of Pension Systems - Variations of the "Three-Pillar" System. * Chile. Chile implemented a revolutionary reform of its pension system in 1981, when it replaced its "pay-as-you-go" pension system, which was facing big financial difficulties, with a fully-funded pension system based on individual capitalization accounts. Recognition bonds were issued to transfer workers who accrued pension benefits under the old system. The pension funds, with management entrusted to specialized pension fund management companies (AFPs), achieved very high real rates of return, averaging 13 percent a year between 1981 and 1990. Participants must affiliate with one APF of their choice and must have one individual account. The right to transfer accounts imparts a strong element of competition among APFs. Participation in the new system is compulsory for new entrants into the labor force (except self-employed). The APF is supplemented by a public pillar which provides a minimum pension guarantee. * Argentina. Argentina, as most other countries, reformed its public pensions plans before introducing a new structure. The retirement age was raised by 5 years and eligibility from 15 to 30 years of service, while the pensionable age base was lowered. Workers were allowed to choose where to place their II percent contribution into the new privatized pillar or the reformed publicly managed system. Currently, about 64 percent have chosen the privatized pillar, while as many as 85 percent of new participants into the labor market making the same choice. In comparison to Chile, Argentina has a relatively large public pillar. The system is supplemented by a flat minimum pension. * Australia Australia has for a long time provided a flat rate pension, subject to income and asset limits, financed from general revenues. The public pillar provides pensions to about 75 percent of the population and is much more generous and expensive than, for example, the Chilean minimum pension guarantee scheme. In 1991, a second pillar of a mandatory funded employer-based retirement superannuation scheme was added. This scheme is scheduled to be phased in over ten years, starting with a 3 percent employer's contribution in 1992, and increasing to 9 percent by 2002. Towards the end of the period, a 3 percent employee contribution will be added, matched by a 3 percent means- tested contribution by government. Total superannuity contributions by boards of trustees that include an equal number of employer and worker representatives, one of the reasons for union support of the second pillar. 64. Short-Term Reforms. Among the short-term measures that would have a direct positive impact on the pension bill, the most effective and easily implementable are: (a) the withdrawal of Circular 44/90 with immediate effect. Reverting to the previous system would then restore the previous scheme (Pension Minutes of September 1, 1990), which is still generous compared to international practice, thus re-linking pension payments to years of service and reinstating a retirement age of 55 years; (b) reducing the lump-sum payments from 24 months to 12 months of pensionable base; and (c) scaling back.the survivor benefits (100 percent of a worker's pension) to international standards since survivors should be able to maintain their living standards with a lower income. The combined savings from these measures could be in the range of Rs 0.6-1.0 billion in 1997 and 1998. 24 65. There are three additional measures that could be considered for immediate implementation: (a) the age for pension eligibility could be raised, especially as recent improvements in life expectancy have undermined further the sustainability of Sri Lanka's pension system. This is envisageable in the light of the fact that retirees in other countries receive 15-20 years of pensions on average compared with a minimum of 25 years in Sri Lanka; (b) the relatively high replacement rate (80-90 percent of final salary) for retirees could be scaled back, especially for those with only 10 years of service - other countries impose a penalty of 2-3 percent for each year falling short of minimum; and (c) it is possible to reduce pension outlays by linking the pension base to an average of the salary of the previous 5-7 years, compared with the current practice of the last month's drawn. 66. Medium-Term Reforms. As mentioned in the previous sections, Sri Lanka's pension system is expensive, difficult to plan and results in undesirable incentives. The short-term measures described in the two previous paragraphs would alleviate the fiscal burden of pensions, but would not suffice to establish long-term sustainability of the whole system. This second objective could be achieved only through a comprehensive reform program that takes into account what is available in the non-government sector and the social welfare system in place. The World Banks suggests that countries contemplating pension reform consider the "three-pillar" system as the best prospect for financial security for the old and sustainable economic growth (see Box 5). 67. Sri Lanka already has a mandated savings pillar that is supported by a means-tested back-up for those with limited labor participation. This mandatory savings pillar, the Employees' Provident Fund (EPF), for workers in the private sector as well as in the extended government and the state-owned enterprise sectors,24 is relatively large and, assuming that there is some after-tax real rate of return, EPF's 20 percent contribution rate should be sufficient to provide adequate retirement income for most participants, according to standard calculations.25 68. However, Sri Lanka's system falls short of the recommended package in three respects. First, only civil servants have the benefit of pensions as annuities in Sri Lanka. Non-civil service workers either have access to the Provident Fund lumpsum payment or to private insurance. Second, the savings pillar (EPF) does not offer pension annuity-only the lumpsum is available. Third, investment returns on pension funds are not maximized: provident funds are centrally managed assets, as opposed to the system of competing private managers, and the sole investment medium is domestic government securities. This means that there is no diversification of the inter-generational burden outside the governmental sector. These three broad issues need to be addressed to develop a pension system that protects workers in a financially sustainable manner. 23 Averting the Old Age Crisis, 1994, Washington D. C., World Bank. 24 In addition, there is since 1980 the Employees Trust Fund (ETF). 25 A veruing Old-Age Crisis, 1994, Washington D. C., World Bank, pp. 74-75 and pp. 293-302. 25 Box 5:. The "Three Pillar" Pension System. A recent World Bank study suggests that countries contemplating pension reform consider the "three-pillar" system as the best prospect for financial security for the old and sustainable economic growth. In the "three-pillar" retirement system, there are two different mandatory pillars - one publicly managed pillar, financed through taxes, to address redistribution, and a second privately managed pillar which is fully funded to carry out the income-smoothing or savings function - and one voluntary pillar which, by definition, is fully funded. The key design issue revolves around what should be the content of what is mandated. Namely, should there be a single program on a pay-as-you-go basis, or should there be a two-pillar blend of a public and private pillar? The World Bank recommends that countries consider having both a publicly-financed pillar and a capitalized pillar. The first pillar would be publicly-managed, financed out of payroll taxes or general revenues, and focuses on redistribution -providing a social safety net for the old. Benefits could take one of 4 forms -they could be flat (uniform for every old person), flat per year of employment (a uniform benefit for each year of contributing employment), means and asset-tested, or a minimum pension guarantee tied to a mandatory saving plan in the second pillar. Argentina and the Netherlands use a flat pension in that first pillar, UK uses a flat benefit tied to contributory years, Australia uses a means and asset-tested plan and Chile uses a minimum pension guarantee. These are different ways of achieving the same end- keeping the old with low lifetime incomes out of poverty. In contrast, the second mandatory pillar would link benefits closely to contributions, probably through defined contribution plan (in which benefits ultimately depend on total contributions plus investment income) to reduce evasion and political manipulation of benefits. It would be fully-funded to increase national saving and avoid intergenerational redistributions. The funds would be privately and competitively managed, because experience has shown that this leads to a higher rate of return and more efficient allocation of capital than occurs with publicly managed funds. Investment diversification across public and private securities, stocks and bonds, and even international diversification, is important to minimize risk and maximize yields. The investment manager can be chosen by the workers for their own individual accounts, as in the case of Chile and several other Latin American countries; workers, as in Australia and several other OECD countries. In either case, substantial government regulation is needed, to prevent fraud and excessive risk-taking. The relative size of the two pillars depends on the weight given to redistribution version saving, but probably each pillar should be somewhere between one-third and two-thirds of the total. The third pillar in this system in voluntary, designed to provide greater old age protection to those willing to pay for it in their younger years. Source: Averting the Old Age Crisis, 1994, Washington D. C., World Bank. 69. Medium-Term Reforms -Unifying the treatment of government and non-government employees. As discussed earlier, the defined benefit pension system for civil servants has led to escalating costs and constant legislative changes in favor of ever more generous pension rights for a selected group of the population. Given that Sri Lanka already has the general features of a three-pillar system in its general pension system, the first most desirable action would be to merge the civil service workers into the EPF, as there is no obvious reason for a separate, or even a supplemental, pension structure for civil servants in Sri Lanka. 70. Merging the PSPS into the EPF would also serve to improve short-term fiscal discipline and labor incentives. The government would no longer be tempted to adjust the compensation of the civil servants by making the pensions more generous, thereby deferring costs to some later date. If the government faces difficulties in attracting a competent work force, higher compensation could be made instead of making generous retirement promises that may be difficult to honor in the future. Furthermore, by integrating the PSPS into the nation's overall pension regime, greater labor force mobility between the public and private sectors will be encouraged, and the large reward in the current system for public sector work later in life will be removed. 26 71. Offering pension annuity. The merger of the civil service scheme with the EPF would require that both options, lump-sum and pension annuities, are offered to retirees. In this context, the first task would be to introduce actuarial calculations in EPF's system (see Box 6). If that is not possible or while it is being implemented, civil servants could be provided a provident fund with parameters compatible with the EPF, and with exchange agreements between this fund and the EPF. The provident fund could purchase an annuity for the retiree from a life insurance company with the funds. accumulated or, as is the practice in some countries, the fund balance is divided by the average life expectancy in months for individuals at the age of retirement, and that amount is given as a monthly pension. Box 6 Other Eristing Pension Schemes in Sri Lanka The EPF Act (no. 15 of 1958), also contains a provision for the creation and maintenance of private (approved) Provident Funds by employers. By the end of 1994, there were 217 such funds with a membership of 197,800 contributors and an outstanding contribution of Rs 5.4 billion. The contribution rate varies, but with a minimum of 8 percent by the employee with the rate for the employer 1.5 times that of the employer. Two separate pension schemes have been established over recent years for self-employed farmers (1987) and fishermen (1991), both administered by the Agricultural Insurance Board-which is itself a statutory board established in 1974 under the Ministry of Agriculture. The scheme only covers those not under the EPF-ETF (270,000 contributing farmers and 30,000 fishermen), is based on voluntary contributions, and because of its social objectives, receives a subsidy from the Govemment equivalent to 22-50 percent of the premium paid mainly to cover the administrative costs of running the scheme. The scheme offers a pension calculated on actuarially sound assumptions but their investments are restricted to government securities. It is too early to say how good these schemes are, but it is clear that they have some of the features that are being recommended to Sri Lanka, e.g., actuarial calculations, transparent subsidy, pension option. " Act ofParliament no 17, 1996 established a Social Security Board, which aims at providing pension and social security benefits to self-employed persons. It is not operational yet, but will likely to follow the principles of the farmers' and fishermen's pension schemes. 72. Enhancing the remuneration of pension funds. A key parameter in establishing long-term sustainability of pension funds is the rate of return on the contributions that are accumulated by employees. Sri Lanka would need to liberalize investment opportunities for pension fund managers in order to increase their financial strength and reduce their dependency on the budget. Experience with public management pension funds internationally has ranged from tolerable to disastrous. Long-term financial fund management expertise is rarely available in the government sector and the accumulated funds often become the target for funding projects, which may be politically popular, but are financially unsound. The funds are often used simply to finance budget deficits, which means that when the pensions fall due, the government has to identify new budget resources. Even in Singapore, with its substantial financial market expertise, the publicly managed fund has paid poorly, and in response to employee complaints, individuals are now allowed to divert some of their accumulated funds to private mutual funds. 73. As a result, the World Bank recommends establishment of a funded system that uses private management, with competition among the managers, for these funds. This approach is similar to what is being done in Sri Lanka already under the ETF. The institutional apparatus for this system varies from country to country, from individual employees choosing their own fund among a variety of regulated options, from government agencies in conjunction with employees choosing a fund for the whole agency, to centrally administered funds who hire private fund managers to manage the portfolio on a competitive basis. 27 C. Transition Issues 74. Moving toward a funded pension system however entails short-run costs. The government would have to pay current pensioners while at the same time setting aside contributions for current workers. Combining this strategy with the short-term reforms discussed above would reduce this cost substantially. In addition, over time the costs will fall dramatically and eventually they will be lower than what would be paid under the current system. 75. The most modest transition solution would be to close down the PSPS for new entrants and enroll them in the EPF. If, during the transition, the employee contribution rate for the PSPS is set at 8 percent (i.e. the same rate as the EFP), then moving new entrants into the EPF would have no real fiscal effect, except to the extent that recorded public debt will rise somewhat in accordance with the government's 12 percent contribution to the EPF .2 But there will be no real economic effect in the sense of a change in government sector spending as a percent of GDP. However, what would have been increases in implicit public debt for civil servants will now be recorded as an increase in explicit government debt. 76. A more ambitious scheme would be to move all employees below a given age into the EPF. This may be done by depositing into such employee's EPF accounts an amount equivalent to what would have been deposited if they have been covered by the EPF. Again, this would cause explicit government debt to rise, although with a more than offsetting decline in implicit government debt. A special bonus deposit may have to be added to these EPF accounts to recognize that accrued rights in the PSPS - even after assuming the short-term reforms of a withdrawal of Circular 44/90, reducing the lump-sum, scaling back the survivor benefit etc. - have higher values than the newly deposited amounts into the EPF. This bonus calculation would be difficult to calculate with complete certainty, although some alternative scenarios could be constructed to create upper and lower bounds. In any event, any bonus would result from political negotiations. 77. An intermediate strategy would be to cover new entrants (and may be some relatively newly hired employees) and allow, but not require, already employed civil servants to switch to the EPF along the lines described in the previous paragraph. The uncertainty of the PSPS - particularly if the short-term measures are substantial - may induce many of the existing staff to switch to the EPF. None of these methods are without cost. However, the gain is that the civil servant pension system in Sri Lanka will be permanently put on a path which is financially sustainable and secure. V. Social Transfer Programs 78. Over the years, Sri Lanka has established an impressive record in social development. Its social indicators compare favorably with countries at similar income level: in 1993, the adult illiteracy rate was 12 percent, compared with 41 percent on average for all low-income economies; life expectancy at birth was 72 years, compared with 62 years for all low-income economies; and infant mortality was 17 per thousand live births, compared with 64 per thousand live births on average for all low-income economies. Primary education is provided to boys and girls uniformly and basic health services cover the whole country. An array of social transfer programs is in place to assist the less favored segments of the population. These remarkable achievements reflect the high priority and long-standing commitment of public expenditures to these sectors. 26 Under the current systern, these funds are immediately borrowed back by the government. 28 79. Traditionally, the social welfare programs have taken the form of cash transfers. food stamps. and commodity subsidies. Budgetary outlays on these programs have constituted a large share of total expenditures, 10 percent of total expenditures or 3 percent of GDP in 199527. Commendable achievements were made in consolidating some of the smaller transfer schemes -- the main Janasaviya, the mid-day meal, and the food and kerosene stamps (which were limited to the North and Ehst of the country) -- into the Samurdhi program when it was introduced in 1995. However, there remains considerable scope for reducing the cost of the overall program without jeopardizing the objective of assisting the poor. This section provides recommendations for fiscal savings and medium-term strengthening of the social transfer programs, starting with the un-targeted food subsidies (where the main fiscal saving exists), followed by the Samurdhi program, and the other smaller programs. Tables 1-3 of Annex 4 present the government's social welfare program in 1995, 1996 and 1997 (including the recommendations of this chapter). There is still much room to advance up the hierarchy of cost-effective targeting so that adequate support is delivered to the poorest fraction of the population at a fiscally sustainable cost. A. Food Subsidies 80. Generous food subsidy policies have been in place since the second world war. Originally, the entire population received transfers through the distribution of rice rations at subsidized prices and through bread and wheat subsidies. By the late 1970's the government concluded that the cost of these subsidies were unsustainable (about 5 percent of GDP) and, in 1978, the rice ration was restricted to the poorest half of the population, followed by a removal of the bread and wheat subsidies in 1980. A major reversal occurred in 1994, when general commodity subsidies were re-introduced for flour as part of the electoral promise. This subsidy increased to about Rs 7 billion in 1996 or 1.2 percent of GDP, well above the 1996 budgetary allocation of Rs 3.7 billion. The subsidy benefits all consumers and, as indicated by recent analyses, benefits disproportionately the richest segments of the population.2' Lastly, the subsidy, has induced substitution away from consumption of rice, exacerbating the financial problems of the Paddy Marketing Board. The subsidy should be eliminated at the earliest. 81. The 1994 policy reversal also included re-introduction in the 1995 budget of the fertilizer subsidy (Rs 1.3 billion in 1995). The fertilizer subsidy benefits all farmers, including large profitable tea estates. In addition to the concerns on targeting, this subsidy encourages use of chemical fertilizers that have adverse environmental effects, on water mainly. The fertilizer subsidy could easily be phased out during 1997. B. Samurdhi Program 82. The Samurdhi ("prosperity") program was launched in June 1995 and was developed gradually in tandem with the consolidation of other transfer programs. The Samurdhi uses community-based targeting methods to reach the poorest 1.2 million households or about 35 percent of the population (compared with 1.5 million under the previous food stamps program).29 Through surveys of beneficiaries, conducted by 24,000 Niyamakas ("mobilizers") at the village level, these households are divided into four groups, receiving Rs 100-1,000 per month in the form of a combination of food stamps, cash payments, and a compulsory group savings component depending on the household poverty level. The compulsory saving accumulates in a group savings account at a state bank. The beneficiaries are expected to be re-screened at periodic intervals (6 months) and progressively removed once graduated from poverty. The allocation in Poverty alleviation is also carried out through programs recorded in the capital budget These include the National Development Trust Fund project (NDTF), the Integrated Rural Development Program (IRDP), and, to some extent District Level Capital Projects. The total estimated cost in 1995 for these programs was Rs 3.6 billion (or 0.5 percent of GDP), including a capital allocation of Rs 64 million for the Samurdhi program. See Sri Lanka In the Year 2000 - An Agenda for Action, 1996, Washington, D.C., World Bank, pp 34-35. 29 This program does not cover the families who benefited from the Janasaviya program totaling about 400,000 households. 29 the 1996 budget for the transfer component is Rs 8.4 million (or 1.1 percent of GDP). In addition to this component, the program also launches small village-based projects -- such as minor roads, culverts, bridges, irrigations and common wells --, and health and nutrition activities. An-amount of Rs 232 million has been budgeted in 1996 for these purposes. 83. Consolidation of the various small transfer programs into the Samurdhi has improved targeting significantly. However, there are a number of concerns related to the Samurdhi program that needs to be addressed quickly. First, the targeted number of mobilizers to be recruited (36,000) is excessive in light of Sri Lanka's past failures in the training of similar mobilizers and their recruitment is based on weak hiring procedures.o The mobilizers, who have been appointed among unemployed youth, are not trained to undertake the broad range of assigned duties. Rather than embarking on a costly training program, these tasks could be undertaken more effectively within the framework of other on-going projects. It would be prudent to contain the number of mobilizers at the present level of 24,000 until a reliable review of their operations is completed. In addition many of the tasks being carried out by the mobilizers could be done in a more cost-effective way by NGOs. Second, the mobilizers should be kept on a one-year employment contract to avoid imposing a long-term burden on the budget. Third, there is a strong concern about its high cost. It would be advisable to impose at an early stage hard budget constraint on this program, and, in a first attempt, its budgetary allocation for 1997 could be kept at its 1996 budgeted level of Rs 8.4 billion. Finally, time limits for transfers to households should be enforced strictly. Their absence create distortions in the labor supply.31 C. Other Transfer Programs 84. The Janasaviva program ("strength of the people") was introduced in 1989 to provide poor households with monthly cash grants (Rs 1,458) for a period of two years. Identification of beneficiaries was to be conducted by mobilizers through successive geographical rounds (11 in total). The beneficiaries were required to work for 20 days a month or be placed in training. A second element of this program was an "interest payment" of Rs 250 for an indefinite period.32 Beneficiary households were admitted from the food stamp program after some screening procedures that excluded many recipients. The screening method was based on community meetings and through inspections of households. As a result, the Janasaviya program successfully improved targeting methods over time, especially with respect to food stamps. However, the overall program was costly and difficult to administer. Excluding the interest component, it was discontinued in January 1996 (only 5 out of the 11 rounds were undertaken) and was superseded by the Samurdhi. 85. Janasaviya interest payments amounted to about Rs 1.2 billion in 1995. These transfers, which support households regardless of their economic position and entail unnecessary administrative costs, could be integrated into the Samurdhi program, within the overall target of 1.2 million households. They should be discontinued by end- 1996 at the latest. 86. There are several other smaller programs that remain in place. There is the food stamp program, which aims at improving the nutritional status of the poor through monthly coupons (Rs 150) that can be exchanged for food in cooperative stores. Beneficiaries under this program who do not have access to electricity also receive a monthly kerosene stamp (Rs 48). The two programs, which initially covered almost half of Sri Lanka's population, included many of the poor but excluded some of the poorest 30 The plan is to increase the number of Niyamakas to 36,000 by end-1996, and most of the recruitment is scheduled to take place under the control of Members of Parliament. " Sahn, David E. and Harold Alderman, 1995, "Incentive Effects on Labor Supply of Sri Lanka's Rice Subsidy", in Dominique van de Wale and Kimberly Nead, eds., Public Spending and the Poor. Theory and Evidence, Washington D. C., World Bank. 32 Initially, the recipient was supposed to get a lump sum of Rs 25,000 at the end of the two year program, but because of fiscal constraints it was subsequently converted into a monthly "interest payment" of Rs 250 for an indefinite period. 30 households. In September 1995, they were both restricted to the North and East of the country. The emergency assistance program. Rs 2.1 billion in 1995, provides assistance to families who are displaced from their homes by the civil conflict in the North and East, and covers households living outside as well as within the war affected area. However, many of the beneficiaries living outside the war affected areas have received assistance for a long period of time (over 4 years) without any re-screening. These beneficiaries should be re-screened at the earliest and, if necessary, integrated into the Samurdhi program. Such a consolidation of the emergency assistance transfers could yield savings of about Rs 600 million. 87. These three programs - food stamp, kerosene stamp, and emergency assistance - benefiting the North and East absorbed Rs 3.5 billion of budgetary resources in 1996 and all suffer from weaknesses regarding targeting, work incentives or leakages. In the short-term, they could be consolidated into a single program for the North and East. At a later stage, when the war ends and when the poorest segments of the population can be identified in the North and East they could be integrated into the Samurdhi program. 88. The triposha program, which is managed by the Ministry of Health, is intended to provide a take- home dry food supplement to malnourished children under five, pregnant women and lactating mothers. The program (Rs 250 million in 1996) is poorly managed, relatively expensive and dependent on imported maize. Despite more than two decades of implementation, there is no evidence that the program has had a significant impact on the nutritional status of beneficiaries. This program could be phased out gradually while persistent malnutrition is addressed by alternative programs, such as the demonstrably successful nutrition component of the IDA-supported Poverty Project. 89, At about the same time as the Janasaviya program started, distribution of school uniforms and textbooks were introduced for all households with school children. These two programs, with a total cost of Rs 1.0 billion in 1995, could be improved by targeting households of the lowest income group. However, enhanced targeting for these programs would imply considerable administrative costs and could be undermined by self-exclusion. In this context, it would be advisable to integrate them within the Samurdhi program so as to make more explicit the trade-offs in assistance for households. These programs could be eliminated by 1998. 90. The proposals described above would yield substantial budgetary savings in 1997. The estimated cost of the social transfer programs, including the wheat and fertilizer subsidies, would decline from Rs 23.9 billion in 1996 (or 3.1 percent of GDP) to 1.6-1.7 percent of GDP in 1997-98, as shown in Table 6. Should peace materialize, additional budgetary savings could be generated as recipients in the North and East are re-screened, and, if necessary, integrated into the Sanurdhi program. 31 Table 6: Social Transfer Programs in 1995-1998. (Billion Rs) 1995 1996 1997 1998 Est. (w/measures) (w/measures) Samurdhi 2.3 8.4 8.2 8.2 Food Stamp 1.4 0.6 0.6\a - Kerosene Stamp 0.4 0.2 0.2\a - Janasaviya (round and interest payments) 2.8 0.8 - - Mid-Day Meal 1.8 - - - School Uniform and Textbooks 1.0 1.6 1.6 1.6 Triposha (supp feeding) and Infant Milk 0.2 0.3 0.3 - Emergency Assistance 2.1 2.1 1.5\a 2.1\c Public Assistance and Other 1.0 1.0 1.0 1.2 Wheat Flour Subsidy 5.0 7.4\b - - Fertilizer Subsidy 1.3 1.5 1.5 0.8 Total 19.4 23.9 15.1 13.9 Memo: Total as% of GDP 2.9 3.1 1.7 1.6 a/ A consolidation of these programs into one program is expected to achieve additional savings not quantified in the table. b/ Assumes a substantial phased reduction in the subsidy from the present wholesale price of Rs 11.65 per kg. c Consolidation of the Food, Kerosene and Emergency Assistance Program Source: Government ofSri Lanka and Bank staffestimates. VI. Public Enterprises 91. The study reviewed the financial health of Sri Lanka's largest public enterprises and the government's privatization policy. It appears that public enterprises do not represent a financial burden on the Government's recurrent expenditures; however, there is considerable misclassification of expenditures by several of these enterprises that underestimate their operational deficits. More generally, the public enterprises provide a relatively low quality of service and their ability to meet demand is inadequate. They also result in large budgetary capital transfers (including onlending from multilateral institutions). There are some areas where short-term fiscal savings could be achieved, in particular through the review and consolidation of existing statutory boards. This section offers some recommendations to strengthen the government's policies in these sectors." A. Role, Size and Composition 92. The public enterprise sector has played an important role in the economic history of Sri Lanka. Since independence in 1948, successive governments have increased their ownership of economic resources, resulting in a large public sector which employed over 20 percent of the country's labor force by 1976. Prompted by poor economic performance over three decades, Sri Lanka initiated an economic liberalization program in 1977. This resulted in the deregulation of manufacturing and a gradual reduction in public sector's presence in the economy. The contribution of public enterprises to GDP was reduced from 35 percent in 1981 to 22 percent in 199 l. Thereafter, this ratio has continued to decline steadily. This analysis does not examine the efficiency of public enterprises, which would require a different set of criteria. The focus is solely on their fiscal impact- In 1991, the public sector represented 65 percent of the GDP of the services sector, 50 percent of the transport sector, 45 percent of the banking, insurance and real estate sector, 11-13 percent of the manufacturing and agricultural, livestock, fisheries and mining sectors (World Bank, Private Sector Assessment, 1995) 32 93. The public enterprise sector in Sri Lanka consists of 42 commercial corporations, 49,fully-owned government companies, 38 subsidiary/associated government companies. In addition, there are 104 statutory boards. Together these entities employed approximately 176,000 persons and absorbed 7 percent of total domestic credit in 1995." The commercial corporations are set up under Acts of Parliament, but are self-financing. They include monopolies, for example in the large utilities and petroleum, and marketing boards which are set up to regulate prices of essential commodities. Government companies operate under the Companies Act and include the state owned banks and insurance companies, and manufacturing enterprises producing cement, steel, paper, etc.16 Statutory boards are also set up under Acts of Parliament; most of their recurrent and capital expenditures are met by the Government budget. They assist the Government in carrying out infrastructure and irrigation development, perform regulatory functions, undertake research, etc. The largest are the Road Development Authority and the Mahaweli Authority. B. Corporations and Companies 94. The study focused on the financial operations of some of the largest and most important corporations and companies. These included three utilities (Ceylon Electricity Board, National Water Supply and Drainage Board, Sri Lanka Telecom), three transport entities (Sri Lanka Railways," Sri Lanka Port Authority, Air Lanka), three public manufacturing enterprises (Ceylon Steel, Ceylon Petroleum and Ceylon Fertilizer), and two trading institutions (Paddy Marketing Board and Co-operative Wholesale Establishment). Their overall strength was assessed on the basis of (i) their financial performance, as measured by the ratio of operating surplus to sales; and (ii) their fiscal impact, namely the financial flows between them and the government's recurrent budget, including levies, taxes, and interest; the amount of budgetary capital transfers plus onlending they receive; and commercial borrowing guaranteed by the Government. 95. Financial Performance: The financial accounts of these 12 large public sector entities were consolidated into one financial statement. In 1994, their total operating revenues were Rs. 92 billion (14 percent of GDP) and their operating expenditures were Rs. 62 billion (10 percent of GDP) (Annex 1, Table 5). Most enterprises were generating operating surpluses, and the average ratio of their operating surplus to sales was 0.33, ranging from a high of 0.67 and 0.58 for Ceylon Electricity Board and Sri Lanka Telecom, respectively (1995 data), to a deficit of 65.5, 0.06 and 0.37 for Paddy Marketing Board, CWE and Sri Lanka Railways (1995 data), respectively (Table 8)." 3 In 1995, the main borrowers were Ceylon Petroleum Corporation (to finance repair of damaged storage facilities) and Cooperative Wholesale Establishment (to finance increased wheat purchases at higher world prices). Companies include Fully-Owned Government Companies, where government or any other public corporation /statutory board owns 100 percent of the shares, Subsidiary Government Companies where government holds the majority shares, and Associated Government Companies, where the government owns 20 -50 percent of the shares. Sri Lanka Railways is operated as a government department and its operating losses are covered by the Budget. 1995 balance sheets for CWE and Paddy Marketing Board were not available at the time of the mission. Table 7: Profitability of Public Enterprises (Ratio of Operating Surplus to Sales) 1992 1993 1994 1995 Ceylon Electricity Board 0.54 0.69 0.72 0.67 National Water Supply and Drainage Board 0.51 0.55 0.50 0.47 Sri Lanka Telecom 0.86 0.81 0.64 0.58 Sri Lanka Railways -0.59 -0.45 -0.37 -0.48 Sri Lanka Port Authority 0.46 0.42 0.41 0.42 Airlanka 0.14 0.19 0.28 0.12 Ceylon Steel Corporation 0.19 0.18 0.21 n.a. Ceylon Petroleum Corporation n.a. 0.21 0.27 0.19 Ceylon Fertilizer Corporation 0.07 0.17 0.24 n.a. Paddy Marketing Board -1.89 -0.77 -65.54 n.a. Cooperative Wholesale Establishment 0.05 0.01 -0.06 n.a. Source: Financial Statements ofSOEs 96. Fiscal Impact: Apart from the Paddy Marketing Board and Sri Lanka Railways, all the institutions reviewed were net revenue earners for the Treasury.3" Between 1992 and 1995, they contributed approximately Rs. 42.6 billion (at about 2 percent of GDP) to the Treasury in the form of interest on Government loans, income tax, defense levy and turnover tax (Table 9). In addition, their cash surpluses were tapped at irregular intervals by the Government to meet budgetary needs: the Government raised Rs. 7.4 billion in special levies from public corporations between 1992 and 1995, mostly from SLT, CEB, and CPC. Table 8: Public Enterprise Contributions to Government Revenues 1/ (million Rupees) 1992 1993 1994 1995 Total Ceylon Electricity Board 2,479 2,783 3,402 5,001 13,665 National Water Supply & Drainage Board 101 202 224 286 813 Sri Lanka Telecom 788 1,991 1,386 3,280 7,445 Sri Lanka Railways (373) (341) (304) (365) (1,382) Sri Lanka Port Authority 697 869 468 1,363 3,396 Airlanka n.a. n.a. 69 19 88 Ceylon Steel Corporation 95 159 219 n.a. 473 Ceylon Petroleum Corporation n.a 4,125 7,021 6,318 17,464 Ceylon Fertilizer Corporation 7 42 25 n.a. 75 Paddy Marketing Board (3) (147) n.a. n.a. (149) Total (in million Rs.) 4,032 9,963 12,738 15,903 42,635 GDP 425,300 499,800 578,800 661,900 Total (in percent of GDP) 2% 1/ Includes interest on Government loans, taxes, dividends and special levies. Note: For loss-making enterprises, includes current transfers from budget to cover losses. 97. These companies have large capital expenditure needs, which amounted to Rs 56 billion between 1992 and 1995, representing 58 percent of their total operating surpluses. Over that period, Rs 37 billion CWE would be a net revenue earner if its foodgrain (wheat) operations were excluded from the analysis. 34 or 66 percent of these expenditures was financed by the Government budget (Table 10). This component was in turn mostly financed by external savings, namely foreign loans from multilateral institutions which were onlent to the enterprises. In 1995, bilateral/multilateral loans amounting to Rs. 11 billion were onlent to public enterprises from the Government budget.0 In all cases, the institutions were servicing these loans on schedule. The remaining 34 percent of capital expenditures was financed by internally generated sources. Table 9: Impact ofPublic Enterprises on Government's Capital Budget 1/ (million Rupees) 1992 1993 1994 1995 Total Sri Lanka Telecom (297) (836) (2,019) (1,901) (5,053) Ceylon Electricity Board (1,391) 11 (791) (838) (3,009) National Water Supply and Drainage Board (2,737) (2,154) (3,051) (2,704) (10,646) Sri Lanka Port Authority (670) (2,265) (3,401) (3,443) (9,779) Airlanka - - - - 0 Ceylon Steel Corporation - - - - 0 Ceylon Fertilizer Corporation - - - - 0 Paddy Marketing Board - - - - 0 Cooperative Wholesale Establishment 180 180 113 - 473 Sri Lanka Railways (1,642) (3,192) (1,998) (3,185) (10,017) Total (6,557) (7,963) (10,833) (11,712) (37,064) 1/ Includes capital transfers from budget and onlending of multilateral/bilateral loans Note. Positive numbers indicate repayment of past loans. 98. Government assistance to the public sector also takes the form of guarantees on loans extended by domestic banks, a practice that is facilitated by the fact that the two largest banks are state owned. As a result, the Government has accumulated a significant contingent liability on the budget. For the enterprises reviewed above, this liability rose from Rs. 6 billion in 1992 to Rs. 29 billion as of December, 1995 with Air Lanka, Paddy Marketing Board, and CWE being the main beneficiaries. The Government's total contingent liability at end-1995, however, exceeded Rs. 42 billion (over 6 percent of GDP). The failure of some enterprises to meet their obligations to commercial banks has resulted in successive transfers of nonperforming bank loans to the Treasury. In 1993, the Treasury provided Rs. 24 billion in bonds to People's Bank and Bank of Ceylon, of which Rs. 5.5 billion covered losses of the Paddy Marketing Board and the State Plantation Corporations. An additional Rs. 20 billion is being given to the same two banks in 1996, of which Rs. 6 billion will cover losses of the Paddy Marketing Board and various bus companies. 99. The financial data presented above however did not reflect some other important aspects of performance that are much less commendable for Sri Lankan public enterprises. A considerable amount of expenditure misclassification exists in Sri Lankan public enterprise financial accounts which seriously underestimate current expenditures and overestimate capital expenditures; as a consequence, the financial situation of several of these enterprises would actually be weaker than indicated by the financial accounts. This is particularly the case for Sri Lanka Railways, where the proportion of capital expenditures that actually finance current expenditures may be as high as 40 percent, the Road Development Authority, and the Mahaweli Authority. These misclassifications have two major consequences: first, operating deficits appear to be more common than apparent, with the result that corrective management actions have been lagging, especially with regard to cost recovery; second, the enterprises concerned have been less 40 Source: Central Bank Annual Bulletin, 1995, Table 69. 35 capitalized than capital expenditure data show, with the result that their ability to meet demand has diminished considerably over time. 41 This undercapacity, which has been costly to the economy, is quite evident in the case of the CEB--brownouts have been extensive in 1996 and 75 percent of domestic enterprises have developed stand-by power generation--and in the case of SLR which is currently heading for a crisis with regard to passenger transport. Beyond these misclassifications, Sri Lankan public enterprises, and the public sector in general, suffer greatly from an insufficiency of budgetary expenditures on operations and maintenance as mentioned in chapter 4. 100. The weaknesses described in the above paragraphs indicate that Sri Lanka public enterprises are not in good overall condition, especially with respect to their ability to satisfy domestic demand. The sector would benefit greatly from: (i) an assessment of whether there is a clear rationale for their existence (covering all existing enterprises); (ii) a review of the role and operations of each of them; and (iii) an evaluation of their overall financial situation in a consistent manner, emphasizing the need to enforce stricter financial accounting rules with regard to current/capital expenditures. Several actions have been identified that would strengthen the sector in a permanent way. These include: A quick review of the operations of the poorly performing public corporations, such as the Paddy Marketing Board and the Central Transport Board. Divesting the assets of the CTB (including retrenchment of personnel) should be an important objective for the future. This would result in fiscal savings of about Rs. 300 million per annum. All CWE's operations (other than wheat imports) are provided by the private sector and the rationale for its agricultural marketing operations should be reconsidered. Most public enterprises would benefit from this kind of review, focused especially on their employment and pricing policies. See Annex 5 for additional information on Sri Lankan enterprises. * A global review of the transport sector. Sri Lanka Railways is in a very difficult financial situation, with large operating losses, and its assets are in an alarming state. Such a review will help determine the role of railway transport vis-a vis other modes of transport, and rationalize the operations of the SLR.42 * Tightening the rules for granting government guarantees on bank loans. Given the history of bank recapitalization in Sri Lanka, this tightening would impose greater financial discipline on enterprises, and to enhance transparency, information on these guarantees should be published in official documents. * Strengthening and accelerating the privatization program. C. Privatization 101. The GOSL is already engaged in a strong program of privatization. Starting in 1990, and supported by the ERC and PMEAC, the Government has made good progress in privatizing public enterprises. By 1994, about 78 public enterprises (including plantations) had been divested or leased. The program gained new momentum in 1995, as the Government realized its inability to fund capital expenditures of public corporations and companies. The Public Enterprise Reform Commission (PERC), was set up under a Presidential Task Force to identify companies for privatization and to negotiate with domestic and foreign investors. Companies are sold on an "as-is" basis, but buyers of major public 41 For example, the RDA has rehabilitated an unusually low amount of road connection in 1995 (1,000 km). 42 The World Bank Transport Sector study (forthcoming) provides a first attempt at addressing these issues. The need to raise additional revenues to finance the budget deficit was also an important factor behind this inteisification 36 companies are required to provide simultaneously with their bids, an investment plan for the following years. These investment plans usually avoid the need for the employer to retrench labor or reduce salaries. In some cases, 10 percent of the Government's shares are distributed to employees mainly to garner their support for ownership transfer. 102. Several successes have already been registered since 1995, with the privatization of Colombo Gas, five plantation estates and Orient Lanka (a duty-free shop). Privatization revenues amounted to Rs 3 billion in 1995. Enterprises to be privatized by end-1996 include Air Lanka and Sri Lanka Telecom, and the 1996 budget projected Rs 21 billion in privatization proceeds. However, actual receipts from privatization remain uncertain because of the delicate nature of negotiations with foreign partners. 103. There is evidence that privatization is having a beneficial impact on the economy, by strengthening government finances as well as by increasing the levels of enterprise efficiency. For example, Shell is investing US$50 million in the Sri Lanka Gas company, which it recently purchased to finance the development of increased storage and improved port facilities. Moreover, the Treasury envisages a large increase in future turnover taxes to be collected by the newly-privatized gas company. Eighteen of the twenty-two plantations whose management was privatized have improved their financial performance despite a large, Government-mandated wage increase and a drought that reduced output by 15 percent. If one excludes the wage increase, the plantations were profitable within a year of privatization. 104. The Government's privatization program has been commendable to date, and this program should be intensified, with the assistance of "privatization professionals" where necessary, to achieve quick results. Although proceeds from the privatization program are partly intended for budgetary purposes, they should not be used to by-pass the need for fiscal adjustment. For planning purposes, it would be prudent to forecast privatization proceeds on more conservative assumptions. Implementation of the privatization program provides an opportunity for the Government to intensify structural reforms in other areas. More importantly, using these proceeds to retire costly outstanding domestic debt stock would create fiscal space for additional restructuring of budgetary expenditures. In the medium term, the program can be expanded to include complete divestiture of partially-owned government companies. D Statutory Boards. 105. Excluding universities and educational institutes, there are 75 statutory boards in Sri Lanka, engaged in activities as diverse as building roads and conducting rubber research. Combined, they employ 44,000 people and are sustained through current and capital transfers from the budget. Current transfers have remained relatively constant at about 1 percent of total budgetary revenues-Rs. 1.8 billion in 1996. Capital transfers as a percent of revenues have fallen from 36 percent in 1986 to 9 percent in 1996. However, a large share of these transfers finance current activities, as evidenced by the cases of the Road Development Authority and the Mahaweli Authority which together are budgeted to receive 45 percent of total capital transfers to statutory boards in 1996. These two authorities carry out a number of activities that can easily be performed by private entities. There is also some duplication of activity among statutory boards. For example, two boards exist for the regulation and promotion of coconut cultivation (Coconut Development Authority and Coconut Cultivation Board) and another for research (Coconut Research Board). Together these three boards employ 1,410 persons and receive annual budgetary transfers of Rs. 215 million. 106. Statutory boards are in urgent need of rationalization. In view of the fiscal crisis, current and capital transfers to these boards in the next budget should be frozen in nominal terms to generate immediate fiscal savings. The restructuring of Mahaweli, being planned under a Bank-supported project, is expected to reduce its expenditures by 30 percent in four years, though in the short term it may result in increased expenditures due to voluntary departure packages. In the medium term, consolidation of statutory boards, 37 limiting their existence to universities, educational institutions, and those that are essential for performing regulatory functions, should result in reduced budgetary transfers. VII. Fiscal Devolution A. Background 107. Sri Lanka inherited from the colonial period an essentially center-focused bureaucratic system of administration which the Thirteenth Amendment to the Constitution (of 1978) changed in November 1987, by introducing sub-national units of administration for the first time in the modem history of Sri Lanka. The units of administration, eight in number, were the pre-existing Provinces and were called Provincial Councils. Provincial Councils (PCs) were given jurisdiction on a number of subjects, ranging from public order and implementation of provincial economic plans to utility (housing, roads, etc.) and social services (education, health, etc.). They were also vested with a certain amount of revenue collection authority and could handle a number of functions concurrently with the Center. A few subjects were reserved to the Center, principally national policy on all subjects, foreign affairs and defense, etc. 108. The constitutional provisions were strengthened by further legislation, the Provincial Councils Act No. 42 of 1987, and the basic principles that translated these legal changes into action were elaborated by the Administrative Reforms Committee (ARC) in 1988. The Committee enunciated two basic guidelines to be followed in designing devolution policies. These were: (i) predictability in regard to the resources which the Provincial Councils; and (ii) avoidance of the system of financial relationships being used as a mechanism of control of the periphery by the Center. The ARC also established a Finance Commission, consisting of the Secretary of the Ministry of Finance and Planning and the Governor of the Central Bank, as ex-officio members, and three other persons representing the three major communities to be appointed by the President, to make recommendations on the economic and fiscal relations between the Center and Provinces. 109. The experience with the devolved system in place is not very conclusive. Many weaknesses emerged over the years that have resulted in a confusing situation and actual devolution turned out to be much less than originally intended. A new devolution package has been presented by the Government in early 1996 in the form of a draft law that is now subject to public scrutiny and the objective is to obtain a national consensus. The draft is currently before a Select Committee of Parliament, and if approved, would be presented to Parliament to obtain its assent with two-thirds majority. Finally, it would be placed before the public at a referendum for approval. 110. The driving force behind the Government's new devolution package is the need to resolve the domestic security situation. This package is viewed by the Government as an essential element in its efforts to bring a peaceful solution to the conflict. Of course, the peace dividend would be significant in Sri Lanka and justifies the political imperative. The question is how best to achieve the predetermined level of decentralization, minimizing the costs of achieving the political objective. 111. The paragraphs below discuss the issues that arise in the design of a devolution package, focusing prnmarily on its possible adverse effects on macroeconomic management. These issues, which are described with reference to the weaknesses of, both, the current devolved system and the proposed package, are even more relevant that Sri Lanka now faces the need to effect a fairly large fiscal adjustment in the next few years. 38 B. Political Autonomy and Accountability 112. The degree of autonomy conferred by the Thirteenth Amendment to Provinces was relatively small and, on the ground, the institutions of government were not adequately remolded to serve the devolution objective". For example PCs remained subordinated to the political authority of the central government, as exercised by a Governor appointed as the local agent of the President, while the territorial structures of the different central line ministries remained in place. This situation was exacerbated by the ambiguity of the concurrent list which gave rise to frictions in certain fields, such as schools and hospitals. Provinces complained about their lack of autonomy. The new proposal palliates some of this weakness and strengthens the decentralized aspect of economic management by reducing central supervision of regional authorities.45 The two central government officials are to be removed from the Finance Commission and new authority is given to the Chief Minister of the Region. In addition, the Secretary of the Finance Commission, a position held by the Secretary of the Cabinet, is to be replaced by an independent official and the concurrent list is being eliminated. These changes are welcome improvements, but the real test will be in strengthening the role and capabilities of the Finance Commission with respect to fiscal management and imposing its authority on Regions." A strengthening of the Finance Commission, through increased support staffand financial resources, together with a clarification of its role with regard to supervision and monitoring especially is crucial, given its central role in determining transfers. 113. However, like the experience of Brazil shows, greater political autonomy may turn out to have adverse fiscal implications. In this context, several offsetting changes are needed in transfer and revenue policies to place clear political accountability for the consequences of resource decisions on the regional governments and to provide adequate incentives to induce decentralized political authorities to act in the interest of the whole economy. The devolved system in place does not emphasize accountability mechanisms and, in reality, decentralization turned out to be limited: the Center maintained tight control on Provincial finances because it provided the greater share of provincial expenditures, but this situation has also allowed Central Ministries to preempt funds (especially investment expenditures) at the detriment of Provincial Councils. The new proposal does not address the accountability issue adequately and would benefit greatly if mechanisms were introduced to improve both accountability and transparency offiscal operations at the provincial level. At the outset, rules need to be established requiring regions to provide information on their actual expenditures more promptly and to ensure that funds provided for certain purposes are.not diverted to other purposes. Additional rules could be considered that impose: (i) strict reconciliations between ex-ante and ex-post aggregate spending; (ii) tight sanctions against overspending; (iii) quick publication and dissemination of auditing operations to the public; and (iv) greater use of client surveys. C. Expenditure and Revenue Assignments 114. The Thirteenth Amendment assigned major expenditure responsibilities to provinces in many areas of education, local government, infrastructure, social services, agriculture, health, etc. In addition, other aspects of expenditures were assigned concurrently to the provinces and central government. These represented fairly large expenditure assignments, which most regional governments could not finance. Such imbalances are common across countries and are easily palliated. But difficulties arose in Sri Lanka because of ambiguities in responsibilities and because of the absence of control mechanisms in expenditure management. There was considerable mismanagement of funds by Provincial Councils due to the lack of In theory, political autonomy is considered an essential element for local govenments to maximize the social utility of their economic policies. 4 The new proposal refers to regions as opposed to provinces. 6 Currently, the Finance Commission has no visible staff, and the line ministries as well as the Ministry of Provincial Councils and Indigenous Medicine continue to have large roles in conflict with the objective ofdevolution. 39 qualified staff (notably civil engineers), non-enforcement of financial procedures and regulations, and a proper system of auditing. The new proposal resolves some of the assignment problems, for example, by eliminating the concurrent list, but ambiguities remain--probably due to the fact that expenditure functions have many dimensions and different levels of government may legitimately be involved at some stage. In the Sri Lanka context, the expenditure assignment issue is less important than the basic question ofweak expenditure management, because from an economic perspective, what matters is not Who legally is empowered to do what, but rather What is done and how well it is done. 115. Tax revenues devolved to PCs have represented only 4 percent of total taxation in the country, and these have represented 20-23 percent of their total revenues47. The balance was received as grants from the Center. It is not clear if the changes brought by the new package will increase or decrease provincial revenue authority, although it is most likely not to affect the status quo. Nonetheless, fiscal retrenchment in the future will require additional revenue efforts on the part of Regions which will be possible only if they have the right incentives and human resources to do so. Tax assignment in Sri Lanka would need to take into account several important considerations that have not been fully recognized yet. These considerations relate to the raising of revenues, the mobility of the tax base, the existence of economies of scale in tax administration and income distribution.4 Also, Sri Lanka's central government has large spending responsibilities (debt servicing, pensions and defense) which cannot be easily compressed and for which a stable tax base should be established. Should these considerations be misapprehended, macroeconomic problems are inevitable, like in the cases of Brazil, India, and Russia. 116. There is no revenue-sharing arrangement in place under the current decentralized system. The new package proposes to introduce such an arrangement for a few taxes. The main problem is with the GST because it is not efficient to assign the revenue from the distribution chain to the region in which such revenue is leviable. Another problem is that taxes on mineral rights should be a central subject. In defining the modalities of revenue-sharing arrangements, Sri Lanka would need to ensure that taxes are assessed and administered in the most efficient way taking into consideration the reasons for weak tax performance at the provincial level during 1989-95 (see Box 7). Tax revenues of PCs were Rs 4 billion in 1994, of which 50 percent was from the turnover tax; 21 percent from stamp duties; 18 percent from motor traffic fees; and the remaining 11 percent from various instruments. For example, if a tax base can escape taxation at the local level by easily moving to another jurisdiction, that base is not a good candidate for local taxation. The more important are economies of scale in tax administration for a given tax, the stronger is the argument for leaving that tax to the national government. Economies of scale may depend on informational requirements, for example the need for a national taxpayer identification number on technical requirements, such as the use of large computers, etc.; cross-check, etc. This consideration would call for the VAT and for the global income tax to be nationally collected taxes. 40 Box 7 Revenue-Sharing: The Experience ofArgentina The sharing of tax revenues has raised serious problems in certain countries, most evident in the case of Argentina. Major tax reforms were introduced in that country in the late 1980's and succeeded in sharply raising the share of taxes to GDP. However, part of the potential impact of this effort to reduce the public sector's fiscal deficit was dissipated by the revenue-sharing arrangement, which stipulated that 57 percent of any addition to tax revenues be shared with the provincial governments. The latter immediately spent the additional revenue. The attitude of the subnational government has been that economic stabilization is a national public good and is thus the sole responsibility of the national government. The effort of the Argentine central government was also directed at reducing its own spending through, inter alia, reductions in civil service employment. However, at the same time that this was taking place, in part as a result of the additional tax revenue received, the provincial governments were increasing their employment and their total spending. Furthermore, the provinces with the lowest own-revenue mobilization were the ones that increased the size of their civil service the most. Across all provinces, the correlation coefficient between these two variables was about -0.70. It is likely that, at the margin, the benefits lost by reducing central government employment exceeded the benefits gained by the increase at the provincial level. In conclusion, the Argentine sharing arrangements have magnified the effort necessary at the central level to reduce the country's fiscal deficit and have probably reduced the efficiency of public spending. Source: Tanzi (1995). D. Intergovernmental Transfers 117. The key to intergovernmental finance in Sri Lanka will remain the design and implementation of central-regional transfers. The current system of transfers has three components all of which suffer from major weaknesses, especially in that they do not stimulate financial management at the local level. The first component, the block grant, which represents about 65 percent of total grants, is allocated to meet the recurrent expenditures of PCs for programs implemented in respect of functions devolved by the Constitution, and is basically equal to the salary bill associated with these devolved functions less the estimated revenues collected directly by the Provinces. As a result, there is no provincial discretion as to how these transfers can be spent, and there is a direct disincentive for Provinces to collect any revenues of their own. The second component, the matching grant, which represents about 17 percent of total grants, is a matching element to recompense PCs for their revenue collection efforts beyond certain benchmark, namely the 1992 performance. However, in practice no budgetary allocation had been made for this grant until 1995. It has remained fixed since then. The third component, the criteria based grant, representing the remaining 17 percent of total grants, is meant for the discretionary capital expenditure of PCs apportioned on the basis of criteria developed by the Finance Commission. Unfortunately, however, the merits that this component may have on distributive grounds are muted by the fact that it has been frozen (Rs I billion) for some time and no common guidelines have been issued to provinces. The new proposal does not elaborate on transfers, a subject that is left to the new Finance Commission. In the design of these principles in the future, the Finance Commission would need to recognize that while self-financing, a concept at the heart of the justification of the decentralization, is not feasible in Sri Lanka, it is critical that transfers be designed to ensure that adequate incentives for "fiscal responsibility" are provided. The key issue is to link marginal increases in services provided by the Region to additional local taxation. This would ensure the smooth functioning of the welfare maximizing mechanism. For this reason, it may be best to envisage, in a first stage, an amount of transfer "fixed" in advance, which implies that, at the margin, local actions to raise revenues or lower expenditures will directly affect outcomes. 41 E. Borrowing Rights 118. The new proposal departs in an important way from the current system of devolution in that it gives Regional Councils the power to borrow as well as to set up their own financial institutions; only international borrowings above a prescribed limit would require the concurrence of the Center. Cross- country experience on this question indicates that borrowing by subnational jurisdiction can seriously exacerbate a country's macroeconomic difficulties. This danger can be contained only if there are constitutional limitations to local government borrowing, ithe market is able to impose a discipline on the borrowing by these governments, and/or if the national governments never intervene when subnational governments get into financial trouble. Unfortunately, few countries have such strict constitutional limitations; markets have proven remarkably unable to discipline borrowing in part because of informational deficiencies (Lane 1994) and, often for political reasons, central governments are unable to refuse to assist subnational governments. (See Box 8). Box 8 Borrowing by Local Authorities in Certain Countries In Argentina, all levels of government can borrow both domestically and abroad, and the Argentine provinces were able to finance a deficit of about 0.7 percent of GDP in 1994. In Brazil, they can borrow from many sources. Sao Paulo alone is reported to have accumulated a debt of some $40 billion dollars. In India the states can borrow domestically subject to approval by the Center. In Pakistan, there are constitutional controls over provincial borrowing, but they do not seem to have been effective. The finance of Mexico's 32 states have been reported as precarious and some states are bankrupt. In Italy, borrowing by subnational jurisdiction has contributed to the deterioration in the fiscal situation and has led to the emergence of "hidden debt". Some municipalities and regions have been essentially bankrupt. 119. Two important reasons weigh against allowing borrowing rights to Sri Lanka's subnational governments. First, regional authorities will feel it legitimate for them to borrow because their revenue assignments do not match their expenditure responsibilities. Second, the socio-economic context of Sri Lanka is such that most regional authorities will tend to assume implicitly that the central government will step in when they are in financial difficulties. The most important reasons for this risk to materialize are: lack of good expenditure management systems within the subnational jurisdictions, and absence of a good budgetary system capable of preparing competent projections for revenue and expenditures; lack of incentives for local policy makers not to borrow (giving immediate benefits to those in power while the costs are paid later). There is no reason to believe that Sri Lankan local authorities will increase their spending excessively financed by credit from self-created institutions; however, borrowing by local authorities have an unhappy record across countries that have devolved fiscal powers, and the "moral hazard" problem with respect to implicit central guarantees is real. In addition, it will be difficult for the Central Government to restrict borrowing by Provinces if it does not lead the way itself in containing its recourse to bank financing. In this context, it might be advisable for the Law to exclude all possibilities for "off-budget "financing through such regionalfinancial institutions that have marred the recent history of countries like Brazil and Argentina. The right to establish financial institutions should remain within the control of the Central government and these should be subjected to national regulations on financial activity. F. Other Issues 120. Actions to strengthen the institutional aspects of devolution will not yield the expected results without accompanying efforts to strengthen the management capabilities of the regional councils. The weak management capability of PCs was at the source of numerous Center-Province conflicts in the recent devolution experience. It is recognized that most PCs were unable to prepare their required legal statutes 42 for a long time, which partly explains why they mobilized only about half of their revenue potential.49 A legal unit in the Ministry of PCs is available to help provinces, but because of lack of financial resources, it has not been able to grow to an adequate size. As has been observed in many other countries, provinces have not been able to attract more qualified personnel because they could not offer more promising careers and advancement. As a result, line Ministries have mostly continued to implement projects through centralized offices instead of going through the Provincial Council structure. The Government's endeavors to mitigate this problem through secondment of senior staff from the Center and the imposition of common salary structures did not suffice. In the light of this experience, it would be most useful for the new system to be accompanied by a comprehensive effort to strengthen human resource development at the local level. It would also be useful to reduce the influence of political connections in the recruitment and promotion processes. 121. Finally, the Thirteenth Amendment, art. 154G (4), contained a good provision to the effect that any Provincial Council could, in effect, at its choice opt to have central law apply even in areas where power had been constitutionally devolved. In Sri Lanka, as in other countries - Spain, Canada - some regions of the country are clearly much more interested than others in having and exercising regional autonomy in particular fields. It is not clear why they should not be allowed to do so. Such a provision would be useful to permit those who want to exercise their powers (as devolved in the Constitution) to the full to do so without forcing other regions, which may be quite happy to live with central rules in many areas, to do the same. Omission of this provision from the Draft Proposals should be reconsidered. 122. To some extent, much the same result may be achieved - with some assurance in this case that those regions that choose to exercise specific powers or functions with national implications (e.g. education, health) have sufficient capacity to do so - by introducing some process of "certification" as in Columbia. However, care must be taken to ensure that this is not used as an excuse for central retention of powers. Another possible approach to accommodate diverse regional interests and capacities might be to establish a system of "contracting" under which some regions could, if they wished, contract with central departments to provide services at specified levels. Political factors may perhaps preclude discussion of such matters in Sri Lanka; however it is unlikely to be economically or administratively efficient to force everyone to follow the same path. All regions may be created politically equal: but this does not meant that all regions have to do the same thing in the same way at the same time. 123. As the above discussion indicates, the design of an efficient system of devolution is an arduous task and depends on each case. This task cannot be successfully accomplished without two essential elements: (i) establishment of an adequate fiscal and financial framework for the evaluation of alternative schemes for decentralization of public sector activities; and (ii) assigning responsibility for the design of the new devolution package to a central unit, preferably in the Finance Ministry. A common failing in decentralization in many countries has been the lack of an adequate central government institutional structure to bring together the many ways in which decentralization cuts across a wide variety of public sector activities. Some central authority must be established for designing the policy as a whole. It is also likely to prove critical to establish good working relations at the central-regional level from the beginning through intergovernmental committees. The Bank stands ready to provide the Sri Lankan authorities the needed assistance is this task. In a first stage, this assistance could aim at building institutional capacity in the Ministry of Finance to assess the financial implications of alternative devolution schemes, formulating the right policies, drafting and implementing regulations, training of staff, etc. For example, regulatory responsibilities for private provision of road passenger services are currently diffused between a national body-The National Transport Commission-provincial road passenger transport authorities, and what remains of the Central Transport Board. Table 1 of 3 SRI LANKA: SUMMARY OF CENTRAL GOVERNMENT OPERATIONS, 1990-96 1990 1991 1992 1993 1994 1995 1996 1996 la Actuals Budget Proj. (in millions of Sri Lanka rupees) Total revenue 67,963 76,179 85,781 98,339 110,038 136,258 148,206 152,089 Tax 61,206 68,157 76,353 85.891 99.417 118,543 132,626 133,009 Nontax 6,757 8,022 9,428 12,448 10,621 17,715 15,580 19.080 Total expenditure and net lending 99,814 120,368 119,823 141,660 170,759 203,483 229,485 233,065 Current expenditure 71,770 83,756 89,638 102.288 127,079 154,159 169.781 178,605 Capital expenditure 19.161 25,305 24,949 33.662 30,391 41,722 43,203 41,372 Net lending \b 8,882 11,308 5,236 5,710 13,289 7,602 16,501 13,088 Overall deficit (before grants) (31,850) (44,189) (34,042) (43,321) (60,721) (67,225) (81,278) (80,976) Grants 6,697 7,870 8,280 8,025 8,257 9,028 9,000 9,000 Net foreign borrowing 11,644 19,329 7,361 9,855 11.778 21,224 19.078 12,816 Net domestic borrowing \b 13,509 16,990 18,402 25,441 40,686 36,973 53,201 51,898 of which: Banking system 258 35 (2,322) (6,079) 1.156 7,065 .. 9,676 (In percent of GDP) Total revenue 21.1 20.5 20.2 19.7 19.0 20.6 19.1 19.8 Tax 19.0 18.3 18.0 17.2 17.2 17.9 17.1 17.3 Nontax 2.1 2.2 2.2 2.5 1.8 2.7 2.0 2.5 Total expenditure and net lending 31.0 32.3 28.2 28.3 29.5 30.7 29.6 30.4 Current expenditure 22.3 22.5 21.1 20.5 22.0 23.3 21.9 23.3 Capital expenditure 6.0 6.8 5.9 6.7 5.3 6.3 5.6 5.4 Net lending \b 2.8 3.0 1.2 1.1 2.3 1.1 2.1 1.7 Current account balance -1.2 -2.0 -0.9 -0.8 -2.9 -2.7 -2.8 -3.5 Overall deficit (before grants) -9.9 -11.9 -8.0 -8.7 -10.5 -10.2 -10.5 -10.6 Grants 2.1 2.1 1.9 1.6 1.4 1.4 1.2 1.2 Net foreign borrowing 3.6 5.2 1.7 2.0 2.0 3.2 2.5 1.7 Net domestic borrowing \b 4.2 4.6 4.3 5.1 7.0 5.6 6.9 6.8 of which: Banking system 0.1 0.0 -0.5 -1.2 0.2 1.1 1.3 GDP (SL Rs. Mill) 321,800 372.300 425.300 499.800 578,800 661,900 775,820 767,280 Privatization proceeds (SL Rs mill.) 841 2,850 1.200 2,996 3,001 21,000 10,000 \a Including measures being proposed by the authorities during the second half of 1996. \b Include privatization proceeds Source: IMF 44 ANNEX I Table 2 of 3 SRI LANKA - CENTRAL GOVERNMENT REVENUES BY COMPONENT, 1990-96 1990 1991 1992 1993 1994 1995 1996 1996 la Actuals Budget Proj. (in millions of Sri Lanka rupees) Tax Revenue 61,206 68,157 76,353 85,891 99,417 118,543 132,084 133,009 Income 7,337 9,722 10,967 12,543 15,277 17,148 17,149 20,820 olw Corporate 4,380 6,183 6,900 7,308 9.656 9,790 11,762 11,458 Stamp duty, Taxes on T bills 5,758 6,574 5,588 4,566 4,857 6,212 5,820 5,820 Taxes on goods and services 28,770 32,107 38,158 47,963 56,685 70,810 80,957 81.069 olw Turnover Tax 20,291 21,430 24,095 29,663 32,300 36,429 40,745 39,134 Excises 8,170 10,597 10,232 11,655 14,632 19,406 21,875 23,598 Defense Levy .. .. 3,764 6,589 9,693 14,408 16,090 16,090 Taxes on international trade 19,341 19,754 21,640 20,819 22,598 24,373 28,700 25,300 Non-tax revenue 6,757 8,022 9,428 12,448 10,621 17,715 15,580 19,080 olw Property income 4,192 5,964 6,262 8,154 7,640 13,073 9,931 13,431 Total revenue 67,963 76,179 85,781 98,339 110,038 136,258 148,206 152,089 (In percent of GDP) Tax revenue 19.0 18.3 18.0 17.2 17.2 17.9 17.0 17.3 Income tax 2.3 2.6 2.6 2.5 2.6 2.6 2.2 2.7 Taxes on goods and services 8.9 8.6 9.0 9.6 9.8 10.7 10.4 10.6 Taxes on international trade 6.0 5.3 5.1 4.2 3.9 3.7 3.7 3.3 Non-tax revenue 2.1 2.2 2.2 2.5 1.8 2.7 2.0 2.5 Total revenue 21.1 20.5 20.2 19.7 19.0 20.6 19.1 19.8 GDP (SL Rs. Million) 321,800 372,300 425,300 499,800 578,800 661,900 775,820 767,280 \a Including measures being proposed by the authorities during the second half of 1996. Source: IMF 45 ANNEX I Table 3 of 3 SRI LANKA - CLASSIFICATION OF EXPENDITURES, 1990-96 1990 1991 1992 1993 1994 1995 1996 1996 la (Actuals) Budget Proj. (In millions of Sri Lanka rupees) Current expenditures 71,770 83.756 89,638 102,288 127,079 154.159 169,781 178,605 Wages/Salaries \b 15.748 17.985 20.793 16.639 18,700 22.811 26.598 25.110 Other Goods/Services Ib 14,419 17.598 18.898 7.793 9,551 10.176 11.704 11.229 Interest payments 20.668 22.073 25.940 30.203 38,031 38.225 45,457 44,936 Foreign 3,678 4.113 4,739 5.102 5.511 6,162 7,351 6,830 Domestic 16.990 17.960 21.201 25,101 32.520 32.063 38.106 38.106 Subsidies/transfers 16.446 17.268 15,908 16.674 21.082 25.430 30.683 31.971 To Corporations (exci. wheat) 2,639 2.009 1,428 1,571 2,634 1.021 2.638 2,513 To Institutions 1,811 1,404 1.649 2.031 1,976 2.182 2.739 2,739 To other levels of Government 1.237 1,050 1.111 896 1,307 1.157 1.512 1.300 To Households 10,442 12.112 11.230 11,846 14.992 20.907 23.562 25.187 of which: Food/kerosene stamps 3,812 2.999 2,971 2,749 2.690 1.799 989 989 Midday Meal/lUniformJSP 4,556 4,809 4.997 5.664 7,665 5.201 3.091 2.600 Samurdhi .. .. .. 7 2,336 8,703 8,703 wheat .. .. .. .. .. 6.820 3.670 7.371 To private institutions and abroad 317 693 490 330 173 163 232 232 Defense n.a. n.a. n.a. 20.463 26.444 42,460 37,980 48.000 Pensions 4.490 8.832 8.099 10,516 13.271 15,057 17,359 17,359 Capital expenditures 19,161 25.305 24,949 33.662 30.391 41,722 43.203 41.372 Net lending %c 8.882 11.308 5.236 5.710 13.289 7.602 16.501 13.088 o/w Privatization proceeds 841 2.850 1.200 2.996 3.001 21,000 10.000 Total expenditures & net lending 99,814 120.368 119.823 141,660 170,759 203.483 229.485 233.065 (in percent of GDP) Current expenditures 22.3 22.5 21.1 20.5 22.0 23.3 21.9 23.3 otw Wages/Salaries %b 4.9 4.8 4.9 3.3 3.2 3.4 3.4 3.3 Other Goods/Services lb 4.5 4.7 4.4 1.6 1.7 1.5 1.5 1.5 Interest 6.4 5.9 6.1 6.0 6.6 5.8 5.9 5.9 Transfers to HH (incl wheat) 3.2 3.3 2.6 2.4 2.6 3.2 3.0 3.3 Defense n.a. n.a. n.a. n.a. 4.6 6.4 4.9 6.3 Pensions 1.4 2.4 1.9 2.1 2.3 2.3 2.2 2.3 Capital expenditures 6.0 6.8 5.9 6.7 5.3 6.3 5.6 5.4 Net lending %c 2.8 3.0 1.2 1.1 2.3 1.1 2.1 1.7 ow Priv proceeds .. 0.2 0.7 0.2 0.5 0.5 2.7 1.3 Total expenditures and net lending 31.0 32.3 28.2 28.3 29.5 30.7 29.6 30.4 (In percent of total expenditures and net lending) Current expenditures 71.9 69.6 74.8 72.2 74.4 75.8 74.0 76.6 olw Wages/Salares \b 15.8 14.9 17.4 11.7 11.0 11.2 11.6 10.8 Other Goods/Services \b 14.4 14.6 15.8 5.5 5.6 5.0 5.1 4.8 Interest 20.7 18.3 21.6 21.3 22.3 18.8 19.8 19.3 Transfers to HH (incl wheat) 10.5 10.1 9.4 8.4 8.8 10.3 10.3 10.8 Defense n.a. n.a. n.a. n.a. 15.5 20.9 16.6 20.6 Pensions 4.5 7.3 6.8 7.4 7.8 7.4 7.6 7.4 Capital expenditures 19.2 21.0 20.8 23.8 17.8 20.5 18.8 17.8 Net lending 8.9 9.4 4.4 4.0 7.8 3.7 7.2 5.6 olw Priv proceeds .. 0.7 2.4 0.8 1.8 1.5 9.2 4.3 Total expenditures and net lending 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 GOP (SL Rs. M.) 321.800 372.300 425.300 499.800 578.800 661.900 775.800 767.280 %a Including measures being proposed by the authorities during the second half of 1996. \b These numbers include expenditure on defense until 1992. For 1993-1996, defense expenditure is excluded from this item and itemized separately below Ic IMF Definition - includes privatization proceeds. Note: n.a. is not available. Source: IMF m %enciexpendx2is juiy 22196 46 Annex 2 Page 1 of4 Sri Lanka Fiscal Sustainabilty' 1.. The intertemporal budget constraint equation (solvency test) has been applied on Sri Lanka's economic aggregates for recent years. On the basis of this test, it appears that Sri Lanka's fiscal stance over 1991-95 was sustainable. Sri Lanka was able to incur primary deficits of 2-4 percent of GDP without exacerbating the debt/GDP ratio. In fact, the latter has declined since its peak of 109 percent of GDP in 1989. The reasons for this are the relatively high rates of growth enjoyed by the country (5-7 percent per annum), the low rate of interest paid by the Government on its foreign concessional financing (2.5 percent per annum), and an element of nonmarket domestic interest rates in as much as two-thirds of budgetary financing usually come from State controlled savings institutions. Average interest paid on government securities has been in the range of 12- 14 percent per annum over 1990-95, which was slightly above inflation, but well below average commercial bank lending rates. 2. Looking ahead, the factors that have helped maintain Sri Lanka's debt/GDP ratio constant (concessional aid and high growth) cannot be taken for granted in the future (IMF, 1995). Economic growth is projected to fall below 3.5 percent in 1996, and even if external concessional financing were to remain unchanged, the balance of payments pressures that have emerged in 1996 (from declining tourism and transfers, and export slowdown) would inevitably create pressures for a reversal of the liberal policies introduced earlier, thus endangering further growth prospects. 3. With the above-mentioned considerations in mind, the same solvency test has been applied to the Sri Lankan fiscal situation on an ex-ante basis. 2 That is, the fiscal stance has been examined in the overall macroeconomic framework in which Sri Lanka is likely to operate and the objectives that the Government has been assigned, as spelled out in the 1995 and 1996 Budget speeches. The principal goal is to accelerate growth to 7-8 percent by 1997, in a context of low inflation (5-6 percent) and a reduction in the external current account deficit to about 5 percent of GDP. Identifying an appropriate fiscal strategy on the basis of these objectives is complicated by the uncertainties that attend the adjustment process. 4. However, it is possible to lay out the macroeconomic considerations underlying an assessment of the sustainability of fiscal policy in future years. Based on past experience, it is This section has prepared in collaboration with the IMF which has presented similar papers in its 1995 and 1996 Article IV Consultation reports. 2 The primary deficit of the public sector, as a share of GDP, is constrained to not exceed the sum of three financing sources: revenue from seignorage, the excess of domestic growth over the relevant real interest costs of domestic and foreign debt, and increasing demands for monetary and non-monetary debt Primary deficits are sustainable if they do not entail ever-increasing debt and money to income shares. The accounting approach to public solvency defines sustainability in the restrictive sense of constant debt and money to output ratios, consistent with steady state (constant) inflation and interest rates. The following equation, which relates the change in the debt- to-GDP ratio to the sources of deficit financing, can be estimated: db = (r-g)b +(p-s) - (r*e-g)b* where db is the change in the ratio of government debt to GDP; r is the effective real interest rate on domestic debt; r* the interest rate on foreign debt; g is the real GDP growth rate; b is the ratio of public domestic debt to GDP; b* the ratio of foreign public debt to GDP; p is the central government primary fiscal deficit; and s is the seigniorage revenue as a share to GDP; and e is the rate of real exchange rate depreciation. To keep the debt to GDP ratio stable, the following condition must hold: p = s + (g-r)b + (g-r*-e)b* 47 Annex 2 Page 2 of 4 possible to project an upper bound for seignorage revenues of about 1.0 percent of GDP.' and to assume that the interest rate on outstanding government debt rises to 5 percent in real terms, the current marginal cost of borrowing.4 Based on IMF's World Economic Outlook projections, the real interest rate on Sri Lanka's foreign debt will be in the range of 1-2 percent a year (the Japanese consumer price index is projected to be less than I percent a year). A 2-3 percent real depreciation of the Sri Lankan Rupee is also assumed over the next few years, based on productivity differentials between Sri Lanka and its main competitor countries. On the basis of these projections, Sri Lanka's public sector deficit would have to decline to 7-8 percent of GDP to stabilize the debt/GDP ratio. However, due to the higher cost of debt service, the primary deficit would have to be eliminated at the earliest (See table below). Sri Lanka: Sustainable Primary Deficits Under Different Scenarios (In percent) Domestic Real Real Exchange Sustainable GDP Interest Domestic Foreign Real Rate Foreign Debt Primary Balance Growth Rate Debt Stock Interest Rate Depreciation Stock /a /b High Case 7.0 4.0 43.5 1.0 2.0 52.7 -4.4 6.0 5.0 43.5 1.5 3.0 52.7 -2.2 Base Case 5.0 6.0 43.5 2.0 4.0 52.7 0.0 4.0 6.0 43.5 2.0 4.0 52.7 0.9 Low Case 3.0 7.0 43.5 3.0 5.0 52.7 3.4 /a As a share ofGDP, using debt ratios as of 1996. lb The sustainable primary fiscal balance has been calculated using the conventional budget constraint equation indicated in footnote 5. There are two complications to this scenario. First, stabilizing the debt/GDP ratio would probably not suffice to ensure medium-term sustainability. As noted earlier, Sri Lanka's public debt is high in comparison with other indebted countries. Under the debt stabilization scenario, interest payments would consume over one-third of current spending, squeezing out other spending vital to ensuring a strong supply response to the reforms. The high domestic interest rates would have adverse consequences on the economy by hampering deficit reduction and unduly penalizing private sector activity. Second, the solvency exercise does not take into consideration possible shocks to the fiscal position over the next few years. These could come from the Treasury's contingent liabilities accumulated through guarantees accorded to banks for lending to public enterprises (see Chapter VI), or because of financial sector reforms. The EPF has already been promised greater latitude in its investment decisions and the State Mortgage Investment Bank (SMIB) is being privatized. Nonperforming assets exceed 16 percent of banks' portfolios, Average long-term seignorage has been estimated in the range of 1- 2 percent of GDP in developing countries. The estimate for Sri Lanka is in this neighborhood (Easterly/Schmidt-Hebbel 1995). 4 Typically, in countries in which interes rates are market determined, interest rates exceed the rate of growth so that there needs to be a primary surplus to avoid the debt ratio rising inexorably. 48 Annex 2 Page 3 of 4 capitalization of overdue interest has increased, and the large interest rate spread has now become difficult for enterprises to bear. Similar financial sector reforms are known to have shifted deficit financing from implicit financial repression to explicit debt issuing in many developing countries (Argentina, Columbia, Chile, Thailand,'etc.) and impacted the government's debt service obligations strongly (ranging fronm 05 to 2 percent of GDP). 6. The question of reducing the debt ratio came out strongly in the analysis on fiscal sustainability carried out by the IMF.s The exercise analyzed the two alternative macroeconomic scenarios that could unfold over the next five years (1997-2001) based on the Government's objectives and the external environment. It examined the appropriateness of fiscal policy with respect to growth, inflation and, more importantly, debt targets, emphasizing the strong linkages between fiscal, monetary, external and other macroeconomic policies. The first scenario assumed that strong fiscal adjustment and bold structural reforms were implemented beginning in 1997, which is consistent with the Government's medium-term strategy outlined in the 1996 budget speech. The driving force underlying this scenario is a major fiscal consolidation to reduce the government's large claim on resources. Specifically, the decisive fiscal adjustment is reflected in a reduction in the primary fiscal deficit from a projected 2.5 percent of GDP in 1996 to under half of a percent in 1997; thereafter, fiscal policy is assumed to target primary surpluses rising to about 1.2 percent of GDP by the year 2001 (See table below). With strong adjustment, especially in the early years of the scenario, fiscal consolidation would be facilitated by a reduction in the burden of domestic debt service over the medium term arising from declines in inflation and, hence, nominal interest rates. The scenario, however, also incorporates a secular increase in the average real interest rate on public domestic debt. This is a consequence of declining financial repression and the change in the composition of the public debt stock; as the maturing portion of public debt is rolled over at market rates, the share of debt at higher interest rates increases. Nevertheless, the interest bill is expected to decline relative to GDP, and the overall fiscal deficit would show an improvement of about 6 percentage points of GDP over the medium term. (The interest bill would also decline from over one-third of tax revenues to less than a fifth). Public debt would decline markedly from nearly 95 percent of GDP in 1996 to about 70 percent by 2001, indicating sustainabililty of the fiscal situation in this scenario. The high investment and growth rates in this scenario would be sustained by relatively high levels of domestic and foreign savings. The domestic savings rate increases by more than 7 percentage points of GDP between 1997 and 2001. Although this is mainly due to the improved public savings performance, private savings are also assumed to have an upward trend due to income growth and financial sector reform. Foreign savings are assumed to remain fairly high throughout the period, reflecting a substantial increase in non-debt creating flows such as foreign direct investment from 1997 onwards. 7. In the absence of strong fiscal adjustment and structural reforms, the second scenario, the medium-term macroeconomic outlook is likely to be significantly different as the Government continues to crowd out private investment. Specifically, primary fiscal deficits are assumed to persist for the next five years. Although a primary surplus could be achieved in 2001, this adjustment path would yield only a modest reduction in the overall fiscal deficit because of continued high interest costs as a proportion of GDP. Interest payments would absorb over one- third of tax revenues, suggesting continued vulnerability of the budget to changes in macroeconomic conditions. In the context of lingering inflationary pressures and high interest rates, growth would likely stagnate. Background paper for the 1996 Article IV consultations, July 17, 1996. Annex 2 Page 4 of 4 Figure 1. Fiscal Deficit (Before Grants) Percent of GDP 12 ...... -Overatl deficit 6 -- - -- ----- - --- - ------ ----- 4 ----- -- -- -- --- CI Primary defic 1990 1991 1992 1993 1994 1995 * 1996** *Provisional ** Projected Source: World Bank and DefFEstimates Figure 2. Composition of Current Expaulitures, 1996 Wages/ Others Salares Defense 10% 0% 14% 27% Pensions 10% Transfers Interest a/ 25% 14% Source: World Bank and IMF Estimates a/Include, %*eat subsidiser but ecludes pensions Annex 3 50 Page 1 of 9 Sri Lanka Civil Service Background Information 1. The Public Sector. The Sri Lankan public sector comprises employees of: the core Government civilian ministries; Provincial Councils created by the 13th Amendment to the Constitution in 1987; the security forces (military and police forces); semi-autonomous statutory boards mainly funded by transfers; the plantation sector and commercial public companies. Charts I and 2 compare the size and structure of public sector employment in 1988 and 1993, the last year in which comprehensive information on the public sector was available. Fgre 1A SnLa*a, PuMi Secor, 198 gme 18. SW Lata, Pubr Secir, 199 Tool Staff : 1,2!%814 Tool Sff: 880,818 1/ Planta- Teachers tions state 1%2% stat Enterp. b Enterp. Teachers 21% 1%20% central oard 24% Prov. Councils Secutity 12% Planta- Forces tions Securit 19% c r gForces Central 9% Gvt. 28% Boards 3% Sources: Sri Lanka Budget Estirmies 1988-93 and 11 Note that the latest public sector rmnpover survey, Dept. of Pubic Enterpdses Surveys of te arch 1994 esitrmtes public sector eployrmnt at 812354. Since he survey excludes official Gnguresof78, ysolers int niry,te esi. Ts are consistent with t reported ligure above. 2. In 1988, public sector employment totaled some 1.3 million staff Since then, the government has undertaken a major public enterprise reform which has reduced the number of employees by some 540,000 employees - 410,000 in plantations 1 and 130,000 in other enterprises. l A second major structural change occurred in 1992 as a result of devolution which shifted more than half of the civil service from the central government to provincial councils. 3. Civil Service EmployMen. While the size of the public sector has been dramatically reduced, employment in the government has continued its inexorable expansion. Between 1985 and 1996, the Government (central and provincial) added 150,000 jobs to the public payroll (13 7,000 since 1990). Thus, in 1996, the civil service was 37 percent larger than it was in 1985, when the Government commissioned Except for 42 estates of the Janatha Estate Development Board and the Sri Lanka Plantation Corporation, all other estates are managed by 22 management corporations. A key condition of the management contracts is an embargo on shedding plantation workers. Plantation staff transferred to the management companies numbered 410,000. For the purpose of this analysis, state enterprises with government equity of 50 percent or less are not considered pars of the public sector. 51 Annex 3 Page 2 of 9 the ARC report (33 percent since 1990, when the administrative reform was undertaken). As a share of population, government employment, including the military, increased from 3.1 to 3.4 percent. Fig. 2A: Total Civil Service Employment, 1985-96 Fig. 2B: Employment by Level of Government, 1990-96 Thousands and Percentages of Population Thousands 700 3.5 400.0 %of Thousand Central Prov. Population employees 350.0 Gvt Council 600 3.3 300.0 2.0100 x : 32 I20.0 200 1.09 100 2 ..2.8 50.0 1985 1988 1990 1992 1995 1996 1990 1992 1995 1996 x Sources: Budget Estimates, Quarterly Survey. 4. Job creation in the Central Government since 1993 is summarized in Table 1. In the Provincial Councils, it mainly involved hiring teachers- 14,000 in 1992 and 35,000 in 1994-95. 3 The main increases occurred in: 4 (a) the police department with the integration of 33,000 part-time police officers into the civil service to strengthen internal security and free up soldiers for front line duties; (b) the education and technical training sub-sectors. This is, in part, a result of the transfer of secondary school teachers from the Provincial Councils to the Center, the hiring of new teachers, and the creation of a technical training department in the Central Government; (c) the health establishment. In most years, the number of doctors increase by 500-700 per year, as all interns and graduates from medical school in-country and abroad are regularly hired by the Government. Nurses are also given permanent jobs after their internship and the large increase in health staff in 1996 reflects this policy. Two new hospitals were created in 1995 which further sustained the rise in health sector staff, (d) the Agricultural and Irrigation Departments. Increases in staff resulted from the transfer of agricultural agents previously working on donor-supported projects. They include redundant staff of the Mahaweli development authority; It should be noted that trainees of all types (teachers, medical interns, nurses) are not recorded as employees until they graduate and are integrated into the civil service. In the budget, the receive allowances rather than salaries. For this reason, figure 2A and 2B underestimate government staff by at least 35,000, equivalent to the numbers of teacher trainees who were waiting to be integrated into the civil service when the 1996 budget was formulated. No information is available on recruitment in the military. 52 Annex 3 Page 3 of 9 (e) the Divisional and District Secretariats. Major increases in the number of cadre positions occurred in 1992/93 as a result of the creation of Divisional Secretariats. Many of the new positions were left vacant, particularly in the Secretariats located in the war zone. In 1995, a number of vacant positions were taken out of the cadre. In 1996, the 270 Divisional Secretariats were merged into 25 District Secretariats. Among other effects, this resulted in the transfer of about 13,000 front-line village political officials (Grama Niladharis) from the Provincial Councils to the Central Government ' and a strengthening of supervisory staff, and (f) the Auditor General's office and the Ministry of Finance, reflecting programs to improve financial management and accountability. Reductions in the number of budgeted positions occurred: in the Railway Department, reflecting the government's policy of reducing excess manpower through attrition; and, the Divisional Secretariats (1995) and the Survey Department (1996), most likely reflecting the existence of vacant positions. In 1996, there are indications that the Departments of Agriculture and Labor retrenched 7,000 redundant staff. 6 Table 1: Central Government, Changes in Budgeted Positions, 1993-96 1993 1994 1995 1996 Auditor General's Office - 537 - - Police Department 33,656 -1,965 -123 - General Education 657 3,398 5,492 4,359 Dept. of Higher Education - - -2,216 - Dept. of Technical Training 2,071 167 508 - Hospital Staff 110 1,909 2,054 5,229 Ministry of Finance 775 105 146 - Dept. of Railways -993 -2,006 - Dept. of Agnculture - 4,985 -1,180 - Survey Dept. - - - -2,000 Dept. of Irrigation 1,325 462 422 Postal Dept. 275 689 109 Government Printing Office 1,007 - - - Divisional and District Secretariats 8,308 -216 -5,100 14,832 Miscellaneous 1,240 -235 2,449 -966 Total 48,693 7,059 3,141 21,533 Sources: Budget Estimates, 1993-96, Government of Sri Lanka. 4. Actual employment is difficult to quantify since there is no central management of the civil service and, thus, an absence of a central data bank of Government staff From 1990 to 1995, a quarterly survey was carried out by the Ministry of Policy, Planning and Implementation (MPPI), which monitored approved positions versus actual employment for most staff in the Provincial Councils and for about 65 percent of budgeted positions in the Central Government. The results of the survey (Figures 3A-3D) show that the total number of vacant positions peaked at 61,000 in March 1993. As of March 1995, there were only 18,000 empty positions, mostly in the Provincial Councils. Among non-reporting government units, the most important was the police department. At end September, 1996, the Government reports and estimated 20,000 vacant positions. 5 Note that the Grama Niladharis consist of village headmen who were transferred to Provincial Councils after devolution in 1987. This has been confirmed by officials from the Departments of Agriculture and Pension. It points to the fact that selective retrenchment of staff is possible. Every four years, the Department of Census and Statistics carries out a census of the staff in the public sector, including semi-independent institutions and corporations. The last published census dates back to 1990. The results of the 1994 census are currently being processed. While periodic surveys and census are useful tools for understanding the historical developments in the public sector, they are not adequate as management tools to manage Sri Lanka's large civil service labor market. .AnnexS 53 Page 4 of9 Figure 3A. Quarterly Survey, Central Government, 1990.95 Figure 38. Quarterly Survey, Provincial Councils 1/, 1990-95 Thousands Thousands 4 o ade( Cadre Positions 140 s n0 ,.D 90 Dec-91 Dec-92 Mar-93 Dec-93 Dec-94 Me-95 D90 Dec91 Deo-92 Mar 93 De.3 Dec-94 Ma-95 -20 -2 11 Not including teachers. FIgure 3C. Quarterly Survy, Teachers, 1990-95 FIgura 3D. Approved Psaldens & Actual Enployment Thousands 1990-95 150 350 Approved Emploryees 0 Approved VaatEmployees 0 Cadre 00 Cadre P go- ~ Vacat 4 6-Positions 20VcD 40 De Deo-91 Dec-92 Mar-93 Dec-93 Deo-94 Mar-95 0 *50 e"Dec-90 e91 Dec-92 Mar-93 Dec-93 De-94 Mar.95 so .ces: MPP2Ia y suwys,1980-199& 5. After March 1995, the survey was discontinued and development in employment are unclear. Recruitment has apparently continued unabated and may even have accelerated. In all likelihood the bulk of vacant positions in the Provincial Councils have been filled, while at least 35,000 teacher trainees were recruited. As of end October 1996, about 25,000 trainees had been inducted into the newly created teacher service and the remaining 10,000 were actively lobbying the Government to be integrated into the civil service. The Government indicated that there were some 20,000 vacancies in the whole civil service. 6. The most egregious example of a recent increase in public employment is the Samurdhi program. This initiative at first recruited some 24,000 mobilizers among unemployed youth on a one-year contract basis at a cost of about Rs 1 billion. By end-October 1996, the number of mobilizers had increased to 38,000. Hiring is on the basis of recommendations from local members of Parliament and the recruits are generally perceived as party activists. No training is provided. A renewal of the contracts for a second year is planned and there are expectations that these contractuals will eventually be made permanent civil servants. In addition, there are plans to support the mobilizers with a large corps of volunteers who would receive a small stipend and expenses. 7. Contractual Employees. In the Central Government, about 2,500 contractual staff are Although not shown in the tables, the number of contractual staff employed by the central government has declined since 1991 (no information on contract staff is available in provincial councils). This is mostly the result 54 Annex 3 Page 5 of 9 of allowing contractuals to join the civil service. 8 In Sri Lanka, it appears that contractual staff have a legal right to a permanent position in the civil service after ten years or 2,000 working days of service. Staff Groups 8. The government of Sri Lanka broadly categorizes staff grades as follows: grade A consists of senior managers, from Minister to Director General; grade B, middle managers; grade C includes policemen, clerical workers as well as most primary school teachers and equivalent grades in health services and most secretarial support staff, grade D consists of manual workers, drivers, office cleaners, etc. 9. Figure 4A shows trends in civil service employment by staff grades since 1985. ' The number of manual workers (category D) declined in absolute terms but there was a surge in staff numbers in other categories. The number of mid-level managers (grade B) as well senior officials (grade A) doubled during the same period. The most significant increase occurred at grade C reflecting the rapid rise in the number of teachers, health workers and policemen. R9W 44 OWSume SAWGu= 1aS95 FR 4c BaWGovAW&mentSuffGoePSbysev, "W /Tormandr; tihousexbJ B: 1. C: 273.7 300 Pat Eco Mgmt A 1W SODaSlY 100 1I36 D:134 0 10 20 30 40 50 so 70 0 90 Sourc: Budget Estrus, 198546 Soes: Bidget Esgrmts. 186 10. Figure 4B provides a breakdown of staff grades by functional groups of ministries and departments in 1996. 30 percent of the staff are in the country's security forces and 21 percent are employed in political and general economic management functions. The infrastructure and services groups account for 20 percent of the total, mostly the Railways and Postal Departments. Social sectors represent about 18 percent. The composition of staff groups by category reflects the nature of the services being provided: in the security forces and the political and economic management groups, most of the staff consists of policemen and clerical level staff at grade C; in the production, infrastructure and services groups, grades C and D staff are about equally represented; and in the social sectors, where teachers, medical staff and nurses constitute the majority of the staff, grade C staff is the most important cohort. I1. From the figures abovS the sectoral composition of staff by grade has implications for future efforts at trimming the size of the civil service. Grades C and D are clearly the main groups where there The policy of taking contractuals into the civil service is counter-productive over the longer run since it reduces the government's ability to adjust the skills mix as government functions change. 9 In this chart, the security forces only include civilian staff, mostly in the police department. The size has been quoted as standing at 78,000 for more than a decade. 55 Annex 3 Page 6 of 9 are likely to be redundant staff Grade C staff consist mainly of policemen and teachers (60 percent of all C level staff). Reducing the number of police officers will be difficult until the security situation stabilizes. As for teachers, a staff rationalization program based on attrition is currently being implemented (IDA- financed TTRP) which is expected to gradually reduce over-staffing over a period of ten years while strengthening qualifications overall. This program is only in its first. year of implementation and its long- term success has yet to be demonstrated. About 50,000 grade C staff are employed by Provincial Councils, whose cadre is currently being reviewed not knowping whether it will be expanded or contracted. Out of the remaining 100,000 level C staff, about 25,000 are employed in the Railways and Postal Departments. The Postal Department is currently undertaking a program to rationalize its operations, including selective adjustments in the size of the staff. In the Railway Department, the number of budgeted positions have been reduced over time. However, the Railway still remains over-staffed. "o Both the Postal and Railway Departments would benefit from being transformed into commercial or quasi- commercial entities with consequent adjustments in the size and composition of their staff a major focus for reducing costs and improving efficiency over time. Thus, for grade C staff, across-the-board staff adjustments would need to be selective. One approach could be to focus on lower-level staff categories (secretarial and clerical functions) and develop common staffing ratios for general administrative functions. The staff rationalization strategies should take into account the streamlining of clerical activities and the introduction of informatics, the main avenues of productivity improvements available for Sri Lanka at this juncture. 12. Grade D staff in the Central Government total about 70,000. They represent another large cohort where there exist obvious redundancies. Since many of the activities provided by this category of staff are already standardized, a possible approach to reducing staff would be to "contract out" routine work performed by minor officials to the private sector. Statutory Boards 13. Statutory boards are government institutions created by parliamentary statutes to deliver services to various categories of beneficiaries. Sri Lanka distinguishes several categories of boards, including educational (universities and affiliated colleges), technical training institutes, regulatory and promotional agencies, and development authorities. There were 104 boards in existence in 1993, the last year for which there exist comprehensive statistics, and the average age of existing boards was 17.5 years. Since then, four new marketing boards have been created. The longevity of boards is indicative of the boards' legal standing: once created, they are seldom, if ever, abolished. " 14. Total employment in the boards has stabilized at about 45,000 employees since 1993 (Figure 5). Within this number, there have been substantial variations among the main sectors, with development authorities losing staff (mainly Mahaweli) and others gaining staff, notably the university sector, reflecting the creation of colleges affiliated to the main universities since the early 1990s. 12 One factor which may have limited the growth of employment in the boards is their dependence on government grants. This imposes an annual hard budget constraint on employment during the fiscal year. Also, the boards' pension system follows practices in the private sector: contributions to the EPF raise the cost of employment and thereby provides clear signals to managers on the actual cost of staff employment. 10 The ARC Report of 1987 compared the Sri Lankan Railways to the Malaysian system and estimated that its staff was double the size of Malaysia with the same size transportation system. 11 Note that the various transport boards, which were peoplelized by the government in the early 1990s are considered commercial boards and are not covered in this section. 12 In Sri Lanka, the college, university and post-graduate schools system consists of a number of semi-independent institutions. Thus, the higher education system is not integrated into the civil service. The system is regulated by a University Grants Commission, through which Government grants are allocated. A notable change in the higher education system, of late, has been the integration of the network of affiliated colleges into the main universities. An effort to rationalize the university network has been undertaken, including mergers and the creation of new institutions. A constant concern is the apparent disconnect between the curricula of the universities and the needs of the emerging market economy. Overcoming this increasing irelevance of universities had been the main reason for the creation of a network of small affiliated colleges since the early 1990s, which are now being absorbed into the mainstream of university teaching. 56 Annex I Page 7 of 9 Figure 5. Statutory Boards, Enploy7mnt 1988-93 Figure 6. Sta tutory Boards, Conpensa fion, 1988-96 Saffnumber Rs per year 45.000 40.000 100 35oo Education 80 3o.oon Regulation 2M00M Promotio P mo m0 Develop- 2oconmenit 1988 100 103 Sources: Department of Public Enterprises, Annual Surveys 15. While compensation policies follow those of the central government, management has some leeway in rewarding staff and adjusting compensation to reflect market conditions. In Figure 6, average compensation appeared to have been nearly identical across the main functional groups in the 1980s. In the 1990s, however, substantial differences have emerged: universities and other teaching establishments have outpaced other groups in improving staff compensation. 16. The government's administrative reform program did not include a review of the effectiveness of the boards. Such a review is overdue since many of the boards were established to support government past interventionist policies. In particular, a review of promotion and development boards should be placed high on the agenda of a future reform program. A major review of the Mahaweli Development Board is currently underway and a program to restructure this large development authority is being discussed with the Bank. Compensation 17. Compensation in the civil service consists of salaries, monetary allowances and in-kind subsidies (housing and cars), the latter accorded to a limited number of senior staff members. Civil servants do not pay taxes, although a special levy has been collected since 1995 to help finance the war. Civil servants benefit from low interest housing loans. They do not contribute to their pension scheme, except for the Widows and Orphans Scheme (see Chapter IV). 18. As of end 1996, the pay system consists of 94 separate pay scales within 12 broad bands, with a minimum salary of Rs. 25,500/year for the lowest level staff upon entry to a maximum of Rs 174,600 for the highest paid official in the All-Island Service (Secretary to a Minister). The highest salary paid in the civil service is to the Chief Justice of the Supreme Court (Rs 234,000/year). Each pay scale has annual increments (except for the highest level of judicial officers), with a maximum 23 increments for one category of manual worker. Increments are accorded regularly to staff, apparently without taking performance into account. The overall effect on the wage bill is to increase it by three percent per year. Annex 3 Page 8 of 9 19. Basic pay has been regularly supplemented by a number of monetary and non-monetary allowances. There is a long history in Sri Lanka of according allowances in lieu of regular salary increases (cost of living allowances--COLA, devaluation allowances, interim allowances, efficiency bars, transportation, and a number of special allowances to reflect differences in technical requirements of various jobs in the government, negotiated on a case-by-case basis by each technical service). The 1988 salary reform, recommended by the ARC, resulted in the integration of all allowances into basic pay, except for a COLA accorded to lower level staff. After its introduction in 1988, the COLA was increased in 1989, 1990 and 1992 for certain categories of staff. Generally, higher level staff were not given allowances so that their relative compensation lost ground to that received by lower-level staff throughout 1988-93. 20. Some staff categories, such as judges and doctors have received additional allowances. Non- monetary allowances, such as low housing rent, use of official vehicles (minus a small deduction for personal use) and product subsidies (fuel allowances) are also common for senior officials. Some senior officials also receive entertainment allowances, depending on their function. The COLA and other special allowances, which compensate for increases in the cost of living, are mostly pensionable. 21. Sri Lanka has practiced a policy of infrequent salary adjustments, which vacillate between a "clean" salary structure with no allowances (except for a COLA for junior staff) and a pay structure to which are added a variety of allowances. Since independence the practice has been to adjust compensation after inflationary pressures, in many instances with a lag of several years. Another trend has consisted of adding salary scales to the general salary structure to reflect the specific circumstances of various categories of staff. In 1986, the 150 different pay scales were reduced to 40. Since then, the number has increased gradually to 94 in 1996. Such a large inventory of pay scales reduces the transparency of the system and opens the way for competitive pressure by various services to secure their own preferential pay scale. 22. The absence of a transparent, unified pay policy is apparently the main factor behind a major muddle in compensation practices which occurred prior to and immediately after the 1994 elections. Starting with a major salary adjustment granted to teachers and to mid-level technical service staff members, several discrete modifications to individual pay scales were accorded to specific services. Higher-level staff obtained allowances in lieu of salary increases. These adjustments contributed to upset relative compensation packages, created widespread discontent in the civil service and exacerbated working relations among the various staff groups. A National Salary Commission convened in 1995 to examine the distortions created and to recommend a policy to restore balance in compensation. After examining numerous representations by various interest groups, representatives of the various services and their trade unions, the Commission recommended a first adjustment which would consist of adjusting compensation scales (salaries and allowances) to reflect the relative position of the various staff groups recommended by the ARC. Since these relativities were reflected in the 1993 pay adjustment, it was recommended that a pay adjustment be undertaken to restore the relative pay of the various staff groups to the 1993. From this revised structure, which would be achieved by integrating all compensation components into basic pay, a new salary structure would be established, built up from a base pay of Rs 4,000 per month for the lowest level entry staff member. Civil service staff would continue to be tax exempt and would also continue to benefit from the present pension provisions. The base cost of the adjustment was estimated at Rs 14 billion. Table 2 shows the main trends in the evolution of wage scales by grade since 1988. 1 The table shows: * a persistent increase in the use of allowances to augment salaries of higher level staff throughout 1988- 96; The average wage rates throughout 1988-96 were calculated using 1996 budget staff levels. Annex I 58 Page 9 of9 * a persistent decline in compression ratios, with negative consequences on monetary incentives for higher level staff members; * the very large corrections required in most staff grades to re-establish relative salaries to adequate levels as well as to provide for minimum standiids of living for staff in the lowest pay grades. Table 2: Sri Lanka, Developments in Civil Service Pay Rates, 1988-96. 1988 1993 1996 Salary Commission Staff Average PEs AllowlPEs Average PEs Allow/PEs Average PEs Allow/PEs Average PEs % Inc % Inc Grades Rs'000 Rs'000 Rs'000 Rs'000 from 1996 from 1993 D 23.0 13.5% 38.3 36.9% 42.9 24.0% 64.0 49.1% 67.0% C 28.4 4.6% 46.5 24.9% 61.8 15.4% 84.4 36.6% 81.8% B 50.8 0.00/0 73.9 8.0% 93.2 22.3% 143.1 53.6% 93.7% A 99.3 0.0% 131.6 0.0% 168.1 18.6% 259.0 54.1% 96.7% Total 29.6 5.3% 47.6 19.2% 61.1 17.7% 86.1 41.1% 81.1% Compression Ratios based on: Wages 6.8 6.5 5.8 5.8 PEs 5.8 4.6 5.1 Sources: Reports of the Salary and Cadre Committee, 1988, 1993 and of the Pay Commission of 1995. 23. Sri Lanka has yet to carry a comprehensive wage comparison survey which would take into account all elements of compensation. However, a partial survey conducted in 1995 by the Salary Commission confirms that the Sri Lankan civil service remains un-competitive relative to private sector employers, except for lower grades. At the senior management level, the difference in pay is as high as 300 percent. It remains high at mid- and junior management level. At the lowest entry level, the civil service offers better conditions. Although the wage comparison took into account all wage deductions in the private sector (taxes and contributions to social security) it did not factor into the comparisons the real benefits provided by the government's generous pension system. In effect, the government social safety net package may be a key security factor which makes employment in the civil service attractive to all but the highest level staff. 24. Among key recommendations of the Salary Commission, one merits special attention: the creation of a permanent National Commission on Civil Service Salary and Cadre Issues, which would provide a much needed high level focus for investigating critical issues pertaining to changes in the status of the various professional groups and assist the government in adjudicating differences which may arise from time to time as the market for the skills required by the Government evolves over time. November 5, 1996 59 Annex 4 Page I of 3 Sri Lanka -- Social Transfer Programs in 1995. Estimate Program 1995 Budget 95 1995 Baseline /a Program Number of Period Monthly Cost Total Cost Total Cost Total Cost Total Cost Benficiaries la (Rs) (million Rs) (million Rs) (million Rs) (million Rs) Samurdhi Total 3,507 500 2,336 - Mobilizers (Niyamaka) 22,000 Jun.-Dec. 2,000 308 - Housholds ultra poor 100,000 Aug.-Dec. 1,000 500 Households of > = 3 members 982,000 Aug.-Dec. 500 2,455 - Households of 2 members 158,000 Aug.-Dec. 200 158 - Households of 1 member 120,000 Aug.-Dec. 100 60 - Managers 1,500 Aug.-Dec. 3,500 26 Stamps Total 2,621 2,910 1,833 3,095 Food Stamps 1,500,000 Jan.-Aug. 150 1,800 2,100 1,362 2,250 Food Stamps (North/East) 359,000 Sep.-Dec. 120 172 - - - Kerosene Stamps 1,200.000 Jan.-Aug. 48 461 685 437 720 Kerosene Stamps (North/East) 325,000 Sep.-Dec. 48 62 - - Infant Milk Stamps Jan.-Dec. 125 125 34 125 Janasaviva Total 3,143 3.500 2,794 3,136 Interest Round 1 118,000 Jan.-Dec. 250 354 360 Interest Round 2 103,000 Jan.-Dec. 250 309 300 Interest Round 3 100,000 Jan.-Dec. 250 300 Interest Round 4 99,000 Apr.-Dec. 250 223 Janas. Round 4 99,000 Jan.-Mar. 1,458 433 420 Janas. Round 5 120,000 Jan.-Jul. 1,458 1,225 2,056 Janas. Round 5 120,000 Aug.-Dec. 500 300 Other Welfare Program 5,556 6,701 6,119 6,353 Mid-Day Meal (children) 4,300,000 Jan.-Aug. 166 1,328 2,300 1.778 2,500 School Uniforms 4,300,000 Jan.-Dec. 600 600 629 650 School Textbooks 4,300,000 Jan.-Dec. 425 425 409 300 Triposha 750,000 Jan.-Dec. 175 225 169 175 School Season Tickets Jan.-Dec. 150 150 133 150 Emergency Assistance (refugees) 400,000 Jan.-Dec. 500-1,400 2,128 2,128 2,128 2,128 Public Assistance (widows, orphans) 350,000 Jan.-Dec. 750 750 750 450 Assistance to wounded/disabled 123 123 soldiers Sub-Total: 14,827 13,611 13,082 12,584 Wheat flour subsidy \c Jan.-Dec. 6,800 5,000 5,000 - Fertilizer subsidy \d Jan.-Dec. 1,500 1.500 1,345 - Total: 23,127 20,111 19,427 12,584 Memo: Sub-Total as % of GDP: 2.24% 2.06% 1.98% 1.90% Total as % of GDP: 3.49% 3.04% 2.94% 1.90% al Number of households unless otherwise indicated. b/ Part of a Block Grant. c/ Does not include the indirect cost of the Paddy Marketing Board's activities to sustain price of rice. d/ Includes the fertilizers Urea, Sulphate of Ammonia. Muriate of Potash, and Triple Super Phosphate. Source: 1995 Budget and Bank staff estimate. n:Iannisalperlsamurdhi.xis 60 Annex 4 Page 2 of 3 Sri Lanka -- Social Transfer Programs in 1996. Current Plan 1996 Budget 1996 Program Number of Period Monthly Cost Total Cost Total Cost Benficiaries \a (Rs) (million Rs) (million Rs) Samurdhi Total 8,447 8,432 \b Mobilizers (Niyamaka) 24,000 Jan.-Dec. 2,000 576 Housholds ultra poor 100,000 Jan.-Dec. 1,000 1,200 Households of > = 3 members 982,000 Jan.-Dec. 500 5,892 Households of 2 members 158,000 Jan.-Dec. 200 379 Households of 1 member 120,000 Jan.-Dec. 100 144 Housholds transferred from Janasaviya 200,000 Sep.-Dec. 250 200 \c Managers 1,500 Jan.-Dec. 3,500 56 \d Stamps Total 969 1,059 Food Stamps (North/East) 359,000 Jan.-Dec. 150 646 761 Kerosene Stamps (North/East) 300,000 Jan.-Dec. 48 173 228 Infant Milk Stamps 125,000 Jan.-Dec. - 100 150 70 Janasaviva Total 843 \c 1,991 Interest Round 1 118,000 Jan.-Aug. 250 236 Interest Round 2 103,418 Jan.-Aug. 250 207 Interest Round 3 100,935 Jan.-Aug. 250 202 Interest Round 4 99,362 Jan.-Aug. 250 199 Other Welfare Prosram 5,041 5,041 School Uniforms Jan.-Dec. 1,100 1,100 School Textbooks Jan.-Dec. 525 525 Triposha Jan.-Dec. 250 250 School Season Tickets Jan.-Dec. 165 165 Emergency Assistance (refugees) 400,000 Jan.-Dec. 500-1,400 2,128 2,128 Public Assistance (widows, orphans) etc. 350,000 Jan.-Dec. 750 750 Assistance to wounded/disabled 123 123 soldiers Sub-Total: 15,300 16,523 Wheat flour subsidy \e Jan.-Dec. 7,000 \f 3,670 Fertilizer subsidy \g Jan.-Dec. 1,500 1,500 Total: 23,800 21,693 Memo: Sub-Total as % of GDP: 2.00% 2.16% Total as % of GDP: 3.11% 2.84% a/ Number of households unless otherwise indicated. b/ The budget allocation for Janasviya recipients is included in the Jansaviya budget. c/ Janasaviva Program is expected to be scrapped with the repeal of the Janasaviya Act. The poor families will be determined through a survey and those selected will be placed on a monthly payment of Rs 250. d/ Monthly cost of Rs 2,500 January through May and of Rs 3.500 thereafter. el Does not include the indirect cost of the Paddy MarketingBoard's activities to sustain price of rice. f/ Assumes substantial further reductions from the present wholesale price of Rs 11.65 per kg. A continuation of the present scheme would result in a cost of more than Rs 7 billion. 7/ Includes the fertilizers Urea, Sulphate of Ammonia, Muriate of Potash, and Triple Super Phosphate. Source: 1996 Budget and Bank staff estimate. m:Aannisalperlsamurdhi.xis 61 Annex 4 Page 3 of 3 Sri Lanka -- Social Transfer Programs in 1997. With And Current Plan 1997 measures peace \a Program Number of Period Monthly Cost Total Cost Total Cost Total Cost Benficiaries \b (Rs) million Rs) (million Rs) (million Rs) Samurdhi Total 8,854 8,254 8,254 Mobilizers (Niyamaka) 24,000 Jan.-Dec. 2,000 576 576 576 Housholds ultra poor 100,000 Jan.-Dec. 1,000 1,200 1,200 1,200 Households of > = 3 members 982,000 Jan.-Dec. 500 5,892 5,892 5,892 Households of 2 members 158,000 Jan.-Dec. 200 379 379 379 Households of 1 member 120,000 Jan.-Dec. 100 144 144 144 Housholds transferred from Janasaviya 200,000 Jan.-Dec. 250 600 \c - - Managers 1,500 Jan.-Dec. 3,500 63 63 63 Stamps Total 969 969 150 Food Stamps (North/East) 359,000 Jan.-Dec. 150 646 646 - Kerosene Stamps (North/East) 300,000 Jan.-Dec. 48 173 173 - Infant Milk Stamps 125,000 Jan.-Dec. 100 150 150 150 Other Welfare Program 5,041 4,413 2,913 School Uniforms Jan.-Dec. 1,100 1,100 1,100 School Textbooks Jan.-Dec. 525 525 525 Triposha Jan.-Dec. 250 250 250 School Season Tickets Jan.-Dec. 165 165 165 Emergency Assistance (refugees) 400,000 Jan.-Dec. 500-1,400 2,128 1,500 - Public Assistance (widows, orphans) etc. 350,000 Jan.-Dec. 750 750 750 Assistance to wounded/disabled 123 123 123 soldiers Sub-Total: 14,864 13,636 11,317 Wheat flour subsidy Jan.-Dec. - Fertilizer subsidy \d Jan.-Dec. 1,500 1,500 1,500 Total: .16,364 15,136 12,817 Memo: Sub-Total as % of GDP: 1.69% 1.55% 1.29% Total as % of GDP: 3.50% 1.72% 1.46% a/ Assumes improved conditions in the civil war which would enable recipients in the North and East to be screened and, if necessary, integrated into the Samurdhi program. b/ Number of households unless otherwise indicated. c/ Janasaviya Program is expected to be scrapped with the repeal of the Janasaviya Act. The poor families will be determined through a survey and those selected will be placed on a monthly payment of Rs 250. d/ Includes the fertilizers Urea, Sulphate of Ammonia, Muriate of Potash, and Triple Super Phosphate. Source: Bank staf estimate. m:annisalperisamurdhi.xIs 62 Annex 5 Page 1 of 2 Sri Lanka Background Information on Public Enterprises 1. The Paddy Marketing Board (PMB): Latest information available on PMB's financial operations indicate that it incurred a loss of Rs. 175 million on its trading operations in 1994 due essentially to Government's price support policy. Set up in 1972 to ensure a minimum price to paddy farmers, PMB operates the Government's Guaranteed Price Scheme (GPS) for paddy and guarantees a floor price of Rs. 7.42 per kilo to the farmer. It purchases all paddy brought to the Board at that floor price, irrespective of market conditions. The paddy is milled largely by private millers and the rice is sold in the open market.' In times of good harvest, the PMB has purchased large amounts of paddy, financed largely by government guaranteed loans from People's Bank. Recently a poor harvest resulted in high paddy prices for farmers, who sold most of their outputs in the market, rather than to the PMB. This situation was exacerbated by the large subsidy on wheat since 1994, which has switched consumption out of rice. As a result of all these difficulties, PMB cannot meet its loan obligation of about Rs 2 billion to the People's Bank. These loans have been written off by People's Bank and are being repaid by the Government as part of the Rs. 20 billion re-capitalization operation mentioned above. According to a recent World Bank Report, the GPS has been largely ineffective in guaranteeing a floor price to farmers in recent years. 2 2. The Co-operative Wholesale Establishment (CWE) also incurs considerable financial losses. These result primarily from the wheat operations, for which the CWE holds an import monopoly. CWE is the channel for the wheat subsidies which amounted to Rs. 7 billion in 1995. CWE also purchases and markets a number of other food items, such as onions, lentils, chilies etc. For these commodities, CWE's role is to stabilize market supplies and prices against speculative attacks. CWE has traditionally maintained high procurement prices that crowd out private sector activity in the marketing of agricultural commodities, but cutbacks in budgetary transfers have reduced its activities and market dominance. CWE still holds 30, 20 and 10 percent of domestic market share for lentils, chilies, and sugar respectively. But for all the other agricultural commodities, its market share is less than 10 percent. 3. The Central Transport Board (CTB) as it is today represents what is left of the original organization which at the time of Peoplisation had a fleet of 5,000 buses and 50,500 employees. One fourth of the employees retired with compensation while 30,000 were reemployed as employee shareholders of the Peoplised bus companies. CTB's main functions are to operate several repair workshops around the country, oversee the functioning of the bus companies, and administer about Rs. 153 million annually in school transport subsidies from the Ministry of Education to private bus operators. For this the CTB has retained approximately 4000 staff, 800 of whom are in the head office in Colombo, and is supported by Government transfers amounting to Rs. 300 million annually. 4. Sri Lanka Railways (SLR): Large capital expenditures and inadequate revenues from passenger and freight traffic have resulted in continued losses for the Railways. The company The PMB also owns some rice mills. 2 See Sri Lanka: Nonplantation Crop Sector Policy Altematives, World Bank Report No. 14564-CE, 1995. 63 Annex 5 Page 2 of 2 recorded an operational deficit of Rs. 760 million in 1995, representing about 0.5 percent of current budgetary revenues. SLR increased its fares in March 1996 to a level somewhat higher than bus fares. However, there is considerable political sensitiveness to additional fare increases because of the poor quality of service offered. It also received Rs. 3.2 billion from the Government's capital budget, which constituted 8 percent of total capital expenditures in 1995 Budget. However, a large proportion (up to 40 percent) of these budgetary capital expenditures actually finance current operations such as routine maintenance and salaries. Given the large capital investment required to provide rail transport services, Sri Lanka Railways will continue to be highly dependent on the Treasury in coming years. 5. Other: Five other enterprises also have specific problems. The inability to raise tariffs has prevented the National Water Supply and Drainage Board (NSWDB) from covering its investment costs, and the Government is allocating approximately Rs. 2-3 billion annually for its capital expenditures. Air Lanka is highly leveraged, and the Company is expected to face liquidity problems in the near future. There is an urgent need to inject about US$ 70 billion to re-establish financial stability in Air Lanka. The Government has guaranteed US$ 131 million of Air Lanka's total debt of US$ 351 million for the recent purchases of new aircraft. The Road Construction and Development Company (RCDC) is the sole contractor for its parent institution, the RDA, and is able to complete contracts at lower costs due to its access to capital equipment at subsidized rates. As a result, the private sector has not been able to enter the road maintenance and construction market. The Building Materials Corporation (BMC), another trading enterprise, has an insignificant share of the building materials market, and its relevance as a public sector undertaking is questionable. Finally, the conflict in the North has resulted in the closure of CeJlon Cement's plant in Jaffna. Its 730 idle employees continue to receive their wages which are financed by monthly transfers of about Rs. 3 million from the Treasury. 64 Page 1 of 2 BIBLIOGRAPHY Anand Ritu and Sweder van Wijnbergen. 1983, Inflation and Financing of Government Expenditure: An Introductory Analysis with an Application to Turkey, World Bank Economic Review Vol. 3 No. 1, January. Blanchard, Jean Olivier, Suggestions for a New Set of Fiscal Indicators, Working Paper No. 79, MIT, April 1990. Blejer, Mario and Adrienne Cheasty (eds.). 1995, Measuring the Fiscal Deficit, International Monetary Fund, Washington D. C.. Buiter, Willem. 1983, Measurement of the Public Sector Deficit and its Implications for Policy Evaluation and Design, International Monetary Fund StaffPapers 30, June. 1985, A Guide to Public Sector Debt and Deficits, Economic Policy, November. 1990, The Arithmetic of Solvency, in Willem H. Buiter, Principles ofBudgetary and Financial Policy, Cambridge, Mass.. Massachusetts Institute of Technology Press. Boston Institute for Developing Economies. 1964, Report to the Ministry of Finance, Technical Assistance to Improve the Budget Planning Process. Chu, Ke-Young and Richard Hemming (eds.). 1991, A Guide to Public Policy Issues in Developing Countries, International Monetary Fund, Washington D. C.. Coopers and Lybrand. 1990, Administrative Reform in Sri Lanka, Appraisal Report. Development Economics Group. 1989, Final Report and Recommendations on the Public Sector Pension System for Sri Lankan Department ofPensions, Prepared under USAID contract, June. Dinh, Hind T. 1995, Fiscal Solvency and Sustainability in Creditworthiness Analysis, World Bank mimeo, May. Easterly, William, Carlos Rodriquez and Klaus Schmidt-Hebel. 1990, Public Sector Deficits and Macroeconomic Performance, World Bank, Washington D. C.. Employers' Federation of Ceylon. 1992, and 1994, Wages and Fringe Benefits. Gotuaco, Larry. 1991, Report on a Study of the Insurance Industry and the Employees' Provident Fund (EPF) and Employees' Trust Fund, Asian Development Bank, November. Government of Sri Lanka. 1987, Reports of the A dninistralive Reforms Committee (10 in all). 1990, Census of Public Sector and Corporation Sector Employment 1990, Department of Census and Statistics. . Ministry of Lands, Irrigation and Mahaweli Development, 1993, Report of the Working Committee on Rationalization of Structures and Cadres - . Ministry of Finance, 1995, Report on Salary Structure of the Private Sector, Department of Census and Statistics. 65 Page 2 of 2 Guerard, Yves. 1991, Public Service Pension System. Government of Sri Lanka: First Preliminary Report, Prepared under United Nations Development Programme project, September. . 1992a, Public Service Pension System. Government of.Sri Lanka: Second Report, Prepared under United Nations Development Programme project, September. . 1992b, Public Service Pension System. Government of Sri Lanka: Third Report, Prepared under United Nations Development Programme project, December. International Labor Organization. 1991, Report on Conversion of Provident Fund to Social Insurance System, Geneva. Jayaraina, R.S., 1995, New Scheme for Awarding of Pensions, Payments of Commuted Gratuities and Payment of Death Gratuities, Ministry of Public Administration, mimeo, June. Piyasena, R.W. 1994a, Implementation of the Recommendation of the Pension Anomalies Committee. Department of Pensions, mimeo, August. . 1994b, Implementation of the Recommendation of the Pension Anomalies Committee. Department of Pensions, mimeo, September. Prud'homme, Remy. 1994, On the Dangers of Decentralization, World Bank Policy Research Working Paper, No. 1252, February. Tanzi, Vito. 1995, Annual Bank Conference on Development of Economics: Fiscal Federalism and Decentralization: A Review of some Efficiency and Macro Aspects, International Monetary Fund. Shah, A. 1990, Inter-Government Fiscal Relations in Sri Lanka, paper presented at the 46th Congress of the International Institute of Public Finance, Brussels, August 27-30. van Wijnbergen, Sweder. 1989, External Debt, Inflation and the Public Sector: Towards Fiscal Policy for Sustainable Growth, World Bank Economic Review 3 (3). Wanasinghe, Shelton. 1994, Activating the Administrative Reform Process in Sri Lanka, Institute of Policy Studies, July. . 1993, Sri Lanka: Public Expenditure Review. Country Operations, Industry & Finance Division, Country Department III, South Asia Region, Report No.12337-CE, December. 1 1995a, Policies and Strategies for Teacher Education and Teacher Deployment in Sri Lanka, including Statistical Annex, June. . 1995b, Implementation Completion Report, Sri Lanka: Economic Restructuring Credit (2128-CE) and the Public Manufacturing Adjustment Credit (2185-CE), Report No. 14986, September. 1996, Sri Lanka: Sri Lanka 2000 -- An Agenda for Action, Country Operations Division, Department I, South Asia Region, Report No. 154455, March. M:\Annisa\ per/9596\Biblio.Doc July 25, 1996 9:32 AM MICROGRAPHICS CATALOGUERS/FILE CONFIDENTIAL CONFIDENTIAL Report No: 15940 CE Report No: 15940 CE Type: SR Type: SR