Sri Lanka - Review of superannuation benefit programs in Sri Lanka

Document of The World Bank Report No. 20468-CE SRI LANKA REVIEW OF SUPERANNUATION BENEFIT PROGRAMS IN SRI LANKA May 19, 2000 Finance and PSD Sector Unit South Asia Region  CURIRENCY AND EQUIVALENT UNITS Currency Unit = Sri Lankan Rupees (Rs.) US$1.00 = Rs. 69.34 WEIGHTS AND MEASURES Metric System ABBREVIATIONS AND ACRONYMS ACT Advance Company Tax APPF Approved Private Provident Fund CSE Colombo Stock Exchange EPF Employees' Provident Fund ETF Employees' Trust Fund GOSL Government of Sri Lanka IFS Independent Fiduciary Services SLIC Sri Lanka Insurance Corporation NIC National Insurance Corporation PSPF Public Service Provident Fund PSPS Public Service Pension Scheme YOS Years of Service GOVERNNIENT OF SRI LANKA - FISCAL YEAR January 1 - December 31 Vice President: Ms. Mieko Nishimizu Country Director: Ms. Mariana Todorova Sector Director: Ms. Marilou Uy Team Leader: Mr. Joseph Pemia Task Leader: Mr. Christopher Juan Costain  SRI LANKA REVIEW OF SUPERANNUATION BENEFIT PROGRAMS IN SRI LANKA CONTENTS Page No. PREFACE ...............................................................i I. EXECUTIVE SUMMARY .. II. OVERVIEW..........................................................4 Table 1: Aging and Dependency .4 Table 2: Summary of Principal Superannuation Benefit Schemes .7 Table 3: Cost of the Public Service Pension Scheme .10 III. CURRENT SYSTEMS AND ISSUES .15 Table 4: Values and Volumes for the Colombo Stock Exchange .17 Table 5: Employees' Provident Fund Investments .18 Table 6: Employees' Trust Fund Investments .19 Table 7: Sri Lanka Public Service Pension System .24 Table 8: Sri Lanka: Public Service Pension System - Pension Distribution .25 Table 9: Public Service Pension Scheme - Modifications to Pension Formula .26 IV. INTERNATIONAL EXPERIENCE RELEVANT FOR SRI LANKA .31 A. A Coordinated and Consistent Approach ........................................ 31 Table 10: Social Risk Management in Old Age ........................................ 33 B. Taking New Approaches to Reduce Administration Costs ........................................ 36 C. Extending Social Risk Management through Informal Support and Safety Nets ........................................ 37 D. Fund and Unfunded Schemes ........................................ 37 E. Fostering Financial Markets ........................................ 3 8 F. Data Reporting ........................................ 41 G. Tax Reform ........................................ 41 V. PROPOSED REFORM STRATEGY .43 Table 11: Proposed Reform Matrix .53 This report is based on the findings of a mission which visited Sri Lanka in July 1999. The mission consisted of Juan Costain (Mission Leader-SASFP), Sriyani Hulugalle (SASFP), Dimitri Vittas (DECRG), and Alberto Musalem (CMD).  Page No. SUMMARY STATISTICS ........................................................ 58 REFERENCES ........................................................ 60 ANNEXES Annex 1: Mandatory Retirement Savings Ceilings in Selected OECD Countries 1993 ...................................................... 61 Annex 2: Sri Lanka Population Pyramids ..................................................... 62 Annex 3: Economically Active Population, by Age, 1981 ..................................................... 63 Annex 4: Sri Lanka: Ownership of Government Debt ..................................................... 64 Sri Lanka: Ownership of Treasury Bills, Bonds & Rupee Loans ............................. 64 Annex 5: Sri Lanka: Fisca:l Accounts ..................................................... 65  SRI LANKA REVIEW OF SUPERANNUATION BENEFIT PROGRAMS IN SRI LANKA PREFACE This report examines the current system of old age income security in Sri Lanka to determnine whether it provides reasonable pensions for retirees. It also explores the effects of the current system on labor market mobility and the fiscal deficit and looks at the potential for establishing greater demand for debt, particularly long term debt, and equity securities through the strengthening of contractual savings. The focus of this note has been to review the formal systems of retirement savings provision currently in place in Sri Lanka, particularly from the perspective of cost effectiveness, administrative efficiency, and to ascertain policy options available to the Government of Sri Lanka. The report has been produced on the basis of a fact-finding mission to Sri Lanka from July 12 to July 30, 1999, that included the following World Bank staff: Juan Costain, Principal Financial Sector Specialist, SASFP; Sriyani Hulugalle, Economist, SASFP, Alberto Musalem, Adviser, CMD; Dimitri Vittas, Lead Economist, DECRG. During the mission, the team met with government officials, representatives of institutions currently responsible for managing contractual savings institutions, selected research institutes, private and public-sector financial institutions to gather information as to current practices and policies. In its coverage of opportunities for reformn, this note draws heavily from the work of Robert Holzmann, Director, HDNSP. The paper has also benefited substantially from comments from Eric Bell, SASPR, Hermann von Gersdorff, LCSFR, and Olivia Mitchell of the Wharton School, University of Pennsylvania. i   SRI LANKA REVIEW OF SUPERANNUATION BENEFIT PROGRAMS IN SRI LANKA I. EXECUTIVE SUMMARY 1. Sri Lanka has the benefit of a well-established system to provide a mechanism for retirement savings for those employed in the formal sector. While existing programs compare favorably with other countries in the region and of comparable levels of per capita income, there remains significant scope for improvement in terms of the adequacy of many of the schemes in providing sufficient income for retirement. In contrast, the non-contributory pension plan for government employees is relatively generous in the replacement rates it provides. There is reason for concern that the cost of this program could escalate in other than benign economic circumstances. Moreover, certain initiatives undertaken to extend coverage to the informal sector represent a potential material risk in terms of fiscal cost, possibly without providing effective benefits to participants. It is recommended that the Government of Sri Lanka address the principal defects of the existing system with some urgency with a view to pre-empting potential difficulties. Pensions costs derived from the war situation are likely to have an increasingly material bearing upon the budget. 2. It is important to consider an holistic framework for social risk management in old age in Sri Lanka. Informal and family systems are likely to continue to provide the bulk of assistance to the elderly. Other payments and benefits, including severance pay, health benefits, and programs such as Samurdhi play an important role. A significant number of the elderly continue to work into old age while own savings and housing are also important. Formal pension systems represent only one component of the social risk management framework. It is noteworthy that studies show that the elderly in Sri Lanka are in fact less likely to be in poverty than average. Also, a large part of the population aging will be caused by a sharp reduction in the number of young people. Overall dependency ratios for Sri Lanka are likely to remain stable or actually decline. Ultimately Sri Lanka must determine the extent to which it wishes to support a given replacement rate or redistribution of wealth in favour of the elderly. 3. Pensions play a key role in various subsectors of policy analysis: fiscal and monetary policy, labour market efficiency an unemployment, social security, civil service reform, financial sector reform, taxation etc.. It is possible, and necessary, to address this complex topic from the perspective of, and with the support of in depth analysis of the implications for, each of these important subsectors. At the same time, however, it is necessary to take a practical approach to address the most pressing needs for reform. As such, the primary focus of this paper has been to address the imperatives for reform for pensions qua pensions. The most urgent, and straightforward, areas for policy action are summarized below. These and other issues are addressed more thoroughly elsewhere in this paper. The ultimate scope and nature of the reform program undertaken must ultimately be shaped by the views and wishes of the Government of Sri Lanka. It is our proposal that the design of the reform program should be taken to the next stage on the basis of the response of the Government and other interested parties in Sri Lanka to this paper.  4. As necessary, the Bank 'proposes to support this endeavor with further analysis. In particular, with the support of the Government, the bank proposes to undertake a further study of the fiscal consequences of the various systems, including various reform options based upon current and project demographic profiles. This analysis would be carried out using the bank's Pension reform Option Simulation Toolkit (PROST). This study would make use of training in PROST already provided to Government staff and would calculate direct costs emanating from the Public Sector schemes as well as the funding levels of the funded schemes and the indirect cost to Government through interest payments. Further initiatives might also include a deeper analysis of the inter- linkage between the system of retirement income and the existing social safety net. 5. The imperative for reforn stems from the need to improve the standing of the current pension system in the following key areas: Adequacy and Safety - The level of benefit of all but the Public Sector scheme fall short of what is necessary for adequate financial protection in old age. Poor investment returns are a major contributor to this result. The absence of inflation index for Public Sector schemes also erodes the initially generous replacement rate. The absence of a regulatory framework for provident funds and gratuity funds as well as the absence of a dependable market for annuities detract from the usefulness of the existing schemes. Fiscal Sustainability - Historically, inflation has eroded the value of Public Sector pensions, as inflation is settling at lower levels, the real cost of the high replacement rates and generous commutation options for these pensions will increase. The below market rates paid to government provident fund participants represents a substantial tax and is undermining the role of these funds as retirement savings vehicles. The required move to market rates, however, will have adverse budgetary consequences. Furthermore, the additional pension costs incurred as a result of the war situation, both through losses and increased defence staff, are increasingly material. The government sponsored voluntary schemes that have been established in recent years are nominally supposed to be funded. It would appear, however, that these schemes represent more of a potential redistribution mechanism, and a potentially significant future cost to Government. Labor Market Mobility - Not only does the current Public Sector pension scheme encourage workers to choose government employment over the private sector, the absence of any pension portability from the government scheme significantly discourages movement form the public to private sector. Coverage - The formal system of social protection in Sri Lanka still reaches only around half the labor force. 6. While the broad topic of retirement savings and social protection for the aged requires an holistic approach, assessing and integrating the relevant components of the various subsectors, there are real benefits which can be achieved in the short term through improvements to the pension system in its own right. The primary themes of the short term reform agenda would be the need to reduce the role of government in all aspects of economic management and rely more on the private sector, to remove obstacles to labor market mobility and to maintain fiscal soundness. It is anticipated that in most cases the reforms introduced in the short term would apply only to "flows" i.e., to new contributions or new mnembers to schemes and that reform to "stocks" (existing portfolios and members) would be staged over time. The full strategy for reform is provided in detail in the main text. It is to be noted that some of the most important reforms, especially for the PSPS, would follow at a later stage, however a sequencing of the recommended First Stage Reforins would be as follows: 2  Improve and rationalize the provident fund schemes (a) Allow transportability and beneficiary choice as to which scheme to contribute (including PSPF members). (b) Develop regulatory framework and allow new multi-employer APPF's open to all. (c) Wind down the ETF, PSPF and gratuity funds, cap mandatory contributions to Provident Funds. (d) Allow foreign investment (even on an FX neutral basis). (e) Develop government insured annuities market. (f) Allow private sector asset management and liberalize interest rates. (g) Disallow loans to provident fund members. Reform PSPS (a) Reduce vesting period and allow portability for PSPS, (b) Increase retirement age, eliminate or reduce commutation. Review role of pensions in social security framework (a) Rationalize (and limit growth of) voluntary contributory DB programmes 7. A large part of the primary focus falls upon funded contributory schemes in the first instance as they impact the most people, there is an environment which favors reform, and it would be possible to make a real impact upon the economy and upon participant utility in the near term. Moreover, these reforms would set the stage for the unification of the pension schemes and a larger role for the private sector. 8. The low returns of the EPF are a likely significant cause for contributor evasion, the EPF has become more of a tax than a benefit. The administration of the EPF acts as a further deterrent, low cost does not equate to good value for money. The ETF is for all practical purposes a severance system and is basically redundant, ostensibly designed to increase worker ownership of equity it appears to have resulted in real transfers from rather than to employees. Given the fragile and limited nature of the investment market in Sri Lanka, as well as the very poor returns received in recent years, workers are entitled to the greater security which can be provided through diversification into international markets. While the liberalization of interest rates and the freedom to invest overseas are likely to have a material impact upon the funding of the fiscal deficit, it is appropriate funding costs should be recognized as such, and not co-mingled with pensions'. 9. Allowing transferability of the PSPS pension will have an immediate positive impact upon labor market mobility without fiscal cost2. Equalizing the nature of benefits between the EPF and the APPF's (and ultimately allowing movement between them) should serve to improve labour mobility in the private sector. The elimination of less useful schemes such as the ETF and gratuity funds, as well as the elimination of provident fund contributions above a certain level, will serve to reduce the hidden costs of employment and encourage participation in the formal sector (and possibly reduce gross labor costs). 10. The immediate introduction of a gradual increase in retirement age for Government employees and a reduction in commutation rights will have an obvious positive impact upon the cost of the PSPS system and is consistent with global practice. The replacement of the PSPF with the EPF will make a more modest fiscal saving (due to the difference in tax status). The freezing or slowing down of the various voluntary schemes sponsored by the Government will avoid potentially substantial future cost. l Obviously at present the "tax" of sub-market returns is borne almost entirely by EPF members. 2 The full benefit of this measure will only be felt at such time as Civil Service compensation is revised to favor current salary rather than future pension. 3  II. OVERVIEW Background 11. Pension reform is a complex, sensitive topic which has an overarching iniluence on the effective operation of a large section of the economy. In addition to the direct role of superannuation benefits in providing social protection for the elderly, pension reform has major imnplications for labor market mobility (especially in Sri Lanka where retirement savings schemes are employer based), labor participation rates and the cost of labor (and hence unemployment), and financial market development. Pension reform has both a direct and indirect impact upon the fiscal deficit, directly through the cost to the state of the public service and subsidised voluntary schemes, and indirectly through the role of the contractual savings sector as the largest source of funds for government debt. Pensions also play a key role in the social security framework and in poverty alleviation programs. In Sri Lanlca there is a key interface with the existing provision of health services and the Samurdhi benefit program. 12. Sri Lanka is the poorest country in the world to have achieved below replacement rate fertility and is experiencing the fastest aging of its population in the history of the world. Recent population projections estimate that the proportion of those over 60 years will increase from the current level of 9% to about 20% in 2025. Moreover, it will be experiencing this transition at a relatively low level of income, and in a context of a gradual deterioration of the extended family support system and the nuclearization of families where protection for old age is still principally done by investing in children. In considering the potential broader socio-economic impact of measures which might be taken to allay the fiscal cost of an aging population (suchl as the consequences for youth employment of increasing the retirement age) it is important to bear in mind that a large part of population aging in Sri Lanka will be caused by a marked reduction in the number of young people. The age group Cl-14 is predicted to decline in absolute terms for the foreseeable future, translating into fewer youtlhs seeking employment. In terms of overall dependency (the ratio of those below 15 plus those over 65 to those aged 15-64) is likely to remain relatively constant or actually decline (see Table 1). TABLE 1: Aging and Dependency Source: World Ban* 1980 1997 2015 Life Expectancy at Birth 68 73 Lower Middle Income Country Average 65 69 Age dependency ratio 0.7 0.5 (dependents as proportion of working age population) Lower Middle Income Country 4verage 0.7 0.5 Population aged 65 & above (%,0 of total) 6.2 9.1 Lower Middle Income Country Alverage 6.6 8.1 Women aged 65 and above (per 100 men) 102 121 Lower Middle Income Country Average 125 118 3 It should be noted that international experience suggests that the substitutability of young employees for early retirees is limited and that the additional cost of early retirement schemes increases the cost of labor and reduces demand. 4  The Principal Programs 13. Sri Lanka has a long history of provision for retirement income with a number of different programs dating back up to 100 years (see Table 2). The schemes are noteworthy in that all are employment based. The system to which one belongs and the retirement protection which one receives are determined by occupation. With the significant exception of the scheme covering government employees, all the schemes are contributory. Both private and publicly administered programs are in existence, although all schemes established in the last fifty years are government administered. Retirement saving is mandatory in the formal sector of the economy and the formal schemes cover some three million employees, or about half of the formal labor force. A further half million workers in the informal sector are covered by voluntary schemes. 14. By far the biggest retirement savings scheme is the Employees Provident Fund (EPF), established in 1958, which is administered by the central bank on behalf of the Department of Labor. The EPF has a little under two million members active contributors and some Rs. 161 billion in assets4, 98% of which is invested in government debt. The EPF holds some 36% of all the government's domestic debt (see Annex 4). The EPF collects a total of 20% of wages and pays a lump sum at the age of 55. 15. The second largest program is the non-contributory Public Service Pension Scheme (PSPS) which covers some 800,000 government permanent employees and an additional 360,000 pensioners. The PSPS offers replacement rates at retirement of around 85%, but these are subject to ad hoc adjustment rather than being indexed and quickly deteriorate in real terms. The cost of the PSPS represents around 1.9% of GDP and 9.8% of recurrent government expenditure (see Table 3). Although this cost has been relatively stable, and is actually estimated to have declined in 1999, it still represents one third of public sector total compensation (see Table 7 & Annex 5). The minimum retirement age is 55 and the mandatory retirement age is 60. Early retirement has become common which, with increased life expectancy, has led to extended periods of benefit. 16. A significant series of retirement savings schemes are those represented by the 204 Approved Private Provident Funds (APPFs) which are privately administered by individual employers and employer groups. These funds largely predated the EPF Act of 1958 and were grandfathered under the Act, although legislation, passed in 1976 but not made effective until 1996, has called into question their continued existence (this legislation is in the process of being reversed). The APPFs have approximately 164,000 members with average balances of Rs. 152,000, around double that of the typical EPF account. The APPFs are allowed considerable freedom in their investments, but have not been actively regulated and have not always enjoyed the full confidence of their members. Contributions to the APPFs must be as high as those of the EPF, but are often higher. 17. The Employees Trust Fund was employed in 1981 to promote employee ownership of equities. The scheme receives a compulsory employer contribution of 3% of the wages of all employees in the formal sector. At present, the ETF has 1.5 million contributors, which would suggest significant evasion. The ETF has not enjoyed a very successful investment history and there is a widespread impression that many of its investments have been politically motivated. The ETF has an investment portfolio of Rs. 22 bn, which is 81% invested in government debt, despite the original mandate to invest in equities (which represent a rapidly declining share of the portfolio which stood at 10% in 1998. 4 Equal to about 16% of GDP. 5  18. The Voluntary Public Pension Schemes. The government has introduced three pension schemes targeting workers in the informal sector. The first scheme, for farmers, was established in 1986 and has 430,000 members. Additional schemes were introduced for fishermen (1990) and the self-employed (1996) and now have 36,000 and 23,000 members respectively. These schemes are contributory and pay a pension according to years of contribution. The schemes are not self sufficient and it would appear contributions are scarcely enough to cover administrative expenses. 6  TABLE 2: Summary of Principal Superannuation Benefit Schemes Scheme Year Coverage Membership Financings Replacement Administration AsS.tS7 Started Rtate (Rps. NO) Employees 1958 Private Sector 1,900,000 12% Employer 25% final wage9 Depattment of Labour 160,989 Provident Fund 198 Mandatory 8% Employee 2/fnawge through Central Bank ______ Employees 1981 Private Sector 1,500,000 3% Employer Ministry of Finance 21,910 Trust Fund Mandatory_____ Public Service 19471 Public Sector 790,000 Non Ctrbt 85-90% final N/A Pension Scheme Mandatory 358,000 wageNo_otiuoy wg__ Public Service Public Sector 12% Government Ministry of Public Provident Fund 1942 Mandatory 50,000 8% Employee 27% final wage9 Administration and 4,386 Home Affairs 4-7% Employee Ministry of Public Widows & 1956 Public Sector Same as (according to Same as PSPS Administration and N/A Orphans Pension Mandatory PSPS grade) Home Affairs Farmers3 1986 Farmers Fixed payments Fixed payments Ministry of Voluntary 427,783 according to age accordingbto Agricultureto contribution Agiutr1,0 FisherTnen 1990 Fishermen Fixed payments Fixed payments Ministry of Voluntary , according to age contribution Agriculture Fixed payments Mnsr fSca Self Employed 1996 Self Employed 23,000 Fixed payments according to Ministry of Social 30 Voluntary6 according to age contribution Services 3 Approved Private From Private Sector 12% Employer Provident Funds4 19408 Mandatory 164,315 8% Employee10 Department of Labour 25I,000 Private Sector 4.17% final wage InidulE Gratuity 1983 Mandatory Employer' I x years employed Individual Employers Various 1994 Private Sector 61,000 Employee 75-90% final Various. 50% with Sri 40,0009 Private Pensions Voluntary , wage Lanka Insurance Co. 1/ Extending a scheme dating from 1901. 2/ Active accounts only for EPF, ETF & PSPF. For PSPS, the first figure is for "contributors", the second for recipients. All other figures are for contributors only. 3/ Comprising four schemes covering tea small holders, rubber small holders, coconut small holders, and small time cashew planters. 4/ 204 Approved Private Provident Funds as of 12/98. 5/ Percentages refer to total wage income. 6/ For self-employed persons earning less than 3. 7/ As of December 1998. 8/ Mercantile Sector Provident Fund 9/ Estimate 10/ Minimum. Many schemes have higher contributions. 11/ Funding is necessary for tax deductibility but many gratuity plans are not funded. 7  The Issues Adequacy 19. The level of benefit provided by the vast majority of the schemes in place falls well short of what is necessary to provide meaningful financial protection for old age. Benefits are quite uncertain since they are most often provided in the form of lump sums rather than annuities. Thus, the burden of risks is born by workers during the period of accumulation and by retirees during retirement. This shortfall is despite high mandatory contribution levels (e.g., 23% of total wage for EPF & ETF versus only 17% for high income and 16.5% for medium income countries, see Annex 1). 20. It has been estimated that the replacement rate provided by the core retirement savings scheme, the EPF, is only 25% of final wage, much lower than even the 40% believed desirable for lower income countries. This poor replacement rate can be attributed to poor investment returns and, to some extent to evasion and withdrawals5. Fiscal Sustainability 21. In recent years, Sri Lanka has experienced chronic and sizable fiscal deficits (hovering around 8-10% of GDP). The cost of the Public Service Pension Scheme has represented a substantial part of these costs (see Table 3). In contrast, members of the Employees Provident Fund have borne a significant tax burden through the EPF's investments in Government of Sri Lanka (GOSL) securities, largely at below-market rates. Equally, ETF members have been unwitting and involuntary "tax-payers" through the direction of their contributions to uneconomic investments chosen according to political criteria. Any reform in the pension system would require careful assessment on its fiscal impact and its implementation would require fiscal adjustments. Reforming the pension system would likely imply the use of more transparent tax revenue measures, further rationalization of Government expenditures, and acceleration of the privatization process to obtain needed divestiture proceeds to further debt reduction. 5 This result is obviously for participarts, if measured against potential participants the result would be worse. 8  22. It has been estimated that rates paid on provident fund holdings of government securities have been at least 2% below rates paid on Treasury Bills. A move towards market based investments for provident funds, however, would likely to have a greater impact upon the cost of funding for the Government especially if the alternative of foreign investment were to be permitted. The different income and substitution effects would require a more detailed analysis, but a summary of the likely effects is as follows. If the historic rate has been 2% below Treasury Bills, it is to be anticipated that the yield curve between Bills and the long term bonds held by the provident funds would be at least an additional 2%, to give an overall increase of 4% per annum. Equally, in the face of such higher rates, it is possible that alternative investors would emerge and the rate increase would be ameliorated6, alternatively, the market dynamics might be such that it becomes necessary to pay higher rates not only on the debt held by the provident funds but for all holders of long dated instruments. Furthermore, if overseas investment were also to be permitted, the absolute increase in rates required for market based funding may be even higher. As interest in domestic debt makes up around 20% of recurrent expenditure and 5% of GDP, as provident funds hold 40% of domestic debt, a move to market rates of 2% affecting only the provident fund holdings would increase the deficit by a little under 1/2% of GDP. If the rate increase for provident funds were to be as much as 4%, or if the 2% increase in rates were to affect all long term government debt, the increase in the deficit would be around 1% of GDP. 23. Many of the pitfalls usually associated with pay-as-you-go pension schemes in other countries are unlikely to have a material adverse impact in Sri Lanka. The typical comparison made is with national schemes in other countries covering all workers. As the unfunded and non- contributory defined benefit scheme in Sri Lanka is restricted to government employees, the dependency ratio (recipients to contributors7) will increase only as a function of the increased life expectancy of civil servants after retirement (which is likely to be modest) and the extent to which public sector employment declines in absolute terms (which is unlikely in the foreseeable future). Moreover, unlike in national schemes, there is no possibility of evasion as the scheme is non- contributory (essentially included as part of compensation). As the benefits paid under the public sector pension scheme are not indexed but subject to ad hoc revision the benefits are defined only in nominal terms while the government has discretion as to benefits in real terms (within the scope of what is politically palatable). 24. The overall consequences of the public sector pension scheme are discussed in more detail below, and there are certainly issues with regard to its equity and overall economic impact. In strict fiscal terms, it is probably fair to say that the public sector pensions scheme in itself is unlikely to have material fiscal consequences in the medium term8. This being said, it is apparent that the additional costs which are being incurred by the public pension scheme as a consequence of the war situation are material, even in the short and medium term, however, this aspect lies outside the scope of this report. 6 It is also likely that an increase in the rates paid by provident funds would make them appear less like a tax to participants and so reduce evasion, thereby increasing saving and lessening any increase in rates. 7 The term "contributors" is used loosely as the contribution is implicit in the wage, except for the Widows' & Orphans' contribution. A more exhaustive study of this topic, included cost projections, will be carried out using the PROST system. 9  TABLE 3;' Cost of the Public Service Pension Scheme 1991 1992 199 1994 1995 1996 1997 1998 1999 2000* Pension Expenses 8,832 8,099 10,516 13,271 15,057 15,465 17,916 19,477 18,765 21,378 (Rps Mn) as a% GDP 1.9 2.1 2.3 2.2 2.0 2.0 1.9 1.7 1.7 as a % recurrent 10.5 9.0 10.3 10.4 10.3 9.0 9.7 9.8 9.1 9.4 expenditure as a % of budget 21.9 22.8 21.3 25.3 20.7 21.1 22.2 deficit I * Figures for 2000 GOSL estimates Source: Central Bank, IBRD, 2000 Budget Coverage 25. Despite the long history of superannuation benefits in Sri Lanka and the wide range of programs now in place retirement savings still reach only around half the population. The formal systems of social protection cover only around 3.5 million of a labor force of some 6.7 million. However, this is typical for a low middle income country such as Sri Lanka; coverage is much more extensive than other countries in the region (India, for example, has coverage for only 10% of the labor force). Moreover, employment is still heavily dependent upon the rural sector, with 71% of the population living in rural areas and 38.3% of the labor force engaged in agriculture. Typically, providing for the informal sector of the economy has been the most challenging. 26. Sri Lanka remains a country where a large number of people suffer from poverty. According to the World Developmrnent Report, in 1990 35.3% of the population was deemed to be living in poverty. While it is recognized that the greatest challenge lies in extending social protection in old age to the poor, it is equally the case that formal systems relying upon financial savings based processes are unlikely to provide the solution for those who are too poor to save. Furthermore, the transactions cost of interfacing with the formal system would significantly erode the value of savings, especially for those living in rural areas without easy access to financial institutions. Similarly, transaction costs, as well as information asymmetries would also preclude centralized savings being invested in poor and isolated communities. 27. For those in poverty "reti:rement" may not be a recognizable life event, but rather employment continues into old age (at present it is estimated that 37% of elderly men and 15% of elderly women are still participating in the work force, see Annex 3). Equally, some of the most poor are likely to be those, aged or not, who are unable to obtain employment. As such, the needs of this group might be better targeted by a concept of risk management which covers unanticipated eventualities (albeit more frequent with age) such as ill health or the loss of a wage eamer. It is likely, therefore, that solutions to meet the retirement savings needs of those living in poverty are likely to stem from community based micro-insurance structures, as well as means-tested grant funding (as far as this is fiscally sustainable). 28. It is noteworthy that in Sri Lanka not only is labor market participation lower amongst the elderly poor than in other households (which is likely to be an issue of causality) but also the elderly are less likely to be in poverty than other age groups (especially children). This may be attributable to an element of survivor bias (the elderly are those who commanded higher levels of income while working) or that households in which the elderly live have more income providers with respect to dependents than average. Alternatively, households in which the elderly live may be those with higher income earning capacity. I0  Labor Market Mobility 29. The retirement savings schemes covering the bulk of participants are defined contribution, and therefore have a generally limited effect upon labor market mobility. The Public Sector Scheme, in contrast, has a series of characteristics which are quite restrictive upon labor mobility. Not only is the PSPS non-contributory defined benefit, but it also has a ten year vesting period and anyway pays a pension only to those who retire directly from public service9. This situation is exacerbated by the fact that public sector compensation is perceived to be heavily skewed towards the pension. Because of these restrictions, it is very difficult for individuals to move between the public and private sectors. Lessons from International Experience 30. Certain Bank experience in other countries may be relevant for Sri Lanka and provide some insights as to the broad themes that determined successful strategies. i. Adoption of an holistic view on retirement savings provision 31. It is important to consider an holistic framework for social risk management. Pension or provident funds distributions are supplemented by other payments and allowances such as severance payments, gratuities, occupational pensions, payment in lieu of holiday as well as individual savings (perhaps in the form of housing or business or agricultural assets). Individuals may receive support from family members or may simply continue in employment into old age. 32. Systems which function best, producing best social protection at lowest cost and minimum market distortion are those that address all support processes together with issues such as unemployment payments, housing finance, health services, access to annuities and taxation. ii. Balancing individual and social equity with efficiency consideration 33. Mandating too high a replacement rate may lead to overprovisioning and welfare losses. Optimal replacement rates are an inverse function of the discount rate of the time value of money of the beneficiaries. Typically, this discount rate is negatively correlated with wealth and so for lower income countries a replacement rate of 40% or below would be appropriate. Higher rates often lead to evasion and early withdrawal as well as restricting credit to the informal sector. 34. De-linking benefits from contributions has proven to be very costly. Direct budgetary means are more effective in income redistribution. While most retirement savings schemes in Sri Lanka are occupational and closely linked to contributions, the new voluntary schemes for farmers and others appear to reflect this de-linking. iii. Avertingfiscal unsustainability 35. Pay-as-you-go schemes can appear attractive in that they allow for immediate disbursement of benefits without an accumulation period. The PSPS appears insulated from the principal risk to which these schemes are vulnerable, that a rapidly aging population will adversely change the ratio of contributors to recipients. Great caution should be taken in implementing such schemes. It is difficult to implement a worthwhile and affordable scheme in lower income countries. If such a scheme were to be contemplated, it would be advisable to set a low replacement rate for poverty alleviation only. 9 So that a person moving to the private sector after twenty years government employment would have no pension. 11  36. For any defined benefit scheme, it is important that accurate actuarial assessments are maintained. And that adequate clata and projections regarding fertility, wages, employment, interest rates etc., are maintained. It is best that projections extend for at least 75 years and that findings and contexts are discussed in public. iv. Roles for public and private s'ector pensions, funded and unfunded schemes 37. Non-contributory defined benefit schemes such as the PSPS are at odds with modem civil service management, privatization objectives and the need for mobility between public and private sectors. Many governments are acting like any large enterprise and offering transferable defined contribution schemes. 38. One area of innovation which has been seen in other countries is allowing international diversification for provident funds. International diversification reduces risk, often increases returns and avoids many issues of governance and capacity which are associated with investing in a small capital market such as in Sri Lanka. Emerging market equities in general have proven to be highly volatile with weak returns. Allowing diversification to include developed rnarkets would provide better security for contributory systems. 39. Increasing coverage to poor rural communities through formal, mandated systems is difficult, due to high transactions costs, the high time value of money of the poor and the frequent wish to direct savings to uses within the community. Systems which follow the principles of micro-finance may be more successful. It is not clear if pay-as-you-go or funded schemes are more effective, but it is probably appropriate to avoid schemes targeted at any given occupation (such as farmers). 40. Worldwide only 21% of mandatory pension schemes are fully funded and 25% are partially funded (including the U.S.). The trend, however, is very much toward funding. Advantages of funded schemes include that they increase transparency and fiscal responsibility, they limit incentives for evasion, they are more transferable and favor labor market mobility and promote the development of capital markets. Disadvantages are that funded schemes pass the market risk to beneficiaries. v. Reducing administration costs 41. Centralized systems (like the EPF or ETF) generally have lower administrative costs (and no marketing costs). In most countries, however, these systems offer poor service, suffer political interference, and receive low returns on investments. Multipillar systems have not done so well with respect to costs but have faired better in other respects. Experience in other countries is not yet sufficient to draw firm conclus ions but there is evidence that increased costs can offset the higher returns. A major portion of the higher costs come from duplication of administration and marketing expenses. Countries have tried to reduce these additional costs through experiments such as clearinghouses for contribution collection, record keeping etc., restricting the choice of schemes, and even providing group annuities for a whole country through international bidding. 12  vi. Making best use of informal support mechanisms 42. As Sri Lanka has a special resource in the deep cultural roots of family support, it has been suggested that the introduction of formal support mechanisms might "crowd out" family care. This contention has been supported by evidence from Singapore, the U.S., and the Philippines but would appear to apply only to a very small percentage of families. 43. Some countries have tried strategies to "crowd in" family care, these include tax incentives to encourage workers to support elderly relatives, housing policies that favor young people caring for their parents, targeting of social assistance to only the destitute elderly without family support and introducing legal reforms which encourage care of the elderly. vii. Deepeningfinancial markets 44. As centralized schemes are believed to impede the development of new financial instruments, some countries have developed approaches to counter this effect. Policies have included allowing individuals to choose to diversify into external investment schemes and outsourcing of assets to private domestic or international fund managers. Results have been mixed but a trend is developing towards allowing fund participants to choose from a limited number of portfolios, including privately managed funds. viii. Enhancing supervision 45. International best practice in this field is becoming well defined. The consensus is that regulations covering retirement savings should act as an overlay to broad securities regulations governing capital markets. The basics of pension fund oversight are determined to involve oversight of private and public collection, disbursement and data management systems and processes, funds management, and dispute resolution. ix. Tax reform 46. The framework for analyzing taxation of retirement savings considers three distinct phases, the contributions, income on the accumulated funds and the distribution. International tax best practice holds that income which is deferred for retirement purposes should be taxed only once. It is generally held that income generated on accumulations (i.e., the middle phase) should be tax exempt. In most economies it is recommended that the taxation should fall only at the time of distribution, however, for countries such as Sri Lanka, with a weak fiscal position, it is generally regarded as being preferable to tax all income up front, so that contributions to a retirement savings plan would be out of taxed income, but would never be taxed again. Proposed Reform Framework 47. Notwithstanding the various shortcomings, in comparison with many other countries, the situation of Sri Lanka with regard to retirement savings has many positive attributes. The system is not in crisis, there is a well established framework of coverage which reaches a significant portion of the formal sector, initiatives are in place to extend the reach of social protection and, at least for the short to medium term, the fiscal burden on the fiscal deficit is not untenable. As in other sectors of the economy, the absence of effective regulation is seen to be compensated by the overarching presence of the Government of Sri Lanka. In as much as that the bulk of programs in place are on a funded, defined contribution basis with the basis for a system of private sector fund managers, Sri Lanka can be said to be "ahead of the curve" and well positioned for further development. 13  48. There are several core issues outstanding as mentioned above and as detailed below. Sri Lanka must ultimately determine its approach to these issues and its desired position in the social risk management matrix. To whiat extent will support for the elderly continue to be provided through families and the informal system? How much income support can the government provide directly or through the financial markets? How much emphasis can be given to the promotion of equality of treatment for workers of different income levels, genders, occupations and geographic regions? The rapid rise in old age dependency in Sri Lanka suggests that the government will encounter limitations in raising t:he necessary budgetary resources and should limit its role to the provision of basic social insurance and social assistance. These issues would benefit from being addressed, in the short or medium term, by a coordinating body with representation and jurisdiction across various departments of government. 49. The issues relating to old( age security reform are complex with many topics holding relevance for more than one field. A summary presentation of a potential reform agenda for Sri Lanka is presented in matrix fortn in Table II along with a detailed explanation of reform measures. 50. A synopsis of the reform objectives and strategies is as follows: OBJECTIVE STRATEGIES A. Promote Savings for Old Age i. Enhance safety of pension savings. by Making Pensions a More ii. Increase returns on provident funds. Attractive Savings Vehicle iii. Increase participant choice between schemes. iv. Amend tax treatment of retirement savings instruments. v. Reduce administrative inefficiency of EPF/ETF. i. Replace lump sum payment at retirement with lifetime B. Provide Adequate Income income product. throughout Old Age ii. Limit opportunities for early withdrawals. iii. Defer age of retirement, equalize retirement age for men and women. _ Establish a pension for the needy old funded by the general C. Provide a Safety Net for the bugt Poor without Access to budget Pensions _ i. Reform of PSPS to better resemble private sector schemes. D. Reduce Labor Market ii. Establish Equivalence between EPF & APPFs. Distortions i. Reform PSPS. ii. Limit the safety net for the poor through a means-tested E. Improve Fiscal Sustainability definition of poverty. iii. Defer age of retirement, equalize retirement age for men and women. iv. Reform Tax Code. F. Deepen Capital Markets Broaden financial market alternatives to benefit participants through liberalizing investment environment and facilitating provision of new instruments. 14  III. CURRENT SYSTEMS AND ISSUES 51. Reflecting the evolution over time of the provision of old age income protection in Sri Lanka, there exist a number of independent systems, each targeted at a distinct group. All of the schemes are employment related so that the retirement income one receives is a function of where one works. The schemes differ markedly from each other in terms of the basis for contribution and benefit and have not been designed to facilitate transfer from one system to another. The majority of schemes are administered by the government, for private sector as well as government workers. 52. The oldest scheme operating in Sri Lanka is the non-contributory, defined benefit Public Service Pension Scheme, which covers govemment employees and the military, which has been in existence for almost a century. The next schemes to develop were the Approved Private Provident Funds (APPFs) which were established by certain private sector employers on a voluntary basis in the first half of the last century. These schemes are all on a defined contribution basis with a lump- sum distribution. This same basic format was assumed by the government sponsored Employees Provident Fund which was established in 1958 to offer similar benefits as the APPFs to all employees in the formal sector not covered by one of the existing schemes. 53. From 1981 through 1996 a series of additional schemes were introduced by the government which, while still employment based and contributory, incorporated a social agenda. The first of these, the Employees Trust Fund, is a contributory scheme for all employees in the formal private sector which was introduced with the additional purpose of promoting stock ownership by employees. From the mid-1980's to 1990's, three additional government schemes were introduced for workers outside the formal sectors, covering farmers, fishermen and the self- employed. While all of these schemes receive participant contributions, government subsidy will be required to cover costs and to pay benefits. The Employees Provident Fund 54. The Employees' Provident Fund represents the central pillar of the retirement savings system in Sri Lanka dating from 1958 and covering almost one third of the labor force. The EPF operates individual accounts for each employee receiving a contribution of 12% of wages from the employer and 8% from the employee. These amounts are invested in a pool and the proceeds paid in a lump sum at retirement at age 55. 55. In 1998, the Employees' Provident Fund had 7.7 million accounts (of which only 1.9 million were active) and nearly 161 billion rupees (US$2.3bn) in assets. The average balance per EPF account is Rs. 20,900 (although this figure is heavily influenced by the inactive accounts, almost 3 million of which had balances below Rs. 2,500), while in 1998 the average refund per person was 85,000 rupees. It is also possible to make withdrawals from the provident fund in certain restricted circumstances, relating to cessation of employment for reasons of health, marriagel