FINANCE FINANCE EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT Asian Provident Funds Meeting Tomorrow’s Challenges Authors: Richard Jackson and Evan Inglis © 2021 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved. This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO), http:// creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: 2021. Richard Jackson and Evan Inglis. Asian Provident Funds: Meeting Tomorrow’s Challenges. EFI Insight-Finance. Washington, DC: World Bank. Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third- party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to reuse a component of the work, it is your responsibility to determine whether permission is needed for that reuse and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. Cover design and layout: Diego Catto / www.diegocatto.com >>> Table of Contents Acknowledgments iii Abbreviations v Report Overview vii Chapter 1: Asia’s Provident Funds at a Crossroads 1 The Provident Fund Model 2 The Crossroads 5 Chapter 2: Investment and Governance 13 The Stages of Provident Fund Evolution 13 Assessing Investment Performance 18 Three Provident Funds in Focus 21 India’s EPF 23 Indonesia’s JHT 26 Malaysia’s EPF 29 Chapter 3: Benefit Design and Adequacy 33 Adequacy Today 34 Adequacy Tomorrow 36 Chapter 4: Directions for Reform 41 Investment and Governance 42 Benefit Design and Adequacy 44 Technical Appendix 49 About the Authors 55 About the Global Aging Institute >>> Acknowledgments This report was produced by Richard Jackson and Evan Inglis from the Global Aging Institute, in collaboration with Fiona Stewart from the World Bank. The report contributes to the knowledge and research outputs produced under the Joint Capital Markets Program (J-Cap) of the IFC and World Bank, which is kindly supported by the Government of the Grand Duchy of Luxembourg. The authors would like to thank the representatives from the provident funds included in the report for their kind assistance with data provision. Particular thanks goes to the team from the Employees Provident Fund of Malaysia for their feedback and very helpful discussions while the report was being drafted. Thanks also go to Mark Dorfman, Ketut Kusuma, and Katya Gratcheva for their insightful peer review comments. The authors also appreciate the insights and assistance from other World Bank and IFC colleagues, including from the offices in Jakarta, India, Singapore, and Kuala Lumpur. ACKNOWLEDGMENTS <<< iii >>> Abbreviations Administradoras de Fondos de Pensiones AFPs (Chile) [Chilean pension fund management companies] Badan Penyelenggara Jaminan Sosial BPJS (Indonesia) [Indonesian social security agency] BRS Basic Retirement Sum (Singapore) CalPERS California Public Employees’ Retirement System (United States) CPF Central Provident Fund (Singapore) CPP Canada Pension Plan (Canada) CPPIB Canada Pension Plan Investment Board (Canada) EPF Employees Provident Fund (Malaysia) EPF Employees’ Provident Fund (India) EPFO Employees’ Provident Fund Organization (India) EPS Employees’ Pension Scheme (India) ETF Exchange Traded Fund FRS Full Retirement Sum (Singapore) GAI Global Aging Institute GDP Gross Domestic Product GIC GIC Private Limited (Singapore) GPF Global Government Pension Fund Global (Norway) GPIF Government Pension Investment Fund (Japan) ILO International Labour Organization IMF International Monetary Fund JHT Jaminan Hari Tua [Indonesian provident fund] (Indonesia) JP Jaminan Pensiun [Indonesian pension program] (Indonesia) MA MediSave Account (Singapore) MIS Members Investment Scheme (Malaysia) NPS National Pension Service (South Korea) OA Ordinary Account (Singapore) OECD Organisation for Economic Co-operation and Development RA Retirement Account (Singapore) SA Special Account (Singapore) SSGS Special Singapore Government Securities (Singapore) UN United Nations ABBREVIATIONS <<< v >>> Report Overview Across the emerging world, policymakers are grappling with how to build retirement systems that meet the needs of their rapidly developing and rapidly aging societies. Nowhere is the challenge more urgent than in Asia, which is both developing and aging more rapidly than anywhere else on earth. Provident funds, which are fully funded, government-managed, defined contribution systems, have long been the dominant form of retirement provision in much of Asia. The purpose of this report is to assess the strengths and weaknesses of the provident fund model, evaluate the per- formance of three of Asia’s four largest provident funds, and identify steps that they and other provident funds can take to improve retirement security. The funds covered in the report are India’s Employees’ Provident Fund (EPF), Indonesia’s Jaminan Hari Tua (JHT), and Malaysia’s Employees Provident Fund (EPF). Singapore’s Central Provident Fund (CPF), Asia’s other large provident fund, is not covered because it does not invest in financial markets, and so differs fundamentally from the others. The report identifies two key features of the provident fund model that may make it an attractive choice for both governments and workers in emerging markets. The first is that governments can harness provident fund savings to advance national development objectives, which can be a great advantage in countries that may lack the tax capacity to fund development directly through government budgets. The second is that provident funds can serve a wide range of workers’ sav- ings needs beyond the need to save for retirement, from financing home purchases or college educations to providing funds in the event of a medical emergency or the loss of employment. This too can be a great advantage in countries where insurance and credit mechanisms may be underdeveloped and most workers lack precautionary savings. Yet the report cautions that the same features that make the model attractive may also make it difficult for provident funds to ensure retirement security. Investment policies that are intended to advance national development objectives, even when they are well designed and effective, may not be the policies most likely to maximize returns for participants. The fact that provident funds are all-purpose savings programs may mean that retirement savings takes second place to participants’ other savings priorities, which are typically more immediate and pressing. To be successful as retirement systems, provident funds must strike the right balance between their competing goals. What the right balance is, moreover, will necessarily shift over time as coun- tries develop and their populations age. Asia’s provident funds stand at a crossroads. The traditional system of family-based retirement security is under mounting stress from the forces of modernization, and will soon come under intense new demographic pressure from declining family size. Yet adequate government and market substitutes for family support networks are not yet fully developed. Many workers, and in some countries the vast majority of workers, are earning no contributory retirement benefit of any kind. India’s EPF and Indonesia’s JHT only cover about one-tenth of the workforce. And REPORT OVERVIEW <<< vii though the coverage rate is considerably higher in Malaysia’s The report recognizes that the reforms that may be appropri- EPF, it is still far from universal. Moreover, as the report docu- ate for one country may not yet be appropriate for another. ments, many of those workers lucky enough to participate in Some Asian countries, after all, are considerably more devel- provident funds receive benefits that are too small to support oped than others, and some have populations that are also them in old age. Meanwhile, only a tiny sliver of the work- aging more rapidly. Nonetheless, the report identifies several force, just 1 to 3 percent in India, Indonesia, and Malaysia, broad directions for reform that together could greatly improve participates in a private retirement savings program. Although benefit adequacy in most provident funds. all three countries have some kind of noncontributory income support program for the indigent elderly, only a small fraction One critical goal should be to improve investment performance. of the population in need actually receives benefits. The result The indispensable first step is to develop explicit guidelines is growing retirement insecurity. for balancing economic development and retirement security objectives. Managing the tension between these objectives There are many reasons why provident fund participants often is a fundamental challenge for provident funds, and effective end up with insufficient retirement savings, including withdraw- investment policy and governance must be built on guide- als for nonretirement purposes, low contribution density, early lines that spell out the relative weight to be given to each. The retirement ages, and the lack of provision for lifetime income. balance between the objectives should be made clear to all Historically, the real investment returns earned by most provi- stakeholders; there should be regular evaluation of whether dent funds have also been lower than those earned by most policies and outcomes are consistent with the guidelines; and large pension funds and sovereign wealth funds. At the same the balance between the objectives should be reexamined time, rapid real wage growth has compounded the problem of and updated over time as the country’s economy develops low returns, making it difficult for savings to keep pace with and the provident fund grows. All of this will require developing incomes. It is possible that shifts in some of these variables “social impact” indicators that allow provident funds to track will help to boost benefits in the future. Contribution density the benefits of their investment policies to society as a whole may rise along with development, while real wage growth may and compare them with financial returns to members. As an slow. Today’s workers, who may not be able to rely on their interim measure while more robust indicators are being devel- extended families when they grow old to the same extent that oped, provident funds could increase investments in sustain- today’s retirees do, could also decide to preserve more of their able, green, and other types of “labeled bonds,” which have savings for retirement. Still, without significant reforms, retire- the explicit objective of furthering social and development ment insecurity is likely to remain widespread. goals while also delivering reasonable returns. Projections prepared for the report suggest that gross replace- Improving investment performance will also require diversify- ment rates for average earners entering the workforce today ing investment portfolios, at first domestically and then glob- could, under realistic assumptions, be as low as 26 percent ally. Malaysia’s EPF has already moved decisively in this in Malaysia, 31 percent in Indonesia, and 51 percent in India. direction over the past ten to fifteen years. However, many In Indonesia and India, moreover, these projections include provident funds, including India’s EPF and Indonesia’s JHT, benefits from government defined benefit pension programs remain heavily invested in fixed-income securities, especial- in which provident fund members also participate. Excluding ly government debt, and prohibit foreign investment. While those benefits, the replacement rates for Indonesia and India these investment policies may make sense in the early stag- would be just 11 percent and 19 percent, respectively. es of provident fund development, when the focus is on do- mestic development, diversification becomes increasingly If Asia’s provident funds are to meet tomorrow’s challenges, important as the focus shifts to ensuring retirement security. they will need to evolve into something closer to dedicated re- In the end, requiring investment portfolios to remain locked tirement systems whose primary objective is to maximize the up in government debt risks turning provident funds into retirement income of participants. Advancing broad national “captive investors” that finance government activities by im- development objectives may remain an important function of posing below-market returns on members. some provident funds, especially in the region’s less devel- oped economies. Provident funds may also continue to serve There are other investment strategies that provident funds multiple savings purposes. The balance between priorities, could pursue to increase returns on member savings. As most however, will need to shift toward ensuring retirement secu- provident funds now operate, asset values are not marked to rity. Successfully bringing about this shift will require changes market and unrealized gains and losses are held in a reserve in both investment policies and benefit design. account. Member accounts are credited with administratively viii >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES determined returns that may bear only a loose relationship expanding coverage to those workers who do not. Low cover- to market returns. All member contributions, moreover, are age is of course largely a function of high labor-market infor- pooled in the same identical investment portfolio, regardless mality, and afflicts all types of government retirement systems of the member’s age, and all members are credited with the in emerging markets, not just provident funds. Until recently, same rate of return. While this approach has the advantage the obstacles to expanding coverage beyond the formal sec- of allowing provident funds to smooth returns and may help to tor were almost insuperable. However, advances in digital IT, promote social solidarity, it precludes individual customization financial inclusion, and national ID systems, as well as the use of the asset portfolio, which has the potential to significantly of financial incentives like matching contributions, are opening improve investment outcomes for members. Some provident up new ways to make government retirement systems more funds may want to consider moving toward a more market- inclusive. Although a detailed discussion of policy initiatives based approach to accounting for and crediting investment re- that might extend the reach of provident funds was beyond turns that allows for customization of the asset portfolio along the scope of the report, it stresses the critical importance of lifecycle lines, perhaps using a multifund model. Evolving in ensuring that a broader cross-section of the workforce saves this direction would also allow the adoption of liability-driven for retirement, whether through provident funds themselves or investment strategies, where asset allocation, including the through parallel voluntary retirement systems with more flex- duration of fixed-income securities, is aligned with the objec- ible contribution and withdrawal rules that are tailored to the tive of providing income in retirement. needs of informal-sector workers. As for benefit design, the place to start is for provident funds to The research for the report was largely completed before the preserve more of members’ total savings for retirement. In de- COVID-19 pandemic began, and the analysis and recommen- fined contribution systems, a reasonable rule of thumb is that dations do not address the additional threats to retirement se- workers need to save 10 to 15 percent of wages for retirement curity it poses. Clearly, the pandemic makes pursuing the re- each year in order to replace one-third to one-half of their pre- form agenda outlined in the report more challenging. Several retirement income. Among the three provident funds covered countries, including India and Malaysia, have enacted emer- in the report, only Malaysia’s EPF earmarks this large a share gency measures that temporarily reduce contribution rates of wages for retirement savings. In India and Indonesia, there and liberalize access to provident fund savings. Although is no earmarking at all. Retirement savings is simply the re- these measures may be necessary, they will further under- sidual account balance left over after withdrawals for housing, mine future benefit adequacy. The economic fallout from the education, unemployment, and other nonretirement needs. pandemic could also stall movement toward diversification of Improving retirement security will require reordering current provident fund portfolios. With borrowing needs rising, policy- savings priorities. makers may be reluctant to reduce provident fund holdings of government debt. Declines in foreign direct investment could It will also require reforming outdated design parameters, such also reinforce the domestic investment bias built into the tradi- as retirement in the fifties and 100 percent lump sum payouts, tional provident fund model. that are incompatible with lifelong financial security when peo- ple live as long as they now do in Asia. To its credit, Indonesia As they navigate today’s turbulent waters, policymakers would is raising the JHT retirement age in stages from 55 to 65. But do well to keep their eyes on the long term. The forces of de- in India’s and Malaysia’s provident funds, the retirement age mography and development are inexorably reshaping Asian is still 55. All three provident funds, moreover, allow retirement societies, and the need to develop more adequate and more savings to be withdrawn entirely as a lump sum payout. All inclusive retirement systems remains as important today as it provident funds should follow Indonesia’s lead and gradually was before the pandemic began. As policymakers look to the raise the retirement age to 65. All should also require at least future, they will find that the provident fund model continues to the partial annuitization of account balances, something that have many important advantages over alternative retirement Singapore’s CPF has done. If this is not possible, a second- system models. But they will also find that the model needs to best option would be to require phased withdrawals. evolve in important ways if it is to meet tomorrow’s challenges effectively. While the pandemic may slow the necessary evo- These reforms would not only improve retirement security for lution, it should not and need not derail it. those workers who already participate in provident funds. They would also make provident funds more robust platforms for REPORT OVERVIEW <<< ix 1. >>> Asia’s Provident Funds at a Crossroads There is a large literature on the strengths and weaknesses of pay-as-you-go pension systems, and in particular the financing challenges they face as populations age. There is an equally large literature on fully funded and privately managed personal account systems. Yet the dominant type of retirement system in much of Asia—the fully funded but government-managed provident fund—has received comparatively little attention from policy analysts. This is unfortunate, since two key features of the provident fund model may make it an attractive choice for both governments and workers in emerging markets. The first is that governments can harness provident fund savings to advance national development objectives, which can be a great advantage in countries that may lack the tax capacity to fund development directly through government budgets. The second is that provident funds can serve a wide range of workers’ sav- ings needs beyond the need to save for retirement, from financing home purchases or college educations to providing funds in the event of a medical emergency or the loss of employment. This too can be a great advantage in countries where insurance and credit mechanisms may be underdeveloped and most workers lack precautionary savings. Yet the same features that make the model attractive may also make it difficult for provident funds to ensure retirement security. Investment policies that are intended to advance national develop- ment objectives, even when they are well designed and effective, may not be the policies most likely to maximize returns for participants. The fact that provident funds are all-purpose savings programs may mean that retirement savings takes second place to participants’ other savings pri- orities, which are typically more immediate and pressing. To be successful as retirement systems, provident funds must strike the right balance between their competing goals. What the right bal- ance is, moreover, will necessarily shift over time as countries develop and their populations age. In the early stages of their evolution, the natural focus of provident funds is on national economic development. Portfolios are usually heavily invested in government debt, governance needs are important but basic, and a large portion of savings flows to nonretirement purposes. As provident funds develop, however, the focus should shift to ensuring retirement security. Along the way, portfolios should be diversified, governance procedures and safeguards should become more sophisticated, and a growing share of savings should flow to retirement. This report turns the spotlight on Asia’s provident funds—or more precisely, government- managed provident funds.1 The term provident fund is sometimes used more broadly in Asia to describe any kind of funded retirement system. In Thailand, for instance, employer pensions are called provident funds. Hong Kong’s government-mandated retirement system is also called a provident fund, even though it is a privately managed personal account system. In this report, the term is used exclusively to refer to government-managed provident funds, of which there are many, from Bhutan’s and Nepal’s to Brunei’s and Sri Lanka’s. The report focuses on three of them in particular: India’s Employees’ Provident Fund (EPF), Indonesia’s Jaminan Hari Tua (JHT), and Malaysia’s Employees Provident Fund (EPF). (See box 1 at the end of the chapter.) 1. For the major data sources used in the report, see the Technical Appendix. This appendix also includes a discussion of methodological issues related to GAI’s evaluation of provident fund investment performance and benefit adequacy. CHAPTER 1: ASIA’S PROVIDENT FUNDS AT A CROSSROADS <<< 1 These three provident funds were chosen primarily because The purpose of the report is to assess the strengths and weak- they are among Asia’s largest, whether measured by assets nesses of the provident fund model, evaluate the performance under management or number of participants. (See figures 1 of India’s, Indonesia’s, and Malaysia’s provident funds, and and 2.) The three countries also represent a wide range of identify steps that they and other provident funds can take to economic development, and so the funds themselves reflect improve retirement security in a rapidly developing and rapidly a wide range of provident fund evolution. The other potential aging Asia. This chapter first takes a closer look at the provi- candidate was Singapore’s Central Provident Fund (CPF), dent fund model. It then discusses the changing demographic, which is even larger, at least in terms of assets under man- economic, and social environment for retirement security in agement. But the CPF, whose assets consist entirely of non- Asia, and why it makes building more inclusive and more ad- marketable government securities, does not invest in financial equate retirement systems such an urgent policy priority. markets or directly fund development, and so differs funda- mentally from Asia’s other provident funds. (See box 2 at the The Provident Fund Model end of the chapter.) Although the report makes occasional ref- erence to the CPF, it is not included in the analysis. > > > Government provident funds are mandatory, fully funded, F I G U R E 1 - Assets in Billions of U.S. Dollars at the government-managed, defined contribution systems. Govern- End of 2018 or Most Recent Year Available ments may establish them for special categories of workers, such as civil servants or the armed forces. They may also es- 250 Assets as a Percent of GDP tablish them for private-sector workers, in which case they are, Indonesia 2% 201 India 3% at least aspirationally, national retirement systems. The three 200 Malaysia 56% provident funds on which the report focuses cover private- sector workers—or, more precisely, private-sector workers in 150 formal employment. In Malaysia, the provident fund is the only 91 contributory government retirement program for private-sector 100 workers. In India and Indonesia, private-sector workers are also covered by defined benefit pension programs. 50 19 0 In general, provident funds function as all-purpose savings pro- Indonesia India Malaysia grams, rather than dedicated retirement systems. Some, like JHT EPF EPF Malaysia’s EPF, allocate members’ contributions to separate Note: The figure for India is an estimate for end of March 2018. accounts earmarked for retirement and nonretirement purpos- Source: EPFO, BPJS, EPF (Malaysia), and GAI calculations es. Others, like India’s EPF and Indonesia’s JHT, channel all contributions to the same “retirement” account, but allow the > > > early withdrawal of much or even all of account balances for a F I G U R E 2 - Active Members in Millions at the End wide range of reasons. During the accumulation phase, mem- of 2018 or Most Recent Year Available bers usually earn an administratively determined rate of return 45 on their savings, which may be only loosely related to actual 41.2 40 investment returns on provident fund assets. Some provident 35 funds, including, Malaysia’s EPF, also feature minimum rate 30 of return guarantees of one kind or another. During the payout phase, benefits are typically paid as a lump sum, with most 25 provident funds, Singapore’s CPF being a notable exception, 20 14.6 making no provision for lifetime income. 15 10 7.4 The provident fund model has important strengths, though it 5 also poses some significant challenges. One strength is that 0 provident funds are fully funded retirement systems. At least Malaysia Indonesia India EPF JHT EPF at the macro level, funded retirement systems have decisive advantages over pay-as-you-go systems. While countries are Note: Data for Indonesia are for the end of 2017; data for India are still demographically young and economically developing, an average for 2016-17. Source: EPFO, BPJS, and EPF (Malaysia) they can help to broaden and deepen financial markets, a criti- 2 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES cal component of the development agenda in any emerging Many retirement policy experts would agree that the fact that market. By purchasing government debt, they can also reduce provident funds are defined contribution systems is also a government borrowing costs and support government spend- strength. Almost by definition, defined contribution systems ing on other development priorities. Moreover, if funded retire- cannot have unfunded liabilities. They are therefore much less ment systems are government managed, as provident funds likely than defined benefit systems to burden future genera- are, savings can be directly steered toward infrastructure and tions of workers and taxpayers. Because benefits in defined other social capital investments. Unless pay-as-you-go sys- contribution systems are directly proportional to contributions tems are at least partially funded, they do not have any of paid in, they are also less likely to distort labor-market de- these benefits. cisions. Moreover, by rewarding longer work lives they may encourage later retirement, a potential advantage that will be- Later, as countries age, funded retirement systems can take come increasingly important as emerging markets age. It is pressure off government budgets, which would otherwise be true that all three of the provident funds covered in the report burdened by rising pension expenditures, while also helping now allow the withdrawal of some or all of retirement savings to maintain adequate rates of savings and investment. Across beginning in the mid-fifties.3 But early retirement ages are a the world, from Germany and Japan to Brazil and China, gov- design parameter that can be adjusted, not a fundamental ernments with pay-as-you-go systems are being compelled to feature of the provident fund model. It is also true that defined make large cutbacks in the generosity of retirement provision contribution systems, which do not ordinarily provide for redis- as their populations age and the ratio of retired beneficiaries tribution, need to be backed up by separate poverty protection to contributing workers rises. South Korea, which had the poor programs, such as means-tested supplements, a flat benefit, timing to establish its pay-as-you-go National Pension System or a social pension. While these programs are not well devel- in the late 1980s, just before its birthrate collapsed, is a case oped in most Asian countries, that is a failing of social policy, in point. Promised replacement rates have already been cut not a defect of the provident fund model. twice, from the original 70 percent to 40 percent—and with the system still facing large long-term deficits, may have to be cut To be sure, some retirement policy experts fault defined con- again.2 Provident funds will not face this problem. tribution systems for shifting risks to individuals. But the no- tion that participants bear no risk in defined benefit systems, It is true that full funding is not always an advantage at the and in particular government-managed defined benefit sys- micro level. In countries where the workforce and wages are tems, is misleading. While participants in these systems may both growing rapidly, as they have been in most of Asia, pay- not be subject to investment risk, they are subject to politi- as-you-go retirement systems, whose implicit rate of return is cal risk. If promised benefits are underfunded, governments equal to the growth rate in taxable payroll, may be able to gen- will necessarily have to cut benefits, raise contributions, or erate higher replacement rates than funded systems, whose shift costs to general taxpayers, who in many cases will be rate of return is equal to the return to capital. As populations the same people as the retirement system participants. In age, economies develop, and workforce and wage growth any case, government-managed provident funds, at least as slow, however, the advantage shifts to funded retirement currently structured, do not shift investment risk to individu- systems. Malaysia is already reaching the demographic and als in the same way that privately managed defined contribu- economic tipping point where the return to a funded retire- tion systems typically do. The account balances of provident ment system is likely to exceed the return to a pay-as-you-go fund members do not automatically rise and fall along with system. In India and Indonesia, this tipping point still lies well investment performance, and though the administratively de- in the future, which means that switching to pay-as-you-go re- termined returns they earn may vary, they are often subject to tirement systems might allow them to deliver more adequate a minimum guarantee. benefits than their provident funds can for some time to come. But sacrificing the considerable macro advantages of funding This brings us to government management, which distinguish- for a micro advantage that is bound to fade over time might not es provident funds from most other types of fully funded defined be a wise policy choice. contribution retirement systems. Government management has a number of potential advantages. It may reduce administra- 2. See OECD, OECD Economic Surveys: Korea 2018 (Paris, OECD, June 2018), pp. 66-68 and Neil Howe, Richard Jackson, and Keisuke Nakashima, The Aging of Korea: Demographics and Retirement Policy in the Land of the Morning Calm (Washington, DC: Center for Strategic and International Studies, March 2007). 3. India’s EPF allows the complete withdrawal of retirement savings at age 55, while Indonesia’s JHT does so at age 57. Malaysia’s EPF allows the complete withdrawal of retirement savings accumulated up to age 55 at age 55, but requires incremental contributions made after age 55 to be preserved until age 60. CHAPTER 1: ASIA’S PROVIDENT FUNDS AT A CROSSROADS <<< 3 tive costs below what they would be in a privately managed ings are not directly linked to market returns. The practice of personal account system. It may improve compliance and in- crediting member accounts with an administratively determined crease participation. It may even help to foster a sense of social interest rate allows provident funds to smooth returns, which solidarity. Most importantly, it is what allows provident funds to may be beneficial. But it can also open the door to hidden serve as direct instruments of national development policy. cross-subsidies from provident fund members to government or vice versa. A related challenge arises from the fact that provi- Along with the potential benefits, however, come some signifi- dent funds pool all member contributions in the same identical cant governance challenges. Ensuring transparency and ac- investment portfolio and, as a rule, credit all members with the countability may be difficult when it is government that is polic- same rate of return. While this practice may promote social soli- ing itself. Investment decisions may be politically determined, darity, it makes it difficult for provident funds to adopt strategies which means that they may be inefficient, and even if they are for optimizing member outcomes that depend on individual cus- not politically determined, the sheer size of provident funds may tomization of the asset portfolio, such as lifecycle investing or give government undue influence in financial markets. Then other more sophisticated liability-driven approaches to invest- there is the question of how provident funds can effectively ment. The rate of return guarantees that some provident funds pursue two objectives: maximizing risk-adjusted returns for par- feature may also hurt members in the long run by forcing asset ticipants and promoting national economic development. When managers to pursue overly conservative investment strategies. the two objectives conflict, it is likely to be the former that loses out. As we will see in Chapter 2, the investment performance of There are other design parameters characteristic of provident the three provident funds covered in the report has lagged the funds that may undermine the adequacy of retirement provi- performance of other large pension funds and sovereign wealth sion, including early retirement ages and the lack of provision funds around the world, in some cases by a wide margin. Al- for lifetime income. But these parameters can be adjusted with- though there are many reasons for this, politically determined out fundamentally altering the provident fund model itself. There investment decisions, along with the inherent tension between is no compelling policy reason, though there may be political provident funds’ dual objectives, may have played a role. reasons, why provident funds could not gradually raise retire- ment ages as populations age, and indeed Indonesia’s JHT has A similar tension arises from the fact that provident funds are already begun to do just that. Nor is there any compelling policy typically all-purpose savings programs. Like their ability to ad- reason why they could not require at least the partial annuitiza- vance national development objectives, their ability to serve tion of account balances. Although none of the three provident multiple savings goals is part of what makes provident funds funds covered in the report has done so, Singapore’s CPF has. attractive. Malaysia’s EPF serves as an important source of fi- nancing for housing and education, while India’s EPF and In- All types of government retirement systems have potential donesia’s JHT function as substitutes for unemployment insur- drawbacks. In the near term, pay-as-you-go systems may en- ance. But just as provident funds’ role in advancing national courage governments to promise overly generous benefits, development objectives creates conflicts between competing creating large unfunded liabilities, while in the long term they objectives, so does their pursuit of multiple savings goals. Not may be rendered unsustainable by the inexorable arithmetic of surprisingly, it is the adequacy of retirement provision that tends rising old-age dependency ratios. Privately managed person- to lose out. Even in Malaysia’s EPF, which has relatively strict al account systems avoid the potential pitfalls of government rules governing early access to savings, nonretirement with- management of retirement savings, but may be plagued by lack drawals account for two-fifths of total withdrawals. In the case of competition, distortionary investment regulations, and high of Indonesia’s JHT, where nearly 90 percent of withdrawals are administrative fees. That other models of retirement provision nonretirement withdrawals, one wonders whether it should be also have drawbacks, however, does not make addressing the considered a retirement system at all. challenges posed by the provident fund model any less impor- tant. Overcoming them while more fully leveraging the model’s The provident fund model poses additional challenges. One strengths will require significant reforms. More fundamentally, it arises from the fact that the returns members earn on their sav- may also require a sea change in philosophy. 4 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES The Crossroads Until recently, governments throughout Asia could assume, Yet as Asian countries develop and modernize, traditional fam- with at least some justification, that workers who reached old ily support networks for the elderly are coming under increasing age without a pension or personal savings would be support- stress. Urbanization, industrialization, and the spread of more ed by their extended families. It was this assumption that has “individualistic” Western values are breaking up extended fami- allowed them to focus on other development priorities, with lies and eroding traditional social and cultural norms. Among building adequate retirement systems often an afterthought. these norms is the ethic of filial piety, which in many Asian soci- But this assumption is no longer prudent today and will be eties requires grown children to care for their aged parents. In even less so tomorrow. the 2015 GAI survey, representative samples of workers and retirees in ten Asian societies were asked, “who, ideally, should To be sure, the extended family continues to play a much larg- be mostly responsible for providing income to retirees.” The er role in retirement security in Asia than it does in the West. possible responses were government, former employers, indi- A 2015 Global Aging Institute (GAI) survey of Asian workers viduals through their own savings, or grown children or other and retirees found that rates of multigenerational living remain family members. In none of the ten did the share of respon- high. In most Asian countries, at least half of the elderly live dents saying grown children or other family members exceed in the same household as their grown children. In the region’s 15 percent. Workers and retirees were also asked, “who, ide- less developed countries, the share often exceeds three- ally, should be mostly responsible for providing personal care quarters. In most Asian countries, moreover, income transfers to retirees.” The share saying grown children or other family within families continue to flow up the age ladder from young members should be responsible for personal care was larger to old, precisely the opposite direction they now flow in most than the share saying they should be responsible for income in Western countries.4 all ten societies. Even so, this was the majority view in just two of them, the Philippines and Vietnam. (See figure 3.) > > > F I G U R E 3 - Share of Respondents Saying Grown Children or Other Family Members Should, Ideally, Be Mostly Responsible for Providing Income and Personal Care to Retirees 70% 63% 60% 60% 50% 48% 40% 32% 34% 29% 30% 30% 26% 20% 21% 20% 11% 13% 10% 10% 11% 10% 8% 8% 10% 6% 6% 0% Thailand Hong Kong Malaysia South Korea Taiwan Indonesia Singapore China Philippines Vietnam Income Personal Care Source: Richard Jackson and Tobias Peter, From Challenge to Opportunity: Wave 2 of the East Asia Retirement Survey (Alexandria, VA: Global Aging Institute, 2015) 4. Richard Jackson and Tobias Peter, From Challenge to Opportunity: Wave 2 of the East Asia Retirement Survey (Alexandria, VA: Global Aging Institute, 2015). CHAPTER 1: ASIA’S PROVIDENT FUNDS AT A CROSSROADS <<< 5 Even as the old order of family-based retirement security be- are doing so at a breathtaking pace, far faster than Western gins to pass away, adequate government and market substi- countries once did. The total fertility rate has now sunk beneath tutes for informal family support networks are not yet fully devel- the 2.1 replacement rate throughout East Asia, and though it oped. Civil servants and the armed forces typically participate is still above replacement in most South Asian countries, it is in special government retirement programs that offer generous falling fast. Meanwhile, life expectancy has soared and in many replacement rates. But many private-sector workers, and in Asian countries approaches, equals, or even exceeds life ex- some Asian countries the vast majority of them, are earning no pectancy in Western countries. The result is a dramatic aging contributory retirement benefit of any kind. India’s EPF and In- of the population. Although the degree of aging varies greatly donesia’s JHT only cover about one-tenth of the workforce. And across Asia, mainly because the fertility rate has fallen much though the coverage rate is considerably higher in Malaysia’s further in some countries than in others, the trend is gathering EPF, it is still far from universal. Moreover, as we will see in momentum almost everywhere. In India, Indonesia, and Ma- Chapter 3, many of those workers lucky enough to participate laysia the share of the population aged 60 and over has nearly in provident funds receive benefits that are too small to support doubled since 1990. In all three countries, moreover, that share them in old age. It is true that all three countries covered in the is on track to double again by the middle of the century. (See report make at least some provision for tax-favored private re- figure 4.) tirement savings, whether in the form of employer or personal pensions. But only a tiny sliver of the workforce, ranging from 1 The aging of the population threatens to put enormous addi- to 3 percent, currently participates in these programs. All three tional stress on family support networks. It is not just that the countries also have some kind of noncontributory income sup- number of elderly is growing. As family size shrinks, the odds port program for the indigent elderly. But only a small fraction of that elders will have a child able and willing to support them the population in need actually receives benefits. The result is will decline, even if the propensity of grown children to support growing retirement insecurity. their aged parents does not change. Meanwhile, the aging of the population will also make it more difficult for governments When rapid development is combined with rapid population ag- to shore up retirement incomes by expanding noncontributory ing, the vulnerability of the elderly grows. Every Asian country social pensions. It is one thing for one-third or one-half of the is now progressing through the demographic transition, the shift elderly to be dependent on social assistance when the elderly from high fertility and high mortality to low fertility and low mortal- constitute 5 to 10 percent of the population. It will be another ity that accompanies development. Many countries, moreover, thing entirely when they constitute 20 to 40 percent. > > > F I G U R E 4 - Elderly (Aged 60 & Over), as a Percent of the Population, 1990-2050 30% 24% 20% 21% 20% 11% 10% 10% 10% 6% 6% 6% 0% India Indonesia Malaysia 1990 2020 2050 Source: UN Population Division, World Population Prospects: The 2019 Revision (New York: UN Population Division, 2019) 6 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES Asia’s provident funds thus stand at a crossroads. If they are These changes would not only improve retirement security to meet tomorrow’s challenges, they will need to evolve into for those workers who already participate in provident funds. something closer to dedicated retirement systems. Advanc- They would also make provident funds more robust platforms ing broad national development objectives may remain an for expanding coverage to those workers who do not. important function of some provident funds, especially in the region’s less developed economies. Provident funds may also The remainder of the report is organized as follows. The sec- continue to serve multiple savings purposes. The balance be- ond chapter discusses provident fund governance and invest- tween priorities, however, will need to shift toward ensuring ment policies, practices, and performance, with a particular retirement security. This will mean reforming investment poli- focus on India’s EPF, Indonesia’s JHT, and Malaysia’s EPF. cies so that they maximize risk-adjusted returns for provident The third chapter discusses provident fund benefit design and fund members. It will mean preserving more of members’ to- adequacy, with a focus on the same three provident funds. tal savings for retirement. And it will mean reforming design The fourth and final chapter offers some broad recommenda- parameters, such as retirement in the fifties and 100 percent tions for improving provident funds’ effectiveness in maintain- lump sum payouts, that are incompatible with lifelong finan- ing retirement security. cial security when people live as long as they now do in Asia. > > > B O X 1 - Provident Fund Profiles: India’s EPF, Indonesia’s JHT, and Malaysia’s EPF INDIA INDONESIA MALAYSIA EPF EPS JHT JP EPF Year Established 1952 1995 1992 2014 1951 Administrator EPFO BPJS Ketenagakerjaan EPF Board Effective Coverage Rate 9% 12% 50% (% Employment) Contribution Total 15.7% 9.5% 5.7% 3% 24% Rate Earmarked for 0% na 0% na 16.8% Retirement Retirement Age 55 58 57 55 Lifetime Income None DB Annuity None DB Annuity None Requirement Gov’t Bonds & Deposits 88% na 72% na 34% Allocation (% Portfolio) Portfolio Foreign Investment 0% na 0% na 27% (% Portfolio) Note: Effective Coverage Rate. The effective coverage rate equals active members as a share of total employment. Data for Malaysia are for the end of 2018; data for Indonesia are for the end of 2017; and data for India are an average for 2016-17. The effective coverage rates for India and Indonesia refer to the EPF and JHT, respectively; effective coverage rates under the EPS and JP differ slightly. Contribution Rate. In Malaysia, the total contribution rate refers to the contribution rate for members under age 60; the contribution rate earmarked for retirement refers to the Account I contribution rate. Portfolio Allocation. Data for Malaysia are for the end of 2018; data for India and Indonesia are for the end of 2017. In Indonesia, government bonds include non-government bonds. CHAPTER 1: ASIA’S PROVIDENT FUNDS AT A CROSSROADS <<< 7 India’s EPF Indonesia’s JHT The Employees’ Provident Fund (EPF), established in 1952, The Jaminan Hari Tua (JHT), established in 1992, is ad- is administered by the Employees’ Provident Fund Organi- ministered by BPJS Ketenagakerjaan, one of Indonesia’s zation (EPFO), which in turn is part of the Ministry of Labour two social security agencies. Prior to the creation of the and Employment. Participation is mandatory for workers at BPJS in 2014, the JHT was administered by PT Jamo- private-sector firms with at least twenty employees, as well stek, a state-owned enterprise. Since coming under BPJS as for workers at some smaller firms in a few specified in- administration, JHT coverage has in principle been ex- dustries. Firms can apply to the EPFO for an exemption that panded to include all private-sector employees, though in allows them to manage their employees’ EPF accounts in practice it is limited to those employed in the formal sec- house. As of 2017, roughly 1,500 mostly large firms oper- tor. Self-employed workers can participate in the JHT on a ated an EPF Private Trust, as this arrangement is called. voluntary basis, while civil servants and the armed forces Self-employed workers can participate in the EPF on a vol- are covered by separate retirement systems. As of 2017, untary basis, while civil servants and the armed forces are 15 million workers, or 12 percent of the workforce, were covered by separate retirement systems. Given the large active JHT contributors. size of India’s informal sector, EPF coverage is limited. As of JHT members also participate in the Jaminan Pen- 2017, 41 million workers, or 9 percent of the workforce, were siun (JP), a defined benefit pension program. The JP, active EPF contributors. however, was only introduced in 2014 and will not be- Most EPF members also participate in the Employees’ gin paying pensions until the first cohorts who meet the Pension Scheme (EPS), a defined benefit program estab- program’s minimum fifteen-year contribution requirement lished in 1995 that is administered by the EPFO in parallel begin retiring a decade from now. Employees contribute to the EPF. Employers and employees each contribute 12 2 percent of wages to the JHT while employers contrib- percent of basic salary to the EPFO up to a contributable ute 3.7 percent, for a total contribution rate of 5.7 per- wage ceiling, which is currently Rs 15,000 per month. Of cent. There is no ceiling on contributable wages. The JP the 24 percent total, 15.67 percent is directed to the EPF, contribution rate is 3 percent, with employees contributing while 8.33 percent funds the EPS. There is also a small 1 percent and employers 2 percent. government contribution of 1.17 percent of basic salary The JHT and JP standard retirement age, originally 55, is earmarked for the EPS, which boosts the total EPS contri- being raised in stages to 65, which it will reach in 2043. As bution to 9.5 percent. of 2020, it stood at 57. When members reach retirement The EPF standard retirement age is 55, after which account age, JHT account balances are usually disbursed in a sin- balances are usually disbursed in a single lump sum pay- gle lump sum. There is no provision for lifetime income. ment. There is no provision for lifetime income. Before age Nonretirement withdrawals make up the great majority of 55, the early withdrawal of most or even all of EPF savings is all withdrawals. Partial withdrawals are allowed to finance permitted under many circumstances, including to purchase home purchases, while complete withdrawals are allowed a home or pay off a mortgage, to pay for medical expenses, whenever workers lose or change jobs. and to finance the education or marriage of one’s children. The JHT is in the early stages of provident fund develop- Members can also cash out the portion of their account bal- ment. Most investments are still in government debt and ances that is attributable to their own contributions when other fixed-income securities, and foreign investment is they quit or are laid off from their current job, provided that not allowed. However, the JHT also has a sizeable posi- they are unemployed for at least two months. The EPS full tion in domestic equities. benefit retirement age is 58, but reduced early retirement benefits can be claimed starting at age 50. Despite being in operation for nearly seven decades, the EPF is still in the early stages of provident fund develop- ment. Although the EPFO has recently begun to make small investments in domestic equities, the EPF portfolio remains heavily tilted toward government debt and other fixed-income securities. Foreign investment is not allowed. 8 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES Malaysia’s EPF prior to retirement for specified purposes, from purchas- ing a home to making the Haj, receives 30 percent. The standard EPF retirement age is 55. When workers turn The Employees Provident Fund (EPF), established in 55, their Account I balance and any remaining Account II 1951, is administered by the EPF Board, which operates balance is transferred to a new account called Akaun55. under the supervision of the Ministry of Finance. Cov- The transferred funds may be withdrawn in whole or in erage is mandatory for private-sector employees with part as a lump sum, taken as phased withdrawals, or left regular employment contracts, as well as for those pub- in the account to earn interest. Any incremental contribu- lic-sector employees not eligible to join the civil service re- tions made by members who remain employed past age tirement system. It is voluntary for self-employed workers, 55 are allocated to another new account, called Akaun as well as foreign workers. There are separate retirement Emas, which cannot be accessed until age 60. systems for most civil servants and the armed forces. As The EPF, which has a globally diversified portfolio with of 2018, 7 million workers, or 50 percent of the workforce, substantial exposure to equities, has progressed well be- were active EPF contributors. yond the early stages of provident fund development. As The EPF contribution rate is 24 percent on wages up to in other provident funds, members earn an administrative- RM 5,000 per month for members under age 60, with em- ly determined return on their account balances. The EPF ployees contributing 11 percent and employers 13 per- Members Investment Scheme (MIS), however, allows cent. On wages above RM 5,000, the contribution rate members whose balances exceed the “basic savings” is 23 percent. Members aged 60 to 74 who are still em- amount to withdraw a portion of their savings and invest ployed contribute at a reduced rate, while those aged 75 it in approved investment funds, where it earns a mar- and over are not required to make contributions. There is ket return. The basic savings amount, which rises along no ceiling on contributable wages. with age, is the amount of savings that the EPF calculates The EPF consists of two accounts: Account I, which is would be sufficient to provide a minimum monthly benefit earmarked for retirement, receives 70 percent of total roughly equal to the poverty line. contributions, while Account II, which can be accessed CHAPTER 1: ASIA’S PROVIDENT FUNDS AT A CROSSROADS <<< 9 > > > B O X 2 - The Special Case of Singapore’s CPF The Central Provident Fund (CPF), established in 1955, is earmarked for retirement. The overall contribution rate operates under the Ministry of Manpower. Coverage is is 37 percent of ordinary wages for members under age mandatory for employees who are citizens or permanent 55 up to a contributable wage ceiling of SG $6,000 per residents. The self-employed are required to contribute to month, with employees contributing 20 percent and em- MediSave, the CPF subaccount that finances health care, ployers 17 percent. (Members aged 55 and over contrib- but their participation in the rest of the CPF is voluntary. ute at a reduced rate.) Over the course of a full career Foreign workers are excluded from coverage. As of 2018, from age 20 to age 64, 54 percent of total contributions 2 million workers, or 56 percent of the workforce, were go to the OA, 27 percent to the MA, and 19 percent to active CPF contributors. the SA. Like most provident funds, the CPF serves multiple sav- Where the CPF differs from other provident funds is in ings purposes. Contributions are split between three ac- its unique approach to investment. Rather than invest counts, with the apportionment varying by age. There is member contributions in financial markets, the CPF the Ordinary Account (OA), which may be accessed prior invests them in Special Singapore Government Secu- to retirement for specified purposes, including purchas- rities (SSGS). These securities, which are nonmarket- ing a home and financing the education of family mem- able government bonds, are similar in nature to the bers; the MediSave Account (MA), which is earmarked nonmarketable bonds that many governments issue to for paying health insurance premiums and out-of-pocket keep track of intragovernmental borrowing. The yields medical expenses; and the Special Account (SA), which on the CPF’s SSGS holdings, which vary by account, 10 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES are set by formulas which peg those yields to the yields that other provident funds in the region would do well on equivalent marketable securities, while also provid- to emulate. Other funds also have much to learn from ing for an interest rate floor. The government transfers the pivotal role that CPF savings has played in helping the SSGS proceeds to the GIC Private Limited, one of members finance homeownership, which is now close Singapore’s two sovereign wealth funds. The GIC, which to universal in Singapore. As is the case in most Asian was established in the 1970s with the objective of secur- provident funds, CPF retirement benefits are often mod- ing the long-term economic future of the country, invests est, which is not surprising given the relatively small allo- the proceeds, along with other government funds, out- cation of total contributions to the SA account. The CPF side of Singapore in a globally diversified portfolio. reports that two-fifths of members turning 55 in 2018 did not have sufficient funds in their OA and SA to set aside The CPF also differs from other provident funds in that the BRS, which is meant to cover basic living expenses it requires the partial annuitization of account balances in retirement. Yet the partial annuitization requirement upon retirement. When CPF members turn 55, a new provides critically important protection against longev- Retirement Account (RA) is created for them using sav- ity risk that other provident funds lack. Meanwhile, as ings from their SA and, if necessary, their OA. Mem- homeowners, CPF members will not need as much cash bers who are homeowners are required to set aside an income in retirement as they would as renters. amount called the Basic Retirement Sum (BRS), while those who are not homeowners are required to set aside It is less clear that other provident funds should emulate an amount called the Full Retirement Sum (FRS), which the CPF’s approach to investment. Singapore’s circum- is twice as large. The funds transferred to the RA are stances, after all, are unique. As a high-income country used to purchase a government provided annuity, called with a fully developed economy, it does not need provi- CPF Lifelong Income for the Elderly, or CPF Life for dent fund savings to finance domestic development. Nor, short. Monthly benefits ordinarily begin at age 65, which as a global financial hub, does it need provident fund is known as the Payout Eligibility Age, but members can savings to broaden and deepen its capital markets. The choose to defer their annuities up to age 70, in which CPF’s approach to investment may serve Singapore’s case they receive higher monthly benefits. special needs. But if other countries were to adopt it, they would sacrifice some of the most critical advantag- The CPF’s partial annuitization requirement is one es of the provident fund model. CHAPTER 1: ASIA’S PROVIDENT FUNDS AT A CROSSROADS <<< 11 2. >>> Investment and Governance The provident fund model allows emerging markets to pursue two critical objectives at once: advancing national development and improving retirement security. The balance between the two objectives, however, should shift over time. While the needs of development policy may out- weigh the needs of retirement policy when countries are demographically young and economi- cally growing, strengthening retirement security should ultimately take precedence as countries become more developed and their populations age. This shift in focus is especially important in Asia, which is both developing and aging faster than anywhere else on earth. An accompanying evolution in investment and governance is essential to support such a shift in focus. This chapter begins with an examination of the stages of evolution in provident fund investment and governance, as well as some of the special challenges that provident funds may face as they evolve. It then evaluates the investment performance of India’s EPF, Indonesia’s JHT, and Malaysia’s EPF relative to the performance of a selection of other large pension funds and sovereign wealth funds around the world. Finally, it takes a step back and offers a more detailed discussion of specific investment and governance issues and concerns in each of the three provident funds. The Stages of Provident Fund Evolution The dual objectives of provident funds create an inevitable tension between different approaches to investment and governance, but also suggest a natural evolutionary path. Figure 5 offers a schematic depiction of the natural evolution in provident fund objectives, while figure 6 offers a schematic depiction of the accompanying evolution in investment and governance. The schemas are both proscriptive and descriptive. They illustrate how provident fund objectives, and with them investment and governance, should evolve over time. But they also illustrate how, to a significant extent, this evolution is actually occurring in Asia’s provident funds. CHAPTER 2: INVESTMENT AND GOVERNANCE <<< 13 When provident funds are first established, they invest almost almost every country with a large funded retirement system, exclusively in domestic government bonds and deposits, an the retirement system and domestic financial markets are in- arrangement that can serve both of their objectives. On the tertwined, the one supporting the development of the other one hand, by becoming major holders of government debt as they both grow. By diversifying their portfolios to include provident funds can reduce government borrowing costs and domestic corporate equity and debt, provident funds can help support government spending. On the other hand, because to broaden and deepen domestic financial markets, a critical developing-country debt offers a significant credit spread or pillar of national economic development, while also helping to risk premium,5 it can provide a reasonable return to provident maintain reasonable returns for their members—at least for a fund members. As countries’ economies develop, the risk time. Ultimately, as provident funds continue to grow and do- premium earned by owning domestic government debt natu- mestic capital markets begin to mature, maintaining reason- rally drops. At the same time, however, the growing size of able returns will require global diversification of investment provident funds, together with the development of domestic portfolios. financial markets, allows them to diversify their portfolios. In > > > FIGURE 5 Economic Development Retirement Security • Support government lending • Reduce old-age poverty • Supply funds for social capital • Provide adequate income replacement Balance investments in retirement should shift • Facilitate the growth of over time • Maintain economic growth by private-sector capital markets minimizing the impact of retirement on consumption > > > FIGURE 6 Provident Economic Fund Retirement Development Objective Security Investment Stages Government Domestic Direct Domestic International ”Mark-to- Lifecyle “Liability- debt infrastructure placement in public investment market” funds driven” and state- corporations markets accounting strategy owned and real enterprises estate Governance Stages Focus on Focus on eliminating Focus on developing Focus on developing eliminating self-dealing and general investment specialized investment fraud, graft, conflicts expertise expertise and theft of interest 5. Governments in developing countries pay higher rates on their debt than those in developed countries to compensate lenders for the risk of default related to economic, polit- ical, and legal uncertainty. See Aswath Damodaran, Country Risk: Determinants, Measures, and Implications—The 2015 Edition (New York: New York University, July 2015). 14 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES The three provident funds covered in the report now find them- as a rule, credit all members with the same rate of return. In de- selves situated somewhere along this continuum. (See table fined contribution systems, however, asset allocation is ideally 1.) India’s EPF is still overwhelmingly invested in government tailored to each member’s age. In the early years of a member’s debt. Although it has recently expanded its portfolio to include career, it is appropriate for portfolios to assume a high level of some exposure to domestic equities, no foreign investment of risk, which most of the time should result in high returns. As any kind is allowed. Indonesia’s JHT also invests heavily in the member grows older and retirement approaches, however, government debt, but has a much larger exposure to domestic portfolios should become progressively more conservative. equities. As with India’s EPF, no foreign investment is allowed. Along with lifecycle investing, as this age-related investment Ten to fifteen years ago, the situation would have been similar strategy is called, customization of the asset portfolio may also in Malaysia’s EPF. Since then, however, it has acquired a glob- involve liability-driven approaches to investment that aim to re- ally diversified portfolio. It is important to note that the place of duce income risk in retirement by matching asset cash flows each of the provident funds in the continuum has little to do with with future income needs through attention to currency risk, in- how long the provident fund has been operating. India’s and flation risk, and bond duration. Although these strategies have Malaysia’s provident funds both date to the 1950s, while Indo- the potential to improve investment outcomes for provident fund nesia’s was established in the 1990s. Rather, its place reflects members, they might challenge the sense of common national each country’s stage of economic development. interest that underlies the whole provident fund model. There is, at least potentially, another evolutionary stage in in- It is true that one of the three provident funds covered in the vestment policies and practices that none of the three provident report, Malaysia’s EPF, already allows for a degree of individual funds has yet reached. During this stage, opportunities open up customization of the asset portfolio. Members whose account to optimize member outcomes through individual customization balances exceed certain thresholds are allowed to withdraw the of the asset portfolio. Taking advantage of these opportunities, excess and invest it in one or more approved investment funds however, would require fundamental changes in the way that of their choosing. But this special program, called the Members provident funds currently operate. Some provident funds may Investment Scheme (MIS), is relatively small. More importantly, determine that the advantages are important enough and will self-directed retirement accounts are unlikely to be the most ef- evolve in this direction, while others may not. fective way to optimize investment outcomes. Individual cus- tomization of the asset portfolio can and should be achieved As explained in Chapter 1, provident funds typically pool mem- within provident funds’ overall collective investment framework. ber contributions in the same identical investment portfolio and, > > > T A B L E 1 - Asset Allocation at the end of 2018 or Most Recent Year Available India Indonesia Malaysia EPF JHT EPF Government Debt 79% 63% 28% Deposits 9% 9% 6% Non-Government Debt 8% 0% 22% Equities 4% 27% 39% Real Estate/Infrastructure/Other 0% 1% 5% Memo: Foreign Investment 0% 0% 27% Note: Data for India and Indonesia are for the end of 2017. Government debt in Indonesia includes non-government debt. Source: EPFO, BPJS, and EPF (Malaysia) CHAPTER 2: INVESTMENT AND GOVERNANCE <<< 15 Progressing to the next evolutionary stage might also require government-managed system, by keeping investment deci- altering the way that provident funds account for and credit in- sions at arms-length. vestment returns. Provident funds typically value assets with- out marking them to market. For bonds, which are accounted Needless to say, governance procedures and safeguards for as “held to maturity,” this means that interest income is need to evolve in tandem with investment policies and prac- recognized as it is received, but no change in market value is tices. In the early stages of development, when provident recognized when interest rates change. For equities, it means funds invest largely or even exclusively in government debt that returns are equal to dividends plus any gains or losses and deposits, simple safeguards against corruption may be realized on sales, with unrealized gains and losses kept in a sufficient. As assets under management grow, as economies reserve account. and financial markets develop, and as portfolios are expanded to include investments in all types of domestic assets, more This approach allows provident funds to smooth returns, elaborate safeguards against self-dealing, nepotism, favorit- which has potential benefits. Crediting member accounts with ism, and manipulation of markets must be added. By the time stable returns may serve to increase public trust and confi- a provident fund has acquired a fully diversified global portfo- dence. It can also shield members close to retirement from lio, the rules governing asset allocation, along with the entire large downswings in financial markets. At the same time, how- investment decision-making process, should have become an ever, the approach can distort investment decision-making integral part of the governance structure itself. in ways that undermine long-term performance. Rather than focus on maximizing long-term returns, provident funds may It is important to recognize that provident funds face some feel compelled to “strategically” sell assets in order to man- special governance challenges that arise less frequently in age near-term returns and meet public expectations. Unreal- other types of funded retirement systems. One such challenge ized investment losses can also build up and ultimately lead is how to minimize political influence over investment deci- to larger reductions in returns for members than if account sions. The most effective approach would be to set up provi- values had moved with the market. In addition, the subjective dent funds as independent quasi-public corporations. The nature of administratively determined returns, relative to the Canada Pension Plan, which takes this approach, provides an objective nature of market returns, creates additional gover- impressive model for how to manage public pension assets at nance challenges and may subject policymakers to criticism arms-length from government. (See box 3.) The reality is that for picking winners and losers. While there may be advan- provident funds are set up as state-run agencies or corpora- tages to an accounting approach that allows governments to tions with boards that typically include, and may indeed be engineer results with participants’ best interests in mind, mov- chaired by, high ranking government officials. All three provi- ing to market-linked returns can streamline governance and dent funds covered in the report appear to have effective safe- improve long-term outcomes. guards against self-dealing by government officials. Some, like Malaysia’s EPF, also have investment committees that, In whatever direction provident funds evolve, investment ex- in principle, are insulated from political influence. But no mat- pertise will become increasingly important. One key decision ter how well such matters are handled, the possibility remains is whether assets will be managed internally or outsourced that governments may use provident funds to influence finan- to asset managers. Most large investment funds around the cial markets, whether by directing investment to politically fa- world move toward in-house asset management as their pool vored industries or by assuming stakes in private companies of assets grows, with only investments requiring specialized that are so large that the government in effect controls them. knowledge outsourced. Developing the requisite expertise, however, takes time. The high compensation and competitive Another related governance challenge arises from the fact private-sector atmosphere that may be most conducive to suc- that most provident funds pursue two objectives that may be cessful in-house management may also be difficult to dupli- at odds with each other. After all, investment policies that are cate in government-managed provident funds. External asset intended to advance national development objectives, even management may thus have some important practical advan- when they are well designed and effective, may not be the tages, particularly as an interim arrangement while provident policies most likely to maximize returns for participants. It is funds are developing their own in-house investment expertise, important for provident funds to develop explicit guidelines a strategy that Malaysia’s EPF has successfully pursued.6 It for balancing national economic development and retirement may also help to reduce political influence, a concern in any security objectives, yet none of the three covered in the re- 6. See William Joseph Price et. al., Case Study on the Employees Provident Fund of Malaysia, The Malaysia Development Experience Series (Washington, DC: World Bank Group, June 2018). 16 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES port appear to have done so. It is also important for them to tute a reasonable measure of the benefits of provident fund have governance procedures in place for resolving conflicts investment policies to members, more sophisticated “social between their dual objectives, yet none of the three appear to impact” indicators are needed to track the benefits of invest- have them. Finally, provident funds should have clear metrics ment policies to society as a whole. None of the three provi- for measuring how effective they are at achieving both objec- dent funds currently have such metrics, though to its credit tives. While returns to assets under management may consti- Malaysia’s EPF is considering how to develop them. > > > BOX 3 - Canadian Public Pensions: Governance Lessons The successful evolution of provident funds into systems with a primary focus on retirement security, and with governance and investment policies that support this focus, will be facilitated by a long-term vision and thoughtful planning. The Canada Pension Plan (CPP), Canada’s national retirement system, as well as other large provincial public pension funds in Canada, provide a model for this kind of evolution. The development of Canadian public pension funds from the 1980s to today has been studied by the World Bank,* and is described here because of the important lessons it teaches about governance. As recently as the 1980s, Canada’s large public pension funds were government entities subject to strict investment limits and often invested entirely in government bonds. Today the funds have globally diversified portfolios with excellent performance track records and an arms-length relationship to government. At the end of the plan’s 2019 fiscal year, the CPP managed close to CA $400 billion, with only about 15 percent of the fund invested in Canada. Because the natural development of provident funds is from a focus on national economic development, with portfolios invested largely or entirely in government bonds, to a focus on retirement security, with globally diversified portfolios, the Canadian experience has relevance. Although this experience relates to defined benefit pension plans, which must man- age more complex financial issues than defined contribution provident funds, the ultimate goal is the same: financial se- curity in retirement. No matter what the benefit structure of a funded retirement system is, governance must be grounded in the same fiduciary objective of achieving optimal outcomes for members. Fully replicating the remarkable success achieved by Canada would be an ambitious goal for Asian provident funds to set. But key aspects of the Canadian experience could certainly be replicated in Asia. Since the time horizon for achieving the kind of change that Canada has achieved is decades, it can only happen if there is a long-term vision and plan in place. The most important lessons from the Canadian experience include the following: INDEPENDENCE. The goal is an arms-length relationship to government. In Canada, even the CPP Investment Board (CPPIB), which manages the investment portfolio for Canada’s partially funded national retirement system, operates independently from government. Achieving this degree of separation requires a concerted effort, and must start with an appropriate legislative and regulatory framework. The lesson from Canada is that it is also critical to develop a high level of trust among the key stakeholders, including government, employers, workers, and financial service providers. This can be accomplished by appointing the right people to the fund board and key management roles, and by making use of the highest-caliber administrators and investment experts. ACCOUNTABILITY AND TRANSPARENCY. Clear and complete financial reporting is an essential aspect of good governance. So is effective communication with a wide variety of stakeholders—policymakers, regulators, mem- bers, employers, journalists, and taxpayers—all of whom have differing but valid interests in understanding the rules, objectives, and performance of national retirement systems. The experience of Canada teaches that prioritiz- ing transparency and accountability will foster the kind of ethical and professional culture which can be both suc- cessful and independent from government. * World Bank, The Evolution of the Canadian Pension Model: Practical Lessons for Building World-Class Pension Organizations (Washington, DC: World Bank Group, 2017). > > > CHAPTER 2: INVESTMENT AND GOVERNANCE <<< 17 INVESTMENT POLICY. The transition from a portfolio largely or entirely invested in government bonds to a globally diversified portfolio requires governance that evolves with or even ahead of changes to investment policy. Even as Canada’s large public pension funds drew on external expertise, they also worked to develop and maintain in-house expertise with appropriate recruiting and compensation and a culture focused on long-term results. The CPPIB now employs over 1,500 investment professionals to manage its portfolio. Key characteristics of the investment policies of the funds include: diversification across asset classes and geographies; significant allocations to infrastructure and real estate, in part because the value of these assets tends to grow with inflation; and liability-driven strategies which assess risk based on expected future benefit payments. While most of the funds invest with the exclusive objective of maximizing long-term returns for their members, some do maintain a mandate to contribute to regional economic development. LONG-TERM PLANNING. A vision for future change is needed, but the ability to adapt to circumstances, recover from failures, and take advantage of opportunities is also important. The following schematic, which is adapted from the World Bank’s report, summarizes the main phases in the development of Canadian public pension funds. Phase Relationship to Government Investment Expertise Most expertise Pre-Reform Integrated Government bonds contracted externally Strategy for independence In-house expertise developing, Solid Foundation Beginning to diversify with buy-in from stakeholders outside experts still prominent Diversified portfolio, Evolving toward Independence Independent limited opportunity set in-house expertise Fully developed in-house Mature Highly diversified, Independent expertise, selective use Organization global portfolio of external experts Source: World Bank, The Evolution of the Canadian Pension Model: Practical Lessons for Building World-Class Pension Organizations Assessing Investment Performance One obvious way to assess the investment performance of higher-return growth assets than the large pension funds and provident funds is to compare it with the performance of in- sovereign wealth funds do. The smaller allocations to growth vestment funds around the world. Figure 7 compares the aver- assets in part reflect the stage of economic development in age annual real rate of return of India’s EPF, Indonesia’s JHT, each country. As development progresses in the future, and and Malaysia’s EPF from 2009 to 2018 with that of a selection as the investment experience of provident funds grows, the of six large pension funds and sovereign wealth funds over exposure to growth assets can be expected to increase, and, the same period. As is readily apparent, four of the other funds along with it, real rates of return. Significantly, Malaysia’s EPF, outperformed all three of the provident funds, in some cases which has the most diversified portfolio and the highest allo- by a wide margin, and all of the other funds outperformed In- cation to growth assets of the provident funds covered in the dia’s EPF. report, had a real rate of return that was much closer to that of the large pension funds and sovereign wealth funds than In- There are several explanations for provident funds’ relatively dia’s EPF did. The real rate of return registered by Indonesia’s poor performance. One explanation is that they allocate small- JHT was also much closer, but that was due in large part to the er shares of their portfolios to equites and other higher-risk, high risk premium on its government debt holdings. 18 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES > > > F I G U R E 7 - Real Rate of Return: Ten-Year Average, 2009-2018 9% 8.5% 8% 7% 6.2% 6% 5.7% 5.5% 5.0% 5% 4.2% 3.7% 4% 3.5% 3% 2% 0.8% 1% 0% India Indonesia Malaysia Korea Japan Chile CalPERS Norway Ontario EPF JHT EPF NPS GPIF AFPs GPFG Teachers PROVIDENT FUNDS PENSION & SOVEREIGN WEALTH FUNDS Note: The ten-year average return for India’s EPF refers to the fiscal years beginning 4/1/2008 and ending 3/31/2018. Japan’s GPIF also has a 3/31 fiscal year; in calculating its ten-year average return for 2009 to 2018, returns for 4/1/08 to 3/31/18 and 4/1/09 to 3/31/19 were averaged. CalPERS has a 6/30 fiscal year; in calculating its ten-year average return for 2009 to 2018, returns for 7/1/08 to 6/30/18 and 7/1/09 to 3/31/19 were averaged. The return for Chile’s AFPs is the simple average of returns for all fund classes. Source: EPFO, BPJS, EPF (Malaysia), Korea’s NPS, Japan’s GPIF, Chilean Superintendency of Pensions, CalPERS, Norges Bank Investment Management, Ontario Teachers’ Pension Plan, and GAI calculations A number of important technical factors also disadvantaged instead of the 2009-2018 period, the relative performance of provident funds relative to the large pension funds and sover- provident funds would have appeared more favorable. eign wealth funds over this particular comparison period. For one thing, slow global growth and massive central bank inter- A final explanation is that provident funds are typically pursu- vention combined to drive interest rates down to record lows— ing two objectives, while the large pension funds and sover- and asset prices up to record highs. Since provident funds eign wealth funds are pursuing just one—namely, maximiz- generally do not mark their asset values to market, these price ing investment returns. If provident fund investment returns gains were not realized in provident funds to the same ex- could somehow be adjusted to include the social returns the tent that they were in the large pension funds and sovereign funds generate by underwriting national development, the pic- wealth funds. For another thing, the comparison period begins ture might appear quite different. It is possible that their social just after the most dramatic part of the downturn in financial returns more than compensate for their lower investment re- markets in 2008-2009, which weighed much more heavily on turns, at least in some theoretical accounting framework. Un- the large pension funds and sovereign wealth funds than it fortunately, the metrics needed to determine whether this is so did on provident funds, with their more conservative portfo- do not currently exist for any of the provident funds covered lios. If returns had been compared over the 2008-2017 period in the report. CHAPTER 2: INVESTMENT AND GOVERNANCE <<< 19 > > > F I G U R E 8 - Real Rate of Return vs. Growth Rate in Real GDP Per Capita: Ten-Year Average, 2009-2018 10% 8.5% 8% 6.2% 5.7% 6% 5.5% 5.5% 5.0% 4.0% 4.2% 4% 3.7% 3.5% 3.2% 2.6% 1.9% 2% 0.8% 0.8% 1.0% 0.6% 0.1% 0% India Indonesia Malaysia Korea Japan Chile CalPERS Norway Ontario EPF JHT EPF NPS GPIF AFPs GPFG Teachers PROVIDENT FUNDS PENSION & SOVEREIGN WEALTH FUNDS Real Return Real GDP Per Capita Return above GDP Per Capita India’s EPF -4.7% Korea’s NPS 1.0% CalPERS 4.7% Indonesia’s JHT 0.9% Japan’s GPIF 3.4% Norway’s GPFG 6.1% Malaysia’s EPF 0.5% Chile’s AFPs 3.6% Ontario Teachers 7.9% Note: See figure 7. Source: EPFO, BPJS, EPF (Malaysia), Korea’s NPS, Japan’s GPIF, Chilean Superintendency of Pensions, CalPERS, Norges Bank Investment Management, Ontario Teachers’ Pension Plan, and GAI calculations Another way to assess the performance of provident funds all of the pension funds and sovereign wealth funds, real rates is to compare the real returns on their investment portfolios of return exceeded growth rates in real GDP per capita, and with real growth in living standards in each country. It is not in five of the six the margin was at least 3 percentage points. enough for funded retirement systems to beat inflation. To One reason for the poor performance of the provident funds generate meaningful benefits, they must also beat real wage relative to the pension funds and sovereign wealth funds is of growth. Figure 8 compares real rates of return from 2009 to course that their real rates of return were generally lower. But 2018 in the three provident funds and the six large pension another reason is that real GDP per capita in the provident funds and sovereign wealth funds with growth rates in real fund countries has been growing much faster. GDP per capita, used here as a proxy for real wage growth, for which it is difficult to compile data on a consistent basis All of this suggests that provident funds will need to improve across countries. In Indonesia and Malaysia, real rates of re- their investment performance as their focus shifts from nation- turn exceeded growth rates in real GDP per capita, but only al development to retirement security. This may require aban- by a narrow margin of 0.9 and 0.5 percentage points, respec- doning old strategies and adopting new ones, some of which tively. In India, the real rate of return lagged behind growth in could involve changes in the way that provident funds current- real GDP per capita by an enormous 4.7 percentage points. In ly operate. To begin with, provident funds may need to relax 20 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES portfolio allocation restrictions, particularly on foreign invest- ducing lifecycle funds or liability-driven approaches to invest- ment. While a primary or even exclusive focus on domestic ment. The steps they take will determine how successfully investment may have made sense in the past, it will make less they manage the necessary transition from a primary focus on sense in the future as economies develop, populations age, national economic development to a primary focus on retire- and returns to domestic investment decline. Provident funds ment security. may also need to reconsider the minimum rate of return guar- antees which some now feature. While such guarantees can protect members against adverse investment outcomes in the Three Provident Funds in Focus near term, most economists agree that in the long term they reduce retirement benefits by compelling asset managers to pursue overly conservative investment strategies.7 More fun- damentally, some provident funds may decide to move toward The remainder of the chapter takes a closer look at investment greater individual customization of the asset portfolio, which and governance in each of the three provident funds covered in turn may require moving from administratively determined in the report. The discussion focuses in particular on invest- returns to market-linked returns. ment policy and investment performance, where a detailed evaluation is possible using published data, available primar- As provident funds prepare for the future, they are likely to ily in the funds’ annual reports and financial statements. It also confront the same lower-return environment that may be chal- pays close attention to accounting practices, since they are lenging retirement systems worldwide. It is almost certain an important aspect of governance and can shed light on how that the high rates of return to financial assets that retirement provident funds operate. As we will see, the financial state- systems have become accustomed to over the past decade ments of the three provident funds vary in their clarity and cannot be repeated over the next, for the simple reason that transparency, with Malaysia’s being considerably more so- the drop in interest rates and corresponding rise in asset pric- phisticated than India’s or Indonesia’s. Although other aspects es that propelled them cannot be repeated. Rates of return, of governance are also important for the success of provident moreover, may fall further in emerging markets than in fully funds, it was not possible to evaluate them, since a thorough, developed economies, since they naturally start out higher “on-the-ground” assessment of governance was not within the there and drop over time as the default risk on government scope of the report. Nonetheless, some key issues of potential bonds declines and equity markets grow and become more concern are mentioned. diverse, less volatile, and more liquid. The section on each provident fund begins with a set of charts. Yet there may also be a silver lining for emerging markets. The first chart, which relates to investment policy, shows how Although rates of return may be lower in the future than they the provident fund’s asset allocation has evolved over the past were in the past, economic growth is likely to be slowing too. If decade. The second chart, which relates to investment perfor- the slowdown were due to failed development, that would be mance, compares the provident fund’s nominal rate of return a problem. But in the emerging markets of Asia, it will not be from 2009 to 2018, labelled “Reported Return” on the charts, due to failure but to success. As their economies develop, the with inflation and the ten-year bond yield, both of which are rapid rates of income growth that Asian countries have experi- important benchmarks. As the price of goods and services enced will gradually converge to developed-world rates. When rises, provident funds must achieve a return above inflation it comes to retirement policy, this means that provident funds in order to provide any value to members. Achieving a return will not need to achieve as high a rate of return in the future significantly above the ten-year bond yield generally indicates as they would have in the past in order to beat real wage that a fund’s portfolio has diversified out of government bonds growth, a bottom line prerequisite for providing adequate re- and is successfully earning a risk premium over time on behalf tirement benefits. of members. The chart also includes an inset that compares the provident fund’s ten-year average real return with the ten- It will be interesting to follow the progress that the less devel- year average growth rate in real GDP per capita. Achieving a oped provident funds make in coming years as they diversify real return significantly above the growth rate in real GDP per their portfolios, both by asset class and outside the country. It capita indicates that members’ account balances are growing may be even more interesting to follow the more developed faster than their incomes, a critical goal if provident funds are provident funds as they continue to evolve, perhaps by intro- to generate adequate replacement rates. 7. See the discussion in OECD, “Design and Delivery of Defined Contribution (DC) Pension Schemes: Policy Challenges and Recommendations” (report presented at the Cass Business School Conference on Defined Contribution Pensions: Guarantees and Risk Sharing, London, March 5, 2013). CHAPTER 2: INVESTMENT AND GOVERNANCE <<< 21 The third chart, which relates to accounting, compares two different measures of investment returns: the provi- dent fund’s Reported Return, already shown on the previous chart, and a measure of return called the GAI Calcu- lated Return. The Reported Return is the rate of return which is reported as credited to member accounts in the provi- dent funds’ annual reports. In India and Malaysia, this number is also widely reported in the press and is the subject of considerable public anticipation and discussion. The GAI Calculated Return is derived from information in the provident funds’ financial statements on beginning-of-year balances, investment return amounts, and other cash flow, such as contribu- tions and withdrawals. To the extent that the information is provided, assets are marked to market value.8 The following for- mula, in which the denominator represents the average balance of the fund assuming even cash flows throughout the year, was used: GAI Calculated Return % = Investment Return $ / (BOYB + 0.5 x CF) WHERE: Investment Return $ = Investment return amount, with assets marked to market if possible BOYB = Beginning-of-year balance CF = Cash flow other than investment returns, including contributions, withdrawals, and expenses This comparison can throw light on important investment and that there are cross-subsidies from the general government governance issues. As explained above, provident funds do budget to the provident fund, or vice versa. Although such not normally mark assets to market value. When the GAI Cal- cross-subsidies would not themselves be an indication of poor culated Return reflects marked-to-market asset values and governance, lack of transparency about them would be. the Reported Return does not, as is the case with Malaysia’s EPF, differences between the two measures reveal the poten- While there are certain investment and governance funda- tial which the provident fund has to smooth returns. But even mentals that apply to all provident funds, other policies and when there is no difference in the way assets are valued, as is practices need to be tailored according to the stage of devel- the case with India’s EPF and Indonesia’s JHT, the compari- opment of each fund. The following discussion tries to take son can still be instructive. In such cases, the two measures into account the unique circumstances in each country. should track each other closely. If they do not, it may indicate 8. None of the three provident funds’ financial statements provide the information needed to mark bonds to market value. For equities, the financial statements for Malaysia’s EPF provide the needed information; those for Indonesia’s JHT do not; and those for India’s EPF, unusually for a provident fund, already mark equities to market value. 22 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES India’s EPF > > > F I G U R E 9 - Portfolio Allocation at the End of 2008 and 2017 4% 1% 8% 9% 2017 79% 43% 2008 56% Foreign Investment: 0% Foreign Investment: 0% Government Debt Deposits Non-Government Debt Equities Real Estate/Infrastructure/Other Note: Asset allocation is as of 5/31/2008 and 12/31/2017. > > > F I G U R E 1 0 - Reported Return vs. Inflation and Ten-Year Bond Yield 12% 10% 8% 6% 4% 2% Real Rate of Return: 0.8% Real GDP Per Capita: 5.5% 0% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Reported Return Inflation 10-Year Bond Note: Returns are for fiscal years ending 3/31. > > > F I G U R E 1 1 - Reported Return vs. GAI Calculated Return 10% 8% 6% 4% 2% 0% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Reported Return GAI Calculated Return Note: Returns are for fiscal years ending 3/31. 10-Yr. Avg. 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Reported Return 8.67% 8.50% 8.50% 9.50% 8.25% 8.50% 8.75% 8.75% 8.80% 8.65% 8.55% GAI Calculated Return 8.2% 6.9% 7.8% 8.0% 8.1% 8.8% 9.2% 8.8% 9.2% 7.6% 8.4% 10-Year Bond 7.8% 7.5% 7.3% 7.9% 8.5% 8.2% 8.5% 8.3% 7.7% 6.9% 7.0% Inflation 7.9% 9.0% 11.2% 11.2% 9.0% 9.7% 9.8% 6.2% 5.6% 4.3% 3.1% CHAPTER 2: INVESTMENT AND GOVERNANCE <<< 23 The Employees’ Provident Fund (EPF) is administered by as high, meaning that real returns have been quite low: just the Employees’ Provident Fund Organization (EPFO), which 0.8 percent. (See figure 10.) With India’s economy growing in turn is part of the Ministry of Labour and Employment. In rapidly, the growth rate in real GDP per capita has exceed- addition to the EPF, the EPFO also administers two related ed the real EPF return by an enormous 4.7 percent per year programs on behalf of its members, the Employees’ Pension over the period, making it extremely difficult for the system to Scheme (EPS), which provides retirement, disability, and sur- generate adequate replacement rates. The good news is that vivors pensions, and the Employees’ Deposit Linked Insur- real returns have improved significantly since 2014 as infla- ance Scheme (EDLI), which provides a life insurance ben- tion has dropped, a trend that will help improve future benefit efit in an amount linked to members’ EPF account balances. outcomes if it continues. The risk premium earned by the EPF GAI’s analysis is limited to the EPF. The financial statements above the yield on ten-year central government securities has and investment return information reviewed by GAI cover the been modest, as would be expected from a portfolio invested period April 1, 2008 to March 31, 2018.9 primarily in government bonds. Measured against the EPF Reported Return it was 0.9 percent, while measured against Investment Policy the GAI Calculated Return it was 0.4 percent. Although the EPF was established in the 1950s, when it Accounting comes to investment policy it is still in the early stages of de- velopment. The overwhelming majority of EPF assets consist The EPFO’s accounting sometimes lacks the clarity and trans- of securities issued by the central and state governments or parency that one would expect from a large investment fund. by state-run institutions. (See figure 9.) To be sure, there has This is especially true when it comes to the central question been some evolution in asset allocation over the past decade. of how investment returns are credited to member accounts. Corporate debt holdings have increased, and for the first time the fund has begun investing in equities. Together, however, Figure 11 shows that the EPF Reported Return is sometimes these two asset classes only comprise 12 percent of the EPF significantly higher or lower than the GAI Calculated Return. portfolio, with equities comprising just 4 percent. The largest The reasons for the discrepancy are not entirely clear, but it movement has been out of deposits, and in particular the cen- may be the result of inconsistencies in accounting. Although tral government’s Special Deposit Scheme, and into govern- the EPF Reported Return for a given year is based, at least ment debt, a category that includes state development loans. loosely, on the fund’s earnings for that year, interest is not The large allocation to government debt, together with a prohi- credited to member accounts until the following year. In the bition on foreign investment of any kind, suggest that domestic interim, the earnings are held in an aggregate “interest ac- development remains a primary policy objective. count,” which functions as a holding account. The interest ac- count gives the EPFO some flexibility in smoothing returns, The EPFO outsources asset management to investment which may be desirable. The problem is that, when it comes firms, which vary in number from year to year. There are cur- time to credit member accounts, the EPFO does not always rently just two: the State Bank of India and UTI Asset Manage- appear to use the same account balance basis. There have ment, both of which are state owned. The investment firms also been occasions when the EPFO has discovered, some- are subject to strict quantitative portfolio allocation guidelines, times years after the fact, that member accounts were never including minimum allocations to government debt and maxi- credited at all.10 The discrepancy between the EPF Reported mum allocations to equities. Equity investments, moreover, Return and the GAI Calculated Return may also indicate the are exclusively in ETFs, with the total split between the two existence of cross-subsidies between the government and the large domestic equity indexes, the Nifty50 and Sensex. As a EPF designed to keep reported returns stable within a narrow result, the yield achieved by the different managers in most and politically acceptable range. Whatever the explanation, it years varies by only a few basis points. is not attributable to mark-to-market accounting. In the case of India’s EPF, the Reported Return and the GAI Calculated Investment Performance Return value assets on exactly the same basis. EPF nominal returns have been high, averaging almost 9 per- The EPFO’s decision to begin investing in equites has created cent from 2009 to 2018. However, inflation has been nearly additional accounting challenges. Although the first equity in- 9. The analysis excludes so-called exempted establishments operating an EPF Private Trust. The EPF accounts of employees at these generally large establishments are managed in house rather than by the EPFO. 10. In 2011, for instance, it was unexpectedly determined that sufficient excess funds were in the interest account to both increase the reported return for the year and update millions of EPF account balances that had not been updated for decades. See “EPFO May Offer 9.5% Interest on PF This Year Too,” The Economic Times, June 23, 2011. 24 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES vestments were made in 2016 and some earnings from those sees the appointment of the other board members. Of these, investments were added to the interest account and allocated five are central government representatives and fifteen are to member accounts in 2018, the ultimate method for allocat- state government representatives. There are also ten employ- ing equity earnings had not yet been implemented as of the er and ten employee representatives. end of the fund’s 2018 fiscal year. There will apparently be two options. Members may elect to have their pro rata share Because the EPF is in the early stages of provident fund de- of the EPF’s market return on equity investments allocated velopment, governance needs are important but basic. As the to separate “equity accounts” opened in their name. If they fund continues to evolve, improving the clarity and transpar- do not make this election, their regular EPF accounts would ency of accounting should be a primary governance objective. continue to be credited with a fixed rate of interest that would So should improving in-house investment expertise, which will presumably factor in returns on the equity component of the be necessary for designing an effective long-term strategy EPF’s portfolio, but would be subject to the same adminis- whether or not the EPFO continues to outsource investment trative decision-making process that currently determines the management. The diversification of the EPF’s portfolio into eq- interest credited to member accounts. uities was an important step, and has been implemented in a way that minimizes governance concerns about self-dealing Governance and conflicts of interest. However, limiting equity investment to ETFs that track the two major domestic market indices may The EPFO is overseen by a forty-three member Central Board not be optimal in the long run. Over time, the EPFO will need of Trustees. The Chairman, Vice Chairman, and CEO, known to take a more sophisticated approach to investing in domestic as the Central Provident Fund Commissioner, are all officials equities. It will also need to consider global diversification of from the Ministry of Labour and Employment, which also over- the EPF investment portfolio. CHAPTER 2: INVESTMENT AND GOVERNANCE <<< 25 Indonesia’s JHT > > > F I G U R E 1 2 - Portfolio Allocation at the End of 2008 and 2017 1% 1% 13% 27% 2017 63% 34% 2008 52% 9% Foreign Investment: 0% Foreign Investment: 0% Government Debt Deposits Non-Government Debt Equities Real Estate/Infrastructure/Other Note: Government debt includes non-government debt. Asset allocation is as of 12/31/2008 and 12/31/2017. > > > F I G U R E 1 3 - Reported Return vs. Inflation and Ten-Year Bond Yield 16% 12% 8% 4% Real Rate of Return: 5.0% Real GDP Per Capita: 4.0% 0% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Reported Return Inflation 10-Year Bond Note: The Reported Return for 2017 is estimated from return amount and cash flow. > > > F I G U R E 1 4 - Reported Return vs. GAI Calculated Return 16% 12% 8% 4% 0% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Reported Return GAI Calculated Return Note: The Reported Return for 2017 is estimated from return amount and cash flow. 10-Yr. Avg. 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Reported Return 9.9% 10.6% 10.6% 10.1% 9.1% 8.7% 13.8% 8.4% 9.5% 9.8% 8.1% GAI Calculated Return 10.1% 11.7% 11.1% 9.7% 9.8% 9.2% 13.2% 8.8% 9.9% 9.8% 8.1% 10-Year Bond 7.8% 11.0% 8.5% 7.2% 5.8% 7.2% 8.1% 8.2% 7.6% 6.9% 7.5% Inflation 4.9% 4.4% 5.1% 5.4% 4.3% 6.4% 6.4% 6.4% 3.5% 3.8% 3.2% 26 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES The Jaminan Hari Tua (JHT) is administered by BPJS Ke- investments between government and corporate securities, tenagakerjaan, one of Indonesia’s two social security agen- and there is no detail provided on either publicly traded equity cies. Prior to the establishment of the BPJS in 2014, the JHT investments or direct placements. This is surprising, since PT was administered by PT Jamostek, a state-owned enterprise. Jamostek typically provided much greater detail on JHT in- In addition to the JHT, the BPJS also administers three other vestments in its annual reports, including allocations to each programs: the Jaminan Kecelakaan Kerja (JKK), which pro- asset class, returns by asset class, and industry exposure. vides work accident benefits; the Jaminan Kematian (JK), There was even a discussion of how the maturity profile of which provides life insurance benefits; and the Jaminan Pen- bonds was matched to liabilities for benefit payments. siun (JP), which provides retirement, disability, and survivors pension benefits. GAI’s analysis is limited to the JHT. The The information that the BPJS does provide indicates that the financial statements and investment return information re- JHT, though still in the early stages of provident fund develop- viewed by GAI cover the period January 1, 2009 to December ment, is beginning to diversify its portfolio. To be sure, like In- 31, 2018. dia’s EPF, the JHT remains primarily invested in fixed-income securities. (See figure 12.) Although the mix is not disclosed, Investment Policy if the allocation is similar to what was reported under PT Ja- mostek the investments are primarily in government debt The BPJS sometimes fails to report basic information on in- and state-owned enterprises, including the four state-owned vestment policy, strategy, and performance. These gaps are banks and Indonesia’s Regional Development Bank. As in In- particularly striking in the area of asset allocation. There is no dia, moreover, foreign investment is prohibited. Yet the JHT indication in its annual reports of the division of fixed-income also has a sizeable equity position, which has grown over the CHAPTER 2: INVESTMENT AND GOVERNANCE <<< 27 past decade from 13 to 27 percent of its total portfolio. In ad- 13, JHT returns have continued to be relatively stable. The ex- dition, there are small allocations to infrastructure, real estate, ception is 2014, when there was a spike in returns attributable and private equity. to what is described as “profit-taking” in equity investments. While 2014 was a good year for equities, it may not be coinci- The BPJS appears to weigh development objectives heavily dental that it was also the year that provident fund assets were in its investment decisions. In recent annual reports, there is transferred from PT Jamostek to the BPJS. discussion of targeting investments that contribute to work- ers’ welfare, particularly in the areas of homeownership, food Figure 14 compares the JHT Reported Return with the GAI security, education, and transportation. There is also a focus Calculated Return for the fund. Since there were no data on infrastructure investment. Although infrastructure currently available to adjust the GAI Calculated Return for marked-to- constitutes a small portion of the JHT portfolio, plans have market asset values, one would expect the two measures to been announced to increase allocations to infrastructure proj- track each other closely. They are indeed well aligned, with ects of state-owned enterprises.11 Investment policy seems only small deviations. This finding is significant. Whatever to be determined and implemented internally, and there is no limitations BPJS accounting may have, there is no significant mention of external managers, except for small allocations to difference between reported investment earnings and the mutual funds. earnings actually credited to member accounts, as there is in India’s EPF. This in turn indicates that there are no cross- Investment Performance subsidies from the government to JHT members or vice versa. The JHT earned an average nominal return of just under 10 Governance percent from 2009 to 2018, slightly higher than the return earned by India’s EPF over the same period. At 5.0 percent, The BPJS is a public legal entity that reports to the President however, the real JHT return was much higher than the real of Indonesia. As is typical in Indonesia, a Board of Commis- EPF return. (See figure 13.) The better results are attributable sioners is responsible for setting policies and oversees a in part to the high credit risk premium on Indonesian bonds Board of Directors, which is more directly responsible for man- and in part to Indonesia’s lower inflation rate. The real JHT aging operations. The members of both boards are selected return also exceeded the growth rate in real GDP per capita and approved by the President and Parliament. Of the seven by 0.9 percentage points over the ten-year period—not a wide members of the Board of Commissioners, two are government margin, but still a better performance on this metric than that representatives. achieved by any of the other three provident funds covered in the report. Relative to the Indonesian ten-year bond yield, Although the JHT is in many ways still in an early stage of JHT’s investment portfolio has provided a 2.1 percent risk provident fund evolution, its substantial allocation to domes- premium. While the BPJS provides no analysis of this excess tic equities, as well as its material level of investment in pri- return, the JHT’s equity investments presumably contributed vate equity and state-owned enterprises, create special gov- much of it. From 2009 to 2018, the Jakarta Composite Index ernance needs. Investment expertise is important for any registered returns in excess of 15 percent per year. provident fund that has diversified its portfolio beyond gov- ernment bonds and deposits. The lack of clear and complete Accounting information on investment policy, strategy, and performance in the BPJS annual reports suggests that further develop- The BPJS does not provide sufficient accounting detail to de- ing this expertise should be a primary governance objective. termine with assurance how investment returns are credited to Given the JHT’s investment in private equity and state-owned member accounts. However, it appears that the BPJS makes enterprises, it will also be important for the BPJS to ensure use of a reserve account to smooth JHT returns. This was that adequate mechanisms for handling conflicts of interest the case when the fund was administered by PT Jamostek, and guarding against self-dealing are in place. The focus on whose annual reports clearly indicated that most investments development objectives in investment decisions similarly indi- were not marked to market value and presented figures for cates that it will be important to monitor and manage possible unrealized capital gains held in the reserve account. Although tensions between those objectives and maximizing invest- no such information is included in the BPJS annual reports, ment returns. these practices presumably continue. As is clear from figure 11. “BPJS Ketenagakerjaan to Boost Investment in Infrastructure Projects,” Jakarta Globe, August 8, 2016. 28 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES Malaysia’s EPF > > > F I G U R E 1 5 - Portfolio Allocation at the End of 2008 and 2018 5% 26% 28% 28% 2018 2008 39% 6% 6% 40% 22% Foreign Investment: 27% Foreign Investment: 1% Government Debt Deposits Non-Government Debt Equities Real Estate/Infrastructure/Other Note: Asset allocation is as of 12/31/2008 and 12/31/2018. > > > F I G U R E 1 6 - Reported Return vs. Inflation and Ten-Year Bond Yield 8% 6% 4% 2% Real Rate of Return: 3.7% Real GDP Per Capita: 3.2% 0% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Reported Return Inflation 10-Year Bond > > > F I G U R E 1 7 - Reported Return vs. GAI Calculated Return 10% 8% 6% 4% 2% 0% -2% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Reported Return GAI Calculated Return Note: The GAI Calculated Return includes unrealized gains/losses on equity investments. 10-Yr. Avg. 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Reported Return 6.2% 5.7% 5.8% 6.0% 6.2% 6.4% 6.8% 6.4% 5.7% 6.9% 6.1% GAI Calculated Return 5.8% 5.8% 9.4% 4.5% 8.2% 7.9% 4.9% 5.0% 4.9% 7.7% -0.2% 10-Year Bond 3.9% 4.1% 4.0% 3.9% 3.5% 3.8% 4.0% 4.0% 3.9% 4.0% 4.1% Inflation 2.4% 5.4% 0.6% 1.6% 3.2% 1.7% 2.1% 3.1% 2.1% 3.9% 0.9% CHAPTER 2: INVESTMENT AND GOVERNANCE <<< 29 The Employees Provident Fund (EPF) is administered by the EPF’s real return averaged a solid 3.7 percent—high enough EPF Board, which operates under the supervision of the Min- to exceed the growth rate in real GDP per capita, though only istry of Finance. The EPF consists of two accounts: Account by a narrow margin of 0.5 percentage points. The EPF earned I, which is earmarked for retirement savings; and Account II, a reasonable risk premium of 2.3 percentage points above the which is earmarked for nonretirement savings. GAI’s analysis average ten-year bond yield, thanks to its substantial equity includes both accounts. The financial statements and invest- holdings. The EPF’s investment performance has not only ment return information reviewed by GAI cover the period been respectable, but also quite stable, with annual returns January 1, 2009 to December 31, 2018. over the period only deviating from the average return by about plus or minus 0.5 percentage points. For a fund with sig- Investment Policy nificant holdings in equities and other risky assets, this pattern of returns seems quite remarkable. One explanation is that The EPF, with USD $203 billion in assets under management the EPF’s sizeable private equity investments are not subject at the end of 2018, is among the world’s largest pension funds. to the ups and downs of a public market. The more important Founded in the early 1950s, it initially invested almost exclu- explanation, however, lies in how the EPF accounts for gains sively in government securities. Beginning in the 1990s it be- and losses on the publicly traded assets in its portfolio. gan to diversify its portfolio, and now has significant holdings in riskier growth-oriented assets. (See figure 15.) Although 56 percent of total assets were still invested in fixed income at Accounting the end of 2018, there were also sizeable investments in Ma- laysian and global equity markets, as well as sizeable direct The EPF Board publishes thorough, detailed financial state- placements in Malaysian corporations. Private equity invest- ments and disclosures that allow important insights into its ments make up almost 10 percent of the portfolio, and include accounting practices. Investments are not marked to market entities classified as joint ventures (some management influ- value. Unrealized gains and losses are held in a reserve ac- ence), associates (significant management influence), and count, whose balance is presented in the financial statements subsidiaries (also significant management influence). Over but is not widely disclosed to the public. Each year, the invest- the past decade, foreign investment has grown rapidly, in- ment earnings allocated to member accounts include interest, creasing from just 1 percent of the EPF’s portfolio at the end dividends, and rental income, which together have provided of 2008 to 27 percent at the end of 2018. about a 3.5 percent return in recent years. In addition, any realized gains on assets sold are also allocated to member The EPF has two rate of return targets. The first, which is a accounts. This has added about 2 to 3 percentage points to legal requirement, is to declare at least a 2.5 percent nominal returns in recent years, resulting in a total dividend that is usu- return or “dividend” each year, while the second is to deliver ally between 6.0 and 6.5 percent. at least a 2.0 percent real return on a rolling three-year basis. Over the period examined, these targets have always been The EPF’s accounting practices allow it to smooth member met. There are policies in place that determine both tactical returns. As can be seen in Figure 17, the GAI Calculated Re- (short-term) and strategic (long-term) adjustments to asset al- turn, which is based on marked-to-market investment returns, location. Assets are managed both internally and by external has risen and fallen significantly from year to year over the fund managers. As EPF’s in-house investment expertise has past decade, while the EPF Reported Return, or dividend, has grown, however, the share of assets managed internally has remained stable. While smoothing returns has its advantages, steadily increased and now stands at about 85 percent of the when reported returns and actual investment returns diverge total. Since 2017, members have had the option of investing greatly, as they did in 2018, it can create problems. Despite their savings in a new Shariah compliant investment portfo- a negative investment return for the year, sufficient assets lio, which to date has delivered only slightly lower returns—50 were sold and investment gains realized to keep dividend basis points in 2017 and 25 basis points in 2018—than the rates (both conventional and Shariah) in the 6.0 to 6.5 percent conventional investment portfolio. range. The asset sales, however, depleted the EPF reserve account, which dropped from RM 41.9B to RM -7.3B at the Investment Performance end of 2018. Thus, at the beginning of 2019, the EPF found itself in a position where, instead of being able to realize gains The EPF’s nominal return averaged 6.2 percent from 2009 to in order to boost dividends, it might have to realize losses, 2018, which may seem low compared with the returns regis- thereby lowering dividends. And indeed, despite large gains tered by India’s and Indonesia’s provident funds. However, the on its equity investments in 2019, when that year’s dividend 30 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES was declared it came in at just 5.45 percent, the lowest level ernment representatives, four employer representatives, four since 2008. employee representatives, and three members chosen for their professional expertise. It is interesting to note that the EPF Board’s interpretation of international accounting standards is a key question ad- Any large, globally diversified investment fund requires a well- dressed by the Auditor General for Malaysia in its opinion on developed governance function. When that fund also holds the 2018 financial statements. These standards establish that large direct ownership stakes in domestic enterprises, as the the deferral of gains and losses for equity securities is only EPF does, that function becomes even more important. Some allowable if an irrevocable decision is made at the time the of the EPF’s direct ownership stakes are in private enterprises, securities are purchased and if the securities are not intended while others are in public ones. The EPF is one of Malaysia’s to be traded in the short term. It is not clear and obvious that seven Government-Linked Investment Companies (GLICs), the EPF’s equity investments meet these criteria. and is thus authorized to invest in Government-Linked Com- panies, such as national banks, utilities, and transportation Governance companies, which together make up over one-third of Bursa Malaysia market capitalization.12 Although the EPF has poli- The EPF Board is appointed by the Minister of Finance. In cies in place designed to insulate investment decisions from 2018, there were eighteen members. Besides the Chairman government influence, ensuring the effectiveness of these of the Board and the CEO of the EPF, these included five gov- policies remains a key governance challenge. 12. Jayan Menon, “Government-Linked Companies: Impacts on the Malaysian Economy,” Policy Ideas no. 45 (Kuala Lumpur: Institute for Democracy and Economic Affairs, December 2017). CHAPTER 2: INVESTMENT AND GOVERNANCE <<< 31 3. >>> Benefit Design and Adequacy At the most basic level, there are two dimensions to the adequacy of government retirement systems: their breadth, as measured by the share of the workforce that participates, and their depth, as measured by the share of preretirement income that they replace. When it comes to the first dimension of adequacy, the performance of the three provident funds covered in the report gives obvious cause for concern. In India and Indonesia, only around one-tenth of the workforce contributes to the EPF or the JHT in a given year. In Malaysia, the coverage rate is considerably higher. Still, just one-half of the workforce contributes to the EPF in a given year. (See figure 18.) Including participants in special retirement schemes for civil servants and the armed forces, the share of the workforce covered by a contributory government retirement program is at least somewhat higher in all three countries and significantly higher in Malaysia, which has an unusually large public sector. But in none of them is the overall coverage rate close to universal. > > > F I G U R E 1 8 - Effective Coverage Rate at the End of 2018 or Most Recent Year Available 60% 50% 40% 20% 12% 9% 0% India Indonesia Malaysia EPF JHT EPF Note: The effective coverage rate equals active members as a share of employment, where active members are defined as members whose employers make regular contributions on their behalf (India) or members who have made a contribution within the past year (Indonesia and Malaysia). Data for Indonesia are for the end of 2017; data for India are an average for 2016-17. Source: EPFO, BPJS, EPF (Malaysia), ILO, and GAI calculations This low coverage problem, however, is not a failing of the provident fund model. Although low coverage constitutes a serious policy challenge, it is no more of a challenge for provident funds than for other types of contributory retirement systems in emerging markets. Low coverage is mainly a function of labor-market informality, and it is little affected by how retirement systems are organized and financed. While there are steps that the three provident funds could take to increase coverage, some of which will be discussed in the next chapter, so long as informality remains widespread progress will be difficult. In India and Indonesia in particular, where between 80 and 90 percent of the workforce labors in informal employment, substantially increasing cov- CHAPTER 3: BENEFIT DESIGN AND ADEQUACY <<< 33 erage will likely require substantially increasing the size of the making the Haj, receives 30 percent. In India and Indonesia, formal sector. The ability of countries to do so will in turn de- there are no separate accounts for retirement and nonretire- pend on the success of their broader development agendas, ment purposes, and consequently no clear distinction between and especially initiatives that increase human capital and re- retirement and nonretirement funds. All member contributions duce inequality. This reality is why it is so critical for emerging flow to the same account, which can be accessed prior to re- markets to back up their contributory retirement systems with tirement under a wide array of circumstances. In India’s EPF, robust social pension systems. these include purchasing a home or paying off a mortgage, paying for medical expenses, and financing the education or This report focuses on the second dimension of retirement marriage of one’s children. Members can also cash out the system adequacy, in part because it receives less attention portion of their EPF balances that is attributable to their own and in part because the outcomes are more closely, though by contributions when they quit or are laid off from their current no means exclusively, related to the design of provident funds job, provided that they have been unemployed for at least two themselves. The evidence suggests that, beyond the low months. Indonesia’s JHT allows partial withdrawals to finance coverage problem, there is also cause for concern about the home purchases, as well as complete cash outs in the event adequacy of retirement benefits in all three provident funds workers lose their job. covered in the report. The chapter begins with a discussion of the factors that tend to undermine benefit adequacy, then There are no longitudinal data available that reveal how much reviews what we know about the benefit levels of current re- of members’ total savings is on average preserved for retire- tirees. Finally, it presents stylized projections of replacement ment over the course of their careers. However, for two of the rates for future retirees. provident funds there are data on current-year withdrawals by type, and these data suggest that nonretirement withdrawals greatly reduce the ultimate adequacy of retirement benefits. In Adequacy Today 2018, nonretirement withdrawals accounted for 42 percent of the total funds withdrawn from Malaysia’s EPF. In Indonesia’s JHT, where the data refer to 2017, the corresponding figure A common problem in funded retirement systems is that contri- is an astonishing 88 percent. Although the figure for India’s bution rates may be too low to generate adequate replacement EPF was not available, it would be surprising if it were not also rates. This is clearly a problem in Indonesia’s JHT, whose com- very large. bined employer-employee contribution rate is just 5.7 percent of wages. It is not a problem, however, in the other two provident Then there is the matter of early retirement ages. In India’s funds covered in the report. In India’s EPF, the contribution rate EPF the retirement age, after which members can withdraw is 15.7 percent of wages, while in Malaysia’s EPF it is 24 per- any remaining account balance, is just 55. In Malaysia’s EPF it cent for most members.13 It is true that contribution rates in India is also 55, although, as of 2017, any incremental contributions and Malaysia were much lower in the past. But most of those made by workers who remain employed past age 55 must be members who have recently retired or are approaching retire- preserved until age 60. Until recently, the retirement age in In- ment age were subject to the current higher contribution rates donesia’s JHT was 55 as well. Indonesia, however, has begun throughout most or all of their careers. raising the JHT retirement age in stages to 65, which it will reach in 2043. As of 2020, it stood at 57. Early retirement ages What does undermine the adequacy of retirement benefits in can greatly lower replacement rates, since they both reduce all three provident funds is that much of what workers con- the number of working years during which contributions are tribute is not preserved for retirement. Provident funds, after made and increase the number of retirement years that ac- all, serve multiple savings purposes. In the case of Malaysia, count balances need to finance. All other things being equal, the allocation of savings to retirement and nonretirement pur- postponing retirement from age 55 to 65 nearly doubles po- poses is, in large part, the result of deliberate policy choices. tential replacement rates. Each EPF member has two accounts. Account I, which is ear- marked for retirement, receives 70 percent of total contribu- There is also the lack of provision for lifetime income. In India’s tions while Account II, which can be accessed prior to retire- EPF, Indonesia’s JHT, and Malaysia’s EPF account balances ment for a variety of purposes, from purchasing a home to can be cashed out entirely as lump sums payouts, which of- 13. In Malaysia’s EPF, contribution rates are lower for members aged 60 and over, as well as on earnings above a relatively high threshold. 34 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES fer no protection against inflation risk or longevity risk. Among > > > Asia’s major provident funds, only Singapore’s CPF requires F I G U R E 1 9 - Active Members as a Percent of Total even the partial annuitization of account balances. Members at the End of 2018 or Most Recent Year Available 52% Beyond design features like nonretirement withdrawal rules or 60% retirement ages, over which policymakers have control, there 40% 50% are two broader economic and labor-market factors that can affect the adequacy of funded retirement benefits. One is low 40% 21% “contribution density,” which is a close relative of low coverage 30% rates. Many workers in emerging markets may start, stop, and 20% then restart contributing to government retirement systems 10% a number of times during the course of their working lives. 0% Some workers, particularly women, may leave the labor force India Indonesia Malaysia EPF JHT EPF altogether to raise children or care for other family members, then later reenter it. Even if workers remain continuously em- Note: For the definition of active members, see figure 18. Data for ployed, they may cycle in and out of covered employment, Indonesia are for the end of 2017; data for India are an average for spending part of their careers in the formal sector and part in 2016-17. In Indonesia, active members refer to JHT members; total members refer to BPJS members, excluding Jakon workers. the informal sector. A substantial literature confirms that low Source: EPFO, BPJS, EPF (Malaysia), and GAI calculations contribution density has played a central role in undermining the adequacy of retirement benefits in Latin America’s person- in Chapter 2, over the past decade the real rate of return on al account systems.14 The limited data available suggest that it investment in the three provident funds covered in the report is also a significant problem in most of Asia’s provident funds. has at best modestly exceeded the growth rate in real GDP per capita, which is used in this report as a proxy for real wage Active members, variously defined as members whose em- growth, and in one of the three has lagged well behind it. Part of ployers make regular contributions on their behalf (India) and the explanation is that provident fund investment performance members who made a contribution within the past year (Indo- has failed to match that of many other large pension funds and nesia and Malaysia), represent just a fraction of total members sovereign wealth funds. But the more important part is that the in the three provident funds covered in the report. In India’s growth rate in real GDP per capita has been very rapid in most EPF, about one-fifth of members were classified as active in of Asia, averaging 3.2 percent per year in Malaysia, 4.0 percent the most recent year for which data are available, in Indone- per year in Indonesia, and 5.5 percent per year in India. To put sia’s JHT about two-fifths were, and in Malaysia’s EPF about these numbers in perspective, even the slowest growing coun- one-half were. (See figure 19.) It is true that the ratio of active try, Malaysia, had a growth rate in real GDP per capita that was members to total members is not a direct measure of contribu- nearly three times the growth rate in U.S. real GDP per capita tion density, and may significantly understate it in some coun- over the same period. tries. In India in particular, the fact that many EPF members have duplicate accounts may make the number of inactive While a higher rate of return raises the replacement rate in a members appear larger than it actually is. Still, these ratios funded pension system, a higher rate of real wage growth low- suggest that contribution density is low enough in all three ers it. Although this dynamic is often not appreciated, it is an countries to have a large negative impact on savings accumu- immutable fact of retirement economics that the faster income is lation. In Malaysia, where more detailed data on contribution growing the larger is the share of income that needs to be saved histories are available, there is no question that it does. A large each year in order to generate the same final salary replace- sample of EPF records reveals that in 2012 just 53 percent of ment rate. Given the same contribution rate, the same contri- individual members aged 16 to 54 made any contributions to bution density, and the same real rate of return, account bal- their accounts, a share that is almost identical to the ratio of ances at retirement relative to preretirement wages, and hence active members to total members.15 replacement rates, would be just half as large at 5 percent real wage growth as they would be at 1 percent real wage growth. The other broader factor that can critically affect the adequacy of retirement benefits is rapid real wage growth. As explained 14. For a comprehensive discussion of the issue, see OECD/IDB/World Bank, Pensions at a Glance: Latin America and the Caribbean (Paris: OECD/IDB/World Bank Group, 2014). 15. The dataset, which includes over 20,000 records, is described in Robert Holzmann et al., “Employees Provident Fund Data for Evidence-Based Social Protection Policies in Malaysia,” SSRC Working Paper Series no. 2016-1 (Kuala Lumpur: Social Security Research Centre, March 2016). Professor Halimah Binti Awang of the University of Malaya, one of the co-authors, kindly shared an extract from the dataset with GAI. CHAPTER 3: BENEFIT DESIGN AND ADEQUACY <<< 35 Assessing how all of this has affected benefit adequacy sufficient, given current Indian and Indonesian life expectancy, is challenging. None of the three provident funds covered to finance an inflation-adjusted annuity equal to 5 percent of in the report publish replacement rate estimates for current preretirement wages. retirees. Nor do they publish the data on account balances by age and salary level that would allow others to calculate It is true that in India and Indonesia provident fund members them. Still, there are sufficient data available to reach some ten- also participate in defined benefit pension programs. Indone- tative conclusions. sia’s program, which is called the Jaminan Pensiun (JP), will eventually deliver replacement rates of between 20 and 35 Malaysia, where the situation is reasonably clear, is a good percent of final salary to full-career workers, which is a mean- place to begin. The EPF recently established a minimum tar- ingful benefit level.16 However, the program does little to help get for retirement savings called the “basic savings” amount. today’s retirees. The JP, which was introduced in 2015, will not For members turning 55, the EPF’s retirement age, this amount begin paying pensions until the 2030s, when the first partici- was RM 240,000 as of 2018, a sum designed to provide a mod- pants who meet the system’s minimum fifteen-year contribu- est retirement income of RM 1,000 per month for twenty years, tion requirement begin reaching retirement age. In the mean- or about two-fifths of the current urban median wage. Yet at the while, retirees will receive lump sum payouts, which currently end of 2018, the average account balance of 54-year-old EPF average just Rp 3 million, or about a month’s wages for an members was only RM 209,862. Moreover, since EPF sav- average earner. India’s program, which is called the Employ- ings is highly skewed by income, most members did not have ees’ Pension Scheme (EPS), is designed to deliver replace- account balances nearly that high. In 2018, just 10 percent of ment rates to full-career workers of around 50 percent of final 54-year-old members accounted for 43 percent of the total sav- salary. But this too lies in the future. Like Indonesia’s JP, the ings of members that age—and just 28 percent accounted for EPS, which was launched in 1995, is still maturing. Replace- 69 percent of it. All told, 72 percent of 54-year-old members had ment rates for most current retirees are quite modest, both an EPF balance of less than RM 250,000, approximately the because no one now retired contributed to the EPS for a full basic savings amount. Twenty-four percent had less than RM career and because the cap on contributable wages has, his- 50,000 in savings, meaning that if they spent their lump sum torically, lagged far behind wage growth. Roughly one-third of payouts at the rate of RM 1,000 per month they would exhaust current retirees receive the subsidized minimum EPS pension them in less than five years. of Rs 1,000 per month, a benefit equal to about one-fifth of the national minimum wage floor. The situation in India and Indonesia is even more worrisome. Although India’s EPF does not publish data on account bal- ances by age, the available data indicate that the average ac- Adequacy Tomorrow count balance of all members was roughly Rs 45,000 at the end of March 2017. Assuming that the average account balance of members nearing retirement age exceeds the average balance The adequacy of retirement benefits can be expected to im- of all members by about the same ratio it does in Malaysia, it prove at least somewhat over the next few decades in all three may have been roughly twice as much, or around Rs 100,000. provident funds covered in the report. Today’s younger workers If so, it only amounted to about one-half of the average annual will probably experience slower wage growth over the course wage for regular urban employees that year. In Indonesia’s JHT, of their careers than today’s retirees did over theirs, making the average lump sum retirement withdrawal was Rp 45 million it easier for their account balances to grow faster than their in 2017, or about one and one-half times the average annual incomes. Contribution density could also increase as younger, wage of formal-sector employees. Without data on preretire- better-educated workers, who may be less likely than older ment balances or retirement withdrawals by salary level, it is workers to cycle in and out of formal employment, climb the not possible to say with any precision what actual replacement age ladder. In Malaysia, where there are data on contribution rates were for typical workers. Nonetheless, these numbers density by age, average density for all EPF members aged 16 clearly indicate that for many if not most workers they were far to 54 was 53 percent in 2012, but for members in their twen- from adequate in both countries. As a point of reference, a final- ties it was 69 percent. Moreover, workers may preserve more balance-to-final-salary ratio of 1.0 at age 55 would barely be of their savings for retirement, both because some provident 16. The JP benefit formula replaces 1 percent of wages per year. But since wage histories are only indexed to inflation in calculating benefit awards, final salary replacement rates will depend critically on the rate of real wage growth over the course of workers’ careers. The high end of the replacement rate range cited here assumes 1 percent real wage growth over the course of a full forty-five year career, while the low end assumes 5 percent real wage growth. 36 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES funds, particularly India’s and Indonesia’s, are attempting to for today’s twenty-year olds would be 83 percent in India, 60 tighten nonretirement withdrawal rules and because today’s percent in Indonesia, and 55 percent in Malaysia.20 (See figure workers do not expect to rely as much in retirement on their 20.) Most policy experts would consider all of these replace- extended families as today’s retirees do. In the 2015 Glob- ment rates at least adequate, and in the case of India and al Aging Institute survey cited earlier in the report, no more Indonesia, where provident fund benefits are supplemented than one in five workers in any of the ten Asian countries sur- by defined benefit pension benefits, quite generous. veyed said that they expect to be financially dependent on their grown children when they are retired, and in some of > > > the more developed countries the share was as low as one F I G U R E 2 0 - Future Gross Replacement Rates: GAI Projection with OECD Assumptions in twenty.17 Meanwhile, in India and Indonesia, the maturation of the EPS and JP programs will also be improving retirement income prospects. India To gain some sense of the range of possible outcomes, GAI EPF & EPS EPS 43% EPF 40% 83% made stylized projections of gross replacement rates using the same basic projection methodology that the OECD uses in its Pensions at a Glance series.18 All of the GAI projections Indonesia refer to workers who enter the workforce in 2020 at the age JHT & JP JP 35% JHT 25% 60% of twenty, earn the economy-wide average wage throughout their careers, and retire at the standard retirement age in each of the three countries. That age is assumed to remain Malaysia unchanged at 55 in India and Malaysia, but to rise from 57 55% EPF to 65 in Indonesia in accordance with current law.19 To fa- cilitate comparability across countries, account balances are 0% 20% 40% 60% 80% 100% converted into inflation-adjusted annuities at retirement. The calculations of account balances assume a 3.0 percent real rate of return while the annuity calculations assume a 2.0 per- Note: Projections refer to workers who enter the workforce at age cent real discount rate, the same assumptions that the OECD twenty in 2020 and retire at the standard retirement age in each country. uses. For simplicity of presentation, replacement rates are cal- Account balances are converted into inflation-adjusted annuities using unisex life tables. All scenarios assume a 3.0% real rate of return and culated using unisex life tables. Because women live longer a 2.0% real discount rate. The GAI Projection with OECD Assumptions than men in all three countries, male replacement rates would assumes 1.25% real wage growth, 100% contribution density, and be somewhat higher and female replacement rates would be 100% savings preservation. somewhat lower. Source: GAI calculations While this scenario provides a useful reference point for what Given these assumptions, future replacement rates will depend is potentially achievable, it would be a mistake to use it as critically on three additional variables: real wage growth, con- a guide to policymaking. The OECD’s assumptions for real tribution density, and nonretirement withdrawals. GAI begins wage growth and contribution density, which it uses in its pro- with a scenario, labeled “GAI Projection with OECD Assump- jections for all countries, both developed and developing, are tions” on the accompanying charts, that adopts the OECD’s unrealistic for Asia’s emerging markets. It is true that real wage assumptions for all three of these variables. This scenario, growth is likely to fall in the future along with development. But which should be considered a best-case scenario, assumes even in Malaysia, where it has averaged nearly 4 percent over that real wage growth will be 1.25 percent per year, that con- the past decade, it is doubtful that it will fall all the way to 1.25 tribution density will be 100 percent, and that 100 percent of percent. In the other countries, where real wage growth has provident fund savings will be preserved for retirement. Under been even higher, it is almost inconceivable. It is also true that these assumptions, GAI projects that gross replacement rates contribution density may rise, but not all the way to 100 per- 17. Jackson and Peter, From Challenge to Opportunity: Wave 2 of the East Asia Retirement Survey. 18. In particular, GAI consulted OECD, Pensions at a Glance: Asia/Pacific 2018 (Paris: OECD, 2018), which includes projections for all three provident funds covered in the report. More recent projections for two of them, India’s EPF and Indonesia’s JHT, are published in OECD, Pensions at a Glance 2019: OECD and G20 Indicators (Paris: OECD, 2019). 19. In India, the fact that the EPF and EPS have different standard retirement ages creates some additional complexity. GAI’s projections for the EPF component of the overall replacement rate assume that retirement savings is withdrawn and converted into an annuity at age 55, while those for the EPS component assume that bene- fits continue to accrue until age 58. 20. Some readers may have noted that, despite using the same key assumptions, GAI’s projected replacement rate for Malaysia is lower than the OECD’s projection published in Pensions at a Glance: Asia/Pacific 2018. GAI discussed this discrepancy with the lead author of the OECD study, Andrew Reilly, who determined that the OECD’s published results significantly overstated Malaysia’s replacement rate. CHAPTER 3: BENEFIT DESIGN AND ADEQUACY <<< 37 cent, a goal which, in practice, no contributory pension system > > > ever attains. As for nonretirement withdrawals, there is simply F I G U R E 2 2 - Gross Replacement Rates for JHT no reason to think that India, Indonesia, and Malaysia can en- Members Entering the Workforce in 2020 tirely eliminate them without major changes in policy. INDONESIA GAI therefore offers a second scenario that is more realis- GAI Projection with tic, though in some respects still quite optimistic. In this sce- 35% 25% 60% OECD Assumptions nario, it is assumed that real wage growth will be 2.5 percent What per year, still much less than the recent average in all three happens if... countries. It is also assumed that contribution density will be 75 percent, which would represent a large improvement in Real Wage Growth is 2.5% 28% 19% 46% all three countries. As for preretirement withdrawals, it is as- sumed that 75 percent of total provident fund savings will be ...and Contribution 21% 14% 35% preserved for retirement, which would represent a significant Density is 75% improvement for Malaysia and an enormous one for India and Indonesia. Together, these changes in assumptions have a ...and Savings 21% 11% 31% Leakage is 25% dramatic impact on projected replacement rates. In fact, they cut them by as much as one-half, to 51 percent in India, 31 0% 20% 40% 60% 80% 100% percent in Indonesia, and 26 percent in Malaysia. (See fig- JP JHT ures 21 to 23.) Although the retirement security prospects for most future retirees in this scenario would still improve relative Note: See figure 20. Source: GAI calculations to the prospects for most current retirees, the improvement would not be nearly as large. This scenario, moreover, is far from a worst-case scenario. > > > > > > F I G U R E 2 1 - Gross Replacement Rates for EPF F I G U R E 2 3 - Gross Replacement Rates for EPF Members Entering the Workforce in 2020 Members Entering the Workforce in 2020 INDIA M A L AY S I A GAI Projection with GAI Projection with 43% 40% 83% 55% OECD Assumptions OECD Assumptions What What happens if... happens if... Real Wage Real Wage 42% 32% 74% 44% Growth is 2.5% Growth is 2.5% ...and Contribution ...and Contribution 32% 25% 57% 34% Density is 75% Density is 75% ...and Savings ...and Savings 32% 19% 51% 26% Leakage is 25% Leakage is 25% 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% EPS EPF Note: See figure 20. Note: See figure 20. Source: GAI calculations Source: GAI calculations 38 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES All of this suggests that the three provident funds may need to enact significant reforms if they are to meet the future retirement needs of existing members, much less serve as a platform for expanding coverage to a broader cross-section of the workforce. The good news is that governments in all three countries are increasingly focused on the challenge. India’s EPF is beginning to tighten up nonretirement withdrawal rules, while Indonesia’s JHT is trying to do the same. Meanwhile, Malaysia’s EPF has made improving benefit adequacy a policy priority, highlighting its urgency in recent annual reports and other publications.21 More will be needed, but these are all steps in the right direction. 21. See, for instance, Ng Say Fen, “Towards Better Retirement Well-Being,” Social Protection Insight, vol. 3 (2018). CHAPTER 3: BENEFIT DESIGN AND ADEQUACY <<< 39 4. >>> Directions for Reform This report has argued that, as Asian societies develop and their populations age, provident funds will need to evolve into something closer to dedicated retirement systems. Although ad- vancing broad national development objectives may remain an important function of some provi- dent funds, especially in the region’s less developed economies, the focus will need to shift toward ensuring retirement security. Similarly, although provident funds may continue to serve multiple savings purposes, the goal of accumulating adequate retirement savings will need to take precedence over nonretirement savings goals. Bringing about this transition will require significant reforms. More fundamentally, it may also require a sea change in philosophy. The intent of this chapter is not to supply a reform blueprint for each of the three provident funds covered in the report, but rather to point to broad directions for reform that can serve as guidelines for policymakers. The three funds, of course, reflect a wide range of provident fund evolution. The stage of economic and financial market development in each country also varies significantly, as does each country’s institutional capacity. As a consequence, not all of GAI’s recommendations apply to all three provident funds. In the discussion that follows, some of the most important distinctions are noted. CHAPTER 4: DIRECTIONS FOR REFORM <<< 41 Investment and Governance Meeting tomorrow’s retirement security needs will require in turn will require developing social impact indicators. adopting policies that maximize risk-adjusted returns and op- These indicators might include increases in wages, re- timize investment outcomes. The place to start is for provi- ductions in inequality, and improvements in a variety of dent funds to develop explicit guidelines for balancing national socioeconomic factors, from health outcomes to educa- economic development and retirement security objectives, tional attainment, that are factored into the UN’s Human put in place procedures for resolving conflicts between them, Development Index. To GAI’s knowledge, none of the and develop metrics for measuring and evaluating outcomes. three provident funds covered in the report have such Provident funds could also improve their performance by con- procedures in place, and none have developed such in- tinuing to diversify their investment portfolios. Some may also dicators, though, to its credit, Malaysia’s EPF is consid- want to consider moving toward market-linked returns and ering how it might do so. As an interim measure while individual customization of the asset portfolio. Retirement se- more robust indicators are being developed, provident curity could be further improved by enhancing the clarity and funds could increase investments in sustainable, green, transparency of financial reporting and better educating mem- and other types of “labeled bonds,” which have the ex- bers about the importance of retirement savings. plicit objective of furthering social and development goals while also delivering reasonable returns. • Develop explicit guidelines for balancing national economic development and retirement security ob- • Continue to diversify investment portfolios. Maxi- jectives. Managing the tension between these objec- mizing risk-adjusted returns requires a well-diversified tives is a fundamental challenge for provident funds, and, increasingly, a global investment portfolio. Yet and effective investment policy and governance must many provident funds remain heavily invested in gov- be built on guidelines that spell out the relative weight ernment debt and prohibit foreign investment. It is un- to be given to each. The balance between the objec- derstandable that governments often prefer retirement tives should be made clear to all stakeholders; there savings to be invested in government debt, since this should be regular evaluation of whether policies and supports direct government spending on infrastruc- outcomes are consistent with the guidelines; and the ture and other social capital projects. Yet as develop- balance between the objectives should be reexamined ment progresses and the returns on government debt and updated over time as the country’s economy devel- decline, diversification into domestic corporate bonds, ops and the provident fund grows. equities, and real estate can mitigate the impact on provident fund returns. It is also understandable that This clarity is essential because investment policies that governments often prefer retirement savings to be in- are intended to advance national development objec- vested in domestic capital markets, since this supports tives, even when they are well designed and effective, financial development, business expansion, and job may not be the policies most likely to maximize returns creation. Yet as provident funds continue to grow, reduc- for participants. In the early stages of provident fund ing domestic investment bias can improve outcomes for evolution, when development is usually the primary fo- provident fund members while still leaving room for de- cus, a failure to achieve market returns on every invest- velopment priorities.22 Global diversification, moreover, ment may not be a significant problem. But over time, will become even more important as economic growth as ensuring retirement security becomes more impor- slows and returns to capital decline due to Asia’s aging tant, provident funds would benefit from putting in place populations. If provident funds fail to diversify their in- explicit guidelines for balancing their sometimes com- vestment portfolios over time, they risk becoming “cap- peting objectives, as well as governance procedures for tive investors” that finance government activities by im- resolving conflicts between them. For these procedures posing below-market returns on members. to be effective, moreover, provident funds need to be able to compare the social returns to the investments Malaysia understands this, and has been steadily di- they make with the financial returns to members, which versifying the EPF’s investment portfolio over the past 22. See, for instance, Juan Pablo Afanador, Richard Davis, and Alvaro Pedraza, International Diversification of Pension Funds: An Index-Based Rating for Countries (forthcoming World Bank Group working paper). 42 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES ten to fifteen years, at first domestically and then inter- als are young, but progressively shifted to fixed-income nationally. India and Indonesia will want to build on their securities as they grow older. This lifecycle approach recent steps toward diversification by increasing EPF allows the level of financial risk related to investment to and JHT investments in domestic capital markets and, start out high, at a time in life when individuals have few eventually, international capital markets. As they do, financial assets and a long career before them, but to they will need to further develop in-house investment be reduced as their financial wealth increases and the expertise, while ensuring that adequate safeguards human capital risk related to their future employment against self-dealing, nepotism, favoritism, and manipu- income decreases. lation of markets are in place. As provident funds are currently structured, all mem- • Consider moving toward market-linked returns. ber contributions are pooled and collectively invested in The lack of mark-to-market accounting, together with the same identical investment portfolio. Some provident the practice of crediting member accounts with admin- funds may decide, as a matter of policy, that the current istratively determined returns, may distort investment structure, which supports return smoothing and return decisions and lead to suboptimal investment outcomes. guarantees, should be retained. Others may evolve Administratively determined returns can create an unre- toward a more market-based approach to accounting alistic expectation that provident funds will always deliv- for and crediting investment returns. For those that do, er the returns to which members have become accus- customizing the asset portfolio along lifecycle lines may tomed. When gains and losses are deferred until assets significantly improve investment outcomes for mem- are sold, provident funds may be tempted to “strategi- bers. Evolving in this direction would also allow the cally” sell assets in order to boost near-term returns and adoption of liability-driven investment strategies, where meet the public’s expectations, even if this could lower asset allocation, including the duration of fixed-income long-term returns. Unrealized losses can also build up securities, is aligned with the objective of providing in- and ultimately lead to larger losses for members than if come in retirement. asset values had moved with the market. At the same time, minimum rate of return guarantees, designed to One workable approach would be for provident funds protect workers against downside risk, may compel to shift to a multifund model like those used in several asset managers to shift investment portfolios toward of Latin America’s personal account systems, includ- lower-risk and lower-return assets. While these guar- ing Chile’s, Colombia’s, and Mexico’s. Instead of one antees help bolster members’ confidence in the near investment fund, provident funds would administer sev- term, they can lead to lower retirement benefits in the eral funds with different age-related risk profiles. It is long term. true that the multifund model would require more so- phisticated administrative systems than most provident For all of these reasons, market-linked returns are worth funds now possess. However, if migration from one considering. Some provident funds, however, may not fund to another is made automatic based on age, with want to sacrifice the ability to smooth returns that they no individual discretion, the additional complexity can now enjoy. One way to preserve some of the benefits be minimized. The approach might also help preserve of smoothing would be to pay an administratively deter- some of the sense of common national interest that un- mined return on account balances up to some thresh- derlies the provident fund model. Although the returns old. Once that threshold is reached, incremental contri- that members earn would vary by age, they would not butions would earn market returns. Interestingly, India’s vary within age cohorts. EPF is considering a similar arrangement in which members would earn a market return on their pro rata To be clear, this is not a recommendation to turn provi- share of provident fund equity holdings, but would con- dent funds into self-directed retirement accounts. As tinue to receive an administratively determined return explained in Chapter 2, Malaysia currently allow mem- on the rest of their account balance. bers whose account balances exceed certain thresh- olds to withdraw the excess and invest it in approved • Consider moving toward individual customization investment funds of their choosing. But few provident of the asset portfolio. In defined contribution systems, fund members have the expertise to allocate their sav- asset allocation is ideally tailored to each member’s age, ings between the large number of funds on offer, and with the portfolio tilted toward equities when individu- this may reduce returns or increase risk unnecessarily. CHAPTER 4: DIRECTIONS FOR REFORM <<< 43 Policymakers should bear in mind that protecting mem- Benefit Design and Adequacy bers against bad choices is likely to result in better out- comes than facilitating individual choice. To the extent that provident funds move toward market-linked returns Improving the adequacy of retirement benefits will require and individual customization of the asset portfolio, they significant changes in benefit design. Most provident funds can and should do so within the overall collective in- will need to increase the amount of savings dedicated to re- vestment framework that is one of the provident fund tirement, raise standard retirement ages, and institute provi- model’s greatest strengths. sions for lifetime income. Most will also need to explore ways to increase provident fund coverage and contribution density, • Enhance the clarity and transparency of financial while at the same time strengthening the safety net for those reporting. Proper accounting is an important aspect of who arrive in old age with inadequate savings. good governance, and all three provident funds have potential to improve in this area. While the financial • Increase savings dedicated to retirement. Provident statements for Malaysia’s EPF meet a high standard, funds often fail to ensure that even full-career workers the use of a reserve account to smooth returns could be set aside sufficient savings for retirement. Replacement disclosed more prominently. The financial statements rates in defined contribution systems depend on many for India’s EPF and Indonesia’s JHT fail to meet reason- factors, from real rates of return and real wage growth able standards in important respects. In recent state- to retirement ages and life expectancy at retirement. ments for Indonesia’s JHT, key information on invest- However, a reasonable rule of thumb is that workers ment strategy and performance is lacking, and there need to save 10 to 15 percent of wages for retirement is no disclosure of unrealized gains and losses. India’s each year in order to replace one-third to one-half of EPF has struggled with allocating interest and equity their preretirement income. returns in a timely and consistent manner, making it dif- ficult to gain a clear, detailed picture of its financial posi- Among the three provident funds covered in the report, tion. As these two provident funds continue to develop, only Malaysia’s EPF requires that this large a share ensuring clarity and transparency in financial reporting of wages be saved for retirement. Of an overall con- will need to be a high priority. tribution rate of 24 percent of wages, 16.8 percent is dedicated to Account I, which must be preserved for • Better educate members about the importance of retirement. In Indonesia’s JHT, the overall contribu- retirement savings. Educating members about how tion rate is just 5.7 percent of wages, and only a small much they need to save in order to enjoy a secure retire- fraction of this ends up being saved for retirement. At ment is another important aspect of good governance. 15.7 percent of wages, the overall contribution rate India’s EPF and Indonesia’s JHT give members access in India’s EPF is seemingly more adequate. But just to information about account transactions and balances, as in Indonesia’s JHT, nonretirement withdrawals are but otherwise offer little guidance in this regard. Malay- largely unrestricted. sia’s EPF does considerably more, including publicizing minimum retirement savings targets and offering retire- If retirement security is to be improved, this will need ment planning advice to members approaching retire- to change. Indonesia should raise the JHT contribu- ment age. While this is helpful, it is still not enough. All tion rate, perhaps doubling it in stages. It should also provident funds should provide members with periodic consider following Malaysia’s example by dividing the customized estimates of the monthly benefits they can JHT into two accounts, one earmarked for retirement expect to receive if they remain active contributors until savings and the other for nonretirement savings. India retirement age. Ideally, these statements would also in- should consider doing the same with the EPF. The share clude illustrative scenarios that show how failing to con- of total provident fund savings preserved for retirement tribute regularly or making nonretirement withdrawals should be determined by informed policy decisions. It can affect benefit levels. should not be the residual left over after withdrawals for housing, education, unemployment, and other nonre- tirement needs. Although the need is less urgent, even 44 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES Malaysia might consider increasing the share of EPF • Raise standard retirement ages. In rapidly developing savings that it allocates to retirement. economies with growing workforces, it may make sense to encourage older workers to retire early in order to It could be argued that preserving provident fund make room for younger ones. With educational attain- savings for retirement is less important in India and ment rising rapidly, the young have the skills to fill high Indonesia than it is in Malaysia, since workers in these value-added jobs in the growth sectors of the economy, countries also participate in government defined ben- while the old do not. As today’s younger adults climb efit pension programs. As explained in Chapter 3, the age ladder, however, later retirement will not only however, the EPS and JP are not yet mature. EPS become feasible, but also necessary. It will become fea- benefits are still modest, and the JP will not even sible because the skills gap between young and old will begin paying pensions util the 2030s. It is also important gradually close, and it will become necessary because to note that the JP is seriously underfunded, and will life expectancy is rising, leaving retirees at a growing not be able to pay promised benefits to today’s younger risk of outliving their savings. (See Figure 24.) As ex- workers unless the current 3.0 percent contribution rate plained in Chapter 3, postponing retirement can greatly is raised.23 increase retirement security. In fact, all other things being equal, postponing retirement from age 55 to 65 nearly doubles income replacement rates. > > > F I G U R E 2 4 - Unisex Life Expectancy at Age 60, 1990-2050 30 24 25 21 21 20 20 18 18 17 15 15 15 10 5 0 India Indonesia Malaysia 1985-90 2015-20 2045-50 Source: UN Population Division, World Population Prospects: The 2019 Revision (New York: UN Population Division, 2019) 23. See Agus Susanto, “Indonesia’s Pension in 2018 under BPJS Ketenagakerjaan,” Nomura Journal of Asian Capital Markets, 3:2 (Spring 2019). CHAPTER 4: DIRECTIONS FOR REFORM <<< 45 Two of the three provident funds covered in the report ing. It is also true that, because annuities are subject to are now moving in the right direction. The laggard is moral hazard and asymmetric information, it can be dif- India’s EPF, which allows retirement savings to be with- ficult to price them fairly. But this is only true if annuities drawn in full at age 55 and has announced no plans to are optional. If they are mandatory, longevity risk can raise that age. Although India still has a relatively youth- be averaged across the entire population, meaning that ful population, it is not too soon to begin preparing for efficient annuities ought to be easy to price. Finally, it is the future. Indonesia, whose demographics are similar true that lump sum payouts are deeply ingrained in the to India’s, has already begun raising the JHT retire- culture of most Asian countries. But the public’s prefer- ment age in stages to 65. Meanwhile, Malaysia recently ence for lump sums may not be as strong as policy- enacted a reform which requires that the incremental makers assume. In the Global Aging Institute’s survey EPF contributions of workers who remain employed of retirement attitudes and expectations, the share of past age 55 be preserved until age 60. While this is a respondents saying that they would prefer to receive good first step, Malaysia should consider locking up all retirement benefits “all in regular monthly payments” Account I savings until age 60, then further raising the exceeded the share saying that they would prefer to retirement age in stages to 65. receive them “all in a single lump sum” by sizeable mar- gins in all ten countries surveyed.24 • Institute provisions for lifetime income. All three provident funds covered in the report allow retirement • Explore ways to increase coverage and contribu- savings to be withdrawn entirely as a lump sum. Lump tion density. Even as governments better align provi- sum payouts may have made sense in yesterday’s dent funds’ contribution and benefit rules with future more traditional Asian societies in which virtually all of retirement needs, they should also seek to extend the elderly were supported by their grown children and their reach to a broader cross-section of the workforce. few lived into their eighties or nineties. They no longer Low coverage and low contribution density are largely make sense in today’s rapidly modernizing, industrial- a function of labor-market informality, and as long as izing, and urbanizing societies in which life expectancy informality remains high progress will be difficult. Still, keeps rising year after year. Although the complete an- there are steps that governments could take today to nuitization of account balances may not be advisable, make provident funds more inclusive. especially in India and Indonesia, where workers also participate in defined benefit pension programs, all In all three provident funds covered in the report, countries would ideally follow the lead of Singapore’s self-employed workers are exempted from manda- CPF and require at least partial annuitization. If this is tory coverage, though they may participate voluntarily. not possible, a second-best option would be to require Although many of the self-employed work at family- phased withdrawals. Provident funds might also con- owned micro-businesses or in other low-skill jobs, some sider combining phased withdrawals and/or partial lump are well-educated professionals for whom it should be sum payouts with a deferred annuity starting at (say) relatively simple to mandate participation. Even when age 80. This would have the advantage of allowing mandating participation is not practical, governments some flexibility in the use of retirement savings while could employ financial incentives, such as matching still providing back-end protection against longevity risk. contributions, to encourage low-skilled, informal-sector workers, whether employed or self-employed, to partici- While there are potential obstacles to annuitizing ac- pate on a voluntary basis. Until recently, the obstacles count balances, they can be overstated. It is true that to extending coverage beyond the formal sector were private annuity markets are underdeveloped in many almost insuperable. However, advances in digital IT, fi- Asian countries. But if private markets are underdevel- nancial inclusion, and national ID systems are opening oped, government can provide annuities, as it does in up new ways to reach informal-sector workers.25 The Singapore. Since doing so in effect shifts longevity risk Indonesian government, to its credit, has set an explicit to government, it has potential budget implications. But long-term goal of extending JHT coverage to the entire from a policy perspective, the cost is well worth assum- labor force. 24. Jackson and Peter, From Challenge to Opportunity: Wave 2 of the East Asia Retirement Survey. 25. See, among others, Yu-Wei Hu and Fiona Stewart, “Pension Coverage and Informal Sector Workers: International Experiences,” OECD Working Papers on Insurance and Private Pensions no. 31 (Paris: OECD, 2009); Richard Hinz et al. eds., Matching Contributions for Pensions: A Review of International Experience (Washington, DC: World Bank Group, 2013); World Bank, Live Long and Prosper: Aging in East Asia and Pacific (Washington, DC: World Bank Group, 2016); Richard Jackson, Voluntary Pensions in Emerging Markets: New Strategies for Meeting the Retirement Security Challenge (Alexandria, VA: Global Aging Institute, 2017); and Parul Seth Khanna, William Price, and Gautam Bhardwaj, eds., Saving the Next Billion from Old Age Poverty: Global Lessons for Local Action (Singapore: Pinbox Solutions, 2018). 46 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES Alternatively, governments could set up separate con- old-age poverty protection in the form of means-tested, tributory retirement savings systems with more flexible noncontributory social pensions. In India, there is the contribution and withdrawal rules that are tailored to the Indira Gandhi National Old Age Pension Scheme; in needs of informal-sector workers. A growing number of Indonesia, the Asistensi Sosial Lanjut Usia Terlantar; developing countries are experimenting with such sys- and in Malaysia, the Bantuang Orang Tua. While these tems, including India (the Atal Pension Yojana) and Ma- programs help some indigent elders, only a fraction of laysia (the i-Saraan and i-Suri programs, which operate the population in need is enrolled and benefit levels under the EPF’s umbrella). Although their track record are quite modest. Strengthening them should be a high is mixed, some have been remarkably successful at in- government priority. creasing participation. It has been less than a decade ***** since China launched two new informal-sector volun- Asia’s provident funds stand at a crossroads. As the region’s tary retirement savings systems, and already several societies develop and its populations age, retirement security hundred million rural and migrant workers have joined. can no longer take a backseat to other policy priorities. With This is an astonishing accomplishment that the World each passing year, the need to build more adequate and more Bank calls “unprecedented in global experience.”26 inclusive retirement systems is becoming more urgent. As policymakers look to the future, they will find that the provident • Strengthen the old-age safety net. However suc- fund model continues to have many important advantages cessful reform efforts are, some significant share of the over alternative retirement system models. But they will also workforce in most Asian countries will be reaching old find that the model needs to evolve if it is to meet tomorrow’s age without adequate retirement savings for decades to challenges effectively. The good news is that policymakers come. This reality underscores the importance of hav- throughout the region understand this, and are beginning to ing a robust old-age safety net in place. India, Indone- make the necessary adjustments. sia, and Malaysia all make at least some provision for 26. World Bank, Live Long and Prosper: Aging in East Asia and Pacific, 151. CHAPTER 4: DIRECTIONS FOR REFORM <<< 47 >>> Technical Appendix The technical appendix is divided into four sections. The first section supplies references to the basic data and informational sources used by GAI; the second offers additional detail on GAI’s financial analysis of the three provident funds covered in the report; the third offers additional detail on the rate of return comparisons with other large pension funds and sovereign wealth funds; and the fourth offers additional detail on GAI’s replacement rate projections. Basic Data Sources Demographic, economic, and financial market data cited in the report or used in the analysis come from standard sources. Population data, both historical and projected, come from the UN Population Division’s World Population Prospects: The 2019 Revision, and are available at https://population.un.org/wpp. Labor-force data come from the ILO’s ILOSTAT database, and are available at https://ilostat.ilo.org. Most economic data, including data on GDP and infla- tion, come from the World Bank’s World Development Indicators database, and are available at https://data.worldbank.org. Most financial market data, including ten-year bond yields, come from Investing.com, and are available at https://www.investing.com. Data on exchange rates come from XE.com, and are available at https://www.xe.com. Data on wages come from na- tional surveys. For India, they come from the Periodic Labour Force Survey (PLFS), available at http://www.mospi.gov.in; for Indonesia, they come from Labor Situation in Indonesia, available at https://bps.go.id/; and for Malaysia, they come from the Salaries and Wages Survey Report, available at https://www.dosm.gov. Unless otherwise indicated, all data on provident fund finances and benefits come from the provident funds themselves, and are available in the annual reports, financial statements, and related documents posted on their websites. Most information on investment and gover- nance practices, as well as system parameters such as coverage rates, contribution rates, and retirement ages, also comes from the provident funds themselves, and is available on their web- sites at, for India’s EPF, https://epfindia.gov.in; for Indonesia’s JHT, https://www.bpjsketenagak- erjaan.go.id; for Malaysia’s EPF, https://www.kwsp.gov.my; and for Singapore’s CPF, https:// www.cpf.gov.sg. TECHNICAL APPENDIX <<< 49 Financial Analysis Each of the provident funds covered in the report is set up entities that administer the provident funds also administer the as part of an entity which could be viewed as a kind of state- other programs, and the financial statements for the entities run corporation. The financial statements that were reviewed include the other programs. GAI’s financial analysis, however, for the report are financial statements for that corporate is limited to the provident funds, whose transactions are entity. The provident funds themselves are just part, although reported separately in the entities’ annual reports. typically a very large part, of the corporate entity. Moreover, provident fund accounts are just part, although once again Just as any corporation does, the entities that administer typically a very large part, of the provident funds, which may provident funds have operating expenses, including for office include reserves or other moneys that are not allocated to space, IT systems, and the compensation of employees. The member accounts. way that these expenses are defrayed varies from country to country. In India, operating expenses are covered by a separate The corporate entity has its own balance sheet with assets, charge to employers. In Malaysia, they are subtracted from liabilities, equity, and a profit and loss statement. The assets investment income before it is allocated to member accounts. of the provident fund make up a large portion of the entity’s Although it is not documented in the financial statements, this assets, but also represent an equal obligation to make benefit appears to be the practice in Indonesia as well. payments to provident fund members. The investment income earned on provident fund account balances does not constitute The “Reported Return” discussed in the report is the rate of profit for the entity because it is allocated to member accounts return which is reported as credited to member accounts in the where, once again, it represents both an asset to the entity provident funds’ annual reports. The “GAI Calculated Return” and a liability to make benefit payments to members. GAI’s is derived from information in the provident funds’ financial financial analysis refers to provident fund accounts, rather statements on beginning-of-year balances, investment return than to the corporate entity as a whole. amounts, and other cash flow. To the extent the information is provided, assets are marked to market value in deriving the India’s and Indonesia’s provident funds are part of broader GAI Calculated Return. social security systems that include defined benefit pension and insurance programs. In these countries, the corporate The following formula was used: GAI Calculated Return % = Investment Return $ / (BOYB + 0.5 x CF) WHERE: Investment Return $ = Investment return amount, with assets marked to market if possible BOYB = Beginning-of-year balance CF = Cash flow other than investment returns, including contributions, withdrawals, and expenses 50 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES The remainder of this section includes some specific notes Notes: (1) The BPJS did not report a return percentage for related to GAI’s financial analysis of each provident fund. 2017; GAI therefore set the Reported Return for that year equal to the GAI Calculated Return. (2) Although the JHT India appears to have an investment reserve account, the BPJS The Employees’ Provident Fund (EPF) is administered by the does not discuss the account or report the amounts in it in its Employees’ Provident Fund Organization (EPFO). In addition annual reports and financial statements. to the EPF, the EPFO also administers the Employees’ Pension Scheme (EPS), a defined benefit pension program, Malaysia and the Employees’ Deposit Linked Insurance Scheme The Employees Provident Fund (EPF) is administered by (EDLI), a life insurance program. GAI’s financial analysis is the EPF Board. The EPF consists of two accounts: Account limited to the EPF. The financial statements and investment I, which is earmarked for retirement savings, and Account II, return information reviewed by GAI cover the period April 1, which is earmarked for nonretirement savings. GAI’s financial 2008 to March 31, 2018. EPFO annual reports, as well as analysis includes both accounts. The financial statements separate “consolidated annual accounts,” are available at and investment return information reviewed by GAI cover the https://search.epfindia.gov.in/OperationStatistics/operational_ period January 1, 2009 to December 31, 2018. EPF annual stats_en.php. reports are available at https://www.kwsp.gov.my/about-epf/ news-highlights/publications. Notes: (1) GAI’s financial analysis excludes EPF Private Trusts. (2) Government debt in GAI’s asset tabulations includes Notes: (1) The EPF financial statements include an investment “Government Securities,” “State Government Securities,” reserve account, whose size, use, and impact are discussed “State Development Loans,” and “Public Sector Financial in some detail in Chapter 2 of the report. Institutions.” (3) Although the EPF financial statements do not include an investment reserve account, there is an aggregate “interest account” in which earnings are held pending crediting to member accounts. (4) The EPFO began investing in equity ETFs in 2016, but earnings from the investments were only credited to member accounts starting in 2018. (5) Final financial statements were not yet available for the EPF’s 2018 fiscal year at the time GAI completed its analysis; account information for 2018 is based on minutes of the midyear meetings (137 and 138) of the Financial Investment Committee, also available at https://search.epfindia.gov.in/ OperationStatistics/operational_stats_en.php. Indonesia The Jaminan Hari Tua (JHT) is administered by BPJS Ketenagakerjaan. In addition to the JHT, the BPJS also administers three other programs: the Jaminan Kecelakaan Kerja (JKK), a workers’ compensation program; the Jaminan Kematian (JK), a life insurance program; and the Jaminan Pensiun (JP), a defined benefit pension program. GAI’s financial analysis is limited to the JHT. The financial statements and investment return information reviewed by GAI cover the period January 1, 2009 to December 31, 2018. BPJS annual reports are available at https://www.bpjsketenagakerjaan. go.id/laporan-tahunan.html. TECHNICAL APPENDIX <<< 51 Pension and Sovereign Wealth Funds Two of the six large pension funds and sovereign wealth funds included in the rate of return comparisons in Chapter 2 have noncalendar fiscal years. Since the comparison period of January 1, 2009 to December 31, 2018 begins in the midst of the Great Financial Crisis, this could skew the results. One of the funds in question, Japan’s GPIF, did not have a significant exposure to equities in 2008-2009, so the issue is unimportant. However, the other fund, CalPERS, did. This issue is addressed by estimating CalPERS returns for January 1, 2009 to December 31, 2018 as an evenly weighted average of ten-year returns from July 1, 2008 to June 30, 2018 and July 1, 2009 to June 30, 2019. That this method roughly captures the impact of the financial crisis on global equity returns is borne out by the following calculations using MSCI All Country World Index, All Cap (MSCI ASCW AC) returns over the two periods: GLOBAL EQUITY (MSCI ASCW AC INDEX) ½ X (7/1/2008 – 6/30/2018 = 4.00%) ½ X (7/1/2009 – 6/30/2019 = 8.13%) ESTIMATED 1/1/2009 – 12/31/2018 = 6.07% ACTUAL 1/1/2009 – 12/31/2018 = 7.56% The annual reports and financial statements of the pension funds and sovereign wealth funds included in the comparison are available online: • CalPERS: https://www.calpers.ca.gov/page/investments/ • South Korea’s NPS: https://fund.nps.or.kr/jsppage/fund/ about-investment-office/ investment-financial-reports prs_e/prs_e_04.jsp • Chile’s AFPs: https://www.spensiones.cl/apps/ • Norway’s GPF Global: https://www.nbim.no/en/ centroEstadisticas/ paginaCuadrosCCEE.php?menu= publications • sest&menuN1=sistpens&menuN2=fondospen • Ontario Teachers’ Pension Plan: https://www.otpp. • Japan’s GPIF: https://www.gpif.go.jp/en/performance com/corporate/ontario-teachers-reporting 52 >>> ASIAN PROVIDENT FUNDS | MEETING TOMORROW’S CHALLENGES Replacement Rate Projections GAI’s projections of gross replacement rates use the same rates derived from country-specific, unisex life tables for the basic projection methodology that the OECD uses in its year in which workers retire. The lifetables used are from the Pensions at a Glance series. UN’s World Population Prospects: The 2019 Revision. The replacement rates presented in the report are thus unisex The GAI projections refer to workers who enter the workforce replacement rates. Because women live longer than men in in 2020 at the age of twenty and earn the economy-wide all three countries, male replacement rates would be average wage throughout their careers. Since there is no age- somewhat higher and female replacement rates would be wage profile, final salary replacement rates and replacement somewhat lower. rates expressed as a share of (wage-indexed) lifetime earnings are identical. Workers are assumed to retire at the standard Within this basic framework, GAI ran two projection scenarios retirement age in each of the three countries. That age remains that illustrate the impact that differences in real wage growth, unchanged at 55 in India and Malaysia, but rises from 57 to 65 contribution density, and nonretirement withdrawals can in Indonesia in accordance with current law. In India, the fact have on benefit adequacy. The GAI Projection with OECD that the EPF and EPS have different standard retirement ages Assumptions scenario adopts the OECD’s assumptions for creates some additional complexity. GAI’s projections for the all three of these variables. Specifically, it assumes that real EPF component of the overall replacement rate assume that wage growth will be 1.25 percent per year, that contribution retirement savings is withdrawn and converted into an annuity density will be 100 percent, and that 100 percent of provident at age 55, while those for the EPS component assume that fund savings will be preserved for retirement. benefits continue to accrue until age 58. GAI’s projected replacement rates for India and Indonesia in To facilitate comparability across countries, account balances this scenario are similar to the OECD’s projections. However, are converted into inflation-adjusted annuities at retirement. its projected replacement rate for Malaysia is lower than the The calculations of account balances assume a 3.0 percent OECD’s projection published in Pensions at a Glance: Asia/ real rate of return while the annuity calculations assume a Pacific 2018. GAI discussed this discrepancy with the lead 2.0 percent real discount rate, the same assumptions that author of the OECD study, Andrew Reilly, who determined the OECD uses in Pensions at a Glance: Asia/Pacific 2018 that the OECD’s published results significantly overstated (Paris: OECD, 2018). No administrative fees are charged to Malaysia’s replacement rate. the accounts during the accumulation phase, which could be interpreted as meaning either that there are no fees or that GAI’s second projection scenario assumes that real wage the assumed 3.0 percent real rate of return is net of fees. growth will be 2.5 percent per year, that contribution density Nor is any load charged when converting account balances will be 75 percent, and that 75 percent of provident fund into annuities. While this assumption would be unrealistic savings will be preserved for retirement. GAI believes for privately provided annuities, it seems reasonable for that these assumptions yield more realistic estimates mandatory government provided annuities, which would of replacement rates than those in the GAI Projection with entail no marketing expenses and would not need to provide OECD Assumptions. for profits. Annuity factors are calculated based on mortality TECHNICAL APPENDIX <<< 53 >>> About the Authors Richard Jackson Richard Jackson is president of the Global Aging Institute (GAI), which he founded in 2014. Prior to founding GAI, Richard directed a research program on global aging at the Center for Strategic and International Studies. He is an internationally recognized authority on global aging and the author or co-author of numerous policy studies on the challenges it poses, including Meeting India’s Retirement Challenge (2018); Voluntary Pensions in Emerging Markets: New Strategies for Meeting the Retirement Security Challenge (2017); From Challenge to Opportunity: Wave 2 of the East Asia Retirement Survey (2015); The Global Aging Preparedness Index, Second Edition (2013); and The Graying of the Great Powers: Demography and Geopolitics in the 21st Century (2008). Richard regularly speaks on aging-related issues and is widely quoted in the media. He holds a B.A. from SUNY at Albany and a Ph.D. from Yale University. Evan Inglis Evan Inglis is an actuary and investment expert with a global perspective who has worked with some of the largest pension funds in the world. Evan is also a respected thought leader in the actuarial profession, as well as a frequent speaker and author on retirement policy issues. He has developed innovative concepts, such as demographic-based investing for pension plans and The Feel Free Retirement Spending Strategy, and has extensive experience working on pension and investment issues in Europe, Latin America, Africa, and Asia. Evan is a former member of the Society of Actuaries Board of Directors and currently serves on the Board of Actuaries for the Civil Service Retirement System. He is also a Fellow of the Society of Actuaries and a CFA Charterholder. ABOUT THE AUTHORS <<< 55 >>> About the Global Aging Institute The Global Aging Institute (GAI) is a nonprofit research and educational organization dedicated to improving our understanding of global aging, to informing policymakers and the public about the challenges it poses, and to encouraging timely and constructive policy responses. GAI’s agenda is broad, encompassing everything from retirement security to national security, and its horizons are global, extending to aging societies worldwide. GAI was founded in 2014 and is headquartered in Alexandria, Virginia. Although GAI is relatively new, its mission is not. Before launching the institute, Richard Jackson, GAI’s president, directed a research program on global aging at the Center for Strategic and International Studies (CSIS) which, over a span of nearly fifteen years, produced a large body of cutting-edge research and analysis that played a leading role in shaping the debate over what promises to be one of the defining challenges of the twenty-first century. GAI’s Board of Directors is chaired by Thomas S. Terry, who is CEO of the Terry Group, past president of the International Actuarial Association, and past president of the American Academy of Actuaries. To learn more about the Global Aging Institute, please visit its website at www.GlobalAgingInstitute.org.