China - Pension system reform

Report No. 15121-CHA China Pension System Reform August 22, 1996 Resident Mission in China China and Mongolia Department East Asia and Pacific Region ;' - * / , s , ffi * 4 _ 4 - % A C ''- ~~~~~~~4 ~ ~ ~a 4 ... ... ~~- ~ ~~ %*4~~~ --'~~~~,* '( ' * "~~~~~~~7~~ " ~~~4 + -~~~~''~~~ - ~ ~ ~ ~ ~ '4'4~~~~',~~*-4 vu ~ ~ ~ ~ ~ ~ A'I'~.& "t 4'' '''- " ~~  ABBREVIATIONS AND ACRONYMS ACFTU - All-China Federation of Trade Unions CD - Certificate of deposit CFA - Certified financial analysts COEs - Collectively-owned enterprises CPI - Consumer Price Index CR - Contribution rate FIEs - Foreign invested enterprises HiMCs - Housing management companies IPD - Implicit pension debt MOCA - Ministry of Civil Affairs MOF - Ministry of Finance MOL - Ministry of Labor MOP - Ministry of Personnel NBFIs - Nonbank financial institutions NPC - National People's Congress OECD - Organization of Economic Cooperation and Development PAYG - Pay-as-you-go PBC - People's Bank of China PFMCs - Pension fund management companies PICC - People's Insurance Company of China POEs - Privately-owned enterprises ROR - Rate of return SCR - Sustainable Contribution Rate SDR - System Dependency Ratio SEdC - State Education Commission SIA - Social Insurance Agencies SICs - Social Insurance Companies SOEs - State-owned enterprises SPC - State Planning Commission SRC - System Refonn Commission SSB - State Statistical Bureau TVEs - Township and village enterprises TW - Total wages  CONTENTS Acknowledgments ............................................................v Abstract ............................................................ vii Executive Summary ............................................................ ix 1. LONG-TERM AND SHORT-TERM CHALLENGES IN THE PENSION SYSTEM .............................................................1 The Challenge ............................................................1I Evolution Of The Old-Age Pension System .............................................................3 Goals Of The Present Reform .............................................................7 Key Problems Of The Current Pension System . ............................................................9 Adverse Implications Of The Present Pension System ............................................................ 14 2. KEY ISSUES AND OPTIONS FOR PENSION SYSTEM REFORM .............................. 19 Unification ............................................................ 19 Reducing Pension Costs ............................................................ 20 Extending The Coverage Of The Pe n sion System ............................................................ 23 Funded Individual Accounts ............................................................ 24 Reforming The Financial Sector ............................................................ 24 Estimating Transition Cost ............................................................ 26 Mechanisms For Financing The Transition ............................................................ 28 China's Advantages In Bearing The Transition Costs ............................................................ 30 3. QUANTITATIVE ANALYSIS FOR A PREFERRED PENSION SYSTEM ................... 33 Introduction ............................................................ 33 Demographic And Macroeconomic Developments ............................................................ 33 Simulation Results With Limited Reforms ............................................................ 36 A Conceptual Framework For The Proposed Pensioni System ................................................. 39 Three Pillars Of The Proposed Pension System ...................................... ...................... 43 Transition Plan ............................................................ 45 The Effect Of The New System On Individual Pensionier ....................................................... 45 Financial Viability Of Pillar I ............................................................ 46 Options For Funding The Transition ............................................................ 54 Conclusion ............................................................ 55 4. MANAGEMENT AND FUNDING OF THE THREE PILLARS ...................................... 57 Legal Framework And Governance Structure ............................................................ 57 Administration Of The First Pillar And The Transition Plan ................................................... 59 Financial Management And Regulation Of The Second Pillar ............................... ................. 60 Development And Regulations Of The Third Pillar ............................................................ 69 Concluding Remarks ............................................................ 73  - ii - Appendix Tables ................................................................. 75 Annex 1.1: The State Council Circular On The Deepening Of Reform Of The Pension Insurance System ................................................................. 93 Annex 2.1: Financing Of The Transition Gap In Chile .............................................................. 103 Annex 3.1: Role Of The State In The Chilean Pension System ................................................. 107 Annex 4.1: Toward An Integrated System Of Social Insurance In China: A Policy Note ...... 111 Annex 4.2: Private Pension Regulation In Indonesia ................................................................. 119 Technical Annex 1: A Summary Of The Projections Of Basic Pension Insurance For Urban Staff And Workers ................................................................. 121 Technical Annex 2: Estimating Implicit Pension Debt For Two Cities: Shenyang And Shanghai ................................................................. 125 Technical Annex 3: Simulating Pension Profiles For Ten Hypothetical Urban Workers .......... 131 Technical Annex 4: China: A Macromodel For Assessing The Financial Outlook Of The Pension System Witlh Structural Changes In The Economy ...................... 143 Glossary ................................................................. 161 References ................................................................. 163 TABLES IN TEXT Table 1.1: Current and Projected Percentage of Elderly .....................................................2 Table 1.2: Growing Imbalance Between Workers and Pensioners in State-Owned and Collective Enterprises ..................................................................2 Table 1.3: China: Features of Pension Schemes of Selected Provinces, Municipalities and Sectors, 1994 ................................................................. 10 Table 1.4: Reported Compliance Rates in Selected Municipalities .................................. 11 Table 1.5: Levels of Pension Pooling in Selected Provinces ............................................ 13 Table 3.1: Projected Population of Working Age Group and the Elderly ........................ 34 Table 3.2: Labor Force Projections ................................................................. 36 Table 3.3: Simulations with Nonstructural Reforms ........................................................ 38 Table 3.4: Pension and Wages for 10 Hypothetical Urban Workers ................................ 47 Table 3.5: Pillar I under Multipillar System-Simulations Under Various Assumptions ................................................................. 49 Table 3.6: Pillar I Only-Recommended Scenario: 9% Contribution Rate .................... 50 Table 3.7: Funding for Transition Account (in billions of 1994 Yuan) ............................ 52 Table 3.8: Pillar I: Sensitivity of Contribution Rates to Retirement Age and Coverage Rate .............. 53 Table 3.9: Pillar I: Sensitivity of Contribution Rates to Real Interest Rates and Retirement Age ......... 53 Table 4.1: Growth of Income and Implicit Rates of Return on Investment ...................... 63 Table 4.2: Illustrative Investment Rules for Pension Funds ............................................. 66 Table 4.3: Pension System Reform Characteristics: China (Proposed), Chile and Argentina (Actual) ............................................................. 74  - 111 - BOXES IN TEXT Box 1.1: Rural Pensions ............................................................... 4 Box 1.2: Why It is Important to have a High Rate of Return on Pension Funds ............. 16 Box 2.1: ILO Findings on Participation Rate of the Elderly and Flexible Schedule on Retirement .............................................................. 21 Box 2.2: What is a Notional Pension Account? .............................................................. 25 Box 2.3: The Concept of "Implicit Pension Debt" ............................................................. 27 Box 3.1: Why Should Governments Get Involved? The Failures of the Marketplace ... . 41 Box 4.1: Chile's Pension Funds and Their Roles in Capital Markets ............................... 61 Box 4.2: How Pension Reforms Benefit Infrastructure Development in Malaysia .......... 62 Box 4.3: Chile: Adjusting Investment Rules as the Capital Market Deepened ............... 65 Box 4.4: Where Domestic Competition Has Succeeded: In Civil Aviation .................... 72 FIGURES IN TEXT Figure 3.1: China's Population is Aging Fast ............................................................... 34 Figure 3.2: Simulations with Nonstructural Reforms ....................................................... 40 Figure 3.3: Pillar I in Multipillar System with Reforms ................................................... 51   - v - ACKNOWLEDGMENTS The report is based on the findings of the World Bank mission that visited China from August 7 to 31, 1995. The mission members were: Ramgopal Agarwala (team leader and task manager), Estelle James, Yan Wang, Sonia Xiaosong Zhao, Monika Queisser and Cheikh Kane (Bank staff), and Barry Friedman and Wu Xiaoyong (consultants). Rajiv Nundy (Bank staff) and Zhou Xiaoji (consultant) made a major contribution by providing timely assistance to develop the models for quantitative analysis in the report. The counterpart agency from the Government of China was State Planning Commission (SPC) and the counterpart team consisted of Messrs. Yao Hong and Wang Jinduo. During its visit, the mission met with and received valuable assistance from officials of Ministry of Finance, Ministry of Labor, Ministry of Civil Affairs, State Commission for Restructuring the Economic Systems, People's Bank of China, People's Insurance Company of China, All-China Federation of Trade Unions and several line ministries in Beijing as well as in the provinces and municipalities (Beijing; Shenyang, Liaoning Province; Changchun, Jilin Province; Kunming, Yunnan Province; Guangzhou, Guangdong Province; Fuzhou, Fujian Province; Nanjing, Jiangsu Province; and Zhengzhou, Henan Province) that the mission visited. Under the leadership of the SPC counterpart team, 34 background papers were prepared by local authorities (both provincial and municipal) and sectoral ministries presenting the current situation on their pension system, as well as the current plans and longer-term prospects. These background papers provided the basic raw material for the study. The report also drew upon the background papers prepared by Chinese experts and foreign experts. Among the former were: Yao Hong, Guo Shuqing, Zuo Xuejin, Kang Huaiyu, Dong Wenjie, Wang Yanhua and Yu Mingde. Among the latter were: Mukul Asher, Wang Shaoguang and Santiago Plant. Contributions were also made by a large number of experts withini and outside the Bank. Among these were: Messrs./Mmes. Dimitri Vittas, Nicholas Barr, Emmanuel Jimenez and Klaus Schmidt-Hebbel (peer reviewers), Chingboon Lee (UJNDP, Beijing), J.V. Gruat (ILO, Beijing), Athar Husain, Bertrand Renaud, Vikram Nehru, Alan Gelb, Songsu Choi and Jed Shilling. Editing for the report was done by Bruce Ross-Larson and Meta de Coquereaumont, and production work by Cherry Wu, Guo Huiying, Rhoda Blade-Charest, Leila Cruz, Joan Grigsby, Jean Ponchamni, and Meredith Dearborn. The report was prepared under the general guidance of Messrs. Pieter Bottelier, Richard Newfarmer, Michael Walton and Nicholas Hope.   - vii - ABSTRACT China today has a golden opportunity to move toward a unified pension system that combines a defined benefit basic public pillar with funded individual accounts. This conclusion is based on the following findings. China's pension system has two severe problems: the urgent and immediate problem of the pension burden of state-owned enterprises, and the longer-term problem arising from the rapid aging of the population. The current pension system is incapable of tackling either of these problems. It also fails to contribute to the economic development of the country. The problem of state-owned enterprises is due not to the overall demographic dependency ratio, which is still low, but to the fragmentation of the system, generous benefit rates, and high rate of exemptions and noncompliance (stemming from the high contribution rates). The solution lies in higher-level pooling and more realistic benefit levels that reduce contribution rates. The long-run problem stems from the rapid aging of the population combined with incomplete pension coverage. The solution lies in expanding coverage and creating a multipillar system that combines a social insurance scheme with funded individual accounts. The current pension system with largely notional accounts is not solving this problem. The current pension system also fails to contribute to economic development, as it does not encourage labor mobility, provide a level playing field for different enterprises, expedite state-owned enterprise restructuring by delinking pension provision from enterprise management, or assist in the term transformation of savings and the funding of infrastructure and other long-term investments. There is a growing consensus in China on the principles of a reformed pension system: unification of basic design and management of the system, lower benefits, extended coverage, and increased reliance on funded individual accounts. However, the Government views the costs of transition to a funded system as prohibitive. The report argues that China's costs of transition are lower, and its capacity to bear these costs greater, than in other countries that have made similar transitions. The report identifies four factors that facilitate financing of the transition in China: the unification of pension system, the structural change in the economy, rapid growth of GDP and the capital gains in state assets. These favorable factors will generate resources that will enable China to pay its obligations to pensioners and workers under the old system while making a transition to a new reformed system. The borrowing necessary to meet the cash flow needs of transition in the short- and medium-term can eventually be repaid out of payroll contributions, taxes, or proceeds from the sales of state-owned enterprise assets, or a combination thereof. The report outlines a recommended unified pension system that includes both mandatory funded individual accounts and a social insurance scheme. It makes illustrative calculations of  - viii - the financial implications for representative individual pensioners and for the pension system as a whole, showing how the financial viability depends on demographic and macroeconomic developments and pension coverage. The report proposes to use a concept of sustainable contribution rate that attaches more importance to the financial viability over a very long term (over 60 years). And risks associated with, for example, low compliance rates and low interest rates are examined. No funding will be required from the budget in the proposed pension system, including transition program. A combination of 1.1 percent additional contribution rate from workers and transition bonds of about Y 1.2 trillion (backed by housing and other unproductive assets) will finance the transition. The ultimate contingent liability of the Government for a mandatory pension system will remain, but a cushion will be provided by the contingency provision in the basic pillar. Moreover, there will be a considerable saving in the budget on the potential pension expenditure over the long-term for civil service and government institution employees, if the proposed pension system reform could be extended to them now. The success of the proposed pension reform depends on accompanying reforms in legal, administration, and financial systems. The report argues that a unified pension management system be created. At the same time, decentralized public and private pension institutions should be allowed to compete on an equal footing, that investment rules for pension funds should encourage diversification along with a strengthened regulatory framework. Pension reform will fail if coverage of the system cannot be extended, or compliance rate decline, or financial sector reform and capital market development do not materialize and fail to provide an adequate real rate of return on pension reserves.  - ix - EXECUTIVE SUMMARY 1. China's pension system has two severe problems. One is the long-term problem of a rapidly aging population. The other is the immediate and urgent problem of pensions for employees of state-owned enterprises. 2. China's population is aging rapidly, a reflection of the one-child policy of the late 1970s and the 1980s and increased life expectancy. The crisis is dramatized by the "1-2-4" phenomenon (families of one child, two parents, and four grandparents): when the people who are currently entering the workforce retire (the four grandparents), they will have to be supported by one couple (the two parents). By 2030 the absolute size of the labor force in China will begin to decline, and by 2050 the ratio of workers to pensioners (age 65 and above) is projected to decline to about 3 to I from the 10 to I in 1995 (Figure 1). The number of the elderly is projected to rise from about 76 million in 1995 to 300 million by 2050. That is the long-term problem. 3. The immediate and urgent problem is the pension crisis in the state-owned enterprise sector. State-owned enterprises have inherited heavy pension obligations from the central planning era. With the transition to a market economy, employment in the state-owned enterprise sector is growing slowly, if at all, while the number of pensioners is rising rapidly. In some cases, the ratio of pensioners to workers is over 100 percent. An associated aspect of the problem is the slowing down of restructuring. The bankruptcy or sale of a state-owned enterprise raises the difficult issue of how the commitment to pensioners (along with other social welfare obligations of the enterprise) will be honored. When alternative arrangements for pensions and other social services are not available, enterprise reform halts in its tracks since liquidations, joint ventures, or mergers cannot proceed smoothly until the social obligations of state-owned enterprises are assigned elsewhere. KEY FEATURES OF THE CURRENT SYSTEM 4. The formal pension system in China is a largely urban-based, pay-as-you-go defined benefit system that covers mainly the state sector in urban areas. The nonstate sector, which now accounts for more than half the employment in many localities, has only spotty coverage, ranging from 20 to 90 percent. Though localities are trying to bring the nonstate sector under the formal pension system, most nonstate firms are resisting because current contribution rates are so high and the benefit system so uncertain. 5. Contribution rates are high in many state-owned enterprises. For 13 provinces and 12 municipalities for which data are available, the simple average of contribution rates in 1984 was 23.5 percent for the provinces and 25.9 percent for municipalities, well above the international norm. The range is quite wide: from 19 percent in Guangdong to 25 percent in Hunan. For 11 sectors that are exempted from municipal pension pools and are allowed to have their own pension pools, the average is just 15.9 percent, with lows of 10-15 percent for civil aviation, construction, banking, electric power, and petroleum and natural gas, and a high of 24.5 percent for coal mining.  - x - Figure 1. CHINA'S POPULATION IS AGING FAST 1000. 35 900. .30 800I 700 II25 600 II20 Million 500 I 1 400 15 300 I10 200 5 100 1990 1995 2000 2010 2020 2030 2050 Year A 2. Population in age group 65 and above (in millions) + 3. Dependency ratio('/.) (2)/(1) Source: World Population Projections, 1994-1995, The World Bank, 1994. 6. Noncompliance and exemptions account for much of the disparity between the cost rate and the contribution rate needed to break even. Moreover, as contribution rates rise, compliance rates are declining. Many municipalities have reported a drop in compliance rates from 90 percent for the early 1990s to 80 or 70 percent in 1994 and the first half of 1995. 7. Recently, responsibility for paying pension bills has been shifted from individual enterprises to groups of enterprises at the county, municipality, or prefecture level. This pooling of responsibility is intended to spread risk-sharing and help ensure payments to pensioners from enterprises with heavy financial burdens. Many localities have separate pools for various ownership forms. Provincial pooling has also begun in nine out of the thirty provinces. Enterprises, however, generally remain responsible for record-keeping and the actual delivery of pension benefits. And in many places, the pooling is only partial: enterprises with larger proportions of retirees in the pooled system have a higher contribution rate than those with lower proportions of retirees. Thus the pooling system has made only partial inroads in the individual- enterprise based system. ECONOMIC AND FINANCIAL IMPLICATIONS OF THE CURRENT PENSION SYSTEM 8. The current pension system is unable to deal with either the short-term problem of pensions in state-owned enterprises or the long-term problem of old age security in China. Furthermore, the current system is financially unsustainable and fails to contribute to reform of state-owned enterprises or to the economic development of the country.  - xi - * It fails to solve the state-owned enterprise problem because pooling is limited, noncompliance and exemptions are high, and the dynamic nonstate sector is not covered. * It fails to solve the long-term problem because the system's partial coverage means that a majority of China's old people will have no pension when they retire. Only small pension reserves have been accumulated in most municipalities, and these reserves earn a low rate of return. * It fails to contribute to economic development (which will determine future wages and pensions) in key ways. Its Economic and Financial Shortcomings 9. Fails to delink social welfare from enterprise management. Enterprises still carry a large share of responsibilities for their retirees. In many cases, an enterprise's contribution rate depends in part on its own pension obligations. And in most cases, enterprises keep pension records, pay pension benefits, and take care of pensioners' needs, including housing and health. Thus enterprises carry a heavy burden of administration for the elderly, and alternative arrangements will have to be made if nonviable enterprises are to be liquidated. Partial reform of the pension system is creating a new municipal-level bureaucracy without achieving any significant economies of scale in administration or relieving enterprises of responsibility for the elderly. 10. Fails to establish a level playing field. Similar enterprises (say, electronics factories) in two provinces may have to pay widely different "payroll taxes" that can range from 20 to 40 percent. This is similar to imposing a sales or value-added tax at widely different rates for the same product. An enterprise may lose competitiveness not because efficiency in its core business is low but because it is situated in a locality with many retirees and so must pay higher contribution rates. 11. Impedes labor mobility. Factors of production (including labor) must be able to move from one sector or region to another if China is to establish a socialist market economy and restructure its state-owned enterprises. If a nonviable steel mill or textile mill has to be closed down in one city, the redundant labor needs to be able to move to other municipalities where similar industries might be expanding. For that to happen, the pension benefits of workers must be portable. The division of the national system into many separate unfunded municipal pools makes portability difficult and will become an increasingly serious impediment to labor mobility. Because individual accounts are largely notional, departing workers cannot take their personal accounts with them. 12. Fails to increase capital accumulation. The individual accounts being set up under the current system are largely notional; they contain few if any assets because almost all incoming revenues are being used to pay current obligations to pensioners. The interest rates that are paid on the balances are also notional. Total pension reserves are now estimated at less than I percent of GDP, and they are not increasing in many localities. Such notional accounts fail to meet the prefunding, or capital accumulation objectives of pension funds. When workers retire and start drawing annuities based on their individual accounts, the annuities will have to be financed on a pay-as-you-go basis out of contemporaneous contributions, which will have to rise dramatically to meet pension obligations.  - xii - 13. Leads to inefficient capital allocation. Present regulations require that 80 percent of pension funds be invested in government bonds and the rest kept in bank balances. Since the rates of interest set by the Government were below the inflation rate in the last few years, these reserves lost value over time. Thus, even these small reserves will not be there to help when the population ages. The little discretion that is allowed in investment portfolios is exercised by local officials who control the reserves and invest them locally, even if this does not maximize returns. Thus, the opportunity to allocate pension capital to the most productive uses is lost. 14. Misses the opportunity for term transformation. Funded pension funds can create an enormous pool of resources for long-term investments in domestic infrastructure investment and other investments with a high rate of return. The present system of unfunded notional accounts, combined with the absence of reliable long-term financial instruments, is missing the opportunity for term transformation of savings. China has enormous needs for infrastructural and other long-term investments. Demand for infrastructure investment alone is projected to be as high as $744 billion for 1995-2004, or 7.4 percent of GDP (World Bank 1995e). However, because of the absence of long-term instruments, most household savings are in short- and medium-term deposits, which do not provide a solid basis for long-term lending. Thus, foreign financial resources are often mobilized for infrastructure investments (often at high guaranteed rates of return) while domestic savings are left inefficiently utilized. The Current System Is Unsustainable 15. The Chinese official team analyzing the pension burden over the long term (1995-2050) came to the following dramatic conclusions: * By 2033, if the pay-as-you-go system is not reformed, it will require a contribution rate of 39.27 percent. * A fully funded system with an additional contribution rate to meet the needs of transition cohorts will require a 34 percent contribution rate from 2004 to 2031. * A combination of social pool and individual accounts can smooth the rise in contribution rates, but it will still require a 28 percent contribution rate for 2001- 2050 if borrowing (from future generations or other sources) is not used. The situation will be even worse for localities with high dependency ratios. The officially projected contribution rate in the year 2020 is 38 percent for Tianjin and 42 percent for Shenyang. Clearly, such rates will make most enterprises nonviable, particularly nonstate sector companies competing internationally and working under hard budget constraints. The regional disparities in contribution rates that these scenarios imply will also be economically and socially unacceptable. 16. Current trends cannot be allowed to continue. Some innovative solution to the problem must be found. PRINCIPLES FOR A REFORMED SYSTEM 17. The problems with the present system of pensions in China are deep-seated and widely recognized. Extensive discussions have been going on among policy analysts about how to solve these problems and design a new, sustainable pension system. Broad directions of reform needed are becoming clear, but implementation has proven difficult, hampered by  - xiii - misperceptions about the costs of transition and by the political implications of the change (change always involves losses for some groups). Unifying the System 18. The master plan for pension system reform proposed in State Council Circular No. 6 of March 1995 calls for unification of the pension system. It stipulates that "all types of enterprises and workers should be treated equally by a 'unified system, unified standards, unified management, and unified fund usage' (four unifications)." Enterprises and workers now covered under separate plans or not at all would be brought into a single system with common standards. Management would be transferred from enterprises to government agencies, and administrative management and fund management would be separated. There would be multiple channels of funding, including contributions from workers and employers. The idea of multiple tiers of benefits was reaffirmed, including supplementary benefits from enterprises and individual savings. 19. Although the target is a unified system by the turn of the century, State Council Circular No. 6 actually proposes two possible models for the pension system now, and this has created a host of new problems. While a final decision has not yet been reached on a unified basic benefit tier, the two plans both involve individual accounts and social pooling, although organized and combined in different ways. Plan 1, based on ideas developed by the State Commission for Restructuring Economic Systems, emphasizes individual accounts, while Plan II, based on ideas of the Ministry of Labor, emphasizes a larger social component. By proposing two models and by allowing the localities to choose between them or any combination of them, Circular 6 has created a hornet's nest. Each municipality is attempting to differentiate its scheme from others in order to retain control over the pension system and its funds. This has led to the creation of hundreds of schemes all over the country. 20. One way to resolve this dilemma is to make a sharp distinction between the mandatory and voluntary parts of the pension system and between the overall structure of the system and its detailed parameters. The basic structure of the mandatory part needs to be unified, while specific regional or occupational needs can be met through the detailed parameters and the supplementary part. This is consistent with practices in most industrial countries that specify the structure of mandatory systems on a nationally uniform basis. Reducing the Pension Burden 21. There is consensus about the need to reduce the pension burden, but no agreement about how to do so effectively. Three variables can be adjusted to reduce pension burdens: retirement age, wage replacement rate, and pension indexation. The retirement age (60 for men and 55 for women) should be raised since it was determined years ago when life expectancy was lower. Today, the average duration of retirement is over 16 years. Both for the viability of the pension system and to avoid losing skilled personnel to early retirement, the retirement age should be increased to, say, 65-and even higher as longevity continues to increase. A major public information campaign is needed to sell this idea to workers. Since retirement age is a key variable affecting system finance, the sooner China moves to a higher retirement age, the lower can be its contribution rate for a given level of benefit. 22. The wage replacement rates in China's pension systems are high by international standards and by Chinese practice before the Cultural Revolution. Replacement rates for cash  - xIv - wages are over 80 percent, and in-kind benefits (housing, medical benefits, and so on) are continued at the preretirement level, resulting in a replacement rate of 80 to 90 percent of total compensation. Replacement rates in most countries are 40 to 60 percent (they were 50 to 70 percent in China before the Cultural Revolution). There is a broad consensus that the replacement rate should gradually be brought down to about 60 percent of net wages (wages minus payroll contributions from workers). 23. Pensions should be indexed to prices rather than wages. Currently, most municipalities index pensions to nominal wages, with an indexation coefficient of 40 to 80 percent. This is a very uncertain form of indexation, leading in some cases to a reduction in the real value of pensions and in others to a real increase, depending on the rate of inflation. A more appropriate adjustment mechanism, in line with international practice and the stated objective in many localities, is full indexation to the consumer price index (CPI). Introducing Funded Individual Accounts 24. There is a growing consensus that a substantial part of retirement income should come from fully funded individual accounts. The advantages of individual accounts for a rapidly aging population are well known. The pay-as-you-go system will run the risk of requiring very high contribution rates by 2030, when economic growth rates in China may be slowing down. This may lead to high rates of evasion by enterprises, a high burden on society, and a large intergenerational transfer. Funded individual accounts might avoid these problems and instill a spirit of self-help among workers who look after their own retirement costs rather than depending on contributions from their children (whether in the context of the family, enterprise, or social transfers). Individual accounts also develop incentives for workers to ensure that enterprises are making their contributions and that rates of return are being maximized by pension fund management companies. 25. Notional individual accounts do not meet all these individual and social objectives. The current notional accounts link individual contributions and individual benefits, but the accounts do not accumulate reserves, do not earn a decent real rate of return, and do not achieve the term transformation of savings. Thus, contribution rates will have to rise dramatically to pay for the promised pensions as the population ages. Many Chinese policymakers recognize these shortcomings but have not yet devised a practical way to fund the individual accounts and pay current pensioners at the same time. Ensuring an Adequate Real Rate of Return on Pension Funds 26. Funded individual account will not yield acceptable wage replacement rates for pensioners unless the real rate of return on pension funds is at least equal to the real growth rate of wages. If real interest rate is zero and real wage growth is 5 percent annually, the replacement rate will be only 10 percent for a 10 percent annual contribution (Box 1.2). Most calculations of replacement rates made by official agencies assume the rate of return and wage growth rate to be equal, yet the actual rate of return is far less than the wage growth rate. Part of the dilemma here is that the rate of return on pensions can be high only if the general interest rate structure is rationalized, a reform that will increase the financial costs of state-owned enterprises and the Government. A short-term solution could be to offer government guarantees on the minimum real rates of return on pension funds. A longer-term solution could be to rationalize interest rates, decentralize management of the pension funds, and allow diversification of the investments, with the objective of allocating capital to the highest productive uses. Given the  - xv - high productivity of capital in infrastructure and other long-term investments in China, and the high expected returns already paid to foreign investors, it should be possible to devise a scheme that allows pension funds to earn substantially higher yields (Chapter 4). LIMITED REFORM IS RISKY 27. Pension system simulation results from the model developed for the report confirm the finding of the Chinese analysts that the current defined benefit system is not financially viable over the long term. Design changes such as extension of coverage, indexing to prices (but not to wages), reducing replacement rates and increasing retirement age can improve the financial viability of the system. However, the pension funds would remain exposed to high risks (such as possible decrease in compliance rates and labor force participation rates as well as lower rates of return on pension funds) if the defined benefit system continues. In the downside scenario, the required contribution rates will rise to about 30 percent, too high a level by international comparison. It is desirable to reduce these risks (by improving the compliance rates and rates of return on pension funds) and share them between society and the individual. That points in the direction of the multipillar system which combiles social pooling with individual accounts, as is indeed the conclusion of the Chinese policymakers. The key issues then relate to the specifics of these various pillars and the method of transition to the multipillar system. The report uses the simulation model to test implications of variants of such a multipillar system and suggests a preferred approach. MOVING TOWARD A THREE-PILLAR SYSTEM 28. This report proposes a three-pillared pension system, based on the directions of reform identified above and recapped here: * A unified structure for the mandatory part of the pension system, with flexibility through supplementary pensions. * A substantial part (more than half) of pensions to come from individual accounts, which will be fully funded. * A publicly managed part of the system to provide a social insurance. * Gradually move towards target replacement rate for pensions of about 60 percent of net wages during the specified preretirement period. * A unified retirement age for men and women to be gradually raised to 65. * Indexation of pensions to the consumer price index. * Pension obligations to current pensioners and workers under the old system to be honored and financed througlh a combination of sources that do not impose an undue burden on the present generation. The proposed transition mechanism does not impose any burden on the budget beyond the contingent liability of a mandatory pension system. * Reform of the financial sector so as to maximize the rate of return on pension funds and allocate capital to the highest productivity uses.  - xvi - * Gradual expansion of coverage of the pension system to include all formal sector workers in urban areas (including civil service and public institutions) and employees in large township enterprises. The Three Pillars 29. The proposed system combines social pooling with funded individual accounts. It provides for a basic pension component to keep retirees above the poverty line (pillar I) and for large mandatory individual accounts (pillar 11) to obtain the advantages of funding, supplemented by voluntary accounts (pillar III) as desired. A transition mechanism will settle the implicit pension debt to current pensioners and workers. Pillar I thus has a redistributive element, since keeping the elderly above the poverty line involves some redistribution from higher to lower wage earners. Pillars II and III handle savings and link benefits closely to contributions. Pillars I and II are mandatory, while pillar III is based on voluntary private insurance or annuities. The transition mechanism will vanish in the very long run, but its handling is a key issue in the short and medium run (for details about the objectives, management, payments, and funding of the three pillars, see Figure 2). 30. Pillar I. The first pillar, intended as a redistributive and social insurance scheme, would provide a basic benefit equal to 0.6 percent of the local (provincial urban or rural) average wage per year of covered service. A worker with 40 years of credit earning an average income would get a replacement rate of 24 percent from the basic pension. A low-income worker would get a higher wage replacement rate, while a high-inicome worker would get a lower percentage. This basic pension would help to equalize the incomes of the elderly. The basic pension would initially be financed through contributions of 9 percent of the wage bill paid by enterprises. This low rate is made possible by expanding coverage to sectors that are not now covered and that do not yet have retirees. In the initial years, a 9 percent contribution rate would generate some surplus, which could be used to finance part of the cost of transition from a pure pay-as-you-go system to a partially funded system. There are, of course, many options as to how pillar I would be designed, and each is associated with risks and trade-offs. These risks are examined in the sensitivity analyses presented in Tables 3.8-3.9. 31. Pillar II. The second pillar would conisist of mandatory individual accounts that would be fully funded and financed equally (witlh possible increase in workers' share over time) by workers and enterprises. The combined contribution rate would be 8 percent of a worker's individual wages. If the rate of return on pension funds equals the rate of wage growth, an 8 percent contribution rate would yield a replacement rate of about 35 percent of wages in the final years of retirement. The two mandatory pillars together would provide the target replacement rate of 60 percent for the average worker. A funded pension system, however designed, cannot achieve its objectives without a capital market and a financial system that pays a positive real rate of return to long term savings. This report assumes that China will move swiftly to market determined interest rates that remunerate savings. 32. Pillar III. The third pillar would con1sist of supplementary pensions offered by employers on a voluntary basis, or provided by individual accounts established by informal sector workers (including farmers) in licensed Pension Fund Management Companies, or by life insurance policies purchased from insurance companies. The amount would vary, depending on enterprise preferences and capacities, and the willingness to save for old-age security by informal sector workers. It would be fully funded and portable.  - xvii - Payments and Funding 33. For an average worker who joins the system in the future, the target replacement rate after 40 years of service would be about 60 percent of net wages during some specified preretirement period. For this worker, an average of 24 percentage points would come from pillar I and the rest would come from pillar 11. For a worker whose salary is 40 percent below average, the expected replacement rate would be about 80 percent, with 40 percentage points coming from pillar 1. For a worker with wages 40 percent above average, the replacement rate would be about 51 percent, with 15 percentage points coming from pillar 1. 34. The mandatory component of contributions would be at 17 percent of wages (9 percent for pillar I and 8 percent for pillar 11), which is lower than the average current rate of about 23 percent and in line with rates prevailing in East Asian countries such as Japan and Malaysia. Pillar 1, which would fund the basic pensions of new retirees, will have surplus in current cash flows until 2031, with enough reserves to cover future deficits. Supplementary pensions are financed on a voluntary basis by workers and employers. The transition would be financed by borrowing, via bond issuance, from pillars I and 11 in the short run and by a combination of modest additional contributions from workers and proceeds from the sale of state-owned enterprise assets in the long term (see below). 35. The Government is the last resort implicit guarantor for pillar I and the transition plan. As guarantor, the Government will play a key role in the administration of pillar I and the transition. The Government's responsibility for pillar 11 would be as a supervisor and the guarantor of certain financial instruments. For pillar 111, the supervisory functions would be discharged by the central bank, as part of its role as supervisor of other financial institutions. Transition Plan 36. Current pensioners will continue to receive their current benefits, fully indexed to the CPI. Current workers who retire after the reform would receive the basic benefit, an annuity based on their individual accounts, an accrual rate equal to I percent (which could be slightly higher in the initial years)' of their salary for each year of service before reform, and supplementary pensions that their employers might have purchased. Financing the Transition 37. While there is a growing consensus in China that the pension reform should move in this direction, there is no consensus on how to accomplish this. With funding, much of the current contributions of enterprises and workers would go to the individual accounts. Yet financial resources are needed to pay the current pensioners and the accrued pension rights of workers A gradualist scenario was considered with (a) the accrual rate of 1.2 percent to be reduced to I percent over 20 years, (b) the increase in retirement age to be postponed to 2010, and (c) the coverage rate of TVEs to be initially 10 percent (instead of 20 percent) and to rise to 50 percent by 2020 (instead of 2010). The results from the model indicate that the long-term financial viability of the system is not affected. However, the transition debt increases by Y 120 billion, and the surpluses in Pillars I and 11 after funding transition payments are lower by about Y 16 billion per year in the first 10 years. The ability of the social insurance system to fund infrastructure and other investments is thus reduced. The choice has to be made by political leaders in the light of the balance between the higher economic costs of more gradual transition and its greater political and social acceptability.  - xviii - who had contributed under the old system (the implicit pension debt). The problem is how to make these "double" payments. 38. Some rough calculations by Chinese policy analysts suggest that the implicit pension debt could be three to four times GDP. These higih estimates have discouraged the authorities from explicitly recognizing these debts and reforming the system. Estimates of the implicit debt depend on the promised benefits of the pension system, the age structure and life expectancy of the working population and retirees, and the discount rate used. Reasonable values for these variables suggest that the implicit pension debt is actually less than 50 percent of current GDP and an even smaller proportion of future expected GDP, given China's anticipated growth rate. This debt is smaller (largely because of Chinia's relatively low coverage rate) than those which countries such as Chile and Argentinia had to deal withi when they made the transition from pay- as-you-go to funded systems. The low current coverage and the prospect of expanding future coverage make this a fortuitous time for Chinia to move to a system with funded individual accounts. If China instead maintains its pay-as-you-go system, its pension debt will grow rapidly as coverage expands. 39. Countries that have changed from a pay-as-you-go to a funded scheme have used a mix of instruments to bridge the financing gap. The instruments available are the same as those for other public expenditures: bonds, higher taxes, lower spending, or transfers of public assets. The choice and mix of instruments will determine the impact of the pension reform on economic growth and the distribution of the costs and benefits of transition between and within generations. Most countries that have reformed their pension systems have borrowed to spread the burden across many generations and to smooth contribution rates over time. Some of the borrowing could come (through issuing marketable bonds) from the funds accumulating in the mandatory pillars. The bonds can be redeemed by future contributions or by other sources of revenue, such as proceeds from the sale of state-owned enterprise assets. 40. The report presents illustrative calculations regarding the implications of various methods of financinig the transitioni and proposes a burden-slharing formula. Under this formula, the present pensioners will be paid by additional contributioni rate on all covered workers. As noted in Table 3.7, this requires an additional contribution rate of 1.1 percent. For settling the pension debt of existing workers, interest earning and fully-collateralized bonds worth Y 1,236 billion will be given to the transitionl agency. Up to the year 2010, the transition agency will borrow from Pillars I and 11 to cover the payments to pensioniers over and above the receipts due to an additional 1.1 percent contribujtionl rate. After 2010, the transition agency will start cashing, say, 10 percent of the bonds. These cash flows will be enough to make transition payments and start repaymenit the borrowing from Pillars I and 11 (see Table 3.7). By 2050, receipts from additional contribution rates will substantially exceed transition payments, and the loans to Pillars I and lI will be paid off soon thereafter. With this burden-sharing arrangement, the total mandatory contribution rate will become 18 percent, and transition bonds (for all covered workers including civil service and public institution workers) will be worth Y 1,236 billion, which could be safely collateralized by housing assets (and the associated land-use rights) alone of the state sector. CHINA'S WINDOW OF OPPORTUNITY 41. China's costs of transition to a fuLided system are lower thall those in other economies, and its financial capacity to fund the transition is probably greater, for at least four reasons: benefits of the unification of the pension system, rapid growth of GDP, the structural change in  - XiX - the economy, and the expected capital gains. All these favorable factors will remain operative over the next 15 to 20 years; they will largely disappear after 2030. They cannot be counted on to finance high pension costs permanently, but they are well suited to financing the temporary needs of the pension transition. The Four Facilitating Factors 42. Unification. If the system is unified and coverage is extended to the nonstate sector (including township enterprises), substantial resources would be generated that could be used to fund the transition. Contribution rates would need to be kept low so as not to encourage evasion or discourage economic growth. Simulations show that it should be possible to keep rates low. 43. Growth. China's economy is growing at over 1O percent a year and its domestic savings are over 40 percent of GDP. Such rapid growth makes it feasible to use a portion of the incremental income to fund the transition. An important mechanism for accomplishing this is to use price indexation rather than wage indexation of benefits, to keep the rise in pension costs below the growth rate so that the benefits of growth can in part be siphoned off for pension reform. 44. Structural Changes. Associated with rapid economic growth are rapid structural changes in the economy: labor shifting from rural to urban areas, from agriculture to nonagricultural activities in rural areas, and from the state sector to the more efficient nonstate sector in urban areas. The reserve pool of surplus labor will enable the nonagricultural labor force to expand rapidly, even if the overall labor force grows slowly, thereby expanding the contributory base for the pension system. Eventually, these workers too will retire. But if the transition has been made by then, their pension needs will be met, in part out of their own funded individual accounts and by the higher productive capacity generated by their retirement savings. 45. Capital Gains. Inevitably, as state-owned enterprises are restructured and divested of such social responsibilities as housing, some of their assets will be sold. Some of the proceeds from these asset sales could be used to redeem the bonds that are issued in the early years of the pension reform. The state enterprise sector in China, while facing problems, has not collapsed the way it has in other socialist economies in transition. The sector provides about 70 percent of government revenues and accounts for a significant part of the country's impressive savings, export, and growth performance. The value of state-owned enterprise assets is higher than GDP (compared with less than 50 percent in Latin American and Eastern European countries). The availability of these assets opens up possibilities for transition financing in China that are not available in other economies. Two principles might be honored in this process: any asset transferred to pension funds has to be marketable and carry a value corresponding to the market- determined prices, and the link between asset sales and pension liabilities should be as close as possible, so that the asset-liability swap is easy to understand and implement. MANAGING THE REFORMED SYSTEM 46. Successful implementation of the proposed pension system depends on accompanying reforms in legal, administrative, and financial systems. Reforms in all these areas must go hand in hand, and coordination is crucial. To ensure that the pension reform moves forward across the country, a national agency should be created to have the primary responsibility for coordinating the development of the three pillars and the transition mechanism (see Chapter 4 and Annex 4.1).  - xx - Administering the First Pillar 47. The large diversity in initial conditions and administrative capacity may preclude national pooling in the short run; thus the initial pooling for pillar I should probably take place at the provincial level, with a national adjustment fund. Pooling at the provincial rather than municipal level will provide a larger pool of workers and pensioners to spread the risk. It will also facilitate the incorporation of township enterprises into the system, which is beyond the authority of most municipal authorities. Over time, the system should move toward national pooling, with a uniform national contribution rate for the basic benefit. 48. Particular attention should be paid to the need to reduce noncompliance rates. Improved tax administration capacity should reduce evasion, but the likely growth of the informal sector will increase it. One of the best ways to increase compliance is to create incentives and tax enforcement mechanisms at the designing stage. Managing the Second Pillar 49. The fully funded individual accounts should be decentrally and competitively managed by licensed corporatized state, joint venture or private investment companies, subject to minimum capital requirements, reserve requirements, investment rules, information disclosure, and antifraud rules. To prevent the individual accounts from becoming notional, they must be managed separately from the first pillar and the transitional mechanism. Any borrowing by the transition agency should be explicit, using fully-collateralized bonds with a positive real rate of interest paid and repayment ensured. 50. In the short run, it is likely that most pension funds would be invested in public securities. The Government should provide a guarantee for the minimum real rate of return on pension funds. Funds mobilized in this way should be used for long-term capital construction or infrastructure projects with an economic rate of return that justifies the real cost of capital. 51. Over the longer term, strategies should be developed to encourage greater competition in management and to permit investment diversification across sectors (public and private), financial instruments (equities, bonds, and mortgages), and geographic regions. Decentralization and diversification are important to allow the second pillar to maximize returns (subject to an acceptable risk level), to encourage the best allocation of capital, and to assist in the financial market development needed for China's economic development. To this end, the Government should formulate investment regulations and adjust them periodically as capital markets and the real estate market mature and grow. These regulations should preclude concentration in one company, industry, or locality. Developing and Regulating the Voluntary Third Pillar 52. China needs to develop its institutional and regulatory capacity for providers of supplementary pensions: insurance companies and employer-sponsored pension funds. Two major impediments are the weak incentives on the demand side due to the high replacement rate under the old pension system and the lack of a legal framework for employer-sponsored pension funds and for Pension Fund Management Companies. The recommended reduction in the replacement rate should create space for voluntary plans, and a legal framework should be established that regulates the behavior of the plans. In addition, informal sector workers (including farmers) should be allowed to establish individual accounts in licensed PFMCs or purchase life insurance policies, and enjoy tax benefits for pension contributions up to a  - Xxi - maximum. Furthermore, domestic competition should be introduced into the insurance industry by restructuring the People's Insuranice Company of China (PICC). Contestability of the insurance market should be increased by encouraging new domestic companies and foreign and joint venture companies. Clear and transparent investment rules should be set up for insurance companies, similar to those for pension funds (see Table 4.2). More opportunities should be allowed for portfolio diversification and capital appreciation. Administering the Transition Mechanism 53. A special administrative arrangement would be set up in each municipality to handle the paying off of pension obligations owed to existing pensioners and workers under the old system, to finance the transition to a new system, and to meet the cash flow needs by borrowing from pillar I and fI via bond issuance and bond redemption. CONCLUSIONS 54. Pension system reforn involves major redirection of financial flows and asset entitlements and the livelihoods of millions of people at a highly vulnerable stage of their life. In the process, there *vould be major gainers and losers. Enterprises and localities with younger populations would lose initially, and those with older populations would gain through the pooling proposed here. In the proposed scheme, an effort has been made to reduce the conflict by putting emphasis on ildividual accounts and defined contribution systems. However, the basic pension component would indeed have a redistributive element. Even more important is the transition plan, which may involve major chaniges in the entitlement to assets in society. The treatment of women would also be chanlged. They get access to a basic benefit which is independent of their wage and, in that respect, gain since women tend to be low wage-earners. However, they lose by the rise in their retirement age to match that of men. The pension reforms thus impinge upon many powerful political economy considerations, which may play a decisive role in policy making, but are beyond the scope of the present report. 55. The present report emphasizes the urgency of pension system reform in China. The urgency derives primarily from the need for delinking social welfare responsibilities from SOE management so as to accelerate SOE reforms. From the long-run point of view, it is clear that the costs of transitionl will rise with time. But the reforms should not be done in an atmosphere of crisis. 56. There are many risks involved in the reforms. If the financial sector reforrns and capital market reforms do not materialize, the individual account system will not function properly. Particularly important are interest rates that provide positive real returns to savings. Similarly, if coverage of the system cannot be extended, compliance rates decline or local authorities do not cooperate fully, the reforms will fail. A careful program of consensus building along with mobilization offunds for implementation is neededfor successful reforms. The Social Insurance Law is urgently needed to standardize the basic framework of pension provision and to establish the institutional infrastructure of the system and regulatory framework for the fund management companies. Experiments of the standardized system in selected localities should be started soon, along with careful monitoring of the results. These results should be utilized to improve and implement the unified system of pensions, latest by the year 2000.  - xxii - Figure 2: OBJECTIVES, MANAGEMENT, AND FUNDING OF THE PROPOSED THREE PILLARS Mandatory Mandatory Fully Funded Supplementary Basic Benefit Individual Account Individual Account Transition Pillar Pillar Pillar Mechanism Objectives * Redistribution * Savings plus Savings plus Pay off the plus social social insurance private pension debt to insurance coinsurance existing pensioners and current workers Role of * Implicit guarantor * Guarantor of a * Supervision (by * Implicit guarantor Government of last resort minimum rate of the central bank) of last resort * Administration return on pension * Tax benefit * Administration * Tax benefit funds (subject to a * Supervision maximum) * Tax benefit Management Social insurance * Pension Fund * Private/public * Social Insurance Agency companies Management insurance Bureau at (affiliated to Companies companies municipality level Social Insurance (autonomous * Employer- Bureau) public companies sponsored pension or private funds _ companies) _ Payments * 24% of average * Annuities * Annuities * Payment to provincial wages * Surplus invested * Investment in existing paid to new in government financial pensioners retirees bonds and other instruments * Accrued rights of ; Surplus invested approved existing workers in government instruments when they retire bonds (for both * Disability and transition and survivor benefits investment purposes) and other approved instruments * Disability and survivor benefits _ Source of * Contribution: 9% * Contributions: * Contributions * Receipts from Funds of wage bill 8% of workers' from enterprises additional contributed by wages contributed * Investment contribution rates employers by employers and income (1.1%) from * Pooling at workers on a * Contributions workers provincial level 50:50 basis from informal * Borrowings from and fund * Investment sector workers to Pillars I and II adjustment at income their individual * Encashment of national level accounts opened bonds * Income from at a licensed collateralized by investment of Pension Fund housing assets surpluses Management and land use Company rights * Premium paid to life insurance companies by informal sector workers (including _____ ____ ____ ____ farm ers) _ _ _ _ _ _ _ _ _ _ _  - 1 - 1. LONG-TERM AND SHORT-TERM CHALLENGES IN THE PENSION SYSTEM 1.1 China's program of economic reforms since 1978 has achieved spectacular success. While most socialist economies in transition in Eastern Europe are traumatized by shock therapies, China has registered GDP growth of about 10 percent a year over the last 15 years, with not a single year of decline in output. Its inflation rate over the period has been less than 10 percent a year on average, with the worst inflation rate touching 25 percent in one year. With a savings rate of over 40 percent of GDP, export growth of over 15 percent a year in current dollars, and foreign exchange reserves exceeding $90 billion, China is in a strong economic position. This economic situation puts China in a strong position to bear the costs of transition, including pension reform. 1.2 The economic success of China has its roots in many factors-economic, social, and historical. A notable part of that success is the capacity of policymakers to focus intensively on particular problems as the occasions demand, mobilize domestic policy analysis, solicit foreign advice from various quarters, form political consensus and then move decisively on policy reforms. In recent years, the reforms in the fiscal system, financial system, inflation control, and foreign exchange regimes are illustrations of such efforts. 1.3 Currently, pension system reform is one such area on which the Chinese authorities are focusing attention. Enormous amount of analytical work has been done by Chinese policy analysts, a large number of experiments are being conducted all over the country and various sources of foreign advice are being sought. The report is a contribution to that discussion, which hopefully is coming at a stage where decisive action will now be taken without much delay. THE CHALLENGE 1.4 China has not one but two problems of old-age security: the long-term problem due to rapid aging of the population as a whole and the immediate and urgent problem of pensions for the SOE employees. The Long Run Problem 1.5 The one-child policy of the late 1970s and the 1980s and the increasing life expectancy in China mean that the population will be aging rapidly. By 2020, the elderly (age 60 and above) will make up some 16 percent of the population, close to the 18 percent share in OECD counties in 1990 (Table 1.1). While it took most OECD countries 80 to 100 years to double the proportion of its old people to 18 percent, China will do that in just 34 years. By 2020, this ratio will be much higher in China than in its neighbors in Asia: for example, India (10.3), Indonesia (10.9), Pakistan (6.3), Bangladesh (7.6), Viet Nam (9.0), and Thailand (12.8). The situation is projected by the Government to be even more serious in urban areas, with estimated dependency rates climbing from 14.83 percent in 1990 to 47.29 percent in 2030 (see Technical Annex I ). The crisis is dramatized by the oft-repeated observation in China that when the people currently entering the work force retire, four parents will have to be maintained by one couple: the "1-2-4" phenomenon-one child, two parents, four grandparents.  TABLE 1.1: CURRENT AND PROJECTED PERCENTAGE OF ELDERLY (percent) 1990 2000 2010 2020 2030 2050 China 8.9 10.2 12.0 16.0 21.9 26.1 Korea 7.7 10.7 13.9 19.5 25.5 29.8 India 6.9 7.5 8.3 10.3 13.1 20.4 Malaysia 5.7 6.5 8.0 11.0 14.5 22.1 Japan 17.3 22.7 29.0 31.4 33.0 34.4 OECD Average (simple) 18.6 20.0 23.2 26.9 30.8 31.3 Note: The elderly are defined as the percentage of the population over 60 years old. Source: World Bank 1994. Averting the Old Age Crisis. 1.6 Thus in 2030, China will be facing the problems of a mature economy like Japan today while its per capita income will, in all probability, be about one-fiftlh that of industrial countries today. Moreover, precisely at the time that China's old-age burden increases, its GDP growth rates may be slowing down due to declining opportunities for technological catch-up. In other words, China will have a high income country's old-age crisis with a middle income country's resources for tackling them. Advance planning is needed to avert the crisis by using the opportunity provided by today's high growth and high savings to prepare for old-age security. Short-Term Challenges 1.7 A more urgent and immediate problem is the crisis of pensions in the SOE sector. As a legacy of the plan era, state-owned enterprises lhave heavy pension obligations (Table 1.2). With marketization of the economy, employment growth in the SOE sector is slowing, while the number of pensioners in relation to employees is rising. In some cases, this ratio is over 100 percent. In a planned economy where the profits of enterprises were pooled, and resources were allocated according to plan, the burden on individual enterprises was not their concern. However, with marketization, individual enterprises are becoming responsible for their profits and losses, and the burden of pensions is unmanageable for many of them. Even when the core business of an enterprise is profitable, the overall enterprise may show losses because of having a large pool of pensioners; and its capacity to operate, borrow, and expand may suffer. It is also possible that, under these circumstances, many pensioners will not get adequate payments simply because of their past assignments to enterprises that are now not doing well financially. Thus the system is becoming both inefficient and unfair. TABLE 1.2: GROWING IMBALANCE BETWEEN WORKERS AND PENSIONERS IN STATE- OWNED AND COLLECTIVE ENTERPRISES 1990 1991 1992 1993 1994 Workers (millions) 106.8 109.6 108.7 109.0 105.5 Pensioners (millions) 13.2 14.7 15.9 17.7 19.1 System dependency ratio (%) 12.4 13.4 14.6 16.2 18.1 Wage expenditure (billions of yuan) 223.5 251.3 291.5 351.9 448.8 Pension expenditure (billions of yuan) 22.6 26.6 32.5 42.0 56.5 Pension/wage ratio (%) 10.1 10.6 11.1 11.9 12.6 Source: Ministry of Labor, China.  - 3 - 1.8 An associated aspect of the problem is a slowing down of enterprise restructuring due to the pension system. The state-owned enterprises are like miniwelfare states-where the work units look after workers' welfare "from cradle to grave."2 The bankruptcy or sale of SOE raises the difficult issue of how the commitment to pensioners will be honored (along with the other social welfare obligations of the enterprise). When alternative arrangements are not available, the result is a slowing down of enterprise reform; neither liquidation nor joint ventures nor mergers can proceed smoothly unless the issue of the social obligations of SOEs is settled. In addition, with the accumulated losses of enterprises, the banks are now facing an increasing burden of bad debts that hinders the commercialization of these banks. Thus the dilemma of the policymakers is increasingly evident: SOEs cannot be efficient unless they face a hard budget constraint; the banks cannot be commercialized unless they can enforce hard budget constraints on their borrowers; and neither of these is possible unless some solution is found to detach from the enterprises their social welfare obligations, of which pensions are an important component. As the losses of SOEs mount, there is a great urgency for solving the SOE pension problems. 1.9 In view of the seriousness of both the short- and long-term problems of the pension system, various reform experiments have been conducted in these areas in many parts of China and an enormous amount of analytical and policy work has been done. After 10 years of experimentation, the time has come to take stock of the results of experiments and analyses and to take decisive action on reform, muchi as was done in other areas of reform such as taxation and foreign exchange. EVOLUTION OF THE OLD-AGE PENSION SYSTEM Origins of the System 1.10 A brief history of the modern pensioii system in China will help to put the current discussion in perspective. In 1951, the State Council issued Regulations on Labor Insurance that established a pension system. The initial regulations applied to enterprises with more than 100 workers. The system was administered by the All-China Federation of Trade Unions (ACFTU). Not surprisingly, it was a small program because the enterprise sector itself was small and there were few retirees. One year after the start of the program, there were only 8 million enterprise workers and 20,000 retirees, or over 400 workers per retiree. The system was funded exclusively by contributions from enterprises, and a contribution rate of 3 percent of the wage bill was sufficient to finance it on a largely pay-as-you-go (PAYG) basis. The old age pensions were 50 to 70 percent of workers' wages. In this early system, there was a form of pooling across enterprises on a national basis. Seventy percent of enterprise contributions were retained locally to pay pensions while 30 percent were transferred to a national master fund. The All-China Federation of Trade Unions (ACFTU) managed both the local payment procedures and the master fund. The master fund appeared to have provided a degree of prefunding. 1.11 Subsequent reforms expanded coverage. In 1955, the Temporary Regulations on the Retirement of Employees in Government set up a separate system for employees of government 2 Some amount of corporate welfare provision is not uncommon even in OECD countries. Companies often provide housing, medical facilities and other forms of assistance to employees. The problem in China arises from the extent of these social welfare benefits, lack of alternative (outside of enterprise) avenues of meeting these needs and the poor funding capacity of many enterprises to honor these obligations.  - 4 - units, nonprofit units, and party organizations. The Temporary Regulations on the Retirement of Workers and Staff of 1958 dealt with both enterprise and government workers. Coverage was extended to enterprises with fewer than 100 workers. This system basically continued until the Cultural Revolution, which began in 1966. 1.12 During the Cultural Revolution, ACFTU was abolished, as was the Ministry of Labor (MOL). The pension funds that had accumulated were used for other purposes, thus eliminating any prefunding that had built up. Supervisory responsibilities were transferred to local labor bureaus, while responsibility for managing payments was transferred to the enterprises. Pooling ended and so did prefunding, since each enterprise paid the pensions to its own workers out of its current revenues. Box 1.1: RURAL PENSIONS While most urban workers in China have long been covered by pension plans offered by their enterprises, rural workers have had little access to formal pension plans and have instead relied primarily on the extended family for old age support. Parents had several children and invested in them, hoping for future returns, multigenerational households were the rule, mobility was limited and strong social norms reinforced reliance on the family as a social insurance and informal pension system. Because these forces are generally stronger in rural than in urban areas, formal systems for providing old age security made their first appearance in cities, in China as elsewhere. In recent years, however, the Govenmment has become concerned that the informal system is not enough, even in rural areas. Although the one-child policy is less strongly enforced in rural areas, families are shrinking in size and workers are becoming more mobile as a consequence of the shift to a market economy. In 1991, the Ministry of Civil Affairs (MOCA) introduced a voluntary pension insurance system, aimed at farmers and workers in town and village enterprises. The MOCA plan is now being tried, on an experimental basis, in 1,400 counties spread across the country. So far, about 50 million individuals are participating in this scheme and 4 billion yuan have been accumulated. Participation is highest in the richer rural areas along the coast and nearby, such as Shandong and Jiangsu provinces. The plan encourages workers to contribute voluntarily to their retirement savings accounts. The accounts are fully funded and the funds are turned over to county officials to invest. The investment income depends on their investment strategy; some counties invest in treasury bonds or bank deposits that eam 10 to 13 percent while others use a special trust arrangement with a bank that has greater flexibility and earns as Imuch as 18 percent (nominal). Each year the individual accounts are credited with an interest rate that is set by MOCA; currently the nominal interest rate is 12 percent. The rates earned by the fund and credited to individual accounts have in some recent years been less than the current rate of inflation. Ordinarily, the money cannot be withdrawn until retirement at age 60, when it is turned into an annuity that lasts a person's lifetime. An annuity can also be purchased with a lump-sum payment upon retirement. Currently 510,000 people are receiving pensions, most of them on the basis of lump-sum payments. The annuity is based on an interest rate of 8.8 percent nominal and an expected lifetime of 18 years after retirement. The annuity is not indexed for inflation.  - 5 - Box 1.1 continued Why have so many peasants and workers chosen to join, despite the negative real interest rate and the fact that they lose access to their money until retirement? One reason is that few alternative financial instruments are available to them. Most bank deposits earn only 10 percent, so individuals do better by joining the MOCA plan where they earn 12 percent. Another reason is that education and moral suasion may be used by local officials who sponsor the plans and want to make them work. In Yantai city, Shandong province, one rural areas with a participation rate of 87 percent, contributions are conducted during a fund- raising week held once each year. This is administratively efficient and also makes it easy for local officials to apply such moral suasion. In Yantai, local officials have constructed a table that shows workers exactly what size pension they will get from alternative amounts of contributions over their working lives. Given the interest rates mentioned above, these pensions look quite attractive, especially since the numbers are not corrected for the inflation that inevitably goes along with these high nominal rates. It looks like a good deal to most workers. Finally, villages and township enterprises are supposed to use their tax resources or profits to partially match the individual's contribution, under the MOCA plan. In reality, matching is only carried out by the more affluent villages and towns (such as Yantai), but where present it clearly raises the rate of return to the individual and consequently the incentive to participate. In several provinces, additional counties are anxious to become part of the MOCA experiment. Does this mean that voluntary pension plans are the way to go and that these voluntary plans should be provided by a government agency? A closer look at the results of the MOCA scheme casts doubt on such a conclusion. While many individuals are participating, they are contributing minuscule amounts. In Yantai, the average contribution is 40 yuan per worker per year. In 1995, after three years of operation, the average accumulation per participant across all 1,400 counties was 80 yuan. This is only about 2 percent of annual income of farmers and is unlikely to generate more than 5 percent replacement rate of income after retirement. How can this be avoided? First, private insurance and investment companies should be allowed to operate in rural areas, subject to regulation, offering annuities and long-term saving opportunities with a competitive rate of return and without any implicit or explicit state guarantee. Second, if the extended family is breaking down even in rural areas, and if a sizable group of rural workers is deemed too short-sighted to make alternative arrangements on their own, mandatory coverage should be extended to them-providing the mandatory system has been reformed and is no longer building up a pay-as-you-go pension debt. For example, workers in township and village enterprises that exceed a specified size, such as 50 or 100 workers, might be included in the mandatory plan quite soon. TVEs are the fastest growing type of enterprise in the economy, employing about half the nonagricultural labor force in 1993, and there seems to be little reason to keep them out of a reformed mandatory system that combines individual accounts with a social safety net. In the very long run, as the Government's tax-collecting capacity grows, and if the extended family continues to break down, coverage can be extended to farmners, workers in smaller enterprises, and the self- employed groups that pose the greatest administrative costs and monitoring problems for social security systems in every country. By 2050, farmers are projected to be only about 10 percent of the workforce, and the mandatory pension system will cover the majority of the working population in the country, including rural areas. Reforms of the System in the 1970s 1.13 When the economic reforms began in 1978, the State Council issued new pension regulations (Document 104 of 1978) for state-owned enterprises (SOEs), government workers, and nonprofit organizations. It was recommended that large collectively-owned enterprises (COEs) follow the same rules. The prevailing retirement ages were reaffirmed-60 for men and 55 for women with additional adjustments for those in hazardous jobs. Qualifications were  - 6 - eased, allowing a worker to retire after 10 years of continuous service rather than 20. New, higher benefits were related to length of service and to the final standard wage. A person who had worked for at least 20 continuous years would get a pension of 75 percent of his or her standard wage; someone who had worked for 15 to 20 years would get a pension of 70 percent; and someone who had worked for 10 to 15 years would get a 60 percent pension. There was a minimum guaranteed pension of 30 yuan per month. Disability pensions were related to the final standard wage and the extent of care needed. 1.14 The regulations of 1978 created a number of problems that subsequently had to be addressed. There was a clear intention to encourage early retirement to create jobs for a large influx of new workers into the urban labor force. The minimum years of service required to qualify for retirement were lowered to 10, and benefits were raised to make retirement more attractive. For a time, workers who retired were guaranteed a job for one child. Given these incentives, the number of retirees jumped fivefold between 1978 anid 1985, and pension costs rose from 2.8 percent of the urban wage bill to 10.6 percent. As the increase in pension costs became apparent, the special incentives to encourage retirement were curtailed. I .15 Other problems have persisted longer. Basing the pension on the final standard wage has provided an incentive for workers and employers to jack up the final wage in preparation for retirement. The 1978 regulations provided a generous replacement rate relative to the standard wage, which was the largest part of the wage at that time. Labor reforms have since increased bonuses to the extent that the standard wage is perhaps only half the total. While this should have reduced the effective replacement rate and thus helped to controlled the rise of pension costs, actual benefits may be more generous than the rules suggest. Reforms in the 1980s 1.16 New regulations have been issued several times since 1986, and there have been many experiments in pension design. State Council Document 77 of 1986 established pooling across state enterprises on a limited basis (at the municipal level). Enterprises retained the responsibility for actually distributing the pensions. The pool operated by setting a contribution rate (or formula) for participating enterprises. If the pension costs of an enterprise were less than the contribution rate, it would remit the difference to the pool. If pension costs were higher than the contribution rate, the pool would cover the difference for the enterprise. 1.17 The 1986 pension reforms were accompanied by employment reforms that established contract labor. New workers were to be hired on a contract basis, while current workers would continue as permanent workers. Separate city pension pools were established for contract workers and permanent workers. The contract workers made individual contributions, while the permanent workers initially did not. Enterprises contributed to both pools. In the late 1980s, pooling was extended to COE workers in many cities, although these pools were generally separate from those of SOE pools. During the 1990s, other enterprise types, such as joint ventures, joint stock companies, and foreign enterprises, have been brought into pension pools in some cities, although participation in the pools is generally far from complete. In several provinces there have been efforts to move to provincial-level pooling, generally only for SOE workers, but again provincial pools remain fragmentary. 1.18 State Council Document 33 of 1991 was a major statement of policy. It called for individual contributions by all workers, in addition to enterprise contributions. It also called for an expansion of pooling and the establishment of three tiers in the pension system, a basic  - 7 - benefit, a supplementary benefit to be provided by enterprises in sound financial condition, and a benefit based on individual saving. It also called for experiments including a role for individual accounts. GOALS OF THE PRESENT REFORM 1.19 The most recent official statement on pension policy has been on Deepening of Reform of Pension Insurance System in State Council Document 6 of March 1995 (Annex 1.1). It expresses the goal of establishing a unified pension system by the year 2000. Various types of enterprises and workers now covered under separate plans or not at all would be brought into a single system with common standards. Management would be transferred from enterprises to government agencies, although post offices and banks might assist with payment procedures. Administration and fund management should be separate. There would be multiple channels of funding including contributions from workers and employers. The idea of multiple tiers of benefits was reaffirmed, including the enterprise supplementary benefits and individual savings tier. 1.20 Although the target is a unified system by the turn of the century, Document 6 actually proposes for now two models for the basic tier. The central authorities gave the right to select a reform design to the city and prefecture governments and the right of approval or disapproval of the choice to the provincial government. While a final plan has not yet been reached on a unified basic benefit tier, both plans in Document 6 involve individual accounts and social pooling, although organized and combined in different ways. Plan 1, based on ideas developed by State Commission for Economic Restructuring of Economic Systems (SCRES), emphasizes individual accounts, while Plan II, based on the ideas of the Ministry of Labor, emphasizes the social component more than Plan I. Plan I 1.21 In Plan I, the basic pension system for new workers would be individual accounts. A social pool would be responsible for pensions for those already retired, for current workers not fully covered by individual accounts, and for certain adjustments for the retirees drawing from individual accounts. Contributions into individual accounts would be approximately 16 percent of the total wage and would consist of three parts: * An individual contribution of 3 percent of total wages. * An enterprise contribution of 8 percent of each worker's total wage. * An enterprise contribution of 5 percent of the average local wage. 1.22 The intention is that the individual contribution would be increased over time and that the enterprise contribution would be decreased by I percentage point every 2 years for 10 years until the individual contributes half of the total contribution to the individual account. Both a ceiling and a floor are set for the individual contribution. For the ceiling, the wage base used in calculating contributions will not include individual wages in excess of 200 percent or 300 percent (to be set by each city) of the average local wage. For the floor, every worker will have to contribute at least 60 percent of the average city wage, no matter how low the actual wage. This minimum contribution will in turn establish a floor benefit upon retirement. The local government may set a higher floor for retirees and supplement the pension so that it reaches the floor. Interest will be credited to the personal account each year, based on the bank interest rate  - 8 - and the rate of increase in the average local wage, though the exact mechanics for determining the rate of interest are not clear. 1.23 A worker who contributes for at least 15 years and reaches retirement age will receive a monthly pension equal to 1/120 of the total accumulation in the individual account at the time of retirement. With a life expectancy of 16 years at age 60, this implies an interest rate of 4.5 percent. The individual account is intended to last for the person's life expectancy. If the person lives longer, the social pool will continue to pay the pension. A worker with less than 15 years of contributions will receive the amount in the individual account in one lump sum on retirement. If a person dies, the accumulation based on individual contributions will be paid to heirs, and the part paid by the enterprise will be returned to the social pool. The total enterprise contribution rate will be larger than the approximately 13 percent of wages going into individual accounts, in order to fund the social pool. Each city will set its own contribution rate. Should the contributions to the social pool be insufficient to cover its obligations, the necessary funds will be taken from the individual accounts. To the extent that funds are withdrawn, the individual accounts will be notional rather than fully funded. 1.24 Plan I includes two government guarantees: a longevity guarantee and a minimum pension guarantee. A worker who outlives the individual account will continue to receive a pension through the social insurance pool. Workers who contribute to the new system for at least 15 years or who have had a continuous employment tenure (including the contribution years) of at least 10 years before the reform will receive a monthly pension after retirement. If the pension is less than the minimum pension, the Government (the level of government is not specified) will make up the difference. The level of minimum pension will be specified by local governments. Plan II 1.25 In Plan II, more emphasis is put on social pooling than on individual accounts. The plan is designed mainly for cities that had chosen an earlier Ministry of Labor pension model, with a vesting period of ten years. For those whose payment period is longer than 10 years, the pension will consist of the following parts: * a social pension equivalent to 20-25 percent of the local average wage; * a premium pension equivalent to 1.0-1.4 percent of wage base for each year of contribution; * an individual account pension that can be drawn as lump-sum or annuities equivalent to the funds in the individual account; and * a supplementary subsidy from the social pool, which will be eliminated over time. 1.26 As in Plan 1, there will be special transitional arrangements for those who have already retired and for the transition cohorts of workers not fully covered by the new scheme. The floor and ceiling for contributions will be as in Plan I except that earnings in excess of 300 percent of the average local wage will be excluded from the wage base. 1.27 Plan IIB, a modification of Plan II, eliminates the premium pension so that the only pension in addition to the social pension is based on individual accounts. Pension from individual accounts will be gradually raised over time while the premium pension is reduced.  - 9 - During the transition, the increase in the individual account and the reduction in the premium pension will be coordinated, with the goal of maintaining a 60 percent replacement rate for the total pension. 1.28 Although State Council Circular No.6 calls for unification of the pensionl system by the end of the century, it seems to have contributed to further fragmentationi of the system. By proposing two plans-one emphasizing individual accounts and the other social pooling-and by allowing localities to chose between them or any combinationi of the parameters of the two, it has opened up a "Pandora's Box." Each municipality is attempting to differentiate its scheme from that of others in order to retain control over the pension system and the pension funds. This has led to creation of hundreds of schemes all over the country. KEY PROBLEMS OF THE CURRENT PENSION SYSTEM Low and Variable Coverage 1.29 The formal pension system in China is largely urban-based. In rural areas, families are the chief means for support of the elderly. This arrangement is often associated with "son preference" of the farmers, which often comes in conflict with the Government's "one child policy," especially if the first-born is a girl. Many policymakers in Chinia feel that an adequate pension provision in the old age can improve compliance with the one-child policy in rural areas and alleviate its worst side effects. A small beginning in rural pensions has been made by the Ministry of Civil Affairs, though the voluntary scheme is too small to provide adequate protection for the elderly (see Box 1.1 for Rural Pensions). It is, of course, difficult to provide formal old age insurance for farmers, who have small, fluctuatinig incomes with poor recording systems. But as China industrializes and the labor force in agriculture declines, a formal pension system can be extended to rural areas over the long term. 1.30 Even in urban areas, the pension system has been largely focused on the state sector. The nonstate sector, which in many localities now accounts for over 50 percent of employment, has only spotty coverage, ranging from 20 to 90 percent. Most localities are trying to bring the nonstate sector under the pension system. But the current contribution rates in the pension system are so high and benefit system so uncertain that most nonlstate sector firms resist the pressures for joining in. Since the nonstate sector is the most dynamic part with increasinig contribution to growth in output and employment, the local authorities have to proceed cautiously and not render them uncompetitive through high and arbitrary taxation. Only with a rational and realistic pension system can the nonstate sector be persuaded to become a full participant in the formal pension system. Current Contribution Rates are High and Variable 1.31 Pension expenditures in China are only about 12.6 percent of the wage bill of the covered sectors, and surpluses are less than 0.5 percent of the wage bill. Thus, the effective contribution rates are about 13 percent of wage bill. However, for many state-owned enterprises the contribution rates are high.3 For 13 provinces and 12 municipalities for which data are available, simple average contribution rates in 1984 were 23.5 percent and 25.9 percent respectively, well above the international norm of about 20 percent. Within these averages, the 3 One reason for these high rates is that they relate to previous year's wages.  - 10- range is quite wide: from 19 percent in Guangdong and 21 percent in Shenzhen to 25 percent in Hunan and 30 percent in Chongqing. However, for 1 I sectors that have their own pension pools, the contribution rates are lower. The average is 15.9 percent, with lows of 10 to 15 percent for civil aviation, construction, banking, electric power, and petroleum and natural gas, and a high of 24.5 percent for coal mining. These sectors account for 15 million of the 75 million SOE workers. TABLE 1.3: CHINA: FEATURES OF PENSION SCHEMES OF SELECTED PROVINCES, MUNICIPALITIES AND SECTORS, 1994 (percent) Covered pensioners/ Average pension contribution rate Replace- Gross pension/ Covered contributors Employer Worker Combined ment rate gross wage Provinces Liaoning 27.2 21.9 3.0 24.9 .. 21.9 Jilin 23.2 23.0 2.0 25.0 .. 15.4 Heilongjiang 22.6 23.0 2.0 25.0 90 Jiangsu 21.7 20.0 3.0 23.0 65 14.0 Zhejiang 20.4 22.5 3.0 25.5 80 Fujian 22.7 20.0 4.0 24.0 81 13.5 Henan 16.3 25.0 3.0 28.0 75 17.5 Hubei 18.4 20.0 2.0 22.0 85 17.9 Guangdong 23.2 17.0 2.0 19.0 70 Hainan 26.9 18.0 3.0 21.0 65 9.9 Sichuan 25.4 19.5 3.0 22.5 95 22.3 Yunnan 26.9 21.0 3.0 24.0 71 22.3 Shanxi .. 20.0 2.0 22.0 95 Simple average 22.9 20.8 2.7 23.5 79 17.2 Municipalities Beijing 36.0 22.0 5.0 27.0 75 18.0 Tianjin 32.5 25.0 4.0 29.0 Shanghai 40.5 21.0 4.0 25.0 Harbin 26.2 22.0 2.0 24.0 102 21.3 Changchun 27.7 22.0 2.0 24.0 75 Shenyang 38.5 22.0 3.0 25.0 83 Taiyuan 24.5 20.0 3.0 23.0 100 Wuhan 36.9 26.0 3.0 29.0 90 Chengdu 14.0 22.0 5.5 27.5 83 Chongqing 33.1 27.0 3.0 30.0 60 Guangzhou .. 24.0 2.0 26.0 Shenzhen 3.3 13.5 7.5 21.0 61 Simple average 28.5 22.2 3.7 25.9 81 Sectors Civil aviation 5.9 8.0 2.0 10.0 Coal mining 28.7 21.5 3.0 24.5 88 25.0 Communication .. .. .. Construction 19.7 10.0 3.0 13.0 83 Banking 10.7 12.0 2.0 14.0 Electrical Power .. 13.0 2.0 15.0 80 Hydropower .. 18.0 1.0 19.0 Nonferrous metal 29.3 .. .. .. .. 26.8 Petroleum & natural gas 13.9 12.0 3.0 15.0 85 Railway 29.4 .. .. 17.0 Transportation .. .. 2.0 .. 80 Simple average 19.7 13.5 2.3 15.9 83 Source: Background materials provided by the provinces, municipalities, and relevant ministries.  - 11 - 1.32 Moreover, effective contribution rates can also vary because of the differences in the base used to calculate actual contributions. For example, in Shenyang, enterprise contributions are 18 percent of wages plus 40 percent of the pension expenditures of the enterprise, for an average effective enterprise contribution rate of 31.2 percent. In Jilin province, the enterprise contribution is 21.5 percent of both wages and pensions, pushing the effective contribution rate as high as 50 percent for enterprises with high pension bills. 1.33 Payroll taxes for pensions are below 20 percent in most countries in the world. For example, in 1995 the tax rate was 12.4 percent in the United States, 18.6 percent in Germany, and 16.5 percent in Japan. For developing countries in Latin America, the Middle East, Sub- Saharan Africa, and Asia, the regional averages are about 10 to 12 percent. Singapore had an exceptionally high rate of 40 percent, but the funds were mainly used for housing loans by workers. 1.34 With high and increasing contribution rates for SOEs, compliance rates are declining. Many municipalities and provinces have reported declining compliance rates, from 90 percent for the early 1990s to 70 or 80 percent in 1994 and the first half of 1995 (see Table 1.4). One reason for falling compliance is that an increasing number of SOEs is in financial difficulties. Some are running at a loss, and others have completely stopped production. In Jiangsu, a prosperous coastal province, 40 percent of SOEs are running at a loss, and the situation is worse in the Northeast and inland regions. Many of these SOEs hiave negotiated to delay pension contributions, but it is not clear how many will eventually resume paying contributions. Another reason for worsening compliance is the lack of legal framework and enforcement power of local social insurance agencies. China does not hiave a social insurance law, and consequently, contributions are not defined by law, but by provincial and local regulations. The social insurance agencies at the provincial and local level do not have the legal power to enforce payment of contributions or to take nonconitributinig enterprises to the court when necessary. TABLE 1.4: REPORTED COMPLIANCE RATES IN SELECTED MUNICIPALITIES Compliance rates (%) /b Contribution rates (%) /a Earlier Municipality/Province (enterprises, workers) years Current Beijing various (19-27, 5) 95 95 Tianjin various (20-30, 4) 95 Shanghai 30 (average 25.5.4) 90 Shengyang 21 (18,3) 80 Changchun 23.5 (21.5,2) 91 76.9 Nanjing 21.5 (18.5, 3) 80-90 80 Wuhan 29 (26, 3) 90 Taiyuan various (18-25, 3) 80 Chengdu 24 (22, 2) 80 Chongqing 30 (27, 3) 70.2 Fuzhou various (21-29, 4) 95 90 Guangzhou various (21.5-24.5, 2-3) 96 declining Hainan province 21 (18,3) 70 /a Contribution rates often vary among different ownership. /b Compliance rate is defined as percent of enterprise contributing, as reported by local officials. These may have been overestimated since SOEs in financial difficulties can negotiate for delayed payment, which might not be considered as noncompliance. Source: Background papers and interviews with local officials during the mission.  - 12 - Inadequate Scope for Pooling and Portability 1.35 Currently, most pension pooling is conducted at the county, municipality or prefecture level, and many localities have separate pools for SOEs, COEs, or some other combinations of various population groups (Table 1.5). Nine of the 30 provinces have experimented with some form of provincial pooling. Four of them-the three provincial-level municipalities, Beijing, Tianjin, and Shanghai, and Hainan province-have the most complete form of provincial pooling, with a uniform plan and contribution rate for all urban workers. Five other provinces have implemented partial pooling, either by setting up a provincial readjustment fund or by separating SOEs from other enterprises. The contribution rates in the five provinces still differ across localities and by different ownership types. Most cities and counties have a number of separate pools. The enterprises generally remain responsible for record-keeping and the actual delivery of pension benefits. And in many places, the pooling is only partial: enterprises still pay a higher contribution if they have a larger proportion of retirees. 1.36 Portability of pension benefits is still uncommon and difficult. The cities of Shanghai, Changchun, and Chengdu have portability among their work units, but many other cities do not. Portability across work units within the pool is allowed in several provinces such as Liaoning, Jilin, Zhejiang, Fujian, Hubei, Guangdong, and Hainan, but not in several others such as Heilongjiang, Henan, and Sichuan, and there is practically no portability between provinces. Even where portability is allowed, it usually requires separate negotiations between the old and new employers or pools, and agreement is not easy to reach in an environment where enterprises care about costs and profits. Why is China's Pension System so Fragmented? 1.37 The variety in the pension system in China reflects to some extent the variety and complexity of the country's economy. However, the fragmentation has become excessive, because of a lack of clear direction from the center and the vested interests of the localities in differentiating their schemes. Fragmentation in China's pension system is attributable to the wide dispersion of authority over pension policy making and administration. Multiplicity of central ministries in charge of various types of pensions is leading to inconsistent policies. MOL is responsible for pension provision and administration for employees of, mostly public, enterprises; the Ministry of Personnel (MOP) oversees pension and other benefits for civil servants4 and employees of social (nonprofit) organizations; and the Ministry of Civil Affairs (MOCA) is in charge of social welfare programs, including supplementary pension schemes in rural areas. Each province and county has corresponding bureaus or departments that report to the central line ministries as well as to the provincial or local government. These departments have similar divisions of labor with respect to pension schemes for different population groups. 1.38 This dispersion of responsibility among various government agencies has become one of the major impediments to the establishment of unified pension system. After nearly ten years, pension reform efforts remain largely uncoordinated. In addition to the three ministries mentioned above, other government agencies such as the System Reform Commission (SRC), State Planning Commission (SPC), and Ministry of Finance (MOF) have also contributed to the design and implementation of various reform plans. All these ministries have a vested interest in 4 This practice is not uncommon around the world.  - 13 - implementing their own plans in various localities. Officials in these ministries and their local bureaus not only disagree with one another, but often have a confrontational and uncooperative attitude in their dealings with one another, which makes effective implementation and supervision difficult. TABLE 1.5: LEVELS OF PENSION POOLING IN SELECTED PROVINCES Province/ Separate pools by ownership Municipalities Pooling at which level within each pool Pension administration Beijing Province-level municipality One pool but different Dept. of Labor pooling contribution rates by ownership Tianjin Province-level municipality One pool but different Dept. of Labor; Social pooling contribution rates Insurance Company Shanxi Provincial pooling for SOEs For SOEs only Dept. of Labor in provincial government. Liaoning City/county level pooling Separate pools for SOEs, Unified administration for 6 COEs cities; fragmented for 8 cities Jilin Provincial pooling for SOEs Separate pool for COEs at Social Insurance Company and JVEs counties Heilongjiang City/county level pooling Separate pools for SOEs and Dept. of Labor COEs Shanghai Province-level municipality Uniform contribution rate for Bureau of Social Insurance pooling all enterprises Jiangsu City/county level with Prov. One pool in each city/county Bureau of Social Insurance and Readjustment Fund D of Labor Zhejiang City/county level pooling SOEs and COEs are in the SS Bureau at the local level same pool Fujian Complete Provincial Pooling A pool with different In Luoyuan, a general Social (1984-93) with unified fund contribution rates for SOE and Insurance Company was in use; now a contractual FlEs. COEs has separate pools charge of all subpools provincial pooling with at city level Readjustment Fund Henan Provincial pooling for SOEs Contribution rates differ by Bureau of Social Insurance and and COEs localities and ownership types Dept. of Labor Hubei City/county level Pools covered most SOEs and Dept. of Labor in provincial some COEs government Guangdong City/prefecture level with Unified pool at each Bureau of Social Insurance Provincial Readjustment Fund. city/prefecture, FlEs & private uncovered Hainan Provincial pooling Uniform contribution for all Bureau of Social Security urban workers under Dept. of Labor Sichuan Province pooling with a Uniform at city and prefecture Dept. of Labor pension pooling fund set up level, but not for the whole province. Yunnan City/county level Separate pools for SOEs and Bureau of Social Insurance COEs Note: In many cases, record-keeping and pension a ments are still administered by enterprises; enterprises still remain responsible for "welfare" of the retirees, including health care provision. Source: Compiled based on background papers from provinces.  - 14 - 1.39 Another impediment is the excessive dispersion of authority to lower levels of government in pension reform. Pension policies in China are in effect made by provincial and local governments rather than by the central government. By allowing lower levels of government to select from Plans I and II, the central government is relinquishing its authority in pension provision. Provincial and local governments have modified or combined elements of Plans I and II, essentially making decisions on all system parameters of pension provision. Many localities are inclined to differentiate their schemes from others and intentionally introduce nontransparency so as to retain authority over their own program and the surplus funds that are emerging in the short run. As a result, there are now hundreds of separate pension schemes in different localities in China. Instability of the System 1.40 The frequent changes in basic policy and continuous experiments with various pooling scheme have created uncertainties about future contribution and benefit rates. State Council circular 6 (March 1995), by allowing localities to chose between Plan I and 11, has even increased uncertainty. The situation has deteriorated to the point where, in one province the official position (as stated by the governor) favors Plan 1, but working-level officials believe that Plan II will actually be implemented. Such inconsistencies and uncertainties adversely affect the economy at the individual, enterprise, and national level. 1.41 At the individual level, pensioners and workers are uncertain about the source of their livelihood at a vulnerable stage of their life. Pensioners do not know whether their pensions will be fully indexed to wages or inflation. Workers do not know what replacement rate they will get when they retire. 1.42 The uncertainties about the pension system are also causing major problems of planning for enterprises, particularly in the nonstate sector which cannot count on the state for support in case of financial difficulties. A pension contribution rate of 20 to 30 percent of the wage bill is a major tax. A potential foreign investor will be discouraged not only by the high contribution rates but also by uncertainty about what that rate will be in the future. 1.43 Pension system uncertainty is equally unsettling for the financial situation of municipalities and provinces, which will have to bear the burden of deficits if the pension pools cannot meet the pension expenditures. If fully funded individual accounts are set up (instead of notional accounts), the localities may have a large pool of pension funds to manage, which will require sophisticated fund management skills. Even at the national level, the pension system adopted could have major implications for fiscal and monetary policies. Thus, uncertainty about China's pension system is an impediment to forward-looking programs at all levels of society. ADVERSE IMPLICATIONS OF THE PRESENT PENSION SYSTEM 1.44 The current pension system does not tackle the short-term problem of pensions in SOEs nor the long-term problem of old age security in China. Nor is the system helping in accelerating SOE reforms or in economic development of the country. SOE Problem Remains Unresolved 1.45 In the current system, enterprises still carry a large part of responsibilities for their retirees. In many cases, the contribution rates are linked to their pension payments. In most cases, enterprises keep records, pay pensions, and take care of pensioners needs, including  - 15 - housing and health. These enterprises carry a heavy burden of administration for the elderly, and the liquidation of enterprises will necessitate making alternative arrangements for the pensioners. The incomplete reforms in the pension system are creating a new bureaucracy without really achieving the objective of relieving enterprises of responsibility for the elderly. 1.46 Because of the fragmented system, state enterprises in some sectors (such as banking, civil aviation, and electric power) pay less than 15 percent of wage bill as pension contribution, while others in many localities pay more than 30 percent. The nonstate sector is inadequately covered and generally pays much lower contribution rates than SOEs. The higher and increasing contribution rates of SOEs are leading to lower compliance rates that tend to increase the burden on enterprises that comply. There is thus a vicious cycle in contribution rates and coverage-the higher the contribution rates, the lower the coverage, and the lower the coverage, the higher the rates for those that are covered. Thus the fragmentation of the pension and its uneven coverage are at the root of SOE problems of high pension burdens. The present reform programs are not correcting this basic problem. The Long Run Pension Problem Remains Serious 1.47 The current system of pensions is geared to PAYG principle and is unable to accumulate substantial reserves. The individual accounts that are being set up are largely notional (see Box 2.2); they contain little if any assets. Thus when workers retire and start to draw annuities based on their individual accounts, the annuities will have to be paid on a pay-as-you-go basis, out of contemporaneous contributions. The contribution rate will have to rise steadily at that time to meet those payments. These notional individual accounts will not solve the problem of the rapidly aging population. 1.48 With small beginnings in funded personal accounts and the necessary surpluses in operating accounts, total pension funds in the country are now estimated to have about 30 billion yuan in reserves. However, present regulations require that 80 percent of these funds be invested in government bonds and the rest kept as bank balances. Rates of return on these balances and government bonds were below the inflation rate in the last two years which means that these reserves lost value over time. Such low rates of return create strong incentives for localities to avoid the regulations and invest their funds in other projects. Apparently, only 50 percent of the reserves are invested in government bonds and bank accounts. The rest are invested in various local projects with, sometimes, much higher rates of return. Thus there is a strong incentive for localities to move toward nontransparent forms of investment, which may fail to maximize rates of return on a national basis, by keeping the resources in the localities. A more liberal system would encourage investment in a transparent manner and maximize rates of return on a national basis. 1.49 The implications of the low rate of return on pension funds are serious for retirees who depend on their individual accounts. With a contribution rate of 10 percent of wages, the replacement rates will be over 40 percent if real interest rates and growth rates of real wages are both 8 percent. However, if real wages grow at about 5 percent and real interest rates are 0 percent, the replacement rate will be only 10 percent (see Box 1.2).  - 16- Box 1.2: WHY IT IS IMPORTANT TO HAVE A HIGH RATE OF RETURN ON PENSION FUNDS When a plan is funded, the accumulated funds earn a monetary return that helps to finance future pensions. This is an advantage that funded plans have over pay-as-you-go plans. In a market economy, this financial advantage corresponds to the real return from the productive investments enabled by the pension funds, that increase economic growth. But problems arise if the rate of return is less than the rate of wage growth, and even more so if it is less than the rate of inflation. Suppose that wages in the economy are growing at 4 percent and wages of the typical worker are growing at another I percent because of increased experience and skills as the worker ages. So wages for the typical worker grow at the combined rate of 5 percent annually. Suppose further that the interest rate is only 2 percent. In that case, accumulated pension assets are not increasing as fast as wages and will not yield a high replacement rate of final year salary when the worker retires. Table I shows that a 10 percent contribution rate will only yield a 16 percent replacement rate under these circumstances. If a higher replacement rate, say 32 percent, is desired, the worker will have to save 20 percent of wages each year, instead of 10 percent. On the other hand, if the interest rate increases to 5 percent, the worker will get a 34 percent replacement rate from the 10 percent contribution, because his capital accumulation is growing much faster now, keeping up with the growth rate of wages. In general, a pension system will be in financial trouble if the rate of return on its investments falls below the rate of wage growth (see low replacement rates below diagonal in the Box Table). Workers will be disappointed with the low replacement rate they get after years of retirement saving. If capital markets are operating, the low return on investments also indicates that these investments are not increasing output in the economy by a large amount. The situation is even worse if the nominal interest rate is less than the inflation rate, so the real interest rate is negative. This means that the purchasing power of the capital accumulation is declining rather than expanding. If the inflation rate is 20 percent and the interest rate is 10 percent, next year the pension fund has only 92 percent as much purchasing power as it did this year. The funds are gradually being dissipated. By the time the worker retires, the pension that can be paid will be far less than his wage and will purchase far less than it could have if the money were spent in earlier years. If the interest rate is low because the fund is forced to lend to the Government at specified rates, this constitutes a nontransparent tax on workers to finance government expenditures. Whether these expenditures were good or bad, or whether they increased or decreased economic growth, from the viewpoint of pension plan finance, funding does not make sense under these circumstances. A pay-as- you-go scheme, where the contribution is transferred to retirees as soon as it is received, makes more sense if the interest rate is less than the rate of inflation. Box Table: Replacement Rate From a 10% Contribution Rate /a Real interest rates Real wage growth 0 2 5 8 Replacement rates 0 22 40 103 225 2 16 27 65 160 5 10 16 34 78 8 7 11 21 43 LZ This calculation assumes that the individual works for 40 years and has 20 years of expected retirement. A zero inflation rate or a pension that is indexed for inflation is also assumed. Administrative costs are ignored. Real wage growth = economywide wage growth plus age-earnings growth for the individual worker.  - 17 - 1.50 The Chinese official research group5 analyzing the pension burden over the long term came to the following dramatic conclusions: * A PAYG system will require a contribution rate of 39.27 percent in 2033, when the dependency ratio reaches its peak. * A fully funded system, with additional contribution rates for meeting the needs of transition cohorts, will require a contribution rate of 34 percent from 2004 to 2031. * A combination of a social pool and individual accounts can smooth the rise in contribution rates, but the rate will still remain high at 28 percent for 2001-2050. These projections are somewhat flawed because they do not take into account the effect of rural- urban migration. However, the basic conclusion about the nonviability of the current system is confirmed by more elaborate analysis (see Chapter 3). 1.51 The situation is going to be worse for localities with high dependency ratios (Table 1.3). The officially projected contribution rate in 2020 is 38 percent for Tianjin and 42 percent for Shenyang. Clearly, such rates will render most enterprises nonviable, particularly nonstate sector firms competing internationally and working under hard budget constraints. The regional disparities in contribution rates implied by these scenarios will also be economically and socially unacceptable. The present trends therefore cannot be allowed to continue. Some innovative solution to the problem must be found. Loss of Economic Efficiency 1.52 Impediment for Labor Mobility. If China is to establish a socialist market economy, factors of production (including labor) must be able to move from one sector and region to another. This is particularly important to make restructuring of SOEs feasible. If a steel mill or textile mill has to be closed down in one city, the redundant labor ought to be able to move to other municipalities where similar industries might be expanding. For that purpose, it is essential that the pension benefits of workers be portable. Since individual accounts are largely notional, there are really no funds in personal accounts for a departing worker to take to a new job or location, even if portability were permitted (see also para. 1.36). 1.53 This is a serious handicap for SOE restructuring and industrial relocation, both of which are priority issues in China today. For example, in the Government's program to help the development of western provinces, some textile mills are to be relocated from the coastal provinces to western provinces. While many manual workers for these mills will be found in the host provinces, many supervisory staff may have to come from coastal provinces. If these workers lose their pension benefits when they move, they would be unwilling to do so. Similarly, if the surplus workers in the Northeast are to move to the southern provinces, the southern provinces would be reluctant to provide them with full pension benefits out of their resources. Only if pension rights are portable will workers, particularly those close to retirement, be able to move. 5 "Projections for Primary Pension Insurance for Urban Staff and Workers," Research Group for Research on Social Security System, February 1995. See Technical Annex TA. 1.  - 18- 1.54 Lack of a Level Playing Field. Under the present system, two enterprises (for example, two electronics factories) in two provinces may have to pay widely different payroll taxes that could range from 20 to 40 percent. This is equivalent to imposing a value added tax at widely different rates for the same product. Thus an enterprise may lose competitiveness not because the efficiency in its core business is low but because it is in a locality with many retirees. Thus the system will end up allocating resources to enterprises that are not necessarily more efficient but have the advantage of a location with a younger population. China's Northeast region is already experiencing low growth and has a higher percentage of elderly people. If it has to bear a higher tax for pensions, it will suffer a competitive disadvantage relative to other regions, and its economic performance will drop further. 1.55 Missed Opportunity of Term Transformation. One important objective of pension system reform is to help in capital accumulation and productivity growth of the economy. The present system of pension in China is not helpful for either. Ultimately, an aging population will have to be supported by the output of the active working population. The pension system is important in determining the cost of transfers from workers to retirees (whether through families, the Govemment or the pensioner's own account). Even more important is the effect the pension system has on the growth rate of the economy. Growth rates, compounded over the typical working life of 40 years, can make an enormous difference to the output per worker. This influence workers through the rate of saving in the economy and the productivity of capital. 1.56 China's saving rate is currently very high. However, with stabilization in GDP growth rate, spread of conspicuous consumption, and aging of the population, these high rates are likely to come down in the longer term. A pension system geared to funded individual accounts can help to motivate households to save more, but this effect is muted in China's present system with its small, notional individual accounts. 1.57 Even more serious are the missed opportunities for term transformation of savings. China has enormous needs for infrastructural and other long-term investments. Demand for infrastructure investment alone is projected to be as high as $744 billion for 1995-2004 or 7.4 percent of GDP (World Bank 1995e). However, because long-term instruments are unavailable, most household savings are in short- and medium-term deposits, which do not provide a solid basis for long-term lending. Thus foreign financial resources are often mobilized for infrastructure investments (often at high guaranteed rates of return) while domestic deposits are left underutilized. Funded pension funds can create an enormous pool of resources (4 to 6 percent of GDP) for long-term investments thus giving support to domestic infrastructure investment and a high rate of return for future pensioners. The present system of unfunded notional accounts is missing that opportunity.  - 19 - 2. KEY ISSUES AND OPTIONS FOR PENSION SYSTEM REFORM 2.1 The problems with the present pension system are widely recognized in China. From extensive discussions about how to solve these problems and design a new, more viable pension system, broad directions of reform are becoming clear among policy analysts in China. Implementation has proven more difficult, hampered in part by misperceptions about the costs of transition and in part by the fact that change always involves losses for some groups. This chapter notes these directions of reform and then goes on to argue that not only are the costs of transition manageable but that now is the time for making such a transition. UNIFICATION 2.2 State Council Circular No.6 issued in March 1995, which presents a master plan for pension system reforms, calls for unification of the pension system. It stipulates that "all types of enterprises and workers should be treated equally by 'unified system, unified standards, unified management, unified fund usage' (four unifications)." Various types of enterprises and workers now covered under separate plans or not at all would be brought into a single system with common standards. However, as noted earlier by proposing two models and by allowing the localities to chose between them or any combination of the two (see Chapter 1), the document has led China to the opposite direction. Each municipality is attempting to differentiate its scheme from that of others and there are now hundreds of schemes all over the country, with many serious problems. The most challenging task of China's pension system reform is to reduce the fragmentation and make the system and its administration more unified. 2.3 Most industrial countries specify the structure of the mandatory system on a nationally uniform basis, and most have unified publicly managed PAYG pools, though there are some exceptions (see Table A2. I for description of the systems in various countries.) 2.4 Historically, it has often been the case that pension systems have evolved from a fragmented stage to a unified one. Chile, Japan, Korea, Mexico, the United Kingdom, and the United States all had fragmented pension systems in earlier periods, and then moved to unified systems. For instance, prior to the 1985 reform, Japan had three public pension schemes: the Employees' Pension Insurance for the private sector, the Mutual Aid Association for public sector, and the National Pension Insurance for farmers and the self-employed. Benefits from the different schemes varied a great deal. The 1985 reform unified pension insurance by adopting a basic pension system with more equitable benefits and contributions. A transitional mechanism allowed employees to receive pensions from old systems. 2.5 It could be argued that the experience of other countries-smaller and more homogeneous-is not relevant for China, a vast country with large regional differences. But regional disparity is not unique to China; significant regional differences also exist in large countries such as the United States and India (see Table A2.2). Moreover, China is a unitary country rather than a federal one, and the legal powers of the central authority are more extensive than they are in a federal system (such as the United States). The Communist Party structure  - 20 - provides a powerful instrument for unification when necessary. The introduction of the National Tax Service (breaking a centuries-long tradition of local-based tax collections) and of center- local tax sharing system shows that the central authorities can move on unification when necessary. The need for unification of the pension system is strengthened at this juncture in China, when there are some signs of centrifugal tendencies in economic (and other) spheres. 2.6 Unification is essential for the establishment and implementation of a legal framework. With the pension system being extended to the nonstate sector and contribution rates so high (close to VAT collections), compliance from enterprises (particularly private companies and joint ventures) will be difficult unless there is clear legal authority behind the system. The present effort at drafting a social security law has been frustrated by the fragmentation of the system, and variable coverage and contribution rates across regions. 2.7 One way to resolve this dilemma is to make a sharp distinction between the mandatory and voluntary parts of the pension system and between its overall structure and its detailed parameters. The basic structure of the mandatory part of the system needs to be unified, while the detailed parameters and the supplementary part can take into account specificities connected with regional or occupational needs. This is consistent with practices in most industrial countries that specify the structure of mandatory systems on a nationally uniform basis. Chapter 3 gives some recommendations for China in this regard. REDUCING PENSION COSTS 2.8 Increasing the Age of Retirement. The current retirement ages of 60 for men and 55 for women were determined years ago, when life expectancy was about 50 years. Today it is 71 years. Both for the viability of the pension system and to avoid the loss of skills to the economy implied by premature retirement, the retirement age may gradually be increased to, say, 65 years. The issue is however controversial. Many SOEs are facing surplus labor problems, and it is tempting to argue that the problem can be relieved by retiring elderly workers early. However, the cost to enterprises of retired workers is high. Better to extend the mandatory age of retirement (thus saving on the pension burden) while dealing with labor redeployment and severance pay on a case by case basis. Often, it would be cost-effective to reduce surplus labor by redeploying younger and middle-aged workers rather than retiring elderly workers early. 2.9 The view that retirement age should be increasing is gathering strength in China from the findings of recent surveys that show that a significant percentage of retirees do in fact continue work. These findings are in line with international experience, which indicates a trend toward flexible schedule on retirement. (See Box 2.1) 2.10 Reducing the Replacement Rate. Replacement rates in China's pension systems are high by international standards and comparison with rates before the Cultural Revolution. The replacement rates for cash wages are over 80 percent and the in-kind benefits (housing, medical benefits, and so on) continue at the preretirement level, probably resulting in a 80 to 90 percent replacement rate of the total compensation package. This contrasts with a replacement rate of 40 to 60 percent in most countries (see Table A2.3) and 50 to 70 percent in China before the Cultural Revolution.  - 21 - Box 2.1: ILO FINDINGS ON PARTICIPATION RATE OF THE ELDERLY AND FLEXIBLE SCHEDULE ON RETIREMENT The latest labor force participation rates for older men and women in different age groups for 18 industrialized countries are shown in the table below. As can be seen, the participation rates for men aged 55 to 59 vary greatly, from 94.1 percent in Japan-followed by several other counties with rates of over 80 percent-to 59.2 percent in the Netherlands. Several countries have participation rates between 60 and 70 percent for this age group, that, compared with prime age participation rates of over 90 percent, already represent a sharp decline. Sweden has the highest overall participation rate because 77.2 percent of all Swedish women in that age group are also active. The activity rates for women are also high in other Nordic countries, over 60 percent, but are only around 20 percent in Austria, Ireland, Italy, the Netherlands, and Spain. Box Table: Labor Force Partici ation Rates Country Age Groups Age Groups Men Women 55-69 60-64 65-69 55-69 60-64 65-69 Australia 71.8 48.7 8.4/a 37.0 15.4 2.3La Austria 63.0 12.7 3.6 23.8 5.2 1.2 Canada 73.6 47.6 16.6 47.5 24.8 7.8 Denmark 81.9 47.1 24.6 64.5 26.8 8.0 Finland 62.7 23.9 7.1 62.1 18.0 3.1 France 69.3 18.2 4.5 47.8 15.1 3.2 Germany 81.5 34.9 8.0 45.5 11.9 4.0 Ireland 79.8 59.4 26.5 22.1 14.0 5.9 Italy 68.9 37.2 12.6/b 21.1 10.0 3.9/b Japan 94.1 75.6 55.3 56.4 40.1 28.0 Netherlands 59.2 18.0 - 19.3 4.2 - New Zealand 80.3 39.0 8.7/a 49.3 20.2 2.7/a Norway 81.8 61.5 25.8 61.5 45.8 16.7 Portugal 71.9 53.0 29.9 40.1 25.6 14.6 Spain 73.6 44.8 6.3 24.4 16.2 3.9 Sweden 82.5 57.8 - 77.2 49.1 - United Kingdom 75.7 52.2 13.1 54.5 24.7 8.0 United States 77.4 54.9 25.8 57.0 37.4 16.0 La This age group includes all people over the age of 65. /b This age group covers people between the ages of 65 and 70. Note: The data presented in this table have been drawn from the labor force sample survey and the country concemed, except France (from official estimates). Source: ILO: Year Book of Labor Statistics, 1994, 53rd issue, 1994. It is the 60 to 64 age group, however, that indicates most clearly the extent to which older men have dropped out of the active labor force before reaching retirement age. The differences between countries become more clearly marked: Japan with over 75 percent has the highest rate, followed by a group of countries with rates of around 50 to 60 percent (Ireland, Norway, Portugal, Sweden, the United Kingdom and the United States); at the other end of the scale is a group of countries with rates below 20 percent (Austria, France, and the Netherlands). For the over-65 age group, participation rates are low for all countries. Only in Japan does the rate remain over 50 per cent, and a few other countries have rates of around 25 percent (Denmark, Ireland, Norway, Portugal, and the United States).  - 22 - Box 2.1 continued These statistics on the economically active population in their aggregated form do not provide information on how many older workers are actually employed. Two researchers (Klaus Jacobs and Martin Rein) have desegregated such data for 1989 from seven OECD countries according to sectoral employment, self-employment, part-time work, and unemployment. They found that older men (aged 60 to 64) working in agriculture were particularly likely to remain in the labor force. Such workers accounted for 26 and 35 percent of total employment in that age group in Japan and France, respectively, which represented about four times the average rate for all age groups in the agricultural sector. In Sweden, the United Kingdom, the United States, and Germany only 5 to 14 percent of the 60 to 64 year-olds are in agriculture, which was still double the average rate. Similarly, older workers were overrepresented in self- employment as compared with younger cohorts. The highest self-employment rates for older men were found in France, Japan, the Netherlands, and Germany. Among older wage and salary earners, the proportion of part-time workers was high in Sweden, the Netherlands, and the United States (30, 26 and 19 percent respectively), but below 10 percent in all other countries. One of the first concerns of older workers is to gain better control over the transition from work to retirement. As already discussed earlier, in many cases the transition is abrupt because of mandatory retirement or induced early retirement. The exclusion of older workers in this way causes undue anxiety and can have negative effects on their health and social identity. Phased retirement, a gradual reduction of working hours and transfer to lighter work are all ways of easing the transition from work to retirement. Phased retirement is perhaps the formula best adapted to the needs of older workers because it permits them to work part time on their career job. It is different from the many part-time jobs open to older people in the precarious job market. Phased retirement has been particularly successful in Sweden, where it was introduced in 1976. Attempts to introduce it in other countries have not been successful for a variety of reasons. One of the main reasons is that the replacement income for the partial loss of earnings was insufficient. Furthermore, in countries where the level of old-age pension is based on the last years of earnings, workers are likely to be reluctant to take up such schemes as they would lower their final pensions. In Sweden the partial pension received during phased retirement is assessed as pensionable income for the state earning related pension, which is based on the best 15 years rather than on the last years of working life. As the benefits of phased retirement are now widely recognized, new attempts are being made to facilitate its introduction. sometimes, phased retirement conditions are collectively negotiated. British Airways, for example, introduced phased retirement and overcame the "final salary" rules by allowing pension entitlements to be calculated on the basis of the full-time pro-rata salary. What many older workers want is something like a decade of retirement during which they can choose when to retire, whether to retire gradually and whether to continue some form of work up to definite retirement. Increasing the options for flexible retirement would go a long way towards breaking the dichotomy of society into active and inactive, giving older workers the opportunity to remain active longer and remain a part of the active population. In the long run it could help to reduce possible social tensions. Source: ILO, World Labor Report, 1995. 2.11 There is broad consensus that over time, the replacement rate should gradually be brought down to about 60 percent of wages net of payroll taxes. This would also create incentives for retirees to continue in productive employment in one form or another; encourage workers to save in their own accounts, thus keeping up the national savings rate; and keep some pressure on children to provide support to retired parents, which would also provide the bonding and emotional support so much needed by the elderly. 2.12 Indexation of Pensions. The current approach (also emphasized in the draft Social Security Law) seems to provide for indexation of pensions to wages. This allows retirees to enjoy the benefits of current labor productivity growth and facilitates adjustment to the transformation of in-kind benefits to cash wages. However, in most cases, the indexation  - 23 - coefficient to wage growth is nontransparent. For the provinces and cities for which data were made available by the Chinese authorities, this coefficient ranges from 40 to 80 percent. This is, of course, a very uncertain form of indexation. For example, when nominal wages rose by 30 percent in 1994, and the overall price index by 23 percent, a 40 percent indexation to wage growth meant a reduction in real pensions of 11 percent. If wages were instead growing at 25 percent a year and inflation were 8 percent a year, a 40 percent indexation would mean that real pensions were rising at 2 percent a year. The issue of indexing of pensions is a complicated one. On the one hand, it could be argued that the pensioners should share the increasing prosperity of the workers, (although it may be hard to persuade pensioners to share the decline in incomes of the current workers). This would maintain horizontal equity and the pensioners would not feel deprived in relation to the current workers. On the other hand, it could be argued that the pensions are implicit compensations for incomes withheld (implicitly or explicitly) during the working life of the pensioners and pensions related to real wages during that period are appropriate. In some countries such as Switzerland, pension indexation is based on the arithmetic mean between price and wage inflation, which is equivalent to linking pensions to 50 percent of the real wage growth. In China's conditions, real wages are expected to rise fast ( 5 percent or over for 10 years or more) and indexation to real wages will increase the pension burden (including transition costs) significantly. With the expected sharp rise in system dependency ratios by 2030 and the difficulties of finding finance for transition, it seems advisable to lower the expectations of pensioners. If funding situation improves, additional pensions in line with real wage increases can be granted at a later stage. However, at this stage, a more affordable adjustment mechanism, in line with general international practice and the stated objectives in many localities would be full indexation to the Consumer Price Index (CPI). EXTENDING THE COVERAGE OF THE PENSION SYSTEM 2.13 At present, coverage of the pension system is low for the nonstate sector. In particular, township and village enterprises (TVEs) are covered along with farmers, under voluntary pension schemes organized by MOCA (see Box 1.1). The system provides very little old age security for the TVE workers. Many township enterprises are now located near cities, have fixed employment, and their workers are losing their links with farming. These workers now have all the characteristics of formal employment and should be covered in the formal pension system. This would be desirable for the old age security of these workers, and also for the pension system as a whole, to assist in its transition. 2.14 The problem is that these enterprises with low pension coverage are also the most dynamic ones. At a time when economic development and job creation are major concerns, any policy that would produce a major increase in labor costs for these dynamic enterprises must be pursued with caution. The high contribution rates, currently prevailing in the state sector, cannot simply be extended to the nonstate sector, including township enterprises. The key to resolution of this conflict may lie in making the mandatory contribution rates low (and extending them to the nonstate sector) while allowing for the special needs of certain occupations (such as coal mining or civil servants) through a supplementary pension scheme. 2.15 Administrative responsibility needs to be rationalized as well. MOCA, which is currently in charge of pension issues for TVEs, may be reluctant to cede that responsibility to another existing ministry, such as MOL. If, however, a new national agency for social insurance is created and it assumes pension responsibilities from all ministries, such as MOL, MOP, and sector ministries, it may be easier also to effect the transfer of responsibilities for TVEs from  - 24 - MOCA to the new agency. The coverage to Township Enterprises (TEs) would of course be increased slowly. A beginning can be made with enterprises with more than 50 workers and the coverage gradually extended. FUNDED INDIVIDUAL ACCOUNTS 2.16 There is growing consensus in China (reportedly reinforced by strong policy directives from top economic management) that a substantial part of pensions (say half or more) should come out of fully funded individual accounts. The advantages of individual accounts when the population is aging rapidly and economic growth prospects are uncertain are well known (World Bank 1994), although empirical evidence on these advantages is not robust across countries. As noted by Chinese official forecasts (see Technical Annex 1), the PAYG system, if maintained, will require very high (38 percent) contributions by 2030, when economic growth rates in China may be slowing down. This may lead to high rates of evasion by enterprises and a high burden on society. Funded individual accounts would shift the burden of savings to current workers. This would instill a spirit of self-help among workers, who would be looking after their own retirement costs and not depending on their children's contribution (whether in a family, enterprise, or social transfer context). In the implementation phase, individual accounts create incentives for workers to ensure that enterprise contributions are made and that fund managers maximize rate of return on pension funds. 2.17 However, individual accounts do not serve their full purpose-individual or social-if they are notional (see Box 2.2). The link between individual contributions and individual benefits exists, but there is no opportunity to maximize the rate of return, which becomes administratively determined. Nor is the objective of term transformation of savings achieved. The most important problem with notional accounts is that contribution rates will have to rise dramatically to pay for the promised pensions as the population ages. The problems stemming from the demographic transition will not be avoided. 2.18 Thus the present consensus in the direction of substantial funded individual accounts is appropriate. The system is not without risks, however, both to the individual and to society. Investment fund managers can make mistakes, investment companies can go bankrupt, and individuals may be left without much pension in the end. Also, individuals with low wages or infrequent employment may not have much in individual accounts at retirement. In either case, the society has to provide protection. Such social protection should generally be pitched at poverty level, below which the society does not want any individual to fall. REFORMING THE FINANCIAL SECTOR 2.19 Individual accounts, however, do not provide reasonable level of old-age security unless interest rates are close to the growth rate of wages. Most calculations of replacement rates on personal accounts made by Chinese official agencies assume the two to be equal. It is also recognized that rate of return on pension funds can be high only if the general interest rate structure is rationalized. In other words, special rates on pension funds can be only occasional exceptions, not the general rule. 2.20 The case for rationalization of interest rates and deregulation of financial sector appears more persuasive to Chinese officials from the perspective of pension funds and savers than from the perspective of investors. The link between low interest rates and loss of real value of pensions is arithmetically demonstrable and is more persuasive in terms of plight of the elderly  - 25 - that reduced pensions will entail. Thus funded individual accounts could become a pressure point for rational interest rate policy. 2.21 The problem lies in the impact of higher interest rates on the financial position of SOEs and the Government. A simulation analysis shows that a 10 percentage point increase in interest rates would cause a net loss of Y 100 billion for state enterprises, and Y 88 billion to the budget (World Bank 1995b). A short run solution may lie in raising interest rates on longer-maturity loans which would be used for longer-term investments. Given the high productivity of capital in infrastructure and other long-term assets in China, borrowers can afford to pay high (at least 3 percent) real interest rates. This report assumes that China will move in the medium run to market-determined interest rates that remunerate savings, which is crucial to the success of pension reforrns. Box 2.2: WHAT IS A NOTIONAL PENSION ACCOUNT? In the past, in almost all countries, individual defined contribution accounts were fully fumded. Contributions made by the worker accumulated in the account earning investment income, and upon retirement the entire accumulation was turned into a pension. However, some cities in China, including Shanghai, are now experimenting with "notional" accounts that are essentially pay-as-you-go. The basic idea behind a notional account is that a worker's account is set up as a bookkeeping device, to keep track of contributions plus imputed interest at a rate determined by the Government, but funds are never accumulated in these accounts. Instead, the accounts are notional or empty accounts. When the worker reaches retirement age, the notional accumulation in his or her account is converted into an annuity and paid to the retiree out of contributions that younger workers are making at that time. For example, suppose that a worker has contributed 10,000 yuan and been credited with 2,000 yuan interest, bringing the total to 12,000 yuan. This would be converted into an annuity of approximately 100 per month. However, her contributions have been paid out as fast as they came in, to finance pensions of other people. So when the worker retires, there are no assets in her account to finance her own pension. Instead, her pension is financed, on a pay-as-you-go basis, by the contributions of other younger workers, who are also developing notional accounts. Notional accounts are very attractive to countries that want to reform but, because they already have a large public pension debt, find a shift to full funding difficult to achieve. For this reason Sweden and Italy are also experimenting with the use of notional accounts. A plan based on notional accounts accomplishes some but not all of the objectives of a reform. It produces a close transparent relationship between contributions and benefits, thereby deterring evasion and other distortionary behavior. It eliminates some undesirable redistribution within the same cohort of individuals. It automatically adjusts retirement age up or benefits down as expected lifetime increase, thereby preventing pension costs from rising as fast as they would otherwise. However, so long as the account remains notional it may not be portable when the worker changes jobs. It will not increase long term national saving. It will produce large intergenerational transfers as a result of demographic change. And the notional interest rate is highly subject to political manipulation-it may be higher or lower than the market rate, implying hidden subsidies and influencing future tax burdens in unexpected ways. Since notional accounts do not accumulate assets or generate investment earnings to cover the promised annuity, they leave governrments with the full responsibility to cover the annuity on a pay-as- you-go basis. For this reason they continue to require a sharp rise in contribution rates as populations age. If China wishes to avoid this sharp rise in payroll taxes, it must also avoid notional accounts. Instead, it must find a way to fill the individual accounts with funds that will be used to build the productive capacity of the economy during the worker's active phase and to pay pensions during the retirement phase.  - 26 - ESTIMATING TRANSITION COST 2.22 As noted in the preceding section, there is a growing consensus in China that the pension system should move towards individual accounts, and these accounts should be funded, not "notional." However, there is no consensus on how to tackle the major problem of financing the move from PAYG to fully funded individual accounts. With the fully funded individual accounts, the current contributions of enterprises and workers go to the individual accounts and are not available for paying the current pensions. So additional financial resources are needed to pay the current pensions and the accrued pension rights of workers who had contributed before the start of individual accounts. The problem then is: how to make these "double" payments? 2.23 This leads to the concept of "implicit pension debt" (IPD), which refers to the benefit promises that a pension scheme makes to workers and pensioners. It is an "implicit" debt because it is not written down like an explicit debt would be and in many countries it is not legally binding. Nevertheless, a social and political obligation exists to provide workers with benefits in their old age. The size of the implicit pension debt is measured by adding up the present value of the benefits that will have to be paid to current pensioners and to current workers who would have to be paid if the system were stopped today. Box 2.3 and Annex 2.1 notes the specifics of how much implicit pension debt is usually calculated and how other countries making the transition are paying the implicit debt. 2.24 Some rough calculations of the implicit pension debt made by Chinese policy analysts suggest that these transition costs could be three to four times GDP. These high estimates have apparently discouraged the authorities from recognizing these debts explicitly. They have instead been emphasizing the approach of repaying the debt by increasing the contribution rates and coverage rate of workers under the pension system. The Government's analysis shows that under such an approach the average contribution rates by 2030 would be in the range of 28 to 39 percent-too high by international standards. Such rates will be clearly unsustainable, particularly for nonstate sector firms. That would lead to widespread evasions and loss of viability for many nonstate enterprises. It is thus important to estimate the transition costs carefully and to explore whether they can be paid other than by taxing the current workers. 2.25 The estimates of the IPD depend on the benefits the pension system intends to provide to the pensioners and the current workers, the age structure of the working population and the life expectancy of the retirees. International experience suggests that the IPD is in the range of 20 to 30 times pension payments (see Box 2.3). China's pension payments in 1994 were about 2.3 percent of GDP, indicating an IPD of 46 to 69 percent of GDP in 1994. These broad estimates are in line with the results of more detailed calculations based on explicit modeling of the benefits and age structure of the population in China (see Chapter 3, Table 3.7.) 2.26 A second set of estimates were prepared for Shanghai and Shenyang, two cities in China which both have a serious problem of aging population in SOE sector. It is noted that for Shengyang the IPD was estimated to be between one-third and one-half of the municipal GDP and for Shanghai, about 100 percent of municipal GDP. The details of these estimates are presented in Technical Annex 2. 2.27 A reasonable conclusion from these calculations is that the IPD for the enterprise sector in China is less than 50 percent of GDP. This is less than it was in countries such as Chile and Argentina, which have made the transition from PAYG to funded systems and where the IPD was estimated to be between 80 and 120 percent of GDP.  - 27 - Box 2.3: THE CONCEPT OF "IIMPLICIT PENSION DEBT" The concept of "implicit pension debt" refers to the benefit promises that a pension scheme makes to workers and pensioners. Pay-as-you-go pension schemes are based on the promise that future generations will pay the pensions for today's workers; in return, today's workers pay contributions to finance current pensions. There is a "debt" or liability for the Government or enterprise running the pension scheme to honor this promise in the future. It is an "implicit" debt because it is not written down like an explicit debt would be and in many countries it is not legally binding. Nevertheless, a social and political obligation exists to provide workers with benefits in their old age. Future pension entitlements are usually not included in assessments of the public sector's financial situation. But as aging populations put increasing pressure on pay-as-you-go schemes, particularly in OECD countries, concern is growing that contribution rates might have to rise to unsustainable levels and that govermments will have to come up for an increasing share of the pension expenditure in the future. To illustrate the magnitude of the pension burden and to compare pension promises across countries, the concept of implicit pension debt has been introduced. The results depend strongly on many economic and demographic assumptions and are thus subject to much debate, but they do give an indication of the financial burden that economies will face in the future due to their pension schemes. The size of the implicit pension debt is measured by adding up the present value of benefits that will have to be paid to current pensioners plus the present value of pension rights that current workers have already earned and would have to be paid if the system were stopped today. Its magnitude depends on several important factors. First, it depends on the coverage of the pension system, i.e., on the number of pensioners and workers to whom benefits have been promised. In countries where pension systems have a low coverage relative to the total labor force, the pension debt is a small proportion of GDP. As coverage of a pay-as- you-go scheme expands, the implicit pension debt will also grow, because many additional workers will be promised future benefits. Second, the implicit pension debt depends on the age distribution of the population. The more pensioners and workers close to retirement there are, the higher is the pension debt. If the workers are young, the debt will be low, both because young workers have not accrued much pension rights yet and because the promised benefit will be paid many years later and its discounted value is therefore small. Third, the debt depends on the level of benefits. Generous pension promises mean higher payments in the future. But, since the debt is implicit and the terms have not been written down, the promises can be cut gradually. Raising the retirement age, reducing the wage replacement rate and changing the indexation mechanism are three methods that countries have used to reduce their pension debt. Finally, the size of the pension debt depends on the discount rate. Discounting simply recognizes the fact that money received next year has less value than money received today, so future monetary streams should be "discounted" to obtain their present value. A high discount rate results in a lower present value of the pension debt. There is much debate on which discount rate should be used for the calculation of the implicit pension debt. For many countries, calculations have been conducted using a rate of 4 percent which is currently the average long term real interest rate on government bonds in the major OECD countries. Some experts argue, however, that the long-term interest rate is higher and that a rate closer to the returns of private pension funds should be used. If beneficiaries regard promised benefits as risk bearing assets, it might be more appropriate to use the average real rate of return on equities. At the same time, the case can be made for a discount rate lower than the capital market returns due to the public provision of annuities in social security schemes. The results of pension debt calculations are very sensitive to the assumed discount rate which should be taken into account in the interpretation of the numbers obtained. Calculating the unfunded liabilities of a pension system further requires a set of assumptions about various factors such as economic and population growth, wage growth and the future rules of the pension system. Vesting rules, i.e., regulations about a minimum contribution period before pension rights accrue, should also to be taken into account. In many countries, however, adequate records on the contribution history of the workers do not exist. One solution to this problem is to use the average pension for new pensioners as the basis to calculate the workers' accrued rights. These rights are then prorated according to the workers' years of service to estimate the amounts that the pension system owes to the workers.  - 28 - Box 2.3 continued The most commonly used method of calculating the pension debt is based on the so-called termination hypothesis, which assumes that the unfunded system would be terminated immediately and all pensioners and workers would have to be compensated for their future pensions and accrued rights. It does not take account of possible new obligations or income from future contributions or interest. Other calculation methods assume that the system is closed immediately but continues to exist until the last current contributor dies taking into account future contributions from and benefits to all current contributors, or assume an open system by estimating the present value of all fiuture pension payments including those to new entrants. Using the terrnination hypothesis, the pension debt ranges between 100 and 200 percent of GDP in most OECD countries with an average of about 130 percent of GDP in the seven major economies. In Hungary and Uruguay, countries which both have pension systems with a high coverage, high system dependency ratios, i.e., relatively few workers to support a growing number of pensioners, and generous benefit fornulas, pension debt corresponds to more than 200 percent of GDP. Estimates of the Chilean pension debt at the time of refonn ranged between 40 and 130 percent of GDP, depending on the discount rate used. In other Latin American countries which have moved from unfunded to funded pension schemes, the pension debt was estimated to be more than 40 percent of GDP in Peru, for example, and around 90 percent in Colombia. In fully funded schemes, implicit pension debt does not exist, since each generation saves for its own retirement. Funded schemes accumulate assets from the start, sufficient to cover fiture pension liabilities. Changing from a pay-as-you-go to a funded pension system makes the implicit debt explicit. Workers' contributions can no longer be used to pay for current pensions; instead, they must be saved for the workers' future retirement. This means that the "debt" owed to current pensioners has to be financed from a different source, unless the Government wants workers to pay a double burden. The financing of the implicit debt, however, can be stretched out over a long period depending on the design of the transition process and on the source of financing the transition gap. In China, the low implicit pension debt should make it relatively easy to reform the system and to finance the benefit obligations to pensioners and workers nearing retirement. If, however, the system is not reforned soon, the pension debt will grow rapidly. As coverage expands and workers age, it will become much more difficult to move from the pay-as-you-go to a more funded system. With total pension expenditure amounting to only about 2 percent of GDP, China is today in a very good position to move to a pension system which is partially or fully funded rather than pay-as-you-go. MECHANISMS FOR FINANCING THE TRANSITION 2.28 Countries that changed from a PAYG to a funded pension scheme used a mix of instruments to pay off the pension debt and bridge the financing gap. The instruments available are the same as those used to finance any other public expenditure, bonds, higher taxes, lower spending and the transfer of public assets. The choice and mix of instruments will detertnine the impact of the pension reform on economic growth and the distribution of the costs and benefits of transition betweert and within generations. 2.29 In China, experiments with different sources of financing for pension expenditures are already under way, although they are meant to provide pension pools with additional resources rather than to finance the transition gap. In Luoyuan county in the province of Fujian, part of the tax on land-use transactions is transferred to the pension pools. In Zhengzhou city, pension pools receive part of the proceeds from the sale of shares of corporate SOEs. In Shantou in the province of Guangdong, enterprises pay a tax of 1.6 per thousand in retail sales or 2.7 per thousand in wholesale reveuL-tes in addition to the contribution rate to the pension pool. In Tongling city, some shates o3 previously slate-owned enterprises were given directly to the pension pool and some were given to workers individually. In Nangyang city, poorly perfortning enterprises pledged a certain proportion of the revenue accruing from the lease of fixed assets,  - 29 - equipment, factory and warehouse space for the payment of pensions and the redeployment of workers. 2.30 Beyond these specific measures, the Government may decide, as Chile did, to cover the financing gap with government bonds. That would mean exchanging the old implicit debt for new explicit debt (see Annex 2.1). If the Government issues bonds to pay for the pension promises to retirees and workers, future generations will have to pay higher taxes to repay the bonds. Debt financing is thus similar to the PAYG system except that it allows the financial burden to be spread out over a longer period of time. 2.31 Much of the demand for new government bonds would come from the pension scheme itself if it is funded, especially in China where the financial sector is still underdeveloped. In Chile, the new pension funds were invested only in government bonds and bank deposits in first years after the reform. As the financial sector developed and more instruments became available, pension funds were allowed to diversify their investments. Debt financing can pave the way for pension funds to contribute to financial sector development and play an increasingly important role as institutional investors. 2.32 Another approach used by Chile that is probably not open to China was to use its budget surplus to cover part of the transition costs. China has some budget surpluses in its current account, but deficits in the overall account of about 2 percent of GDP. Since the Government intends to further reduce its borrowing program, it is unlikely that the budget can provide much funding for transition. 2.33 Another approach is based on the recognition that the IPD went to creation of enterprise assets and that if the debt is to be redeemed, it should be done by using those assets. This principle is now recognized by the Chinese Government in regulations that provide that when an enterprise is liquidated, the proceeds from sale of assets are first used to meet pension obligations. Some municipalities have in fact been considering such uses of assets for solving the pension debt problem. The Government, worried about this phenomenon, has issued instructions prohibiting the use of state assets for pension funds. The exact reason for this concern is not known. However, one possibility is that the Government is concerned that this process will lead to state asset-stripping and an undue portion of state assets will be siphoned off for pension funds, not leaving enough for meeting other obligations and for smooth functioning of enterprises. 2.34 The assets and liabilities of SOEs should indeed be evaluated carefully and as a package rather than piecemeal. The process is already initiated and the results should be available soon. Preliminary assessments suggest that production assets of SOEs are over Y 2 trillion, while "unproductive assets" such as land and housing constitute over Y 5 trillion. The explicit debt of SOEs to banks is less than Y 2 trillion, while pension debts and unemployment debt6 should be less than Y 2 trillion. Thus putting together assets and liabilities, SOEs are seen to be in a solvent position. Moreover, since the state owns the assets and owes the liabilities, an asset- liability swap will be easier now than later when multiple ownership of assets and liabilities will develop. 6 Unemployment debt is defined as the payments that would have to be made to surplus workers if they were to leave the enterprises. For estimation of this debt, see Guo Shuqing (1995).  - 30 - 2.35 While keeping open the options about the sources of funding the transition costs, special attention should be focused on how SOE assets could be used to help in the transition from a PAYG system to funded individual accounts. Two principles may be used in this connection: * First, any asset transferred to the pension funds has to be marketable and carry a value corresponding to the market-determined price. The scheme should not require cashing in a large proportion of assets in the short run. Rather, enterprise assets should be sold gradually, over a prolonged period of time. * Second, the link between asset sales and beneficiaries should be as close as possible. If enterprises are seen as ultimately responsible for pensions, their assets should be the first claim for settling their pension debt. If the assets and pension liabilities are located in the same place, the swap would be easier to understand and to implement. CHINA'S ADVANTAGES IN BEARING THE TRANSITION COSTS 2.36 Not only are China's costs of transition to a funded system lower than in other economies, but its financial capacity to fund the transition is probably greater, for at least four reasons: the unification of pension system, rapid growth of income, structural change of the economy, and capital gains in state assets. These facilitating factors will be present over the next 15 to 20 years, but they will largely disappear after 2030. They cannot be counted on to finance high pension costs indefinitely, but they are ideally suited to finance the temporary needs of the pension system transition. 2.37 Unification. If the fragmented pension system is unified and coverage is extended to the nonstate sector (including township enterprises), this would generate substantial resources to fund the transition. In expanding coverage, care would have to be taken to keep contribution rates low so as not to discourage growth. 2.38 Rapid Growth. China's economy is growing at over 10 percent a year and its domestic savings rate is over 40 percent of GDP. Such rapid growth increases the size of the "pie" of the economy and make it feasible to use a slice of the incremental income to fund the transition. An important mechanism to accomplish this is to use price indexation rather than wage indexation of benefits, so that pension costs do not rise commensurately with growth and the growth dividend can be siphoned off to pay for pension reform. 2.39 Structural Change. Associated with rapid growth of GDP is the rapid structural change of the economy, with labor shifting from rural to urban areas and from agriculture to nonagriculture activities within rural areas and from the state sector to the more efficient nonstate sector in the urban area. The reserve pool of surplus labor will enable the nonagricultural labor force to expand rapidly, even if the overall labor force grows slowly, thereby expanding the contributory base for the pension system. Eventually, these workers too will retire. But if the transition has been made by then, their pension needs will be met, in part, from their own funded individual accounts and the higher productive capacity generated by their retirement savings. 2.40 Capital Gains. The SOE sector in China, while facing problems, has not collapsed the way it has in other socialist economies in transition. The SOE sector in China provides about 70 percent of government revenues and accounts for a significant part of the country's impressive performance in savings, exports, and GDP growth. The value of assets in the SOE sector is several times the GDP, while it is estimated to be less than 50 percent of the GDP in Latin  - 31 - American and Eastern European countries. The availability of these assets opens up possibilities for transition financing in China that are not available in other economies. 2.41 The above discussion indicates various components of reform in China's pension system that seem advisable as well as the broader macroeconomic factors that make these reforms financially viable. There are, however, risks and uncertainties associated with the institutional and political feasibility of adopting various components of these reform as well as about macroeconomic environment. It is therefore necessary to consider quantitative implications of various packages of reform and macroeconomic assumptions. It is these issues that the report turns to in the next chapter.   - 33 - 3. QUANTITATIVE ANALYSIS FOR A PREFERRED PENSION SYSTEM INTRODUCTION 3.1 The preceding chapters highlighted some of the major weaknesses of the current pension system in China and noted the suggestions made by Chinese policy analysts and others to reform the system. In this chapter, an attempt is made to pull these thoughts together and suggest an outline of a preferred system of pension in China. The suggested system is in line with international experience and, in the judgment of the Bank mission, conforms to the economic realities of China. However, it is fully recognized that there is no unique solution to the pension problem in China and final decisions have to be made in the light of social and political considerations, which are beyond the scope of the current mission. 3.2 Experimentation in design and gradualism in implementation have been the hallmark of China's successful policies of transition from planned to market economy. However, experiments with the pension system have been going on for nearly ten years. The new plans, whatever they are, should be put in place by the year 2000 in order to end the uncertainties for this vital sector and to take advantage of the special opportunities for reform that exist today. What follows is a set of recommendations on a master plan for a pension system. Some specific numbers are postulated for ease of computations, but they should be treated merely as illustrative. In practice, these numbers would have to be tuned to local circumstances. Implementation will no doubt proceed in a gradual manner with experiments in certain localities. DEMOGRAPHIC AND MACROECONOMIC DEVELOPMENTS 3.3 The financial viability of the pension system is crucially dependent on the developments in the economy with regard to growth of labor force, wages, interest rates and such other economic factors. It is beyond the scope of the present report to make detailed investigations for projections on these macroeconomic variables. However, some basic assumptions about these variables are essential for assessing the financial outlook of the pension system and the assumptions made in the report are presented below. These projections are, however, only illustrative; the model set up for the report is capable of generating quick responses to alternative sets of assumptions and is intended to be an interactive tool for discussions with policymakers. 3.4 Population Projections. Table 3.1 presents the World Bank projections on working age population and the elderly in China over the period 1995-2050. It is noteworthy that in China, working age population is projected to become stagnant between 2020 and 2030 and then decline after 2030. On the other hand, the population in age group 65 and above continues to increase steadily and rapidly up to year 2050. As a result, the demographic dependency ratio that was only 8.7 percent in 1990 rises to 31.2 percent by 2050 (see Table and Figure 3.1).  - 34 - TABLE 3.1: PROJECTED POPULATION OF WORKING AGE GROUP AND THE ELDERLY 1990 1995 2000 2010 2020 2030 2050 1. Population in age group: 762.0 808.3 845.8 955.9 988.6 989.4 962.2 15-64 (in millions) 2. Population in age group: 66.1 75.9 86.6 104.2 153.6 214.9 300.4 65 and above (in millions) 3. Demographic Dependency 8.7 9.4 10.2 10.9 15.5 21.7 31.2 ratio(%) (2)/(1) FIGURE 3.1: CHINA'S POPULATION IS AGING FAST 1000 - 35 900-- 30 800-- 700-/ 25 600-- 20 Million 500 % 400 15 300- - 10 200 - - -5 100 0 0 1990 1995 2000 2010 2020 2030 2050 Year Sue 1. Population in age group: 15-64(in millions) A2. Population in age group 65 and above (in millions) + 3. Dependency ratio(%) (2)1(1) Source: World Population Projections, 1994-1995, The World Bank, 1994.  - 35 - 3.5 Labor Force Projections. While the demographic dependency ratio shows the economywide need for supporting the elderly, what matters for the financial viability of the pension system is system dependency ratio, i.e., the number of the elderly covered under the pension system to the number of workers in the system. The key baseline assumptions made in the report to obtain the projected system dependency ratios are noted below: (a) Participation rate. Labor force participation rate depends on the legal retirement policy and a host of other socioeconomic factors. In this report, it is assumed that the legal retirement age will be gradually increased to 65 for both men and women. This change will substantially increase the participation rate. On the other hand, with increase in period of schooling as the economy develops, the participation rate in the age group 15 to 20 may decline. On the whole, the participation rate in China is high by international standards and for projection purposes, it is assumed that the participation rate will increase slightly from about 77 percent in 1994 to stabilize at 80 percent. (b) Declining share of employment in agriculture. Between 1990 and 1994, the share of labor force in agriculture in China declined from 60 percent to 54.3 percent, implying an annual rate of decline of 2.5 percent in the share. This rate of decline is projected to continue and the share of labor in agriculture becomes 17 percent by 2030 and 10.5 percent by 2050. These projections are consistent with experience of countries such as Korea (which had 15 percent of labor force in agriculture in 1994) and Japan (where the corresponding figure was about 7 percent in 1991). (c) Urban-rural distribution. Between 1990 and 1994, employment in urban areas grew by 3.4 percent per year, while that in township and village enterprises (TVEs) grew by 6.9 percent per year. The TVEs are beginning to face some difficulties, and in 1994, there was a decline in employment in TVEs. However, the government policy is one of strong support to development of small and medium towns, where TVEs are the main sources of employment. For projections in this report, it was assumed that the growth rates of employment in urban areas and in TVEs will converge to the same level and the distribution of employment between urban sector and TVEs will stabilize at 60:40. The long- term implications of this trend are interesting. As in other countries, the ratio of labor force in urban areas will increase in China. However, with govemment policy of avoiding excessive concentration in cities, the share of urban employment in China is projected to stabilize around 50 percent, significantly below the levels of 70 to 80 percent reached in countries such as Korea, Japan and the United States. (d) Share of nonstate sector. Between 1990 and 1994, employment in state sector (defined to include government organization, public institutions, SOEs and COEs) has grown only marginally (by I percent per year), from 139.0 million to 145.0 million. The recent government policy statements suggest that the role of government will be reduced over time, and employment in the state sector may go down in the future. For projections in this report, it was assumed that the employment in the state sector (including COEs) will stabilize at the 1994 level. Within the context of this assumption, it is projected that the share of state sector in nonagricultural employment will decline from about 52 percent in 1994 to 21 percent by 2050.  - 36 - TABLE 3.2: LABOR FORCE PROJECTIONS Population in Share of Employment age group Participation labor in Urban outside public Employment 15-64 rate agriculture employment sector la in TVEs (millions) (%) (%) (millions) (millions) (millions) 1990 762.0 74.5 60.0 147.3 8.3 92.0 1994 799.2 76.9 54.3 168.1 27.1 120.0 Projections 1995 808.3 77.0 52.3 172.8 27.8 115.2 2000 845.9 80.0 42.7 224.4 79.5 149.6 2010 955.9 80.0 29.1 315.5 170.5 210.3 2020 988.6 80.0 21.9 359.7 214.7 239.8 2030 989.4 80.0 17.0 382.6 237.6 255.0 2040 950.2 80.0 13.7 381.8 236.8 254.5 2050 962.2 80.0 10.5 400.9 256.0 267.3 La Defined as urban employment minus employment in state-owned units and collectively owned enterprises. A 3 percent unemployment rate is incorporated in the projection. Source: See Technical Annex 4, Table TA4.1-4.3. 3.6 Labor Productivity, Real Wages and GDP. Between 1990 and 1994, average productivity of labor (defined as GDP in nonagricultural sector divided by labor force in nonagricultural sector) grew at an annual rate of 8.9 percent, while real wages grew at 6.4 percent per year. The rapid growth in productivity was associated with high level of capital formation and the large opportunities in China for technological catch-up with more advanced countries. A continued high level of investment in China is both feasible and desirable over the medium term, and the opportunities for technological catch up will remain significant for some time. Therefore, over the medium term (up to 2010), the growth rate of labor productivity and GDP should remain high. However, over the longer term, these growth rates will come down substantially, as savings rates decline with rising dependency ratios and the scope for technological catch-up diminishes. For projections in this report, it is assumed that average real wages will grow at 4 percent per year during 1995-2020, with a decline to 3 percent during 2021- 2030 and 2 percent thereafter. (For individual workers, a merit increase of I percent per year of service is assumed in addition to the average wage increase.) On real rate of return on funds in pension system, it is assumed that these rate will be equal to the growth rate in real wages including merit increases. 3.7 The projections clearly suggest that by the time young worker aged 20 today retires around 2040, China would have many of the characteristics of old age crisis of the mature OECD countries of today. The system dependency ratios will have risen to 50 percent and growth rates of labor force, labor productivity and of GDP would have slowed down considerably (Technical Annex 4, Table TA4.6). Unless new systems are set up for the current young workers for their retirement, they would face significant uncertainties regarding income security in their old age. It is therefore essential to analyze quantitatively the financial outlooks of the present pension system and the various suggested reforms. SIMULATION RESULTS WITH LiMITED REFORMS 3.8 The pension system model designed for the report (see Technical Annex 4) is geared to working out the quantitative implications of various sets of assumptions on demographic and macroeconomic variables, as well as pension system related policy variables. The possible sets  - 37 - of these various assumptions are very large in number and the report does not attempt to work out the implications of all these assumptions. Instead, the model is intended to be an interactive tool for discussions with policymakers with regard to the implications of their preferred policy sets. This section presents only a few of such sets and illustrates how the model could be used to assist in policy discussions on pension system reforms. On the benefit side of the pension system, the key indicator is the replacement rate as a percentage of net wages and on the cost (and affordability) side, the key indicators are: system dependency ratio, sustainable contribution rate and the accumulated reserves in the pension system. Sustainable contribution rate is defined as the rate that will meet the financing needs of pension system until 2050 and at that point the accumulated reserves would be enough to cover 16 years of deficit in the system (the present life expectancy at retirement)7. This concept differs from the balanced contribution rates in that it attaches more importance to the financial viability over a very long term. For details on sustainable contribution rates in industrial countries, see IMF 1996. The results of these simulations with limited reforms are presented in Table 3.3 and Figure 3.2. 3.9 Case I. In this example, the pension system largely retains its current characteristics. It covers mainly the state sector with a small (10 percent) coverage of nonstate sector and none at all of TVEs. The pensions are indexed to the average of increase in prices and in nominal wages. Replacement rate is 80 percent and retirement age is 55 for women and 60 for men. Compliance rate is 85 percent and the projected growth rates for real wages and real rates of interest are: 5 percent per year for the years 2000-2010, 4 percent per year for 2011-2030 and 3 percent per year thereafter (Scenario 0 in Table 3.3). 3.10 The simulation results confirm the views expressed by the Chinese analysts that the current system of pension is not viable in the long term. The system dependency ratio rises to 76 percent by 2050 and the sustainable contribution rate is 45.9 percent. (see Figure 3.2) 3.11 Case II. The situation improves somewhat if the pensions are not indexed to real wages (Scenario 1). However, the sustainable contribution rate is still too high at 40.8 percent. 3.12 Case III. More significant improvement occurs if the coverage rate is increased to 50 percent by 2010 for both TVEs and nonstate urban sector (Scenario 2). The sustainable contribution rate declines to 37.4 percent. 3.13 Case IV. If the replacement is reduced to 60 percent, there is a significant drop in sustainable contribution rate to 28.4 percent (Scenario 3). 3.14 Case V. If the above reforms are combined with an increase in retirement age to 65 by the year 2040, the situation changes dramatically (Scenario 4). As a result of this change, the number of retirees goes down and that of workers goes up. The system dependency ratio goes down from 26 percent in 1995 to 15 percent in 2010 and then rises to 47 percent by 2050. The sustainable contribution rate goes down to 19.7 percent, which is moderate in the light of the average contribution rates of over 20 percent for most enterprises that prevail in China today. (See Figure 3.2) The system thus looks viable except for the risks noted below. 7 The model uses a commentation factor of 10 and the required reserves in 2050 are 10 times the deficit. The sustainable contribution rate changes only marginally if this factor is changed. For example, if the reserves are to be only five times the deficit, the sustainable contribution rate in Case I will be 44.6 percent instead of 45.9 percent.  Table 3.3: Simulations with Non-Structural Reforms Scenario 0 Scenario I Scenario 2 Scenario 3 Scenario 4 Scenario 5 Scenario 6 Scenario 7 Scenario 8 Scenario 9 System Outcomes: Sustainable Contribution ratel 45.9% 40.8% 37.4% 28.4% 19.7% 20.4% 23.1% 22.9% 24.3% 30.4% System Dependency Ratio (%) 1995 26 26 21 21 21 21 21 21 21 21 2000 27 27 19 19 18 18 18 18 18 18 2010 34 34 17 17 15 15 15 15 15 15 2020 49 49 27 27 18 19 19 19 19 19 2030 73 73 57 57 29 31 31 31 31 31 2040 78 78 78 78 50 53 53 53 53 53 2050 76 76 77 77 47 50 50 50 50 50 System Characteristics ( changes effective in year I9951: Scenario 0: Scenario 1 - 9: Real Wage Indexation 50% Scenario I Scenario 0 with no wage indexation Coverage rate: TVE 0% Scenario 2: Scenario I with increase in coverage rate for TVE and Other Enterprises to 50% (by 2010) Other Enterprises 10% Benefits(Rcplacement rate): Scenario 3: Scenario 2 with decrease in replacement rate: retirement : 60%, disability: 30% Retirement 80% Disability 40% Scenario 4: Scenario 3 with increase in retirement age for male and female to 65(by 2040) Survivor (Lump sum) 100% Scenario 5: Scenario 4 with decrease in participation rate to 70% Retirement Age Female 55 Scenario 6: Scenario 5 with decrease in compliance rate to 75% Male 60 Participation rate 80% Scenario 7: Scenario 6 with changes in wage increase rate: 6%,5%/6(20ll),4%(203 1) Compliance ratc 85% Wage Increase rate 5%,4