China - evaluation of the Liaoning social security reform pilot

38183 Report No. 38183-CN China Evaluation of the Liaoning Social Security Reform Pilot December 2006 Human Development Sector Unit East Asia and Pacific Region Document of the World Bank   CHINA Evaluation of the Liaoning Social Security ReformPilot The World Bank   Table of Content Acknowledgements ................................................................................................................. 1 Executive Summaly ................................................................................................................ 2 Evaluation of the Liaoning Social Security Reform Pilot ...................................................... 9 I BACKGROUNDOFTHELIAONINGPILOT..................................................... . 10 1 Pilot Objectives........................................................................................................ . 10 3 Economic Background and Initial Conditions o f the Pension Program ..................11 2 ..Pilot Components and PensionPolicy Expenments................................................ 3.1 Labor Market Development and Outcomes ....................................................... 12 13 3.2 Challenges Faced by the Liaoning Social Security System ............................... 14 I1 PILOT ACHIEVEMENTS...................................................................................... 18 . . 2.ReducingLabor Market Distortions ........................................................................ 1 Strengthening the Credibility of the System............................................................ 19 20 21 4. ImprovingPension ProgramAdministration ........................................................... 3. EnhancingEfficiency inksk Management............................................................. 22 I11 POLICY AND INSTITUTIONAL CHALLENGES............................................ 23 1 Developing a Sustainable PensionInsurance Program............................................ .. 25 1.1Long-term Prospect of theLiaoning Pilot Scenario .......................................... 25 1.2Reform Options to Improve Sustainability ......................................................... 27 1.2.1 Parametnc reforms:..................................................................................... 27 28 1.2.3 Gradual fundingo f individual account: ...................................................... 1.2.2 Separating the impact o f legacy cost: ......................................................... 32 1.2.4 Reducing contribution rate: ......................................................................... 40 1.2.6 Expanding coverage: ................................................................................... 43 1.2.5 Adjustingthe longterm target replacement rate: ........................................ 42 1.3 Complete Policy and Regulations beyond Enterprise Pensions ........................ 45 2. Strengthening Investment Management o f IndividualAccount Savings 2.I Pension Funds Investment under the Liaoning Pilot ......................................... .................46 46 2.2 Alternative Investment Choices and Implications.............................................. 2.3 Improving Governanceto Realize Investment Potentials .................................. 49 51 3. Improving Administration andFinancial Management ........................................... 3.1 Complete Implementation of the Reform Agenda Specified in Pilot Design .....53 53 3.1.1Improvethe legislative basis for pension administration............................ 53 3.1.2 Reorganize pension administration and move to provincial level pooling .54 3.1.3 Regularly publish information on revenue and expenditure to improve transparency ............................................................................................. 54 3.1.4 Complete development o f the information system..................................... 55 3.2 Institute Additional Administrative Reforms ...................................................... 56 1  3.2.1 Financial management improvements......................................................... 3.2.2 Create performance standards. measure performance. andreport results 57 ...56 3.3 Pilot TestMore ComplexReforms ..................................................................... 58 3.2.3 Standardize operatingprocedure................................................................. 57 3.3.3 Issue annual statements............................................................................... 3.3.2 Improve collections..................................................................................... 59 3.3.1 Streamline processes................................................................................... 58 59 I V IMPLICATIONSFORNATIONALPENSIONREFORM STRATEGIES 1.Challenges ................................................................................................................ . .....60 60 2 Recommendations .................................................................................................... . 63 Reference.............................................................................................................................. 65 11  Tables Table 1 Liaoning: PensionInsuranceContribution Rate (%)................................. 16 Table 2 Total Social Security ProgramsContribution Rate (%).............................. 17 Table 3 RevenueandExpenditureofLiaoning PensionInsuranceProgram (inpercentof LiaoningGDP) ............................................................ 18 Table4 (based on Liaoningdata).................................................................. Summary ofProjectionResults inVanous Policy Scenanos Table 5 Implications for Risksand Returns of InstrumentClasses inChina................44 50 Table 6 Summary of ProjectionResults inVanous Policy Scenanos (based on sample national data)......................................................... 62 Figures Figure 1 Declining Urban Employmentand Informationization of Labor Market...........13 Figure 2 The Structure of EmploymentChanged ................................................ 14 Figure 3 DependencyRatiosand Coverage ....................................................... 16 Figure 4 The Aging of Liaoning andthe Demographic Dividend .............................. 25 Figure 5 ImplicitPensionDebt andthe CurrentBalanceunder the Pilot Scenario..........26 Figure 6 Impactof Parametric Reforms ............................................................ 28 Figure 7 Decomposingthe Cost..................................................................... 30 Figure 8 Comparisonofhistorical growth -wage, inflation, bank deposit................... 37 Figure 9 Relationshipbetweenwage growth, investmentreturns, contribution rates and replacementrates ........................................................................... 42 Figure 10 FundedRatiounder CoverageExpansion .............................................. 44 Figure 11 Distribution of Fixed-incomeInstrumentsin China. 2003........................... 51 Boxes Box 1 34 Notional DefinedContribution (NDC) System........................................ Savingsin China and Its Relationto PensionReform................................ Box 2 36 Box 3 The FinancialLandscape in China ...................................................... 39 Box 4 I s PensionContribution Affordable to Industrial Firms inLiaoning? ..................41 Box 5 What aDiversified InvestmentPortfolio Can Potentially Achieve?..................... 47 Box 6 The 2004 Jilin andHeilongjiangPilots andthe Policy Changes Introduced in2005 ....................................................................................... 61 TechnicalAnnex 1 Analysis of PensionLiabilities and FiscalImplicationsunder Reform Scenarios. Mun . Yee Sin 2 Fiscaland EconomicAspects of Public PensionsReform. Kuijs Louis 3..I s 20 Percent PensionInsuranceContribution Affordable to Industnal Firms inLiaoning Province. Chunlin Zhang 4. Evaluationof LiaoningPilot: PensionAdministration . LawrenceThompson ... 111   Acknowledgements The report i s the output o f a collaborative effort between the World Bank and China Development Research Foundation. It was prepared by a team consisting o f Xiaoqing Yu (task team leader), Bert Hofman, Louis Kuijs, Yee Mun Sin, Lawrence Thompson, Jun Wang, and Chunlin Zhang. The team benefited from valuable inputs o f Shaomin Cui, Ping He, Bingwen Zheng, and Northern Trust Investor Services, the World Bank's Poverty Assessment team, and from the qualitative survey analysis led by Qiang Li and Chuanjin Tao. Lansong Zhang provided support in facilitating field visits and formatting the final document. Marcin Sasin provided some research assistance. The team appreciated the very helpful comments and suggestions from peer reviewers and World Bank colleagues including Louise Fox, Indermit Gill, Robert Holzmann, Tamar Manuelyan Atinc, and Ana Revenga. The work was done under the overall guidance o f Bert Hofman and Lu Mai o f China Development Research Foundation. The World Bank team benefited from strong support o f the Liaoning Development Research Center, and fruitful discussions with many government officials and researchers who were generous with their time and knowledge. The team appreciates the collaborations o f officials o f the Liaoning Provincial Government, including Liaoning Provincial Labor and Social Secunty Bureau, Bureau o f Social Insurance Administration, Finance Bureau, Tax Administration, and government officials in Fushun and Dalian municipalities. We are also very grateful to colleagues o f the Social Secunty Department o f the Ministry o f Finance for sharing their views and to the International Department o f the Ministry o f Finance for their support. The team would also like to thank inputs and feedback o f participants o f vatlous workshops and the international conference in June 2005. 1  Executive Summary The Liaoning Pilot has been an important step for China's social secunty reforms. These far-reaching reforms started inthe 1980s and resulted inthe 1997 State Council Document 26 which established a system based on a mandatory pay-as-you-go (PAYG) basic pension and a mandatory funded individual account. Together with the unemployment insurance program and the urban minimum living allowance program (dibao), the basic features o f China's urban pension system can meet the demands put on such a system, namely poverty relief, income redistribution, insurance against the nsks o f longevity, and consumption smoothing. However, in implementing SC Document 26 a host o f issues arose. The Liaoning pilot and the subsequent pilots in the northeast provinces were designed to address those issues. The Liaoning Pilot has met many o f the objectives o f State Council Document 42 that detailed the pilot design. The achievements are remarkable given the short time span o f implementation of the pilot and given Liaoning's initial conditions that were inmany ways less advantageous than in other parts o f China. In particular, the pilot achieved the following: 0 Fundedindividual accounts were established separate from the social poolingbasic pension, and were funded with 8 percent o f payroll taxes. Some RMB 13.8 billion was accumulated by the end o f 2004. 0 The pension benefit formula was adjusted so as to provide better incentives to continue to contribute to the pensionscheme beyond the minimum 15 years. 0 Pension administration was significantly improved with collection by the tax administration, a new information system, and a shift o f administration from enterpnses to government. 0 Labor market distortions were reduced as pension contribution rates became more unified, and portability o f pensions improved through better administration and partial fimding. Survey work indicated that the people o f Liaoning currently have more confidence intheir pension system than before. Data also shows that household savings in Liaoning, in contrast to those o f the rest o f the country, are falling. While the extent to which the credibility o f the system has been strengthened i s difficult to venfy, these findings suggest that people inLiaoning are more confident that their future pension is beingtaken care of. The reforms under the Liaoning pilot need to be enhanced for the pension system to become financially sustainable, yield adequate pensions, become more credible androbust, and create fewer distortions in the labor market. The pension system as implemented in the Liaoningpilothas the following main problems: 0 The system remains financially unsustainable, which, ifleft unattended, would lead to very high future contribution rates that would undermine employment creation and competitiveness, or leadto significant additional fiscal contributions. 2  The number o f people covered by the pension scheme and the number o f people and enterpnses contributing to it is relatively low, aggravating the issue o f financial sustainability, undermining old-age income security for all workers, and distorting competition among enterpnses. The pension premiums paid remain relatively high and are not yet unified across municipalities, creating a bias against hiring labor in the formal labor market, and against workers moving from onejob to the other. Pooling o f pensions remains effectively only at the municipal level, or even below, undermining the insurance function of the pension system and creating barriers to labor mobility The investment returns on the funded individual accounts remain too low to provide for the targeted level o f overall benefits under a matured system. Investment policies and practices leave the hnds in the individual accounts at nsk o f loss and potential abuse. This report recommends a range o f reforms to restore financial viability and credibility o f the system and increase compliance with the system, resulting in more secure pensions, lower fiscal costs, and more jobs. Broadly, the recommended reforms as detailed below can be classified in three categones: (i)reforms to restore financial viability; (ii) administrative and financial management reforms; and (iii) reforms to improve pension fundgovernance and investmentpolicies. Many o fthe recommendations apply equally to Liaoning as to the rest o f China, but while some are within the control o f the provincial authonties, most require action o f the national authonties as well. Most o f the recommended policy reforms can be implementedonce consensus i s reached on them, but some o f the reforms, notably some administrative reforms, require further experimentation to get it right. Ifimplemented, the recommendedreformswouldrestoretheviability ofthepensionpolicy as decided in SC Document 26 o f 1997, namely a basic pay-as-you-go pension plus funded individual accounts. This report does not investigate in-depthalternative pension systems currently debated inChina, such as the Notionally Defined Contribution (NDC) system or a pure pay-as-you-go system. Many o f the reforms recommended for the current system would also be needed for alternative systems to work. The report notes that given China's demographics, a certain amount o f pre-fundingo f future pension obligations is desirable, even if returns on balances are currently low Such pre-funding can be achieved through individual accounts, or through a social security trust fund. The report discusses the option o f a specific method o f crediting interest to the individual accounts on a notional basis, usingamacroeconomic indicator. There is some urgency to further pension reforms. China's demographics leave only a little over a decade before the number o f pensioners starts to rise rapidly, and the number o f workers supporting them starts to decline. The ratio o f pensioners over workers in Liaoning i s projected to increase from 40 percent now to over 100 percent injust 20 years, and similar trends will occur in the rest o f China. Building a sustainable pension system that can accommodate this nse independency i s therefore becoming increasingly urgent. 3  Reforms to restore financial viability The pension policy as implemented under the Liaoning pilot i s financially unsustainable, and the designers o f the Liaoning reforms seem to have paid only scant attention to this cntical element o f the system. An illustrative scenano suggests that, in the absence of government transfer, under the current system pension premiums would over time have to rise to 48 percent o f wages, a level almost double the current level, and double that commonly found in other developing countnes. Other indicators o f financial health o f the system, the implicit pension debt and the financial deficit, are also very high. Inpart this reflects the province's initial conditions, but much results from decisions under the control o f central and local authonties. Pension Policy Reforms. To establish financial sustainability o f the pension insurance program, a set o f reforms can be considered to improve policy design outlined in the Liaoning Pilot. The main ones are: Gradually increasing statutory retirement age to 65 for both men and women. The retirement age can be raised by increasing the statutory retirement age gradually over the next 20 to 30 years to 65, while early retirement for hazardous occupations can be phased out as worlung conditions improve. Restoring actuarial balance in individual account pensions. This can be achieved by basing the individual accounts pension on the balance in the account at the point o f retirement and actual life expectancy at retirement, rather than the current low factor o f 120. Adjusting contribution ratefor small business and the self-employed. Small businesses and the self-employed, almost a third o f the current contributors to the Liaoning's pension program, contribute only 18 percent o f wages, even though benefits are the same as for those who pay 28 percent. Raising the contribution rate to the same level as other workers therefore seemsjustified. Changing the method to index pension benej2s. China's policy is to index pensions based on 40-60% o f average wage increases. Indexing allows current pensioners to share in the benefit o f current growth, but it can become quite expensive and does not protect pensions against wage declines. One option i s to partially index pension to wages as a transitional arrangement and switch to pnce indexation over time once the desirable level of pensionbenefits has been achieved. The combined impact o f the four policy reform measures above would lower the required contribution rate by a third. The implicit pension debt would decline by one sixth, while the financing gap would be reduced by three quarters. Together, these reforms therefore strongly improve the financial health o f the system. Expansion o f coverage o f the pension system to workers not yet covered can further improve the financial health o f the system, but only ifthe expansiontook placeunderthe reformedpensionpolicies. Managing the legacy costs. A significant part o f the financial difficulties o f the Liaoning pension system is not due to the new system in place but due to past policy decisions. It includes cost resulted from early retirement used as a way to reduce unemployment and mitigate social impact o f SOE restructunng. Another part o f the cost stems from the 4  difference between the pension level committed in the past and what the new system can deliver. Together, these "legacy" costs account for more than 27 percent o f the implicit pension debt. The consequence is that current contribution rates have to be higher to maintain financial viability, and as a result, hiring people becomes more expensive. Government needs to decide on a more equitable formula in which these costs could be shared between the different levels o f government through fiscal transfers as well as current and future generation o f contributors by way o f higher than required contributions for a steady state, matured system. Already, central and local governments contribute 1.1 percent o f Liaoning GDP to the pension scheme, but a fiscal contribution that i s required to cover the legacy and transition costs would amount to some 2% percent o f Liaoning GDP in2004. ImprovingAdministrative Efficiency and FinancialManagement A key objective o f the Liaoning pilot was to improve pension administration. Despite the major improvements in administration, and successful implementation o f separate collection o f social secunty contributions, weaknesses remain that result in inadequate coverage and collection o f pension contributions, higher administrative burden on enterpnses, delays in the transfer o f contributions into the investment account, and lower accuracy o f the information on which pension payments are made. Reforms to overcome these weaknesses are crucial for the Liaoning pilot to succeed, as well as essential for a national system to take hold over time. Among the reforms neededare: Improving theLegal Basisfor Social Security Administration The current legalbasis is insufficient to effectively enforce social security contribution payments, and collection o f arrears. In other countnes, the collection o f social secunty funds i s based on laws that have similar legal andenforcement power as the tax code, andChina should stnve for such administrative legislation even before national pensionpolicies are decided upon. Onpension policies, many areas also require more clanty, including coverage and pension entitlement, particularly the status o f pnvately-owned enterpnses registered in rural areas, the status o f the township andvillage-owned enterpnses, the status o f workers registeredin urban areas but worlung in rural areas, and the status of enterpnses that receive tax concessions from local governments. Central government should provide more guidance inthese and other areas by means ofregulations to avoid a situation where administrative decisions at the local levelundermine intended unity inpension policies. Reorganizing Pension Administration The social secunty administration operates in many ways still as an arm o f local governments at the various levels. The province should develop plans to complete the vertical integration o f the social insurance agencies by merging the municipal level social insurance agencies with the provincial level agency Once contribution rates have been equalized and all municipalities have established integrated social insurance administrations, it i s a relatively easy step to move to provincial level administration. 5  A key decision in scaling up reforms to the national level is which organization should collect pension contributions. InLiaoning, this responsibility was by andlarge successfully transferred to the local tax bureau, which is a province-level organization down to the county level. Ifthe long term goal i s to develop a national system, the national tax administration may be the appropriate agency to collect pension contributions. The feasibility o f such an arrangement could be tested in one or more locations as part o f a pilot. Improving Administrative Processes In contrast to the Liaoning reform objectives, there i s still considerable vanation in processes among localities. Standardizing procedures requires developing detailed, step by step instructions for handling all important transactions, prepanng wntten documents outlining these procedures in a senes o f operations manuals, distributingthe manuals throughout the organization and monitonng operations to assure that the approvedprocedures are being followed. Examples o f cumbersome procedures in the Liaoning program include the procedure enterpnses are required to follow monthly to pay their social insurance contributions, the process for registering a new enterpnse or renewing a business license, for transfer a worker's account from one pooling area to another, and for applying for a pensionbenefit. Developing new processes that reduce the reliance on paper forms and that automate data transfers among agencies would reduce the burdenon enterpnses andimprove compliance. Enforcing Compliance A stronger legal basis and less cumbersome processes will improve compliance, but further action is needed. In particular, processes should be designed and implemented to ensure that (i) all covered enterpnses are registered, (ii) all registered enterpnses accurately report their contribution liabilities, and (ii) all liabilities are paid. Some potentially p o w e f i l enforcement techniques are already being tested in Liaoning, and in particular the processes o f Fushun Municipality's tax bureau deserve further consideration. Many enterprises do not pay pension contributions because o f financial difficulties. However, allowing arrears in pension contributions for these reasons undermines the financial viability o f the system. If government decides that certain enterprises with financial difficulties cannot be closed or bear the burden o f pension contributions, the best option is to provide an explicit subsidy from the budget, while enforcing compliance with regardto pension contributions. Ideally, such a subsidy would be time-bound, and based on clear cntena o f financial difficulty, so as to force enterpnses to adjust and restore financial viability over time. Improving the Information System Several important steps in implementing the new information system in Liaoning are yet to be done, including: (a) completing the entry o f the work histones of current contributors in the remaining 10 to 20 counties and distncts for which this has not yet been done; (b) converting the remaining 4 municipalities to the new system; (c) automating data transfers between the tax bureaus and social insurance agencies where it is still done manually; (d) finishing the clean up o f errors in the provincial data base on individual accounts to ensure data consistency and allow 6  generation o f improvedprogram statistics, particularly on earnings. Over time, developing common databases, and electronic exchange o f information among local agencies, and with the national tax authonties for enforcement reasons, is highly desirable. Improving Financial Management Better financial management can improve the financial results o f the pension system, reduce the nsk o f errors and abuse, and be improved by streamlining several cash management processes, initiating reconciliation, and by improving transparency Specific steps include: 0 Establishing a full treasury cash management system. The two track accounting system has some o f the attributes o f a modem treasury system inthat contributions are deposited directly into accounts controlled by the financial bureau. Internal controls would be improved if the financial bureau retained control over the funds and transferred them directly to the bank accounts o f the program beneficianes, while payment orders are drawn upbythe social insurance agencies. 0 Accelerating the deposit of funds into interest bearing accounts. At present, it can take more than a month to move enterprises contributions from the special accounts controlled by the local financial bureau to the accounts for investment transactions. The provincial government could use the treasury account network in the central bankto centralize the funds before transfer to the investment account or the special social secunty account to speedup deposit ininterest neanng accounts. 0 More effective pooling of financial resources. At present, separate accounts are maintained in each county and distnct even in municipalities that officially pool at the municipal level. Authonties should evaluate the relative efficiency o f merging all o f the financial operations into one single provincial level account. 0 Improvzng Transparency. The provincial social secunty agency should regularly publishinformationon the revenue and expenses o fthe social insurance system and the composition o f its investments. Regular audits o f the financial statements by outside auditors should be arranged. ImprovingInvestmentOutcomes A cntical element o f the pension system as piloted in Liaoning is the financial return on the invested individual balances. The higher the returns, the higher the pension that can be achieved, and the lower the contribution rate and fiscal subsidy needed for the pension system to be financially viable. The limited expenence under the pilot is not promising: investment returns have been low and ranged between 1.6-2.7 percent per year between 2001 and 2004. As a result, rather than the targeted 26 percent o f wages, the individual accounts are likely to only replace some 12 percent o f wages. In addition, invested hnds inbank deposits are at considerable nsks: it is the weaker banks that are willing to offer higher interest rates, and some of Liaoning's pension reserves are kept at banks with negative capital. In the event o f bank failure, this could lead to additional fiscal costs or lower pensions, because institutional deposits are not covered under China's nascent deposit insurance scheme. 7  Improving governance and regulation of pension investment. To improve the investment outcome o f pension reserves, the authonty needs to improve the transparency o f management and investment o f pension funds, further develop investment regulations, and strengthen internal controls. Over time, improved regulation and supervision o f the investment industry should allow for investment o f pension funds in a broader range o f investment instruments, including overseas' assets, and funds to be managed by outside, professional funds managers. This is likely to lead to higher returns on invested funds. Broader Financial Sector Reforms Over time, financial sector reforms could lead to higher returns on invested funds. Development o f a corporate bonds market, restructunng infkastructure financing away from the current heavy reliance on bank financing, and revamping the struggling stock market could open up classes o f investment instruments particularly suited for pension funds. Combined with strict enforcement o f better investment rules, this is likely to lead to higher returns. Special Bonds Issue Improving regulations and supervision and development o f the financial market i s likely to take considerable time, and meanwhile the returns on pension funds is to remain low, probably too low to achieve the targeted pension level. To address the concern of low return to individual account savings, one option that the government could consider i s to issue special government bonds to be purchased by the management o f pension funds on behalf o f individual members. These bonds could be indexed to GDP growth or growth in GDP per capita, or wage growth, and thus provide the returns needed to fund pensions from the individual accounts in the future. This option would be similar to declanng a notionalinterest, with backing assets inthe investment accounts. This could enhance the credibility o f the pension system in that it can deliver a decent replacement rate, and when over time these bonds could be traded, allows the investment portfolio to be optimized. 8  Evaluation of the Liaoning Social Security Reform Pilot Social secunty reform has been an integral part o f China's overall economic reform inthe last decade. Earlier reform efforts were largelymotivated by the needto facilitate the SOE reforms. Over time, more attention has been focused on developing a sound social security system that can address the needs o f a market economy and the Chinese society About a decade after the major reformprogram was initiated, pension reformremains an unfinished agenda. There i s lively debate among policy makers, practitioners, andthe academic circle as to what should be the structure o f China's future pension program and what the transition path should be. To accelerate the reform, the State Council launched an urban social secunty refom pilot program in Liaoning Province in July, 2001. The Pilot covered social insurance programs (old-age pension, health insurance, and unemployment insurance), urban minimum living allowance program (dibao),and active labor market programs. Inaddition to some policy expenments, the pilot also devoted substantial attention and resources to improving social secunty administration, including social security fund management, contribution collection, and informationmanagement. While the pilot phase ended at the end o f 2003, the Liaoning expenment has implication for the hture development o f the national social secunty policies. There have been discussions as to whether the Liaoning policy could and should be replicated inthe rest o f the nation. The Liaoning government conducted a review o f the pilot experience, but systematic evaluation based on ngorous analysis seems to be lacking. In spring 2004, the State Council made the decision to extend a modified pilot policy to Heilongjiang and Jilin, the other two northeast provinces. In December 2005, the State Council decided to adopt some features tested in the northeast pilots as national pension policies, and to gradually fund individual accounts inselected provinces. This report focuses on the main component o f the Liaoning Pilot - the pension expenment. It is based on a program review conducted during field visits, some background analysis, and projections analysis usingLiaoning data. The report argues that with substantial support o f the central government, Liaoning has fulfilled most of the requirements set out in the Pilot program. Funded individual accounts were established; adjustment to the pension benefit formula was introduced; pension administration was significantly strengthened; and labor market distortions were to some extent reduced. However, Liaoning pilot did not address many o f the flaws that existed in the overall pension system. Legacy costs - from economic restructuring and pension system transition -heavily burdenthe Liaoning system. As a result, pension contributions remain very high and distortive; effective retirement age is low; a significant part o f the labor force is still not covered or contributing to the program; nsk management remain inefficient due to the low pooling level; the governance structure for pension fund investment i s under-developed and the return on investment i s low; legal basis for effective enforcement i s not fully established. 9  The report recognizes that many o f the challenges faced by Liaoning can only be addressed over a longer time period, or be addressed at the national level. As a national pilot, Liaoning offered valuable expenence and great opportunity to further analyze and develop the national pension policies. The report does not assess the subsequent national pilots implemented in Heilongjiang and Jilin provinces which included some modifications to the Liaoningpolicies. It does not investigate in-depth alternative pension systems currently debated in China, such as the ones based on the Notionally Defined Contribution (NDC) or a pure pay-as-you-go principles. The desirability and viability o f those models are subjects for future research. Section Io f the report provides a brief overview o f the Pilot and its background. Section I1highlights the main accomplishments in implementing the pilot policies with regard to public pensions. Section I11 summarizes the main research findings and highlights the policy and institutional challenges that Liaoning program faces. Section IV discusses the implication o f the Liaoning experiment for future national policy Section V summarizes the main recommendations. I.BACKGROUNDOFTHELIAONINGPILOT The Liaoning social secunty reform pilot was initiatedby the State Council and was under implementation from July 2001 to December 2003. The reform program i s far reaching, but the mainelements that differedfrom the national programs are the pension expenment and the component that supports the transition o f laid-off workers from Reemployment Service Centers to the unemployment insurance program, a process that i s commonly referredto as "binggui" 1. PilotObjectives The Liaoning social security system reform was initiated to achieve multiple economic and social objectives. First, it was motivated by the need to further enterprise reforms and to foster economic growth. The policy measures supported by the pilot would help to ease the social secunty burden on enterpnses, facilitate the separation o f laid-off workers from their enterprises, and assist the closure of the re-employment centers. Second, the pilot was seen as a key policy intervention to maintain social stability It would help to address and honor histoncal commitments and support the poor and vulnerable. Third, the pilot would deepen a wide range o f social protection reforms and develop a broad-based social protection system for the urban population. Inparticular, it aimed at developing a social secunty system that i s independent from enterprises and public institutions, has diverse sources o f funding, follows standard operational procedures, and is administered by the government not the employers. Liaoning was chosen as the national pilot to test the viability of the reform measures in localities facing most severe difficulties during the economic transition. As the following section will illustrate, economic restructuring in Liaoning had led to high 10  unemployment, and declining labor force participation. Many SOEs, the backbone o f the pension system, went into bankruptcy or near bankruptcy Early retirement was used as a way to limit unemployment resulting from enterpnse closure and restructunng. As a result, the pension contribution base was weakened and pension expenditure increased substantially The pension program was in deficit on a cash flow basis in most o f the municipalities, and was financially unsustainable in the longer term given the demographic trend. No real accumulation took place in individual accounts. Combined with other systematic problems, the credibility of the system was fundamentally undermined. 2. PilotComponentsand PensionPolicyExperiments The Liaoning social security reform pilot covered pensions, health insurance, social safety net and labor market programs for the urban population. It consists o f eight major components'. (a) adjust and improve basic pension insurance policy for enterpnse employees inurban areas; (b) analyze and develop pension insurance policies concerning employees o f public institutions; (c) accelerate the development o f urban basic health insurance programs; (d) promote the transition from basic subsistence support program for SOE laid-off workers to unemployment insurance; (e) strengthen and improve urban minimum living allowance programs; (f) realize the socialization of social security program administration and service provision; (g) strengthen the financing and management o f social secunty funds; and (h) accelerate the development o f social secunty legislation. The Liaoning pilot largely followed the same pension policy as developed at the national level2 First, pensions consist of two main components - (i)Basic Pension ("social pooling") that i s defined as a percent o f average local economy wage, and funded on a pay-as-you-go basis, and (ii) individual account pension that i s funded on a mandatory basis, and its monthly level i s equal to the total accumulated balance at retirement divided by a factor o f 120. For workers who were not yet retiredbut were already contributing to the pension insurance system in 1996, their pension benefits will have an additional "transition pensions" component. Second, the minimumyears o f contributory service in order to be eligible for basic pension is 15, while the statutory retirement age i s 60 for men and 50 for women (age 55 for white collar female employees). Third, pension increase is proposed by the Ministry o f Labor and Social Security and the Ministry o f Finance, and approved by the State Council every year. Finally, investment o f contributions under the individual accounts i s restricted to government debt instruments or term deposits with financial institutions. Besides implementing the reform program in a compressed timeframe and in a difficult economic environment, the Pilot differed from national pension policy in the following areas: State Council Document #42, passedon December 25,2000. State Council Document #26, issued in 1997 11  (a) Contribution rates: Contribution to the PAYG component was targeted at 20 percent o f payroll and would be borne solely by the employers. Contribution to individual accounts was set to be 8 percent and would be borne solely by employees (instead of 11 percent into individual accounts, shared by employers and employees nationally); (b) Funding and managing individual accounts: Eightpercent of individual's payroll would be accumulated and managed separately from funds pertaining to the Basic Pension component. (c) Pension beneJitformula: The formula for basic pensions was modified to allow for additional benefit built-up beyond the 15 years o f contribution. Those with contributory service o f 15 years would be entitled to a replacement rate o f 20 percent o f the local average economy wage. Additional years o f contributory service would be credited at the rate o f 0.6 percent per year; The Pilot was also designed to support a range o f social security administrative reforms. In particular, a range of functions previously performed by employers (such as establishing pension claims, maintaining individual record, processing pension payments, etc.) were to be shifted to local social secmty administrations and street level government offices. Substantial effort was expected dunng the Pilot phase to develop the management information system for social secunty programs. In Liaoning, responsibility o f social security contributions collection would be transferred to the local tax authonty, with the exception o f Dalian municipality where the responsibility remained with the social insurance agency3 While some preliminary studies were carned out dmng the design o f the Pilot, in-depth analysis o f the potential impact o f the pilot policies seemed inadequate at inception. A clear set o f measurable outcomes towards achieving the broad policy objectives and indicators to monitor its progress were also laclung. This makes proper monitonng and evaluation o f the Pilot progress and impact difficult to accomplish. 3. EconomicBackgroundand Initial Conditions of the PensionProgram Liaoning started the social secmty pilot under very difficult initial conditions. As one o f China's old industrial bases, Liaoning underwent major economic restructuring which significantly affected labor market conditions. The Pension insurance program was indire state on the eve o f the Pilot. The shift of social security contribution collection responsibility to localtax authority was not a component ofthe Liaoning Pilot, but the process overlapped with the Pilot phase. According to the State Council policy, eachprovince can decide whether the social insurance agency or the local tax authority would assume the responsibility of social insurance contribution collection. 12  3.1 Labor Market Developmentand Outcomes Liaoning's labor market was characterized by declining urban employment and high unemployment. Despite the fact that the economy grew at an average rate o f 9 percent per year from 1995-2003, total urban employment and urban formal sector employment declined drastically dunng this penod (Figure 1). Companng with 1995, total urban employment declined by almost 28 percent by 2003, and total urban formal sector employment was reduced by more than half. The trend o f total urban employment is in sharp contrast with the pattern o f the national labor market where urban employment grew by about 35 percent duringthe samepenod. Liaoning faced daunting challenges with respect to urban unemployment. While the registered unemployment rate grew from 3.2 in the 1990s to 6.5 percent in 2003, the estimated urban unemployment rate based on the 2000 census data was about 16.8 percent, the highest level among all provinces in China. Moreover, the labor force participation rate in Liaoning was only about 64 percent, lower than the national average o f 68 percent. Losinghope o fjob opportunities i s considered the main reason for the low and declining labor force participation rate. Figure1. DecliningUrbanEmploymentandInformationtizationofLaborMarket Total employment and formal sector employment Share of employment in theformal sector declined declined despite the GDP growth =Urban Emplopent 0Urban FormalSectorEmployment +GDP Growth 120- 1200 ~ ~ ~~ ~ 11.0- `= m E 900 0 310.0. :9.0. c3 a a3 750 W ..- 0 8.0.. 600 20 -__ ~~~~~~~ 7.07 450 0 - I I I I I I I I I 1995 1996 1997 1998 1999 2000 2001 2002 2003 Source: "Basic labor market situation in Northeast China ", Juwei Zhang 2005. The employment structure has changed significantly during economic restructuring. The share o f urban employment in theformal sector declined from 88 percent in 1995 to 57 percent in 2003. According to the ownership categones, the share o f workers inpublic sector or in collectively-owned enterpnses reduced substantially, while those worlung in "other" ownership types increased considerably Across zndustrzes, one can observe a 13  significant shift from the secondary to primary industnes, a remarkablephenomenonthat i s differentfrom the rest o fthe country (Figure 2). Figure2. The Structureof EmploymentChanged by ownership by sector 100 - 100 90 80% 80 70 60 60% 50 40 40%- 5 9 30 20 20Y 10 i 0 0%, 1 1995 2003 1997 1999 2001 2003 Demographics and welfare situation of the elderly. Liaoning i s the most urbanized province in China. The 2000 census shows that the urban population accounted for more than 54 percent o f the total population, companng with a national average o f 36 percent. In terms o f the demographic structure, the old age dependency ratio in Liaoning for the population at large i s slightly lower than that o f the national level (16.5 versus 29 7). As in the case of the other two northeast provinces, labor migration has not been significant at the aggregate level and Liaoning has seen net labor inflow from other provinces in recent years. Similar to the situation o f most transition economies, elderly in Liaoning are not at the bottom o f income distribution. Survey o f two cities inLiaoning4shows that, usingincome per capita o f 2,247 yuan (about $280) per year as the poverty line, elderly accounted for 16 percent o f the urban residents in these two cities, but 5 percent o f the poor. (In Shenyang, poverty among the elderly according this definition and data was almost nonexistent). Mean income o f the elderly inthese two cities was about 35 percent higher than that o f the total population. Pension coverage among the elderly averaged 81 percent5 While still limited, it was an essential way to ensure against old age poverty inurban Liaoning. 3.2 ChallengesFaced by theLiaoning Social SecuritySystem Empincalresults presented inthis section are based on the an urban householdsurvey o ftwo municipalities inLiaoning - Shenyang (the provincialcapital) andBenxi, where economies havebeendominatedby state owned heavy industnes. The survey was conducted in2005. Pensioners accountedfor 81% o f those between age 50-60, 85% o f those between age 60-65, and 81% of those above 65. 14  The Liaoning pension system suffered the same systematic issues as those at the national level: increasing system dependency ratio as a result o f aging demographics, declining importance o f SOEs, a highly fragmented and decentralized pension system, and flaws in some fundamental pension policy parameters. Exacerbated by the difficult economic restructuring, the program in Liaoning ran into financial difficulties and required fiscal transfers to pay for current pension expenditure. Liaoning 's pension program witnessed an increasing system dependency ratio. While the ratio of those above 65 years old to those between 15-64 increased gradually by about 1percentage point over eight years since 1995, the number o f pensioners as percentage o f total contributors (system dependency ratio) jumped from 48.6 percent in 2001 to 55.5 percent in only three years according to program administrative data (Figure 3). The Bank's projectionshows that this ratio will increase to over 100 percent injust 20 years. While the ratio o f pension beneficianes to covered workers exhibits the same pattern o f growth as system dependency ratio, the absolute level is significantly lower than the stnctly defined system dependency ratio, indicating that while the coverage has expanded, many in covered workforce are actually not making contributions6 The deteriorating system dependency ratio i s a result o f many factors. On the one hand, economic restructunng has led to shnnking o f contributor base due to high unemployment, poor financial situation o f enterpnses, informalization o f the labor market, and weak enforcement o f contribution compliance. On the other hand, the number o f pension beneficianes increased significantly due to the low effective retirement age, which resulted inpart from usingearly retirement to reduceunemployment. A very highproportion of enterprisesdid not make any social security contributions due to financial difficulty and weak enforcement. Firm level data o f all state owned and controlled industrial enterpnses and non-state industnal enterpnses with sales over five million yuan shows that in 2001, about 53 percent o f the enterpnses did not make any social secunty contribution to pension and health insurance programs. Those who paid have a highly diverse range o f contribution rates, with no sign o f concentration around the legally obligated rates. The actual contribution made for pension and health insurance from these 5,504 industnal enterpnses as a whole accounted for about 10.2 percent o f the covered wage bill in2001. In calculatingthe coverage rate, the paper used total urban employment statistics from the China Labor Statistics Yearbookas the denominator. One shouldnote that it includespart of the urbanlabor force that are not coveredby the "enteTprise old age insurancesystem" accordingto currentpolicies, suchas employees of public service units, civil servants, those who have reachedretirement age but are still working, etc.. It also includesemployeeswith morethanonejobs. 15  Figure3. Dependency Ratios and Coverage Dependency ratios Coverage:particpants and active contributors as % of urban employment 90 60 . 770 808 80 81 5 50 r 70 70 0 4o Beneficiary to covered workers r a t i o 4 ::f-------- A ; r 60 Covered workers 63 1 30 654 601 50 Contnbutors Age 65+ to Age 15-64 ratio 0 I I , 1 3 , 2 0 ~ l , 1995 1996 1997 1998 1999 2000 2001 2002 2003 1998 1999 2000 2001 2002 2003 Source: Datafrom Labor Statistic Yearbook and Liaoning social insurance administration, Liaoning's pension insurance program was highly decentralized and fragmented, hindering the efficient functioning of labor market and overall competitiveness. First, the pension program suffered from very high and differentiated levels of contribution rate. Before the pilot was launched in mid 2001, the total contribution rate for pensions varied from 25 to 30.5 percent o f payroll across municipalities (Table 1). Taking into consideration other social insurance programs, the total social secunty contribution rate was as high as 31 to 40.5 percent in 1999 (Table 2). The high and differentiated contribution rates implied severe burden to labor and bamers to labor mobility Second, thepooling level was very low inLiaoningpnor to the pilot. Inmost cases it was pooledat the sub-municipal (Le. district or county) level. This implied very low efficiency of nsk management, limited resource shanng, and perverse incentives o f non-compliance and mis-reporting. Third, the coverage was relatively low As measured by the ratio o f contributors to urban employment, the coverage was about 65.4 percent in 2001. It was also heavily focused on public sector employees, although the trend i s changing with pnvate sector and small businesses being covered in recent years. Fourth, while the pension benefit structure was unified following State Council Document 26, there were still variations across municipalities regarding spec@ subsidies associated withpensions. A lot o f latitude for specific policy parameters and administrative details was left to the discretion o f the local authonties. Finally, social secunty benefits had limited portability, hindenng the development o f a labor market with adequate mobility Table 1. LiaoningPension Insurance Contribution Rate (YO) Employer Employee Total 1998 12 25.5 - 4 16 29.5 1999 16- 25.5 4 - 6 21 30.5 2000 19- 25.5 5 - 6 24 30.5 --- First halfof 2001 19 25.5 5 - 6 25 - 30.5 2002-2005 19 25.5 -- 8 27 33.5 - Sources: Liaoning SIA 16  1999 2005 Employer Employee Total Employer Employee Total Pension 16-25.5 4 - 6 21 -30.5 19-25.5 8 27 - 33.5 UI 1 0 1 2 1 3 Health about 6 2 about 8 about 6 2 about 8 Maternity <I 0 < I < I 0 < I Work Injury different 0 different different 0 different Total 24 33.5 - 6 - 8 31-40.5 28-34.5 11 39-45.5 I 12 21 33 us 17 7 24 6 8 14 Pension insuranceprograms in Liaoningsuffered from some major designjlaws, inpart determinedby nationalpolicies. Perhaps the most detnmental for the financial health of the system was the pension benefit formula for individual accounts. The amortization factor o f 120, the number that is used to divide the accumulated balance at the time o f retirement to determine the monthly benefit, underestimated the average life expectancy at retirement, and consequently the formula overestimated the monthly pension amount under the individual accounts. At the national level, the life expectancy at average retirement age is around 20 years for men and 27 years for women. For both men and women, life expectancy in Liaoning i s higher than the national average. Inparticular, women will live 5.2 years longer in Liaoning than on average in China. Significant deficit would result in individual accounts and the policy was to tap into the social pooling component to cover individual account shortfalls. This only shifts the problem to the social pool, and leaves the government and future generations o f worker to finance the additional cost. Second, statutory retirement ages were too low given life expectancy, while effective retirement ages were even lower. Early retirement has been prevalent. In 2001, average retirement age in Liaoning was 56 for men and 47 for women. This compares with the estimated average retirement age o f 56 for men and 50 for women at the national level. Such a low effective retirement age could be the results o f early retirement provision for specialhazardous occupations, for disability or sickness, as well as the liberal use o f early retirement as a way to facilitate SOE layoff dunng restructunng. Pension expenditure substantially exceeded contribution revenue. Liaoning's pension system was in financialkash flow deficit. In2000, contribution revenue covered only 72 percent o f the pension benefit expenditure and 69 percent of the total program expenditure. The program was only balanced with substantial fiscal transfer from the central government (Table 3). The estimated amount that should have been accumulated in individual accounts inLiaoning was over RMB 20 billion by the end of June 2001, equivalent to about 24 percent o f Liaoning GDP in2001. 17  Table3. RevenueandExpenditureof LiaoningPensionInsuranceProgram (inpercent ofLiaoningGDP) 2000 2001 2002 2003 2004 Revenues 3.5 3.5 4.0 4.0 4.0 Contributions 2.4 2.4 2.6 2.6 2.8 Employers 1.9 1.8 1.8 1.8 1.9 Employees 0.5 0.6 0.8 0.9 Individual accounts " 0.3 0.6 0.7 0.7 Budgetary transfers 1.o 1.o 1.4 1.3 1.1 From central government 0.9 0.9 1.o 1.o 0.9 From provincial government ... ... ... ... ... Earmarked for individual accounts ... ... ... ... ... From central government ... 0.1 0.3 0.3 0.2 From provincial government ... 0.0 0.0 0.0 0.0 Interest income individual accounts 0.1 0.0 0.0 0.0 0.0 Other 0.0 0.1 0.1 0.0 0.0 Unidentified 0.0 0.0 0.0 0.0 0.0 Expenditures 3.4 3.1 3.3 3.2 3.2 Pensions 3.3 3.0 3.2 3.2 3.1 Other payments to individuals 0.1 0.1 0.0 0.1 0.1 Administration ... ... ... ... ... Repayment of arrears ... ... ... ... ... Other 0.0 0.0 0.0 0.0 0.0 Balance 0.1 0.3 0.8 0.7 0.8 Excluding individual accounts accumulation (principal and interest) ... 0.0 0.1 0.0 0.0 Source: Liaoning provincial finance bureau, 2005; World Bank staff estimates. It WorldBank staff estimate. 21 Benefit payments other than pensions from thepension fund. 11.PILOT ACHIEVEMENTS With substantial support from central government, Liaoning has successfully implemented most of the measures required in the pilot design. Individual accounts were funded at 8 percent o f payroll and the accumulated funds were protected from being used to pay current pensions. The benefit formula for the Basic Pensions was modified. The pension administration was strengthened and the quality o f the services delivered to program participants was improved. Collectiono f pension contributions was shiftedto the local tax authonty as envisaged. Progress has been made in the following key aspects o f the system where the Pilot set out to reform: 18  > to strengthen the credibility of the system by establishing funded individual accounts, > separately managed from the PAYG component; to reduce labor market distortions by accelerating the unification process o f the system, expanding program coverage, reducing overall burden on labor, strengthening > the linkage between contribution and benefit, and enhancing benefit portability; to enhance risk management eflczency o f the public pension program by gradually increasing the level o fpooling; P to improve program administration by improving information and financial management, andpromotingmore effective enforcement o f collection. Liaoning Pilot was a milestone in China's pension system development. It supported the implementation o f the policy framework envisaged by the State Council in 1997, and accumulated expenence which could help to inform future reforms. The policy changes introduced in the subsequent pilots after Liaoning and the modifications made to the national policy at the end o f 2005 demonstrated the value o f expenmentation and the learning that took place inthe long and complex reformprocess. 1. Strengtheningthe Credibilityofthe System As required by the Pilot, finds have been accumulated in individual accounts and a clear separation o f these accounts from the social pooling pillar was established. By the end o f 2003 when the pilot phase ended, RMB8.88 billion was accumulated in individual accounts in Liaoning. The amount reached RMB13.8 billion by the end o f 2004. Moreover, separate financial management o fthe funds under the basic pension and those in the individual account was achieved through new contributioncollection procedures and a new accounting system. The accumulation o f individual account savings was facilitated by substantial transfer from the central and provincial government. The pnnciple was for the central and provincial government to finance the gap created in the PAYG program due to the diversion o f 5 percent o f the payroll into individual accounts. Of the 5 percent estimated contributory wage base, 3.75 percentage points were to be financed by the central government and 1.25 percentage points by the provincial government. In reality, the Ministry o f Finance has transferred a flat amount o f RMB1.44billion annually since the Pilot started, and the local government transferred an annual RMB70 million in2002 and 2003. The extent to which the credibility o f the system has been strengthened is difficult to verify However, preliminary findings from a representative qualitative survey seem to suggest that the public ingeneral have confidence inthe government pension program and are aware o f how their pensions are established. About 70 percent o f the 837 people surveyed indicated that they perfer individual account pensions to voluntary pnvate savings as a source o f old age income. About 60 percent o f the contributors claimed to pay 'The quantitativesurvey covered 1200individualsinthree majorcities inLiaoningprovince. Excludingthe current retirees andthose working inthe public institutionsandcivil service(not coveredby the public pensionprograms),there are 837 peoplewho respondedto questionsconcerning individual accounts. 19  close attention to the amount saved in their account, and 92 percent said that they trusted the accuracy o f the figure that they saw about their accounts. While household saving has been affected by the reform, it i s difficult to venfy whether this is due to strengthened credibility of the pension system. In Liaoning, overall gross domestic saving rate increased by 0.7 percentage point in 2002 and 3.8 percentage points in2003. Loolung at the urbanhousehold savings rate, after peaking in 2001, it declined significantly in 2002 and 2003. This contrasts with the development in the rest o f China where the urban savings rate remained unchanged inthese years. The divergent trend may suggest that householdhave taken into consideration the establishment of their individual account and the expected income from the overall public pension program and adjusted their personal savings behavior. However, it is difficult to distinguish this impact on confidence from the direct impact o f the reform on net incomes, which declined as a result o f the increase inmandatory contributions. 2. ReducingLabor Market Distortions There has been gradual but slow progress towards improving labor market efficiency and reducing distortion. Contribution rates across theprovince have been gradually converging. Employee contribution has been unified at 8 percent o f individual's wage. Contribution rates o f employers, however, have not converged significantly In 2004, there were five contribution rates across municipalities, ranging from 19 to 25.5 percent. The provincial government plans to first lower employer contribution rates from those above 20 percent to 22 and 22.5 percent by the end o f 2005, and then to 20 percent in all municipalities (with the exception o f Dalian and Panjin which will remain at 19 percent) by the end o f 2006. Inaddition, the contribution rate of small business and the self-employed was unified at a total o f 18 percent o f which 8 percent goes to individual accounts. Adding to the complication was vanous industnal funds that were decentralized from the central to provincial government. The combined employer and employee contribution ranged from 18 to 41 percent (e.g., coal) in the late 1990s, and has been unified to the current 20 percent. Policies have been modified to allowfor wider coverage of pension insurance during thepilot, but the outcome is mixed. Employees in enterprises o f all ownership types are covered inpnnciple. Coverage has also been extended to the self-employed, rural migration workers, and freelance professionals duringthe pilot period. However, there are few measures to assess how effectively the policy change has been implemented. Moreover, two points worth highlighting when considenng coverage expansion. (a) While the ratio o fparticipants to total urban employment increased, contributors as a percent o f total urban employment actually declined from 65.4 percent in2001 to 60.1 in 2003 (Figure 3); (b) Given the fact that some fundamental pension policy reforms have not taken place, and that a significantly different contribution rate i s applied to 20  certain part o f the newly covered labor force (e.g. small business and self-employed), expanding coverage seems to have been accomplished at longer term cost and additional distortion. Portability of pension benefts has improved. With individual accounts becoming fully funded, individual account savings in pnnciple can be transferred when individuals move between different pooling areas within or across provinces. The amount for which accumulation had not occurred pnor to the Pilot i s to be honored at the point oftransfer by tapping into the resources inthe social pooling pillar. The Pilot policy explicitly allowed for this portability, and the authonties reported that there were no major problems encountered when such transfer o f pension rights took place. Not enough data was obtained to conclude how well such transfers were executed. Linkage between individuals contribution and benefts has been strengthened. By ' crediting contributory services beyond 15 years with additional pensions, the basic pension component explicitly established linkages between contribution and benefit at the individual level. However, there i s little empincal evidence to assess its impact on individuals' decision to participate and contribute to the program beyond the minimum 15 years' requirement. 3. EnhancingEfficiencyinRiskManagement Progress inachieving a higher level poolinghas beenlimited dmng the Pilot. Only 4 out of the 14 municipalities have achieved municipal levelpooling ofpension finances while the other 10 municipalities are scheduled to do so by the end o f 2005. For this purpose, municipal pooling means that contribution rates are equalized throughout the municipality; social insurance management i s vertically integrated in the municipalities, and for the most part, financial flows are adjusted by the municipal government to equalize the burden among the county and distnct governments. The Provincial Adjustment Fund (PAF) wasfurther developed during the Pilot to equalize social security burdens among various municipalities. Established first in 1995, the PAF requires that each municipality contribute 1 percent o f the actual contribution collected to the provincial government and that the fund will be redistributed according to an equalization formula' Dunng the pilot, the size of contribution has been increased to 5 percent, and in 2004, the contribution from the provincial level pool was further increased to 10 percent, allowing for more significant equalization. However, while the equalization formula appears incentive compatible * The redistributionformulaspecifiesthat the amounteachmunicipalityreceivesinredistribution(R) is determinedaccordingto the following formula:R =funding gap *(1fperformance index), where "funding gap" is definedas "estimatedrevenuethat shouldbecollected(according to the estimatednumberofpeople that shouldbe coveredandtargetedcollectionrate specifiedby theprovincial government) -estimated expenditurethat shouldbepaid from the socialpoolingcomponent-contributionmadeto the PAF", all subject to the verificationof the provincialgovernment. 21  for municipalities to effectively participate in the scheme, whether this happens in practice depends on the specifics o fthe implementation o f the rules. 4. ImprovingPensionProgramAdministration If the achievements in the above areas have been somewhat mixed, the administrative changes introduced inthe Pilot have improved both the quality o f the services delivered to program participants and the financial integnty o f the province's social insurance programs. While some measures are to be further implemented, substantial progress has been made inthe following main aspects: Shifting of social insurance record keeping and benefit payment responsibilities from enterprises to social insurance agencies. Essentially all o f the record keeping and benefit payment responsibilities in the pension and unemployment insurance programs have now been shifted to the social insurance agencies, relieving the administrative burden imposed on enterpnses. Transfer of responsibilityfor collecting contributions to local tax agencies. Dunng the Pilot, the responsibility for contribution collection was shifted from the local social insurance agency to the local tax agency in all but one municipality There is yet to be sufficient data to assess the impact o f this change on collection, but this arrangement can help to improve compliance by allowing for the possibility to easily cross check contributionwage base with that o f income tax. Enterprises can also now make social insurance contribution and local tax payments in one stop at the tax agency, reducing the compliance burden. In most municipalities, however, they still need to record changes and venfy payment forms with local social insurance bureau. Development of information system linking social insurance agencies at provincial, municipal and county/district level. To implement the current pension policies which require each individual's lifetime contributory record, an efficient information management system is essential. The Pilot supported the development o f a new social secunty information system at the provincial level which i s capable o f maintaining information on contributions, work histones, individual account balances, and benefit payments. The system consists o f a provincial level database and a set o f municipal level databases, and also allows for the direct shanng o f data between provincial and central governments. The system also has the capacity to automate most o f the steps in calculating social secunty benefit entitlements although for the foreseeable future, pension calculations will still rely, at least in part, on manual checlung through the paper records. Separation of individual account and socialpoolingfunds. To prevent the diversion o f contributions intended for individual accounts to finance current benefit payments, Liaoning achieved the separation o f funds through new contribution collection procedures and accounting systems. Pension contributions from individual workers are segregated at the point o f initial collection, deposited in a special treasury account in  the central bank, and transferred directly and immediately to an account controlled by the provincial-level finance bureau. The rest o f the social insurance contributions are deposited inanother set o f special accounts located incommercial banks andcontrolled by the financial bureaus at the same poolinglevel. Improvementsin internal controls through extension of two track accounting system to all branches of social insurance. Under the two-track system, contribution collections are immediately transferred to special accounts controlled by the respective financial bureaus. Fundsnot needed immediately for benefit payments are deposited in interest-beanng accounts. The segregation o f contributions to the individual accounts from all other contributions has improved the internal controls on contribution revenues to social insurance programs. Unification of social insurance finance and management. Social insurance management was integrated vertically in four municipalities while the remaining ten are to convert to this model by the end o f 2005. Vertical integration means that county and distnct social insurance agencies are transformed from independent agencies attached to local government to departments o f the municipal social insurance agency and that their staff would become employees o f the municipal government. With vertical integration o f the administrative units, equalization o f the burden across the entire municipality, and automation o f many o f the administrative tasks are three important steps toward the goal o f a standardized social security system. Such a standardized system would have greater assurance in the implementation o f uniform costs, benefits and administrative procedures that will provide equal treatment among enterpnses and individuals inthe municipality 111.POLICYAND INSTITUTIONAL CHALLENGES Despite the accomplishments made in a very short timeframe and in a very difficult economic environment, to develop a pension system that can meet the objective o f poverty relief, income redistribution, insurance against the nsks o f longevity, and consumption smoothing, a model that is viable for the nation, major policy and institutional challenges remain to be addressed. Some o f the issues can be tackled at the provincial or local level, some would require adjustment o f national policies. This section makes the following main observations about the pension program in Liaoning: The demographic development indicates that Liaoning's old age dependency ratio will nse rapidly over the next four decades and that the only penod where working age population will experience positive growth is between now and 2009 This presents a very short penod o f favorable demographic structure, often referred to as the "demographic dividend" 23  Under the policy scenano specified in the Pilot, the pension system inLiaoning i s not financially sustainable. However, a closer look at the program would reveal that a significant share o f the cost is due to histoncal legacy and system transition. The pension program has not been able to function as an insurance scheme. Instead, it has been expected to help address a range o f social and economic concerns. Stripping off the legacy and transition cost, Liaoning's pension program can be financially sustainable subject to the adoption o f a number o fparametnc reforms. While substantial accumulation was realized according to the pilot policy, progress has been very limited in developing a sound governance structure for individual account pensions. Financial market is under-developed andnsk adjusted returnto pension fund investment has been very low The significant differential between rate o f return on investment and wage growthwill likely remain over the coming decade. Social secunty administration can be improved significantly based on Liaoning expenments and it will have a major impact on the overall performance o f the pension system. Attention should be placed on developing necessary legal basis, reorganizing the administrative structure and improving administrative processes, as well as on enforcing compliance. This can and should take place even if there may not be full consensus o f future policy framework at the national level. The paper argues for reforms inthree main areas: 9 To establish inherent financial sustainability ofthepensioninsurance program, (a) a set o f parametric reforms need to be undertaken inthe short to medium term to correct the policy design flaws; (b) costs associated with economic and system transitions should be considered separately to ensure the integrity o f the pension insurance program; (c) a transparent and sustainable financing mechanism should be developed while the contribution rate should be brought down to the extent possible to reduce the distortion inthe labor market; the government could consider financing the legacy and transition cost separately; (d) partial funding is desirable to benefit from the demographic dividend, but the current fund management arrangement requires modification; and finally (e) pension program coverage should be expanded under reformedpolicies. P To improve the investment outcome of pension reserves, the authonty needs to improve the transparency o f management and investment, develop necessary regulations, and strengthen internal control. To achieve better investment return, a broader set o f investment instruments and professional investment management should be allowedinawell regulated andsupervised environment. 9 To improveadministrative efficiency, additional steps are needed to further strengthen financial control and overall program management. Some involve completing initiatives that have already been started under the Pilot. Others involve adapting and customizing international best practice o f processes and procedure that are now employed in other countries to the Chinese context. More complex reforms would need to be hrther pilot tested. 24  1.Developinga SustainablePensionInsuranceProgram 1.1Long-term Prospectof theLiaoningPilotScenario The Adne Phenomenon. Results from the population projection show that Liaoning's old age dependency ratio will nse rapidly and that the only penod where worlung age population will expenence positive growth i s between now and 2009 (Figure 4). Starting in2010, its shareofthe working age population is expected to decreaseuntilitreachesthe lowest point around 2050. Meanwhile, the old age population and its share inthe total are projected to grow steadily until it reaches the highest point at around the same time, 2050. As a result, the population old age dependency rate' is expected to nse rapidly and peak at 90 percent in2050. Figure4. The Agingof Liaoningandthe DemographicDividend Theold age dependency ratio will be rising rapidly Thewindow where working agepopulation will andpeak around 2050 experiencepositzve growth is narrow 8%- 0% I 2001 2011 2021 2031 2041 2051 2081 2071 4% Yea1 Year ;ource: WorldBank staffestimates. Financial sustainabilitv of the urowam. The report examines the financial sustainability o f the pension system using several standard indicators: (a) The Current balance and Financing gap - Calculated as the difference between revenue and expenditure o f the PAYGscheme, the current balance over the projectedpenodprovides a dynamic pictureo f the financial health o f the system. The financing gap is calculated by summing the net present values o f the current balances, often expressed as a percentage o f the base year GDP" (b)Implicitpension debt (IPD) -Calculated as the netpresent values ofprojected liabilities accrued to date under the pension system and often expressed as a percentage o f the current year's GDP IPD i s a useful indicator to measure the stock o f all outstanding liabilities and i s commonly used to compare the maturation and fiscal burden o f different pension systems. (c) Required contribution rate -measunng the contribution rate required to allow a pay-as-you-go pension system to manage on a cash flow basis whereby 9 Populationold agedependencyratio is definedas the ratio derivedby the numberof olderpersons (ages 60 andover) dividedby the number ofworking age individuals(ages 15-60). loThroughout this study, we have usedthe GDP statisticsprior to the adjustmentmadeinearly2006. 25  expenditures can be met solely through contributions andmvestment income. The required contribution rate i s often expressed as a percentage o fthe wage bill. Under the pilot policies, the current balance o f the social pooling component of the program will remainindeficit throughout the projected penod'' (Figure 5). Due to the fact that the pilot provision requires the social pooling account to absorb all payments that are due from the individual accounts once the account balance o f any individual has been exhausted, there will be a significant j m p m program deficit under the social pooling portion around year 2030. To ensure that the rogram i s inbalance on a cash flow basis when financed solely by contribution revenueI ,the required contribution rate would be f 48 percent o f contributory wage for the two components combmed (as compared wlth the current 27-33.5 percent). TheJinancing gap i s about 206 percent o f 2001 GDP. Under the Pilot policy scenmo, the implicit pension debt amounts to about 153 percent o f GDP in 2001, o fwhich, liabilities under the basic pension component accounted for 124 percent of GDP Not surprisingly,the IPDunder the Basic Pensiondeclines over time, while the debt under the individual accounts becomes an mcreasingshare o fthe total IPD. Companson between the national and Liaoning projectionresults usingthe same Liaomng Pilot policy parameters highlights the impact o f the unfavorable economic and system conditions. The required contribution rate for the national program i s 35 percent of wage (companng with 48 percent inLiaonmg) and the financing gap is 64 percent, about a third o fthe fmancing gap inLiaoning. The IPD is 132 percent of 2001 GDP Figure 5. Implicit Pension Debt and the Current alance under the Pilot Scenario ImplicitPensionDebt (as %of GDP) Current balance (as %of GDP) ?a0o 0.5% 160 0 I -- 140.0 0.0% I 2001"" " ' ~ ~ 2010 2020 2030 2040 2050 2060 2070 2075 120.0 100.0 BD.0 80.0 40.0 20.0 -2.0% 2001 2010 2020 2030 2040 2050 2060 2070 2075 Source: WorldBank staff estimates. 11 The actualcollectionrate (defined as actualcontributorsto covered workers who axe employed) was 84% in2001,76.1% m2002 and774% in2003. Theprojectionassumesthat the collechonratewill increaseto 80% in2004, and eventually 85% by 2030. 12 No fund accumulationis availableunder the socialpooling, henceno investment incomeis availableto helpfinance ongomgexpenditures. 26  1.2 Reform Options toImprove Sustainability 1.2.1 Parametric reforms: It appears that the long term financial sustainability of Liaoning's old age insurance system under the pilot policy couldbe substantially improved providedfour specific parametnc reforms are undertaken: Maintaining actuarial balance in individual account pensions13- This can be achievedby amortizing the Individual Account balance based on the average remaining life expectancy at retirement. This will immediately put the whole system on a much sounder footing as the pension benefits are now directly linked to the account accumulations without relying on the social pooling finds to make up the shortfalls. The change in amortization methodology will undoubtedly lead to a reduction o f replacement rate. Figure 6 illustrates the impact o f the changes where the annuity converted from the individual accounts reduced from the case where a factor o f 120 was used (red) to the case where average life expectancy at retirement was used (clear). The replacement rate for a male worker retinng at age 65 would be around 43 percent, companng with about 51percent before the modification o f amortization factor. While any reduction o freplacement rate would be difficult, it i s important that the design flaw be corrected as early as possible to minimize dissatisfaction and false expectation over long term. The impact o f this change on the pension entitlement o f the cohorts who will retire in the next few years would be small as the pension amounts derived from these individual account balances will be a relatively small part o f their overall pension. Gradually increasing and unifying statutory retirement age to 65for both men and women while eliminating all early retirementprovisions- The retirement age increase can be phased in by raising the retirement age by six months every year. The process would be very gradual and it would take over 20 years for men and 32 years for women to complete this process. It i s important to note that eliminating special provisions for early retirement for hazardous occupations does not mean that workers in those industnes will not be properly compensated or protected. Instead o f compromising the integnty o f the pension insurance program, the government could consider compensations in the form o f enhanced wage and subsidies, as well as better work conditions. Adjusting the contribution rate for small business and the self-employed - The current policy is that for small businesses and the self-employed (both the business owners and their employees), the contribution rate i s a total o f 18 percent, o f which 10 percent i s dedicated to the social pooling component. In the meantime, these contributors will be entitled to the same basic pension provided that they meet the 15 year minimum contributionrequirement. There is no economic justification for such a subsidy and redistribution. Considenng that almost a third o f the contributors in l3See section IV.1about recent development inthis regard. 27  Liaoning now fall into this category, such a policy also has major financial implications for the system. Adopting a more transparent and predictable method to index pension benefits - Currently, post-retirement pension is not indexed according to any particular formula. A general policy of the State Council was to index pensions based on 40-60 percent o f average wage growth o f the local economy dung the previous year. In practice, however, the adjustment level i s determinedby the central government every year. It is also common that adjustment i s made by a flat amount across the board. Given the fact that the current generation o f pensioners has been earning relatively low wages throughout their worlung lives, it i s reasonable for them to share the benefit o f current growth. On the other hand, wage indexation can become very expensive and would not be able to protect pensions in the case o f negative wage growth when wage becomes more stabilized. Given the pros and cons o f the two indexation approaches and the unique stage o f the economic development o f China, equity could be improved by partially indexing pension to wages as a transitional arrangement and switching to full pnce indexation later. For the purpose of this exercise, the analysis assumed that pensions will be indexed to 60 percent o f wage growth until 2026 and then to 100 percent o f inflationfrom there on. The combined impact o f the four reform measures above would lower the required contribution rate from 48 percent to 32 percent o f covered wages. The IPD i s expected to decline from the onginal level o f 153 percent to 126 percent o f GDP in 2001, while the financing gap drops drastically from 206 percent to 42 percent o f GDP in2001. Figure6. Impactof ParametricReforms Parametric reforms will lead to signi3cant Replacement rates under two amortization - improvement of thefinancial situation of the system methodologies at different retirement age (current balance as % of GDP) (pension as % of averagefinal wage) 0 5% 60%. KZZL Usingfactor of 120 With parametncreforms 0 Usingaveragelifeexpectancyat r 0 0% 50%. From Basic Pensions 40% -0 5% / 30% -1 0% 20% -1 5% 10% -2.0% 1 0% 45 50 55 60 65 Source: WorldBank staffestimate. 1.2.2Seuaratinpthe imuact of leEacv cost: It is important to decompose the total pension liabilities so as to clearly understand the financial requirement o f a mature pension insurance program versus the financial impact o f histoncal legacy and transitions. A 28  sustainable pension insurance scheme should be able to finance the former with insurance contributions, whereas addressing the histoncal legacy is likely to require subsidies either through fiscal transfers at vanous levels o f government or perhaps even higher contributions by current and future contributors. Separating these liabilities conceptually and financially would help to ensure the integnty o f the pension insurance scheme, and clanfy the responsibilities o f vanous levels o f government. 0 Estimating the legacy cost. To illustrate the concept of decomposing total pension liabilities, we start with the system where systemic flaws have been removed. The current total liability consists o f three main components. The first component i s the early retirement uensions for all existing and future pensioners who are below the statutory retirement age. This cost incurred as part o f the policy decision in the past that used pension insurance as a way to address some o f the social impact o f SOE reforms, and to compensate low level o f income o f hazardous occupations. Some half a million workers (roughly 15 percent o f the total stock o f pensioners) retired before reaching the statutory retirement age. The fiscal implication o f such policies needs to be quantified and perhaps addressedseparately from the regular program cost. The second component is the transition cost incurred as part o f the pension system reform. The government set the targeted replacement rate inthe new scheme to be 58- 60 percent14, significantly lower than the prevailing average replacement rate. While the average replacement rate being paid from the old age insurance system has declined from 88.86 percent when the Liaoning pilot started in2001 to 78.56 percent in2004, it will take a long time to reach the target level15 Furthermore, the transition generation (often referred to as the "middle men'? is entitled to supplements in the form o f transition pensions to make up for the "lost years" when they were not contributing to the Individual Accounts. Ina way, these are liabilitiesofthe past systemandshouldbe identifiedas such. Finally is the cost of the mature uension insurance urogram. For the purpose o f this exercise, mature pension insurance program is defined as the total pension liabilities when applying the benefit policy for the new generation16to all program participants. That is, it assumes that everyone who retires from the system (past and future) will be covered by the same benefit formulas under the Basic Pensionand Individual Accounts based on their anticipated years o f service, as if they have contributed the equivalent o f fullcareerunderthe individualaccounts. Preliminary estimates suggest that the cost o f early retirement pensions were around 0.4 percent o f GDP in 2004, and the system transition cost 2.2 percent o f GDP This compares to the total pension program expenditure o f 3.1 percent and budget transfer o f 1.1 percent o f GDP in 2004. This implies that for the years dunng pilot, budget 14 The targeted replacement rate i s estimated by proratingthe target established by the government inthe scenario where 11% o fpayroll is to be allocated to individual account. l5Infact, as shown insection 111-1.2.5, even this targeted rate is not attainable given the estimatedreturn to investment. l6 is, That those who started to contribute to the system after 1996, often referred to as "new men" 29  support was adequate to cover the cost of early retirement pensions, and part o f the transition cost. The rest o f the system transition cost (as well as the cost o f systematic flaws) were borne by the contribution revenues. Measured in terms o f the IPD as a percent o f GDP in 2001, one can see that only 56 percent o f the total IPD was liability associated with the mature system, whereas 17 percent o f the total IPDwas due to systemic flaws. About 4.3 percent o f the total IPD was due to early retirement; the transition cost accounts for about 23 percent o f the total IPD. Inthe mature system, the program balance will be positive throughout the projection period, and a total contribution rate o f 27 percent i s needed to maintain program balance. The total financing gap i s -140 percent o f 2001 GDP (Figure 7). The fact that the transition cost is so significant reflects the difference between what replacement rate the mature system can expect to deliver and the actual current replacement rate. For example, in the mature system, the total replacement rate for a male worker at age 60 is estimated to be 43 percent, and that for a female worker at age 50 is 31 percent, companng with the average replacement rate o f 88 percent in 2001. As it is shown inthe following sections, the targeted replacement rate of about 58-60 percent that the government identified i s not attamable within the current structure and economc conditions. Figure 7. Decomposing the Cost Cost associated with economic and system I n the mature system, current balanceof the PAYG transitions accounts for a malor part of the total system will be in surplus untilpast 2050 (current balanceas %of GDP) 2 0 - 1 5 1 0 0 5 0 0 , -05% I -10% -15% Scenano -2 0% bource: WorldBank staffestimate. It is important to note that there may be vanous ways to define legacy cost and that what i s presented here i s an illustration o f the basic concept. The government would also need to consider potential options to reduce the legacy cost, such as modifying the indexation rule for the transition generations. One should also take a longer term perspective when assessing the impact and size o f different components o f the liabilities. The liabilitiesresulted from the economic legacy cost i s front-end loaded as they capture all the cost that have already occurred (e.g. early retirement). Incontrast, 30  the savings from correcting the systemic flaws (e.g., delaying retirement age and fixing the amortization factor) will only be realizedinthe long run. Developing a transparent and sustainablefinancing mechanism. As part of the Liaoning pilot, pension expenditures are financed partly by government. Budgetary transfer equivalent to between 1 and 1.4 percent o f Liaoning GDP allowed the accumulation o f individual account to matenalize as required, and kept the pension system's overall budget balanced dunng 2001-2004. It appears that the financing arrangement was not made according to a clearly identified rule specifying the responsibilities o f vanous levels o f government. Between central and local governments, to cover the financing gaps created by the diversion o f contributions to individual accounts, the pnnciple was for the central and provincial government to finance 3.75 percent and 1.25 percent o f the estimated contributory wage bill respectively In reality, between 2001 and 2004, transfers were made in annual flat amounts. Concerning the overall budgetary transfer which was significantly higher than this transition cost, the method used to establish the level was not clear. Within theprovince, redistribution across municipalities was inpnnciple facilitated by the ProvincialAdjustment Fundaccording to a redistribution formula. However, some o f the main elements o f the formula were not explicitly spelt out, leaving room for discretion. Looking ahead, budget transfers should be based on a clearly defined rule that indicates the government's strategy to fimding. Such a rule would also limit the government's commitment, and thereby its liability As illustrated above, the inherent financing need o f a mature pension insurance program can be separated from the impact o f economic and system transitions. It would allow the pension insurance scheme to operate in a sustainable and transparent manner. Policymakers could consider separately financing the cost beyond that o f the mature system, but the transfer has to be fiscally and politically sustainable. Key considerations concerning the "economically justified" transfers are the following: a) Does the general government have the fiscal capacity to support the transfer' With central government involvement the requiredtransfer inpnnciple seem to be feasible relative to the fiscal capacity o f general government. This judgment i s based inpart on the presumption that the legacy costs in other provinces are likely to be lower than in those in Liaoning, which, as discussed above, went through a particularly difficult economic transition. This would mean that the nation-wide costs in 2004 should have been lower than the 2.2 percent o f GDP estimated for Liaoning, compared to total general government expenditure estimated at over 25 percent o f GDP However, inprovinces where the fiscal capacity i s limited and the histoncalburden i s heavy, it would be difficult for the provincial government alone to shoulder the "economically justified" transfer. Clarifying the responsibility between the central and local government anddeveloping the mechanism consistent with the general intergovernmental transfer system would be necessary 31  b) Would central government transfers that vary between provinces be sustainable politically7 Economically justified levels o f transfers would vary significantly betweenprovinces. They may invite political and inter-regional issues. c) Apart from interregional issues, decisions on how much to transfer are also questions of distribution over generations. The policymakers may decide not to transfer the full economically justified transfer. That could be a reasonable outcome, as long as the decision is made explicit and after proper evaluation o f all considerations. Funding for histoncal legacy can have multiple sources. General taxation revenue, proceeds from the sale o f state-owned shares or state assets, and debt borrowing are all potential options. The key is to keep it as least distortive as possible. More broadly, the government needs to address some o f the histoncal legacy in more comprehensive social protection programs. Developing basic social pension and safety net programs to assist the elderly poor are measures the government may want to consider. The current urban dibao system in pnnciple provides such a safety net, but its coverage andeffectiveness is severely constrained by the fiscal capacity o f the local government and couldbe further strengthened. 1.2.3 Gradual fundina of individual account: The analysis described thus far confirms that, after the necessary reforms discussed above are adopted, the current pension system -apay-as-you-go basicpensioncombinedwithadefinedcontribution individualaccount - can serve as a viable framework to provide for old age income secunty in China. Although there seems to be a consensus that pre-fundingo f future pension obligations is generally a desirable policy, there have been different views as to whether the individual accounts should be funded, and if so, how much and how quickly should the funding take place. Potential benefits of funding individual accounts. Whether funding is desirable has been a topic o f debate in countnes facing the need for pension reform. Proponents argue that funding is advantageous from several aspects. From the individuals' perspective, funding individual accounts may increase their confidence in the system and therefore it i s more likely for them to participate andcomply Fromthe economic perspective, fundingmay increase national savings (recognizing that this can be at least partially offset by a reduction o f personal savings), improve the allocation o f productive savings through institutional investors operating on deepened capital market, reduce labor market distortions through strengthened linkage between individual's contribution and benefits, and improve labor mobility from enhanced benefit portability From the nsk management perspective, establishing a funded component and allowing the accumulation to be invested in a diversified portfolio would allow for diversification o f economic and political nsks, both nationally and 32  internationally This last benefit may be partly offset by additional risks associated with funding, such as investmentnsks andmanagement nsks. Pre-conditions. There are many conditions, however, for the economy and the public pension system to be able to capture the desirable outcome o f funded pensions. Macroeconomic stabilityi s required to support the reform; financial market needs to be able to productively allocate the resources at an acceptable level o f nsk in a well regulated and supervised environment; pension administration needs to be able to operate efficiently to meet the requirements o f such a program. Partial andgradualfunding in China. Would developing mandatory fundedpensions be beneficial for China? This is a question that cannot be fully answered by evaluating the Liaoning pilot alone. It would require more in-depth analysis o f the pre-conditions, and a national perspective. Analysis o f China's long term demographic and economic trends, the histoncal legacy that was ihented, and the expectation o f the society suggests that a pension system consisting o f social pooling and funded individual account can serve as the basic pension framework in the long run. China can benefit from the potential advantages associated with funding, and inparticular, enhanced credibility o f the reformed system, increase o f savings (Box l),anda deepenedmore sophisticated capital market. 33  Box 1. SavingsinChina andIts Relationto PensionReform One factor commonly taken into account in considering a funding approach is the impact on national saving. This was a key consideration behind moving to a funded system in several Latin Amencan countnes, even though international ernpincal research on the impact on saving is actually not very strong (see Appendix 2, page 10). An argument made in China against funding is that the current savings rate i s already very high, and that the government currently actually aims at reducing saving. Nevertheless, in the context of reform o f the pension system, a longer term perspective is required when examining China's saving trends. Indeed, despite China's current high national saving, China's saving is likely to come under downward pressure inthe coming decades. Ernpincal analysis (Modigliani and Cao, 2004) suggests that China's high household saving is a result o f increases in income growth and changing demographics. In particular, the analysis suggests that household saving has nsen due to (i) economic the reforms since the end-1970s' and the increase in growth and growth prospects they generated; as well as (ii) introduction o f the one-child policy which led to a gradual the increase in the ratio o f employment to total population and undermined the traditional role o f the family in providing old-age support. Another factor that has dnven the savings rate high is the uncertainty that households face dunng transition. Notable examples are the lack o fadequateand credible old age pensions and health protection. Looking ahead, however, several developments are likely to put downward pressure on household and national saving. In particular (a) an eventual moderation o f economic growth and demographic changes are likely to put significant downward pressure on household saving. Purely for illustration, a slowdown in long-term GDP growth of 2 percentage points would, according to Modigliani and Cao, lead to a decline in the household saving ratio o f a whopping 5 ,percentage points. In addition, the projected increase in the "employee to minor" ratio from around 2.4 now to 2.0 in 2075 would according to World Bank projection lead to an additional 4 percentage point decline in the household saving ratio. (b) Pension system reform may increase national saving, depending, amongst others, on the degree o f funding in the eventual system and household's response to these policy changes. However, any pension reform-related upward pressure is unlikely to compensate for the downward pressure coming from a growth slowdown and demographics; (c) Further financial markets development will reduce the number o f credit constrained people and small enterpnses, and the associated need for saving for anticipating purchases o f consumer durables, life-cycle events, and investment. Finally, (d) the reform o f social security programs and better social service provision will reduceuncertainty households face andhousehold savmgs. The employeeto m o r (child) ratio i s their indicator for demographicpressures. 34  In the short to medium term, the policies as represented in the Liaoning Pilot face significant challenges in achieving the desired outcomes. The decision o f funding should depend on whether the government can take appropnate actions to improve investment policies, develop necessary institutions, and carry out parallel reforms. Without such actions, pre-funding would imply not only significant fiscal input, low investment return and consequently low level o fpensions, but also nsk o fpotential loss o f the accumulated fund. If appropnate actions can be taken to reform policies and improve governance, partial funding may indeed help to increase the willingness o f people to participate, and capitalize on the demographic dividend by setting aside reserves pnor to accelerated population aging. Itis important to point out that funding is a continuum. The modelthat China currently adopts i s a partial funding - partial notional account approach. At the provincial level (inthe expenments subsequentto the LiaoningPilot), part of the individual account is funded and earns an investment return in the market, while the remaining part o f the individual account i s credited with notional interest set by the government. At the national level, the establishment o f the National Social Secunty Trust Fundallows for a different form o f pre-funding. Indeed most o f the Notional Defined Contribution (NDC) systems that are implemented in the world have allowed for pre-funding o f different forms and different levels. The level o f funding can gradually increase as capital market and other conditions for a well functioning D C system are met. Whether the reserves should be managed as part o f a national/provincial trust fund or as part o f individual account are subjects to be assessed and decided separately from the fundingissues. Box 2 provides a bnef description o fthe basic concept o fNDC. 35  Box 2. NotionalDefinedContribution (NDC) System Notional contribution accounts are designed to reap some o f the benefits o f a defined contribution plan, where the pension benefit depends on contributions and investment returns, and pension contributions are tracked in accounts that earn a rate o f return. In NDC, the return earned on contributions is a notional one, set by the government and not the product o f investment returns i the markets. This notional interest rate is similar to the interest rate inan unfundedscheme-that is, the growth rate of covered wages in a mature system. Upon retirement, pension benefit is determined by dividing the accumulated notional capital plus notional return by an annuity factor that reflects the remaining life expectancy o f the cohort at the age o f retirement. The benefits are typically price indexed, but can also have a mixed pnce-wage indexation. NDC scheme maintains pay-as-you-go finance o f the system, but a buffer fund is often established to handle the costs of economic anddemographic shocks. Several countnes including Poland, Sweden, Latvia, Italy andBrazilhave undertakenreforms inthe last decade that transformed their defied benefit system into NDC schemes. They introduced the scheme largely because o f the following appealing features o f the NDC structure. First, the NDC benefit structure is simple, transparent, and actuanally fair by design. As in funded individual account schemes, workers can easily verify the amount o f contributions accrued under their name. Second, the simple structure o f NDC can have important incentive effects on labor supply and retirement decisions. There is strong linkage between contributions made throughout one's working life and the pension benefit he/she will receive. Contributions are less likely to be perceived as taxes, and hence distortions on labor market are reduced. In countnes where decisions to postpone the statutory retirement age are difficult to make, NDC offers strong incentive for individuals to retire later which can partly achieve the desired outcome. The changes inpension benefits due to earlier and later retirement are part o f the benefit design and do not require extra political decisions. Third, the approach allows for aneasy andquick harmonization betweendifferent schemes (such as for public andpnvate sector workers) andhelp to improve labor mobility As with other pension designs, NDChas its limitations and whether the benefitofNDC can be filly realized will depend on the particular country's situation and how the program i s designed and implemented. First, while the NDC systems include some automatic stabilizers to improve the system's finances in the face o f adverse shocks, it cannot help solve major liability overhang from the past. Second, applying appropnate notional interest rate and establishing correct life expectancy o f different age cohorts in the NDC formula are critical for the financial sustainability of the program. Failing to do that would lead to sub-optimal design. Third, as with other partially funded programs, substantial administrative requirements would be necessary to ensure that the buffer fund is well managed. Finally, to address redistributive concerns, an explicit and transparent transfer o f fund to the system and individual accounts is needed to avoid excessive handouts of unfunded promises. See "Old Age Income Support in the 21'' Century", World Bank 2005, and "Pension Reform through NDCs: Issues and Prospect for Non-Financial Defined Contribution Schemes", ed. Hohann and Palmer, (World Increasing the return on pensionfund investment. One fundamental challenge that the system faces in the near term i s the low return on investment. Up to now, the interest rate credited to the Individual Accounts lags far behind the rate o f wage growth 36  nationally (Figure 8). Infact, both the fully funded accounts inLiaoning Province and the National Social Security Trust Funds have earned only modest returns, in the order o f 3 percent nominal, or barely above inflation recently This is significantly lower thanthe marginal return on real capital inChina's economy, and also lower than most diversified investment portfolio's. In order to deliver a decent replacement rate upon retirement, the yield on Individual Accounts must be at least equal to wage growth17 Given that China will likely continue to expenence rapid wage growth in the next decade or so, the eventual replacement rates generated by the Individual Accounts will tend to remain low in companson. One reason for the low returns is that pensions are only allowed to be invested in a few financial assets, those considered safe. Another reason is that China's financial sector as a whole generates only low returns due to the direction o f funds into low yielding investments. These actions are likely the results o f an under-developed and repressed capital market, limited availability o f investment vehicles, lax regulation and supervision o f the overall financial sector, and deficient governance in pension fund management - all o f which represent necessary pre- conditions for the pension reform. Figure 8: Comparison of historicalgrowth -wage, inflation, bank deposit 600 1 wage growth 500 1 580 yuan\ i$ 2 400 I .E 300 200 100 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Source: 2002 China Statistical Yearbook To increase investment returns, the government needs to consider allowing for a more diversified portfolio o f pension fund investment. Although topping the high rate o f wage growth is not going to be easy (see Box 5 and the later discussion o f the investment management), potential return to accumulated pension reserve can be substantially higher than that from investing in bank deposit and government bond. Some financial institutions, e.g., domestic insurance companies, have been successful Replacement rate is defined as the pension divided by the final wage. Under a defined contribution arrangement, accumulated contributions are converted to a pension upon retirement. Thus, the accumulation mustat least grow at the same rate as wages so that the final accumulations are ofthe sametime value of money (at par) with the final wage. 37  in boosting retums and profits by investing overseas - a possible venue for social secunty funds to explore inthe intenm. Strengthening financial sector supervision anddeveloping better governance o f pension funds are cntical steps for the government to take in the short to medium term. Without sound governance that sets down objectives and accountability, without proper checks and balances, savings accumulated in the individual accounts cannot be adequately safeguarded. As the government moves towards allowing for pension funds to be invested in a broader set o f financial assets, such policy and institutional development will become evenmore urgent. More broadly, concurrent financial sector reforms and financial market modernization need to take place in order to increase returns on financial assets as a whole. While China's capital market is still under-developed - with its financial system dominated by state-owned banks plagued with non-performing loans, shallow and poorly governed equity market, and limited fixed income instruments, it is important to note that the financial landscape in China i s fast changing andthings are moving inthe nght direction (Box 3). The mutual funds and the insurance industry are undergoing rapid growth, and financial regulation and supervision has been significantly strengthened in recent years. While the regulatory authonties are confronted with formidable reform and development challenges, there i s strong commitment o f the authonty to the creation o f a sound androbust regulatory framework. 0 Options of crediting individual account interest. Considenng the fact that funding is a continuum and that China has a partial funding partial N D C model in place, alternatives to address the concern over low return to individual account savings should also be explored. One option i s for the central government to consider declanng an interest rate that may be higher than the market rates o f return available inthe short and medium term. For instance, the government could issue special government bonds to be purchased by the management o f pension funds on behalf o f individual members. These bonds could be indexed to a particular indicator, such as the growth o f GDP or GDP per capita. This approach has two advantages. First, assuming that the scheme has been set up with sustainable parameters, individuals would benefit from stronger property nghts when they hold government bonds instead o f just knowing that they have "empty accounts" The second advantage is related to the long term changes to the investment portfolio. At such point intime when China's capital market i s ready for the system to be fully funded, the special bonds could eventually be traded in a secondary market allowing for an orderly shift towards a more diversified portfolio. This proposal shares some o f the pnnciples o f notional defined contribution (NDC) system, but has its difference. It is similar to the N D C in the sense that individual account savings earns a notional return set by the government. It differs from the pure N D C inthat actual assets are present inindividual accounts. 38  Box 3. The FinancialLandscapeinChina China's financial system is dominated by the bankng sector, as banks account for more than 70% o f total financial assets. The banking sector, in turn, is dominated by the largest four state-owned commercial banks (SCBs), whose combined assets are about 66% o f total banlung assets. The SCBs are under rapid transformation, as three out o f the four have been recapitalized to pave the way for participation by strategic investors and IPOs. The banks remain largely state-owned and vulnerable to systemic nsks as their NPLs remain large andmaturity mismatch senous. The equities market is in deep trouble, including poorly governed and insolvent securities companies most of which are also state-owned; a large overhang o f non-tradable state and legal person shares in listedcompanies, and wide-spread frauds that constantly shatter investor confidence. Fixed income instruments, which should offer important asset classes to many institutional investors, have not recovered from the chaos o f the early 199O's, when many bond issuers defaulted. The bond market is mainly a market for the public sector, as government securities take up 52% of the market share, while financial bonds account for 35%, and central bank bills 10%. Corporate bonds only account for 3%. The main issues hampenng corporate bond market development is protracted fragmentation o f the regulatory framework, the merit-based approval system, and the central bank regulation over interest rates, with 140% o f comparable bank deposit rates as the ceiling, with the effect that bond issuanceis consistently under-pnced at the expense o fthe captive savers. Securitization, both asset-backed and mortgage-backed, is being expenmented, but the prospect o f such pilot beingreplicated inmass market remains dim, as legal andregulatory constraints remain, and room for commercial sustainability is narrowed by overly low mortgage rates as a result o f central bank interest rate controls. The mutual funds industry is fast nsing, with 127 open-ended funds and 54 close-ended funds. In addition, there have emerged 15 money market funds inrecent years. The open-ended funds have in general performed very poorly, with an average yield o f only 0.33%, compared with the close-ended funds with average returns o f 8.31%. The insurance industry, while still largely state owned, is undergoing rapid growth. As o f end 2001, there were 52 insurance companies in China, while 11more were in the pipeline for licensing. The number includes 19 foreign insurance companies and is equally divlded between life and non-life insurers. Combined assets o f the insurance industry stood at RMB 459 billion and together all the insurance companies holdmorethan 50% o f corporate bonds outstanding. Financial regulation and supervision has a short history inChina. Inrecent years, financial regulation and supervision has been strengthened, following the creation o f three specialized regulators for banking, securities, and insurance industnes. They all have a burden o f the past, but there is strong Commitment for the creation o f a sound and robust regulatory framework.. For their regulation and supervision to be effective, they need to shift regulatory paradigms, buildup institutional capacity, improve coordination and information shanng and formulate enabling regulatory frameworks that are balancedbetweencompliance-based andnsk-based supemsion. Insum, the financial landscape inChina is fast changing, as financial institutions, SCBs inparticular, are under restructunng, accompanied by reforms in financial policies. Complete hononng o f WTO commitment begnning 2007 will improve competition in the financial market. Effectiveness o f financial regulation and supervision is harder to gauge but a firm and lasting commitment to create an enabling regulatory framework is there, andthe development is inthe nght direction. 39  1.2.4 Reducing contribution rate: A key consideration o f the design o f a sustainable pension insurance program and o f the overall financing strategy i s to reduce the contribution rates as much as possible. The high contribution rates in Liaoning result in relatively high labor cost, which may affect the competitiveness o f the economy, and the decisions to seek formal employment. Financially weaker firms may not be able to afford the premium: analysis o f firm level data o f all industnal enterpnses in Liaoning in 2003 shows that there were still many nonviable firms, accounting for 7 percent o f total industnal employment. In addition, financially distressed firms accounted for about one third of total employment. Putting together these two categones of firms implies that nearly half o f Liaoning industnal firms, accounting for 38 percent o f total industrial labor force and 31percent o f total liabilities o f industnal firms, are indeep financial distress and cannot afford to pay the current level o f social insurance contributions (Box 4). Since it has been established that after adoption o f the parametnc reforms and strippingoff the economic and system transition cost, the long term required contribution rate i s 27 percent o f contributory wage (of which 17 percent is needed to fund the basic pension component), one possibility is to adjust the contribution rates to this required level. This will make the system more attractive for workers and employers, and help to encourage participation and compliance. In such a scenano, the IPD i s estimated to be 85 percent o f 2001 GDP, and the financing gap shows a surplus of 80 percent GDP Figure 7b depicts the dynamics o f the current balance. 40  Box 4. I s PensionContributionAffordableto IndustrialFirmsinLiaoning? Ithasbeena senous concern to policy makers that ahighpension contribution rate would result inhighlabor cost andconsequently depnve Chinesefirms ofadequate profitmargin andweaken their competitiveness inthe international market. This isparticular relevant to firms inLiaoning, China's rust-belt region, as they have beenundergoing painful restructunng since the mid-1990s with senousunemployment pressureand social dislocation. Pension insurance would have direct financial impact on firm performance as it will lead to higher labor cost and reduced profit margin. In a well functioning market, however, other indirect effects are at play Wage and employment are endogenously determined and the increase o f labor cost induced by higher social insurance contribution would be shared by employers andemployees, depending on the relative bargaining power. Systematic data are lacking to allow for an in-depth assessment o f firm behavior inLiaoning in the context o f pension reform. Furthermore giventhe ngidity o f the labor market and relatively short penod o f implementation o f the pension reform in Liaoning, the indirect effect is likely to be insignificant. This study therefore focuses on the direct impact o f social insurance contribution on firmprofitability. The analysis is basedon 2001 and 2003 fm level data o f all state owned and controlled enterpnses and non-state industrial enterpnses with annual sales revenue over RMB 5 million inLiaoning. Analysis shows that about half o f the firms did not make any social insurance contributions. This was the case in both 2001 and 2003. Those who paid have a highly diverse range o f effective contribution rates, with no sign o f concentration around the legally obligated rates. Many firms paid substantially higher than the required rates, some even over 100 percent of payroll. More in-depth information is needed to properly interpret the data, but one possible reasonis that these firms paidsignificant arrears inthe reportingyear. To answer the question of whether the reqmred social insurance contribution rate i s affordable, the study examines the question "what would have happened to the firms' profit margin had they paid 28 percent of the payroll for pension, health, and unemployment insurances in2001 and 20037" Interms of the situation in2003, the analysis reveals the following main findings: (a) about 46 percent o f the Liaoning industnal firms (38 percent o f total industnal labor force) were in deep financial distress and many o f them were nonviable. Without successful restrucmng, they could not afford any payroll tax. (b) Incompanson, more than a third o f the firms were operating at the margin and were probably most sensitive to any change o f payroll tax. 1942 firms of this category reported a total profit equivalent to 2.2 percent of total sales revenue, which would have dropped down to 2.0 percent had they all paid a 28 percent payroll tax. Firms in this category produced half of the total sales revenue and employed more than a third of the mdustnal labor force. There were 927 firms who didnot pay pension and medical insurance at all and other 661 firms paid less than 26 percent o f payroll. (c) Finally, about 24.4 percent o f the firms were inreasonably good financial status and enjoyed significant profit. A social insurance contribution rate at 28 percent is probably affordable for them. However, one should note that this group o f companies represented only 27 percent o f total industnal employment, and 32 percent o fsales revenue. 41  1.2.5 Adiusting the long term target revlacement rate: The government may wish to revisit the target replacement rate while takmg into consideration the relative role o f the public and pnvate pensions in the long term. Currently, the government has set a target replacement rate o f about 58-60 percent which seems high in international companson. Among the OECD countnes, Ireland and Mexico have the lowest replacement rates o f around 35 percent. The United Kingdom and the United States have slightly higher replacement rate o f around 38 percent. Although most workers inthese countnes expect to have retirement income that replaces 70 to 80 percent o f pre-retirement income, they must make up the balance through pnvate pensions and personal savings. The Chinese government may want to consider reducing the target replacement rate o f public pensions as enterpnse annuity andother retirement saving schemes are more developed. The current target replacement rate is also unattainable given the current framework and economic environment. To illustrate, let us consider the pension benefit for an average male worker o f 35 years o f service, earning average economy wage throughout his career, and will retire at age 65. Basedon the profile o f such a worker and the formula under the Liaoning Pilot policy, upon retirement, this worker should receive a replacement rate o f 32 percent o f average economy wage from the Basic Pension. To reach the total target of, say, 58 percent, he would be expecting 26 percent o f his final wage as a pension from the Individual Accounts. As Figure 9 illustrates, for a contribution rate o f 8 percent to achieve the 26 percent replacement rate target would require a real return o f 4 percent (net o f investment and administrative costs), or one percentage point higher than the assumed real growth rate o f wages. Since it is highly unlikely for the actual investment returns to exceed wage growth in the medium term under current financial market circumstances, the targeted replacement rate o f 58-60 percent does not seem to be attainable with the current level o f contributions. - Source: World Bank staff calculauons. 1/ Assumes 3 percent realwage growth; 2/ Accumulauon o f 35 years 3/ Single male, r e m g at age 65; 4/ L i a o m g projected pop.mortalitym 2040 5/ ROI assumed to b e net of expenses 42  1.2.6 Exuandina coverage: There are many economic and social reasons to expand the pension insurance program beyond its current reach and cover labor force in a consistent policy structure. Recognizing the desirable labor market outcome o f coverage expansion, the report argues that expansion should be done after necessary policy reforms are introduced. Projection analysis so far has assumed that coverage, defined as the percentage o f the worlung age population (ages 15-59) participating in the system, would remain constant throughout the projection penod. The scenarios with coverage expansion assume that pension program coverage will gradually reach at least 85 percent o f urban worlung age population by 2075. The assumed growth path i s based on cross-country patterns as well as China's time senes evidence o f the relationship between coverage and income growth. The projection examines the outcome o f the scenanos where contribution rate in the reformed mature system was reduced to the required minimum. Not surpnsingly, inboth cases, there will be substantial surplus interms o f current balance. Under this scenario, the IPD i s 91.6 percent o f 2001 GDP, while the financing gap i s -203 percent. A long term required contribution rate represents a stable rate that would be applied to the covered wage bill inorder to maintain fiscal balance" so that no unfundedliabilities would be passed onto the next generations. This rate i s calculatedwith the underlying assumption that any surplus flows will be put aside in a reserve fund to help pay for demographic and economic shocks. The creation o f such a reserve fund i s especially important if coverage were to be expanded such that system maturity would be delayed. Nonetheless, it is debatable as to how much should be put aside for a public system o f this type in order to ascertain that there are reasonable reserves to cover the implicit pension debt and that a growing burden will not be transferred to future generations. To aid inthis decision, it i s necessary to introduce another sustainability benchmark - target funded ratio. Figure 10 shows that, measured as the ratio o f assets to the net present value o f accrued liabilities, funded ratio gradually reaches 50 percent around 2050, indicating that for every RMB o f liability there will be one-half o f a RMB o f assets to pay for it. l8 The definition of fiscalbalance is such that totalrevenues includinginvestmentincomewould not be allowedto fall belowtotalexpendituresthroughouttheprojectionperiod. 43  Figure 10. Funded Ratio under Coverage Expansion 40% 30% 20% i 10% 0% Source: WorldBank staffestimates. Table 4 summarizes the projection results of all key policy scenmos modeled in this analysis, and reports the three sustainability benchmarks. Table 4. Summary of Projection Results inVarious Policy Scenarios (Based on Liaoningdata) SustainabilityBenchmarks Baseline Scenarios Scenario 1 2 3 4 5 6 Implicit pensiondebt (as % of 2001 GDP) 153 126 120 85 85 92 FinancingGap (as % of 2001 GDP) 172 42 -13 -140 -80 -203 Requiredcontributionrate (as % ofwage bill) 48 32 29 21 27 27 Description of theScenarios Scenarios 1 Liaoning provisions under State Council Document#42 Scenarios 2 System with systemic flaws removed Scenarios 3 System with no systemicflaws, no early retirement assumed (past and future) Scenarios 4 Remove transition pension for middle men, assume all middle men (active or retired) get the long term replacement rate based on length of service Scenarios 5 Scenario 4 using long term required contribution rate Scenarios 6 Scenario 4 with coverage expansion at longterm required contribution rate 44  1.3 CompletePolicy and Regulations beyond Enterprise Pensions Inaddition to the issuespertainingto the public pensionprograms discussed above, several additional policy developments are required, as evident from the Liaoning expenment. First of all, there are many important grey areas that require much more clarity inpolicy decision and legislative support. It concerns the vanous coverage and pension entitlement issues, particularly the status o f privately-owned enterpnses registered (andor located) in rural areas, the status o f the township and village-owned enterpnses, the status o f workers registered in urban areas but working in rural areas, and the status o f enterpnses that may have negotiated agreements containing tax concessions from local governments in return for their locating in a particular junsdiction. Liaoning expenence seems to suggest that many policies were either not fully developed or were determined at the local level and are subject to different interpretations o f local program administrators. The level o f understandingamong workers concerned vanes greatly The unification o f the pension schemes for civil servants and employees of public service units (PSU) with the general public pensions (commonly referred to as "enterpnse old age pensions") needs to be completed as part o f the medium term agenda to reduce bamer to labor mobility Dunngthe Liaoning pilot, no major expenment was camed out regarding PSU pension reform. Civil servants and PSU employees enjoy much higher average income and significantly higher pensions companng with the rest o f the urban labor force. Their pensions are covered by a contributory, defined benefit scheme that i s financed on a pay-as-you-go basis out o f general budget revenues. Employees do not contribute to the system. Normal retirement age i s set at age 60 for men, and age 55 for women, but early retirement age i s possible 5 years prior to attaining the normal retirement age or any time after having been inservice for 30 years. Pension entitlements are basedonthe individual's years of service, and the basic wage at time o f retirement, although reportedly, some provinces and municipalities include bonuses when determining the pension entitlement. The PSUpension scheme i s under significant pressure to reform as PSUs are begging to be transformed. Some are increasingly confronting the situation where expenditures that support core functions are being displaced by a growing share o f pension expenditure, which ultimately jeopardizes the survival o f the PSUs themselves. Given the additional cash flow requirement, the wholesale migration towards a framework similar to the enterpnse old age pension systemmay take some time to complete. 45  2. StrengtheningInvestmentManagement of IndividualAccount Savings As discussed above, investment returns to individual account savings under the Liaoning Pilot was unsatisfactory Underlying the poor performance o f pension funds investment were a number o f factors, ranging from overly restnctive investment policy prescribed by the central government which reflected almost zero-tolerance o f risks, to poor institutional capacity charactenzed by unsound governance structure, unprofessional management o f investment and lax internal controls. The external conditions, especially regulation and supervision o f public pension funds, also left much to be desired. This section starts with a bnefreview o f challenges reflected inthe Liaoning expenence in the areas o f investment policies and practices, governance and controls, institutional capacity, as well as investment objectives and pnnciples. It is followed by an analysis o f alternative investment opportunities and concludes with recommendations for reforms needed to improve investment outcome. 2.1 Pension FundsInvestment under the Liaoning Pilot Restrictive investment uolicies set bv the central government. At the outset o f the Liaoning Pilot in 2000, to protect the individual account savings from losses, the State Council Document 42 explicitly restncted individual account savings to be invested only in government securities and bank deposits. Liaoning followed the central government requirement. At the end o f 2004, one third o f the accumulated individual account savings was invested in government bonds and the remainder in negotiated certificate deposits (NCDs) at banks. Within the latter, about 60 percent are in second-tier banks, some o f which were infinancial distress. Calculated annual rate o freturnwas about 2.7 percent. While the State Council sanctioned investment policy was motivated to safeguard the accrued savings under the individual account, it fell short o f achieving its intended outcome. Not only returns were below what would be required to sustain the pension system, but also the individual account savings were jeopardized as increasing amount was investedincertificates o f deposits offered by insolvent banks. Overly restnctive investment policies incurred substantial opportunity costs in terms o f forgone investment retums. Analysis shows that much higher investment returns can be expected should investment restnctions were relaxed (Box 5). Superior investment results, in terms of risk-adjusted returns could have been achieved if certain critical conditions were met (see below). Investment policies should be an integral part o f pension funds investment regime, and centralized investment policies only served to undermine incentives to build up institutional capacity at the provincial level for sound investment management. It was observed that the SIA essentially practiced a "passive investing" strategy, though not inthe stnctest sense o f the term, by investing in whatever became available within the narrow 46  scope o f investment assets allowed. As monotonous investment in government bonds and bank deposits require relatively unsophisticated skills, prolonged implementation o f those investment restnctions removes the incentive for the provincial government to install soundpension funds governance and to ngorously buildup in-house professional capacity Box 5. What a DiversifiedInvestmentPortfolio CanPotentiallyAchieve? Investment opportunities exist both within home marketsand across borders. Ifa pension fund is constrained inits investment options, assuming other conditions remain constant, opportunity costs are incurredwhich canbe expressedinterms o f increasedrisks or lower expected returns. A portfolio constrained to Chinese government bonds only will have an expected return of 3.6% anda nsk level (measuredas standarddeviations) of4.5%. This compareswith the calculated rate o freturn o f2.4% currently achieved by the finded individual accountsunderthe Liaoning pension. Assumingthe constraints are relaxed to allow pension finds to invest inboth Chinesegovernment bonds and equities (baseline scenano), a portfolio consisting no more than 60% o f equities and no less than 40% o f bonds can produce expected returns between 4.17% and 7.02%, with nsks between 6.02% and 24.28%. The most efficient o f the asset mixes in this baseline scenano has a SharpeRatio o f 0.25 and forecasts areturn o f 5.75% with a nsk o f 6.0%. ExpectedNominalReturnfor Risklevels 6.00 9.00 12.00 ChineseEquityandChinese BondsOnly 4.17 4.17 5.19 Noless than40% mBonds, noEmergmgMarkets, no Chinese Equity 5.75 7.19 7.72 Noless than40% inBonds, noEmergmgMarkets 5.74 7.19 7.73 Noless than40% inBonds 5.77 7.24 8.04 Unconstrainedportfolio 5.76 7.24 8.14 47  Governance of uension funds investment. First and foremost, public pension funds management and investment have a fiduciary responsibility to members participating inthe pension. Inorder to free public pension funds from undue influence or negligentdecisions, such funds must be governed by qualified people working in well-organized governing bodies, including vanous key committees, with clear mandates and responsibilities given them by the stakeholders. In many countnes, governance of public pension funds tends to be politicized, exposing the savings to less-than-market returns or higher risks from investing such funds indubious financial institutions. Governance i s sometimes obscured by existing administrative arrangements, with the result that everyone i s supposed to be responsible but inthe end no one can be held accountable for investment results. In order to safeguard and increase the value of savings accumulated in the individual accounts, it is imperative for investment o f pension funds to be underpinned with sound governance that lays down objectives and accountability, information disclosure that provides reasonable degree o f transparency, internal controls that provide proper checks and balances, and compensation schemes that align incentives with responsibilities and performance. An in-house capacity i s also essential for pension funds investment to be managed professionally The Liaoning pension funds investment management will have a long way to go to satisfy those cntena. One cntical feature o f sound governance i s clanty on who should be responsible for what in financial and investment decisions as well as in implementation. The governance structure surrounding individual accounts management and investment under the Liaoning pilot fell short o f such clarity There was no legal foundation for pension funds governance and management. Despite official documents that supposedly gave the provincial Social Insurance Administration (SIA) the authonty to manage the funded individual accounts, in reality SIA seemed to have no explicit autonomy over financial and investment management. Even in day-to-day operational matters, such as timing, amounts and terms and conditions associated with a transaction, SIA routinely sought approval from the provincial Labor and Social Secunty Bureau and Finance Bureau. It was by no means clear who would be heldultimatelyresponsible inthe event o f loss. Internal controls in uension funds investments. As sizeable funds were being accumulated in the fimded individual accounts, their safeguard was no less important than that o f public deposits taken by banks inLiaoning Province. To prevent frauds and undue risks, the management and investment of the savings in the funded individual accounts should be subject to tight internal controls. In a nutshell, effective internal controls are charactenzed by checks and balances to make sure major investment decisions are adopted and executed with clear responsibilities and based on agreed rules and procedures, clear reporting relationships, an independent internal auditing unit adequately staffed by qualified individuals, with authonty to check enforcement and to follow up on corrective actions. 48  Weak controls in pension funds administration are discussed in the following section o f this report. Inthe area of investment management, internal controls were glanngly absent throughout the SIA system. There was no separate internal auditing unit, and internal auditing functions were not explicitly designated within the SIA. There was a unit in charge o f inspections, but its pnmary function was to uncover voluntary avoidance o f participation in the social secunty scheme, and to enforce compliance through on-site examinations and penalty Given wide-spread non-compliance by numerous firms and frequent fraudulent claims, the capacity o f the unit was increasingly stretched. It was unrealistic to expect the same unit to also assume internal auditing functions for investment o f the individual accounts. Institutional cauacitv. As modem asset investment management i s knowledge intensive, investment in human resources is crucial to performance o f financial investment. The Liaoning SIA was hampered by shortage o f qualified professional staff in pension funds management. The resources devoted seemed to be far from doingjustice to the large and nsing volume o f the portfolio, which amounted to more than RMB10 billion in spnng 2005. One reason appeared to be that the SIA was subject to the same staff and position quota as well as budget allocation as other provincial government agencies, therefore having had little political and operational independence. Once again this phenomenon pointed to weakness in governance o f the pension funds. The total budget for 2004 was in the order o f RMB 4.5 million, only a small fraction o f the money it was supposed to pay for custodial services. It was clear that drastic changes were in order should the objective to "protect and increase the value" (baozhz zengzhi) o f the funds inthe individual accounts be achieved. Investment uolicies and urinciules. Inorder for investment decisions and implementation to be prudent and efficient, there should be an arm's length relationship between the pension funds management and the governments. Pension funds should be allowed to formulate their own investment pnnciples and policies, based on objectives and mandates givenby the participating members, as well as their nsk and returnpreferences. 2.2 Alternative Investment Choicesand Implications There are alternative asset classes with vanous combinations of nsks and returns inwhich public pension fund could invest, in addition to government secunties and bank deposits. As analyzed above, concentration o f investment in government secunties and bank deposits, presumably dnven by zero-tolerance o f investment nsks, incur heavy opportunity costs while not necessarily safe at all times, especially if the savings ended up in certificates o f deposits offered by insolvent banks. What are the alternative investment instruments available to public pension funds in China? For illustration, Table 5 lists classes o f investment instruments together with risk implications. 49  Table5. Implicationsfor RisksandReturnsofInstrumentClasses inChina Asset classes Implicationsfor riskandreturn Equity The market is fast risingbut still shallow and highlyrisky, dominated by poor-performing public companies with weak governance and controls; as insider trading and pnce manipulation remain rampant; and politically charged solution o fnon-tradable shares hamper healthy development o f the market. Nonetheless, confidence inthe equities market has been revived to a certain extent. Government securities L o w nsk; low return. Financialpolicy bonds Issued by policy banks (mainly China Development Bank), with implicit state backing. Risks and returns comparable to government secmties. Corporate/enterprise/project These offer higher nsk-adjusted returns than government securities bonds and financial policy bonds, but so far have remained scarce inChina. MBSIABS Mortgage-backedand asset-backedsecuntization is being prototyped in China. Nomassive roll-out is inprospect inthe near hture as the business case o f securitizationremains weak, hampered by legal andregulatory constraints. Venture capital funds Highrisk;highreturn. Factonnglleasing Mediumto highnsks andreturns. Good for SMEs butrequire special expertise to reap gains and mitigate nsks. Property Long-term highreturn, but should beware o fbubbles. Overseas securities Diverse nsks and returns and desirable inview o f demographic factors. Restrictedat present and even ifrelaxed could still be constrained by home-bias o f investment. Among the fixed-income instrument classes illustrated in Table 5, corporate and project bonds offer desirable nsk-adjusted returns. Unfortunately, the fixed-income instruments remain dominated by government secmties and corporate/project bonds compnse a mere 3% market share, as illustrated in Figure 11. Recently firms were allowed to issue short- term debt instruments, with matunty up to one year, o n the interbank market. Reform of corporate bond regulation and supervision, especially replacing the current merit-based system with information disclosure, is on the government agenda. Progress is slow thus far, and understandably so as it takes time to sort out the intncate stake-holding relationships. 50  Figure11. DistributionofFixed-incomeInstrumentsinChina, 2003 Corporate Bonds (RMB114; 3%) FinancialBonds (RMB 1,161, 35%) Government Bonds (RMB 1,749, 52%) Central BankBills (RMB338; 10%) 2.3 Improving GovernancetoRealizeInvestmentPotentials Itis critically important for accumulated savlngs inthe individualaccounts to be managed and mvested prudently and efficiently Two factors determine whether investment o f individualaccounts pensions can generate sufficient nsk-adjusted returns inthe longterm. One i s whether there is a well-developed financial market that offers diverse instrument classes for investors to choose from. The other i s whether the pension funds are governed well and managed professionally to enable them to realize the nsk-adjusted returns and to benefit from the long-term growth o fthe economy. As discussed earlier mt h s report, both pension funds admmstration and mvestment take time to develop, the same as capital market, therefore an under-developed capital market at any point o f time should not be made excuse for lacks o f reform inpension funds. All thngs considered, sound corporate governance determines whether pension funds mvestment i s positioned to consistently beat the market under any circumstances. Sound governance o f pension funds has the . following charactemtics: .. Clearly definedroles andresponsibilities Independence o fpension funds from political interference Objectives andpnnciples formulatedbasedon collective preferences o fmembers Fit andproper test ofindividuals managmgpensionfunds anddisciplines Internalcontrols to prevent frauds and, corruption andmis-management 51  . Subject pension management to ngorous regulation and supervision - no less so than that o f pnvate pension funds Alternative governance structuresfor pensionfunds administration and investment As lessons have been learned since the inception o f the Pilot about 4 years ago, and ifthe government moves forward with a partially funded program, now is a good time to review and relax the investment restnctions based on a thorough review o f the governance and capacity o f the pension administration. Over time, it will become increasingly important for those who are heldultimately accountable for investment performance to be grantedthe autonomy to formulate statement o f investment principles and policies. Expenence inLiaoning argues for alternative governance structure and investment policies concerning accumulated pension fund. To maximize potential return from the capital market and to benefit from stronger fund management capacity currently available, the government may want to consider centralized administration o f pension funds and reserves, even though investment management can and should be decentralized. Investment services can be outsourced to qualifiedprofessional asset managers. The authonties need to give pnority to building a robust governance structure for the management andinvestment, inview o f the sizeable andfast growing amount o f savings in the funded individual accounts. In the final analysis, quality o f governance matters, because governance structure i s crucial in determining investment objectives, pnnciples and strategies, as well as internal controls and implementation, which in turn affect investment performance. The stakes are high. It is clear that governance o f the individual accounts should receive pnonty attention in the sequence o f reforms, since it constitutes a precondition to other reform measures. Time-bound steps should be specified in improving accountability To preserve public confidence and trust in individual accounts management and investment, the individual accounts governing body should be urged to disclose its commercial mandates, pnnciples and objectives o f investments, as well as rules and procedures. It i s not only the financial informationthat counts, but also the way how things are run. Internal controls Considenng the large and nsing amount o f funds inthe individual account, it i s imperative that internal controls be established in the SIA as soon as possible. Pnonty should be given to creating a separate, independent unit for internal auditing, with direct reportingto the Director-General o f the provincial Labor and Social Security Bureau. It is envisioned that once this function i s installed, it will enhance confidence in the ability o f the SIA to invest the savings safely and efficiently, which inturn will serve to reduce the undesirable interferenceby vanous stakeholders inday-to-day operations o f investment. Regulatory issues 52  As public pension reform is still at an early stage, China has no explicit public pension regulatory framework. Financial regulation and supervision i s separate among banking, secunties, and insurance industnes. None o f the regulatory agencies has explicit mandates to regulate and supervise pension funds, public or pnvate. The Ministry o f Labor and Social Secunty has begunto assume regulatory functions over corporate pension funds, but regulation and supervision o f public pension funds, including the funded individual accounts, has so far remained ambiguous. Withinthe LiaoningProvincialGovernment, so far there appeared to be no clearly defined regulatory function. Consequently, regulatory responsibility i s shared among the Provincial Labor and Social Security Bureau, the Finance Bureau, the Provincial Auditing Office, and the Discipline Inspection Office within the Party Committee. Overlapping regulatory functions inevitably give nse to a regulatory vacuum. The perceived inadequacy in oversight frequently leads to installing additional layers o f safeguard. For instance, at each auction for bank NCDs, two officials were invited from the provincial Discipline Inspection Department to exercise on-the-spot supervision. Such monitonng was at best perfunctory If anything, the presence o f discipline inspectors only served to validate the transactions whose integnty cannot be assured simply by the presence o f watchdogs then and there. As indicated above, given the nature o f public interest and inherent fiscal nsks to the government, public pension funds should be regulated and supervised more ngorously than pnvate pension funds. Specifically, they should be regulated and supervised by either an explicit organization with sufficient powers and independence to conduct effective oversight, or by the same organization that regulates and supervises private pension funds. 3. ImprovingAdministrationandFinancialManagement Built on the progress already made in the Liaoning pilot, additional steps are needed to further strengthen financial control and overall program management. Some involve completing initiatives that have already been started or that were part of the onginal vision, but have not yet been completed. Others involve adapting to the Chinese environment processes and procedure that are now employed in other countnes and that have come to represent international best practice. 3.1 CompleteImplementation of theReform AgendaSpecified in Pilot Design 3.1.1 Imurove the legislative basis for uension administration. Tax authorities report several areas where weaknesses in the legislative basis o f the program undermine compliance. One i s the absence o f clear consequences for noncompliance. Another is uncertainty about the obligation to contribute among enterpnses that may have negotiated tax concessions in return for locating in specific jurisdictions. There is also confusion about the status of people working inmany enterprises that are registered inrural areas and people who may not have formal labor contracts. Some local legislation has been 53  developed invanous municipalities inLiaoning, but the most effective approach would be legislation at the national level. The authonties should also consider breahng the link between that portion o f the labor code that specifies the conditions for labor contracts and the coverage o f the social insurance system. Social insurance coverage should be based on a simple test that there was work performed for pay Enterpnses and workers should not be allowed to avoid social insurance contributions by fiddling with the provisions o f labor contracts (or not having one at all) and the tax authorities should not be expected to enforce other portions o f the labor code in order to gain compliance with the social insurance contribution requirements. 3.1.2 Reorganize uension administration and move to urovincial level uooling. The province should develop plans to complete the vertical integration o f the social insurance agencies by mergingthe municipal level social insurance agencies with the provincial level agency Once contribution rates have been equalized and all municipalities have established vertically integrated social insurance administrations, it is a relatively easy step to move to provincial level pooling, at least for the pension program. It i s worth noting that Document 26 specified that provincial pooling i s to be achieved such that there will be unified contribution rate, unified management, unified adjustment and migration towards vertical integration o f administration. However, up to now, most provinces do not conform to the cntena o f adjustments and that redistribution i s inmost part still subject to a number o f nontransparent factors. Moving towards the long term goal o f national level pooling, Liaoning and other provinces need to accelerate the development o f a provincially-pooledsystem. The fact that local tax system i s integratedvertically andreports to the provincial government makes this a viable intermediate step while the basic policy parameters are gradually unified nationwide. Currently Provincial Adjustment Fund i s the mechanism through which redistribution i s made within a province, but in the case o f Liaoning, the degree o f intra-municipal redistribution i s limited. The redistribution formula i s in principle a function o f the expected funding gap o f each municipality, the contributions made to the provincial adjustment fund, and a performance index. While it appears that the equalization formula is incentive compatible, without knowing how the performance index and other expected values are determined, and how the rules are applies, the effectiveness o f the redistribution i s unclear. 3.1.3 Regularlv uublish information on revenue and exuenditure to imurove transvarencv. The provincial agency should improve its regular publication of information on the revenue and expenses o f the social insurance system and the composition o f its investments, as envisaged in the Pilot provision. Regular audits o f the financial statements by outside auditors should be arranged. Outside audits will help to enhance the credibility o f the financial statements and increase public confidence in the government's management o f social insurance funds. A review o f the internal controls as part o fthe audit process will assureaccurate accounting o f all financial flows. 54  3.1.4 Comulete develoumentof the information svstem. A modern informationsystem is one o f the most important requirements for efficient social insurance program administration. The system now being implemented in Liaoning seems to hold the promise o f substantially improving program operations. A well development provincial data base would greatly improve the quality o f program statistics and, together with the individualaccount data base, provide a sound basis for projecting future program costs and analyzing policy options. Several important steps remain in the initial implementation phase, however. These include (a) completing the entry o f the work histones o f current contributors in the remaining 10 to 20 counties and distncts for which this has not been done; (b) converting the remaining 4 municipalities to the new system; (c) automating data transfers between the tax bureaus and social insurance agencies where it is still done manually; (d) finishing the clean up o f errors in the provincial data base on individual accounts to ensure data consistency and allow generation o f improved program statistics, particularlyon earnings. Provincial officials should also be examining ways to improve further the hctioning o f the information system, including: Reviewing the needfor a common data system across each of the different agencies. The vision outlined in Pilot design was o f a single system with uniform software and hardware connecting each o f the vanous agencies involved in administenng the social insurance system. At least with respect to the tax bureaus, this has not happened. The tax bureaus already had a data system that integrated their operations throughout the province and have not participated in the development o f the new social insurance system. Provincial officials need to review the relationship between the tax agencies and the social insurance agencies to see whether improved data exchange protocols can achieve most o f the gains that a fully integrated system would bnng. Capturing all critical information required to establish benefits so that, over time, manual reviews o f the information in the personal archives to establish benefit entitlement and calculate benefit amounts will no longer be necessary Such critical data should be captured contemporaneously and entered into the work records maintained at the social insurance agency The objective i s eventually to eliminate the need to use the personal archive inthe administration o f social insurance programs and rely instead exclusively on data maintained in the agency's data bases or submitted by the retiringworker at the time a claim is filed. Creating a provincial data base covering pension recipients. After the individual account data base has been cleaned up, a provincial level data base covenng pension recipients should be constructed. Itwould also be the first step ineventually moving to provincial level pooling and management o f the pension system. Verifiing the completeness and accuracy of the information system. When the system i s stabilized, it would be advisable to have an outside organization conduct a thorough review o f its operations to assure that the software i s doing all calculations correctly and i s moving and stonng data as intended. A private accounting firm that 55  audits organizations with major computer systems should be qualified to conduct such a review 3.2 Institute Additional Administrative Reforms Several relatively modest administrative reforms could have a major impact on the quality of both financial andgeneral management. 3.2.1 Financial management imwovements. Financial management can be improvedby streamlining several cash management processesand beginning to reconcile the accounting records covenng individual accounts with the balances being carned in the provincial financial accounts. Specific steps include: Converting the current two-track accounting system into a full treasury cash management system. The treasury cash management systems found inmost advanced countnes separate the process o f approving transactions from the process o f managing cash flow Cash transactions are processed through the treasury system in response to payment orders executed by program administrators. The treasury managers do not have the authonty to initiate transactions without a payment order from the program manager andprogram managers never exercise direct control over the fiscal resources. The two-track accounting system has some o f the attributes o f the treasury system in that contributions are deposited directly into accounts controlled by the financial bureau. At present, however, the financial bureau returns the funds to the social insurance agency for the purpose o f making benefit payments, merging the cash management andpayment authonzation function inthe same agency Internal controls would be improved if the financial bureau retained control over the funds and transferred them directly to the bank accounts of the program beneficianes, inresponse to payment orders drawn up by the social insurance agencies." Accelerating the movement of funds from enterprise deposits to interest bearing accounts. It seems that it can take significant amount o f time for the funds to transfer from the special social secunty account heldby the finance department to the collective individual account heldby the social insurance agency for investment. Incontrast, the employee pension contributions are channeled through a different special account managed by the central bank in a matter o f hours. Given the enormous sum involved, this will not only cause substantial interest losses but also constrain the ability o f the social insurance agency to conduct timely investment transactions. The provincial government could consider using the treasury account network in the central bank to centralize the funds before transfer to the investment account or the special social secunty account. l9 Inaddition, within the finance bureau, the authontyto set andexecute investmentpoliciesshouldbe assignedto a different group thanthe group responsiblefor managingthe fiscal flows. 56  More effective pooling of financial resources. At present, separate accounts are maintained in each county and distnct into which contributions are deposited and from which transfers are made to cover benefit payments, even in municipalities that officially pool at the municipal level. The multiplicity o f accounts may be one o f the factors underlying the money management delays just noted. Municipal authonties should evaluate the relative efficiency o f merging all o f the financial operations into one single municipal level account. Reconciling provincial level individual accounting records with the provincial level financial flows. Although accounting records and bank records are matched at the local level at the time that contributions are collected, the individual account information at the provincial level i s not currently matched with the provincial level fiscal balances. This is a potentially senous internal control weakness. Matches will not be possible until the cleaning up o f the provincial level data base i s complete. At that time, regular matches between account records and financial balances should be instituted. 3.2.2 Create uerformance standards, measure uerformance, and reuort results. Provincial and national program administrators need to create a set o f performance standards to measure current service quality and facilitate service quality improvements. Among national social insurance agencies, best international practice i s to create a set o f performance standards, establish goals for each standard, and report regularly to the legislature and the general public about progress inachieving the goals. Common performance standards for social insurance agencies include (a) the time required to process applications for benefits or reports o f changes in beneficiary status, (b) the fraction o f new benefit calculations or continuing benefit payments that have errors; (c) the effectiveness o f recovery o f debt owed to the social insurance agency, and (d) the level o f public satisfactionwith the service delivered by the agency An integral part o f the process o f setting performance standards and measuring performance i s the establishment o f a separate group within the social insurance agency whose pnmary function i s to review internal operations. Such a group would track performance to assure the accuracy o f reports o f processing times, debt recovery, and similar management indicators. It would also select a small sample o f case files and review them intensively in order to construct measures o f the accuracy o f administrative processes. Where performance measures include the level o f public satisfaction, the internal review staff would develop performance measures through household surveys or other appropriate means. An internal review staff also plays an important role inassunng fiscal integnty They are responsible for evaluating the structure o f internal control systems and monitonng the effectiveness o f their implementation. They also provide the resources to investigate any allegations o f fraudulent behavior or mismanagement among agency staff. 3.2.3 Standardize oueratiw urocedure. Standardizing procedures involves developing detailed, step by step instructions for handling all important transactions, prepanng wntten 57  documents outlining these procedures in a senes o f operations manuals, distributing the manuals throughout the organization and monitonng operations to assurethat the approved procedures are being followed. To achieve uniformity o f operations, standardized procedures need to be executed by staff operating under uniform organizational structures occupying jobs with common job descnptions and standard skill requirements. Like the operating policies, the organizational structures, job descriptions and skill requirements needto be written and communicated throughout the organization. Standardized operating procedures are both a reform objective and a necessary precondition for efficient administration. They are a reform objective because they assure equal treatment o f all program participants. They are a necessary precondition because they allow objective analysis o f operating costs and results and the structured development and implementation o f operational improvements. When an internal review finds a problem in a particular program operation, such as erroneous payments or improper financial transactions, those conducting the review can assess whether the problem was caused by a failure to follow standard procedures or whether standard procedures need to be modified in order to eliminate the problem. In either case, the standardization o f internal processes allows similar corrective actions to be implemented throughout the organization. 3.3 Pilot TestMore Complex Reforms Several important administrative reform areas would be good candidates for further pilots. Inthese cases, although the reform objectives are clear, the most effective approach inthe Chinese context is not as clear. Alternatives can be developed and piloted in different locations with the results evaluated to determine which alternative i s the most promising. Three such areas are streamlining current administrative process, particularly those involving data collection and data processing, improving collections, and improving the dissemination o f program information. 3.3.1 Streamline processes: Many o f the current social insurance administrative processes are cumbersome and impose significant and unnecessary costs on enterpnses, potential beneficianes, and the administrative agencies themselves. These processes are no longer appropnate ina market economy where much o f the economic activity i s camed out inprofit-making concerns, where state enterpnses must keep operating costs low inorder to compete, and where modem information technologies are widely available. Cumbersome administrative procedures increase the cost to all participants and are a barner to compliance, particularly among small enterprises andthe self-employed. Examples o f cumbersome procedures in the Liaoning program include the procedure enterprises are required to follow monthly to pay their social insurance contributions, the process for registering a new enterpnse or renewing a business license, for transfer a worker's account from one pooling area to another, and for applying for a pension benefit. Many o f the steps in the current processes are designed either to make sure that information is transferred between different government agencies, that the reported data i s 58  correct and that correct amounts are paid. These are all valid objectives. Reform will require developing new processes that reduce the reliance on paper forms and that automate data transfers but that do not sacnfice (and may improve) the controls built into the current processes. 3.3.2 Zmurove collections: Once the legal basis for collecting contributions has been improved, alternative contribution collection techniques should be pilot tested to address three lunds o f challenges: (1) making sure all covered enterpnses are registered, (2) making sure all registeredenterpnses accurately report their contribution liabilities, and (3) collecting the liabilities reported. Some potentially powerful enforcement techniques are already being tested in Liaoning. The tax bureau in FushunMunicipality matches its records with those o f the Industry and Commerce Bureau to make sure that all enterprises registered with the ICB are registered with the tax bureau; it compares its information with that from the statistics bureau to determine the scope o f its compliance problem; it sent teams o f auditors into selected potentially problematic enterpnses for intensive inspection. Once its municipal compliance ordinance has been adopted, the Fushun Tax Bureau plans additional steps to improve compliance, including publicity campaigns, publication o f lists o f delinquent enterpnses, and denial o f social benefits or honors to delinquent enterpnse mangers. Similar techniques should be piloted inother localities and impacts properly evaluated. Some improvement would require assistance o f the central government. For example, to be able to check the accuracy o f enterpnse reports o f their total payroll, the tax authonty needs to compare enterpnse reports filed inconnection with their profits tax payments with those filed in connection with social insurance contribution payments. The comparisons can not be made currently, and would require revision o f the profits tax reports required by the national tax authonties. Finally, the feasibility o f using the national tax authonty to collect social insurance contributions should be tested inone or more locations as part o f a pilot. Eventually, China should further centralize the responsibility for operating its pension program through national pooling and national databases. At that point, it may also want to shift collection responsibility to the national tax authonties, if that option proves to be feasible in pilot testing. 3.3.3 Issue annual statements: Internationally, an increasingnumber o f national pension systems are sending each worker an annual statement showing the worker's current account balance, informing the worker o f the implications for his future retirement income, and telling the worker how to report potential errors in his account records to the pension agency These statements are a form o f internal control to assure the accuracy o f employer reports and agency records, increase public confidence that their contributions are being used inthe manner intended, andbetter inform workers about how to plan for retirement. Annual statements were tned in 1998 in at least some o f Liaoning's municipalities, but it was decided at that time that their issuance was too expensive. Since that time, however, 59  the individual accounts have become a much more important component o f the pension package, malung them potentially more valuable to contributors. Moreover, the new information system has beeninstalled,which should have made it much easier to issue the statements. Once the data bases have been fully computenzed, the province should once again issue annual statements and assess both the cost and the benefits o f the issuance. With modern computers, the cost o f issuingsuch statements shouldnot be muchmore than the cost o f the postage needed to mail them plus the cost o f answenng any questions that the statements generate or o f correcting any errors they uncover. IV. IMPLICATIONSFORNATIONALPENSIONREFORMSTRATEGIES After more than a decade o f pension reform and implementation, China's policy makers continue to search for the most suitable pension system for China in the long term. The recent Liaoning and Heilongjiang/Jilin expenments were important steps in this long reform process. While evaluating Liaoning and other provincial pilots does not provide a sufficient basis to develop the national pension framework, it provides useful insight on many o f the cntical issues. Many o f the findings and recommendations illustrated above apply also to the national pension discussions. This section highlights some o f the main findings and recommendations relevant to the national pension reform strategy It is based on the Liaoning analysis above and an earlier projection analysis that the World Bank camed out at the national level2' 1.Challenges Despite the progress in unifyingthe national pension policies, China still has perhaps the most decentralized and fragmented public pension program in the world. Contribution rates vary across or even within municipalities; program financing i s pooled often at the municipal level or lower. Pensions for employees o f public service units and civil servants are still separated from the enterpnse pension scheme. The fragmentation leads to labor market distortion and bamer to labor movement. Resources are used inefficiently due to inadequate nskpooling. Pension program has very limited coverage and that leaves a significant part o f the urban labor force unprotected against old age poverty The coverage rates o f selected localities, expressed as the ratio o f contributors to population aged 15-59, ranged from as low as 2.9 percent to 21.3 percent in2001. The national pension program as in State Council Document 26 suffers from some important policy design flaws which are present in the Liaoning pilot. The most cntical ones include (a) the use o f inappropnate amortization factor in calculating individual account pensions, (b) the ad hoc pension indexation method, and (c) the low statutory retirement ages and liberal early retirement provisions. In 2001, the estimated average retirement age under the national program was 56 for men, and 50 for women. The 2o"ChinaPension Liabilities and Reform Optionsfor OldAge Insurance",May 2005, World Bank. 60  Heilongjiang and Jilin pilots introduced inmid2004 and the national policies announced in December 200521 introduced some modifications to the policies and led to significant improvement (Box 6). Two major outcomes o f the Heilongjiang and Jilin pilots are (i) the removal o f the most detnmental flaw o f the system - the actuarially unsound amortization factor o f 120 for the individual accounts, and (ii) establishing an even stronger linkage between individuals' contributions and pension benefits inthe new Basic Pension formula. The State Council document issued in December 2005 extended some o f the Heilongjiang/Jilin pilot policies nationwide. While continuing with the approach o f partial funding o f individual accounts, the pace will likely be more gradual. Detailed decisions regarding fundingare left to be developed inearly 2006. Box 6. The 2004-2005 Jilin andHeilongjiangPilots andthePolicyChangesIntroducedinDecember2005 Withthe experience ofLiaoning Pilot, the State Council launchedJinlin andHeilongjiangpilots in May 2004* These two provinces adopted a modified version o f the Liaoning Pilot, with the key modifications beingthe following: Basic Pension is determined at the rate o f 1% for eachyear o f contributory semce (minimum o f 15years required)to be applied against the average o f last year's local economy wage and an index of individual's average contributory wage over the contributorypenod. Individual Accounts will be disbursed monthly basedon the accumulated balance dividedby the prescribed age specific factor at time o f retirement. As inthe case ofLiaoning, the total size o f individual account was reducedto 8 percent of payroll. However, the account will begradually finded, starting from funding 5 percent o f payroll from January 2004. Contributions to individualaccounts were to be segregatedfrom the Social Pooling component. The modifications represent significant policy improvement. They effectively removed the actuarially unsound amortization factor o f 120 for the individual accounts and moved closer to actuarial reality, though there is still room for improvement. The new formula for determinmg the Basic Pension will help to reduce the incentive o f workers to quit the system prematurely or to under-report their contributory wages. InDecember 2005, the State Council issued a Decision on "Improving Basic Enterpnse Pension Insurance System"** Based on the northeast pilots, the State Council extended the modified pension policies to the rest of the country, and almost all policy parameters remain the same as those in the Heilongjiang and Jilin pilots. The Decision also called for gradual funding o f individual accounts and envisaged that funding will take place in a few provlnces in 2006. The Decision also pointed out that pension will be indexed to a certain percent of the average wage growth o f employees inthe local economy However, the specific level o f funding, the financing responsibilities, the model o ffindmanagement, as well as the choice of the provinces were all left to be Mher specified. * State CouncilDocument 35 and 36 [2004]. **State Council Document 38 [2005]. ''StateCouncil Document 38,2005 61  The national pension system i sfinancially unsustainable.As in the case o f Liaoning, the system at the national level is burdened by legacy cost resultingfrom early retirement and transition. No financial arrangement was made to explicitly take care o f the histoncal burden and separate that from the financial requirement o f a healthy pension insurance program. Resulting from the histoncal legacy and the flawed pension policy design, the national pension system i s financially not sustainable using both the national policy parameters specified in State Council Document 26 and the Liaoning Pilot policy parameters specified in State Council Document 42 (Table 5). Left unattended, the program financing gap will become 95 and 64 percent o f the 2001 GDP, and the contribution rates to keep the pension program inbalance will be as highas 37 percent and 35 percent o f the wage bill, respectively While the Heilongjiang/Jilin policies and the new national policy will significantly improve the financial outlook o f the system due to the correction o f the individual account pension formula, the long term sustainability o f the system i s yet to be established. Table 6. Summaryof ProjectionResultsinVarious Policy Scenarios (basedon samplenationaldata) Sustainability Benchmarks Scenanos A B C D Implicit pension debt (as % of 2001 GDP) 141 132 110 110 Financing gap (as % of 2001 GDP) 95 64 -103 -20 Required contribution rate (as % of wage bill) 37 35 25 25 Description of the Scenarios ScenarioA System in line with State Council Document#26 (national policy) Scenano B System in line with State Council Document#42 (Liaoning policy) ScenarioC System in line with State Council Document#42, with parametric reforms adopted Scenario D Scenano 3 with gradual reduction in the contributionrate The investmentpolicy specified by the State Council document restncted the investment o f individual account savings to government bond and bank deposits. In most localities, no fundwas accumulated as no fimdingstrategy was developed to finance the transition cost. For places where accumulation took place, the return on investment has been too low to provide for the targeted level o f replacement. The governance structure for pension fund investment remains seriously under-developed, While having decided to move towards gradual funding o f individual accounts in a few more provinces beyond the northeast, the State Council Document 38 o f 2005 only states that policies regarding fund management and investment will be further developed by the central government, leavingno indication o f change to the earlier policies that may be contemplated. 62  Finally, necessary legislative basis needs to be developed, and pension administration requires further improvement. Informationmanagement system required to administer the current pension program is under-developed. Over time, there i s a need to move away from the current arrangement where either the tax authonty or the social insurance agency i s responsible for contribution collection, depending on the decision o f the local government, and to consolidate the responsibility into one agency and strengthen its capacity nationwide. 2. Recommendations Consistent with the findings o f the Liaoning Pilot evaluation, the report argues for a set o f policy reform measures to be adopted at the national level. Implementing parametric reforms: Analysis indicates that adopting the set o f pararnetnc reforms alone will greatly improve the financial health o f the system. The measures include gradually increasing statutory retirement age by half a year per year to 65 for both men andwomen, and switching to pnce indexation in the medium term. The recent policy changes o f the benefit formula for individual account pension i s a move inthe nght direction, although further modifications can be made to improve the actumal fairness. The combined impact o f these reforms will result in substantial surplus in current balances. Companng with the scenano under the Liaoning policy parameters, the long term contribution rate required to keep the system in balance would be reduced from 35 percent o fwage bill to 25 percent (Table 6). Separating the legacy cost and developing a sustainablefinancing strategy. Our analysis o f Liaoning illustrated the concept o f separating the cost o f the sustainable pension insurance program from the cost resulted from historical legacy and transitions. A financing strategy can be developed, among the various stakeholders, based on such a principle. It seems prudent to pre-fund the program partially to benefit from the demographic dividend in the short to medium term. However, a sustainable and transparent financing strategy would be required, and cost shmng arrangement between central andlocal government will likely be needed for many provinces. Improving management and investment of individual accounts. Expenences o f the Liaoning pension reform pilot hold useful implications for investment management o f pension funds at the national level. In particular, they strengthen the notion of centralized pension fimds administration. First, as segmentation o f financial markets diminishes and regulation and supervision is centralized at the national government, investment choices by pension funds are not limited to what one province has to offer, as capital flows without restnctions across provincial borders. This fact reinforces the argument in favor o f centralized administration o f public pension funds. Investment services can be outsourced to qualified professional asset managers which do not have to be resided inthe province. Second, as professional investment managers are scarce in China, the scarcity is even more manifest in any province. Building up in-house investment capacity i s possible by a provincial pension fund administration, but it may 63  hardlymake economic sense to do so, given the lack o feconomies o f scale andlack o f available local talents. Developing a nationalpension system. The longer term goal should be the gradual evolution to single national pension system operating on the basis o f a single national data system and financed through a single national pool. National pooling i s the effective way o f evening out the financial consequences o f regional demographic differences on a long-term and sustainable basis. To be implemented effectively, national pooling will require the construction and maintenanceof a national database containing information on collections, individual account balances, individual work histones and individual pension payment records. Such a database will provide the accounting controls needed to protect the integnty o f the national financial pool. Itwill also automatically eliminate the problems associated with worker migration, including the administrative burden o f moving records from one pooling regionto another and the impact on regional fiscal balances o f transfemng either the account balances or the pension payment liabilities from one region to another. National pooling is probably easier to accomplish if the responsibility for collecting social insurancecontributionsis shifted to the national tax system. As the collection processes are streamlined, either the tax authonty or the social insurance authonty will eventually assume responsibility for collecting and processing the information on individual work experience and deposits to individual accounts. Ineither case, as long as the accounting records are carefully reconciled with the financial flows, these data can then be fed into the national data system. Eventually, pensionbenefits will likely also be administeredthrough a national benejit agency For an extendedintenmpenod, however, national pooling could be combined with provincial level administration o f benefit payments. Before such a system is implemented, however, many o f the administrative reforms recommended here would need to be developed and implemented, including the development o f efficient data transfer protocols, the standardization o f administrative processes, the development o f performance standards, and the creation o f internal review units. These steps would be needed to assure the efficient and effective exchange o f information and to allow effective central monitonng o f regional and local administrative actions. Inthe longer run,the operations of social insurance, financial, and tax agencies should also be subjected to a formal process of risk analysis. Such a process involves conducting a structure review o f all o f the administrative operations, assessingthe nsk o f loss associated with each possible problem together with the cost o f addressing the nsk through additional internal controls, and developing a list of the relative pnonties o f implementing different types o f internal control. 64  Reference Cui, Shaomin, "Implementation status of fundingindividualaccounts inthe Liaoning socialsecunty reformpilot", backgroundpaper, April 2004 He, Ping, "Evaluation of Liaoning Pension Reform Pilot", December2004 LiandTao, "Evaluation of LiaoningSocialSecurity Reform Pilot", (public opimon research), backgroundpaper, June 2004 Lim,Edwinet. al. "Social SecurityReform in China: Issues and Options", November2004 Modigliani,Franco& ShiLarry Cao, "The Chinese SavingPuzzleandthe Life-Cycle Hypothesis", Journal ofEconomicLiterature,vol. 42(l), pages 145-170. NorthernTrust Services, Limited, "The Marginal Cost of Investment Constraints in China", backgroundpaper, June 2005 Sin, Yee Mun,"China:Pension Liabilities and Reform Optionsfor OldAge Insurance", WorkingPaper,World Bank, May2005 World Bank, 2005, "Old Age Income Support in the 21StCentury: An International Perspective on Pension Systems and Reform", World Bank, Washington, DC. Zhang, Juwei, "Basic Situation of theLabor Market in Northeast China", background paper, June 2005 Zheng, Bingwen,"Assessment of Liaoning Social Security Reform Pilot", background paper, May 2005 65  TECHNICAL ANNEX1 EVALUATION THE LIAONING OF PROVINCESOCIAL SECURITYREFORMPILOT Analysis of Pension Liabilities and FiscalImplications under Reform Scenarios' I.Introduction The purpose o fthis annex i s to supplement the information presented inthe report "Evaluation o f the Liaoning Social Secunty Reform Pilot" It covers the provisions o f Document #42, data definitions and sources, the projectionmethodology and assumptions used, as well as a summary o fthe projectionresults. 11.Key Provisions Under Document42 (Liaoning Pilot) Since 1997, China has been strugglingto implement the provisions mandated under the policy framework spelled out in State Council Document #26. The provisions of Document #26 are designed to provide old age insurance for the urban population using a two-tired partially funded pensionsystem that combines socialpooling or Basic Pensionwith IndividualAccounts. Inreality, the system operates on a pay-as-you-go basis for the most part. The Basic Pension portion is financed purely on a pay-as-you-go basis out o f the social pooling account. The Individual Accounts are essentially "empty accounts", since most o f the cash flow surplus has been diverted to supplement the cash flow deficits o f the social pooling account. Recognizing the untenable situation o f "empty accounts", the government passed Document #42 on December 25, 2000 which aimed to improve the effectiveness o f the policy framework for old age income secunty in China onginally set out under Document #26, using the province o f Liaoning as a pilot study The following is a summary o fprovisions as set out under Document #42: Coverage. Coverage under the old-age insurance program is applicable to all kinds o fenterpnses andtheir employees as well as individual workers inurban areas. Although current contributors arepnmarilyfrom state-owned enterpnses, anumber o fmunicipalities have successfully extended coverage to include workers in foreign enterpnses, pnvate firms, personal businesses, as well as casual workers employed by urban enterpnses. Contribution rates. Total contributions by enterpnses are not supposed to exceed 20% o f the contributory wage bill For workers employed under the self-employed category3, their employers are requiredto contribute only 10%o fthe contributory wage bill (roughly halfo f those 1 Ths techcal annex was prepared byYee MunSin, Lead Social Protection Specialist, HDNSP For mumcipalities that are contributmg inexcess o f 20%, the objective i s to gradually mgrate towards the 20% target, but no immediate reductionto the contribution rate i s mandated. Definitionofself-employed vanes bylocation, generally refers to establishment that employed less than 5,7 or 10 employees depending on the local d e f ~ t i o n . Al-1  required o f the enterpnses). Contributions from enterpnses will no longer be allocated to the Individual Accounts. Contributionsto Individual Accounts are to beborne solely byemployees at the rate o f 8% o f contributory wages. Contributory wage. Employeecontributions are subject to a minimumo f 60% and amaximum o f 300% o fthe local average economywage. Those who earn less than60% o fthe average economy wage will have to contribute on the basis o f an earnings level equal to 60% of the local average economy wage. Individuals with unconfirmed monthly wages will have to contribute at the average economy wage o f the enterpnses inthe municipality Pensionable wage. The reference wage used indetermining the Basic Pension i s defined as the local economy's average wage including the wages o f the xza gang workers in the year pnor to retirement. Statutory retirement age. Age 60 for menandage 50 for women ingeneral, andage 55 for women who are inmanagenal positions. Special early retirement age. Men can retire at age 50 and women at age 45 due to illness or disability from work injury, as long as they have contributed for 15 years, and with certification from appropnate authonties. Termination benefits. Workers with less than 15 years o fcontributory service will not be entitled to receive any Basic Pension. Accumulations under the Individual Accounts will be refunded as a single lump sum. Normal retirement benefits. Retirement benefits continued to be determined based on each worker's status as o f December 31, 1996. Workers who were already retired and were receiving pensionpayments in 1996 are referredto as old men. They will continue to receive their pension entitlements in accordance with the old defined benefit formula. Workers who started contributing after 1996 are referredto as new men. Their pensionbenefitswill consist o ftwo parts. Forthose with contributory service o f 15 years o f more: (i) amonthly Basic Pensiono f20% o fthe last year's local average economy wage plus 0.6% for each additional year o f contribution in excess o f 15 years; and (ii)monthly pension payable from the Individual Account denved by a dividingthe accumulated account balance at retirement by 1204 For those with less than 15 years o f contributory service, no pension will be paid, but there will be a a lump sum refind on the accumulated balances from the IndividualAccounts. Workers who were not yet retired in 1996but were already contributing to the old age insurance system in 1996 are referred to as middle men. Their pensionbenefits will be determined on the same basis as the new men, and inaddition, they will be entitledto a transition pensionbased on the formula: (P * A Q *M)+K,where * P =the accrual factor (reduced from 1.4% down to 1.2% underDocument #42) 4 Contributions to IndividualAccounts are creditedwith mterest annually based on actual mterestrates earned onthe mvestments. A1-2  A = average economy wage for the year pnor to retirement Q= an index o faverage contributory wage and is calculated as: where XI, X2, X3;."" ".."Xn represent the individual's contributory wage levels for the years 1996, 1997, 1998 through to the year before retirement; and AI, Az, A3,"'"'"'"' A, represent the average economy wage for the same years; and n is the length o f contributory service, i.e., the years between the time Individual Accounts were first established (assumed to be 1996) through to the year before retirement. M=lengthofservice beforethe establishment o fthe IndividualAccount. Forthepurpose ofthis projection, 1996 is assumed to be the year when Individual Accounts were established. K =a supplement fixed at RMB25 permonth(previously, this amount vanedbymunicipality and could range from RMBO to RMB120 per month). Special early retirement benefit. Retirement benefit i s determined in accordance with normal retirement benefit subject to a 2% per year reduction for each year that retirement occurs pnor to the statutory retirement age. Pensionamounts from the IndividualAccounts are exempt from this reduction. Indexation ofbenefits. Indexation o fthe Basic Pensionwillbeproposedjointly bythe Ministryo f Labor and Social Secunty andthe Ministryo fFinance basedonthe local levelo fcost o flivingand nominal wage growth o f employed workers, to be approved by the State Council. Although no specific level o fpost-retirement indexation was mentioned, other State Council Documents made reference to indexation based on some percentage o f increase innominal wage. Histoncally, the level o findexationhadbeensomewhere between40% and 60% o fthe increase inregional average wage dunng the pnor year5. However, like Document #26, Document #42 remains silent as to whether the pension amounts denved from the Individual Accounts would be subject to the same level o fpost-retirement indexation. Fund management. All funds pertaining to the social pooling or Basic Pensionaccount are to be managed separately from those o f the Individual Accounts, and no subsidies for social pooling may be drawn from the Individual Accounts. Upon exhaustion o f the individual's accumulated balance, any firther pension entitlements under the person's Individual Account will become the responsibility o fthe social pooling funds. 5 State Council Documents#33 (1991) and #6 (1995) bothreferredto lndexation based on increases inaverage local wage levels. Inaddition, documents issuedjolntly by the Mimstry o f Labor and Social SecuIlty and the Mimstry o f Finance announced the actual level o f indexation each year. A1-3  111.SummaryofData The results presented inthis report are based on data collected by the Liaoning Social Insurance Agency (SIA) under a project fundedby an ASEM Trust Fund. Some o f the data were extracted from existing databases o f the municipal SIAs, while others were collected through a customized survey designed and undertaken by the S I A under the ASEM Trust Fund. Only limited investigationwas conducted to examine the overall accuracy androbustness o f the data provided. Information on contributors and pensioners was obtained by extracting data from municipalities that have maintained databases and for those municipalities without databases, sample surveys were conducted to collect the required information. Finally, data on 66% o f pensioners (1.8 million out o f 2.72 million) and 62% o f nominal contributors6(4.18 million out of 6.67 million) was collected. Both age-wise and gender-wise distributions o f the contributors and pensioners from the samples were each grossed up to amve at the age and gender distributions for the whole universe o f contributors andpensioners. To the extentpossible, obvious data flaws were corrected or smoothed to avoid distortions. The results presented here were denved based on 2001 data. Contributors: The total number o f nominal contributors in2001 was reported to be 6.67 million accordingto the SIA. The 4.18 million records collected were first tested for internal consistency (such as logical value o fattained age andentry age into the system) andthen cross checked against the validity o fthe information encrypted inthe code o fthe Identification Card. Annex Figure1:Distributionof nominalcontributorsin2001 30% - I 20% - 10% - I 1 n 1WFernalel 0% 2% 5% 8% 110% 110% 8% 2% 0% 10% InMale 1 0% I 1 ! 1 ~ 3% 1I5% I18% 11% 10% 1110% 115% 112% 1% Source: Data collected by Liaonmg SIA staff Effective contributors refer to those mdividuals who are enrolled 111 the old age msurance systemand have contributed to the systemd u n g the current year. Nomnal contributors refer to those individuals who are enrolled m the old age msurance system, but may or may not be contributing inthe current year. A1-4  Age-wise and gender-wise distributions for the 6.67 million employees were then denved by grossing up the sample distributions. Figure 1 presents a sample distribution o f the active employees in2001 separatedby men andwomen andmajor age groups. Pensionable wage: Average amount o f wage was collected and calculated for each single age cohort for menandwomen separately. The statistical bureaureportedthat average economy wage in2001 was RMB7,444 per year. Figure 2 shows the distributionofpensionable wage by single age cohort normalizedto the wage level o fa 20 year old male (proxy for minimumwage). The flat structure o f the age-earnings profile i s highly unusual and corroborates with the perception that enterpnses generallymanipulatethe wage billinorder to underreport onthe requiredcontribution amounts. It is not clear how the age earnings profilewill shift over time, so, for the purpose o fthis projection exercise, this profile has been kept unchanged over the projectionpenod. Annex Figure2: Sampledistributionof monthlypensionablewage Men (solid), Women (dashed) ~- I5O% T Old agepensioners: The total number o f old age pensioners in 2001 was reported to be 2.72 million according to the SIA. The 1.8 million records collected were first tested for internal consistency (such as logical value o f attained age and age at retirement) and then cross checked against the validity o fthe information encrypted inthe code o f the Identification Card. A1-5  Annex Figure3: Distributionof old age pensioners 19 t8 n" m 20% - .-0 E t lis 10%- 3 :a rn O% ! 45-49 50-54I55-59 60-64I 65-69 70-74 75-79 80+ 1 mFemale1 2% I1 6% I1 13% j 18% 11 7% 1 9% 11 3% 11 1% 11 0% OMale 1 0% 1 1% 12% I 4% 1 6% 1 12% 1 11% 1 5% 1 0% Source: Data collectedby Liaonmg SIA staff Average pension: Basedon the pensiondisbursements reported for 2001 (RMB15 7 billion) and atotalnumbero f2.72 millionpensioners, the average pensionfor the year 2001was determined to be RMB5,773. Figure 4 shows the distribution o f average monthly pension for male and female pensioners normalized to the average pensionpayable to those at age 60. Annex Figure4: Distributionof average monthlypensionby age Men (solid), Women (dashed) 160%1 89 if 0 120% 80% B 40% 0Yo 33 38 43 4% 53 58 63 6% 73 7% 83 8% 93 9% Age Source: Data collected by Liaonmg SIA staff Retirement Pattern. Statistics on average retirement age and the length o f service at retirement were collected by the Liaoning S I A staff, calculated from the collected data. Based on the A1-6  statistics collected, menwere found to retire on average around age 56 while women were found to retire around age 47 (hereon referred to as the assumed retirement age), although the statutory retirement ages are 60 for men, 55 for women inmanagenal or professional positions, and 50 for women innon-managenal types o f services. Length o f service at retirement averaged 26.9 years for menand 17.8 years for women. Figure 5 shows the averages years o fservice at eachretirement age. These averages were used as the retirement patterns for future years. Annex Figure 5: Average years of service at retirement Men(solid), Women (dashed) o ! I I . . . . . . . . . . . . . . . . . . . . . . . . . . 33 38 43 48 53 58 Age Souce: Data collected by Liaomng SIA staff A1-7  IV. ProjectionMethodologyandAssumptions This section presents the overall structure o f PROST, the projection methodology, a summary o f the key assumptions used with the simulations, and the financial flows o f Liaoning's old age insurance system. PensionReformOptionSimulationToolkit (PROST) A genenc PC-basedprojectionmodel developedbythe SocialProtectionUnito fthe World Bank -the PensionReform Option Simulation Toolkit (PROST Version 10)-was usedto analyze China's old age insurance system. The PROST model was developed in VBA and can be used withthe Windows 95/98/NT operating systems inconjunctionwith Office 97 or laterplatforms. PROST has been used in almost 90 countnes to provide quantitative input for pension policy discussions. Outputs from PROST were satisfactorily benchmarkedagainst a number o fcountnes where micro actuanal models were used. As well, a copy o f PROST was provided to the International Labor Organization (ILO) for testing against ILO's own projection model. Results from all such benchmarkinghestingproved the PROST methodology to be sufficiently robust. At the same time, its flexibility has permitted easy adaptation to specific country circumstances for sensitivity testing andcompansons under awide range o f economic andpolicy scenanos. As is the case with any simulation model, the quality o f the outcome from PROST depends largely on the nature and quality o fthe input data as well as the types o f assumptions used for the simulations. Later inthis annex, the key assumptions used and the data from which they were denved will be presented. It should be noted that given the nature and quality o f currently available data, the output from this exercise shouldbe viewed as a diagnostic guide rather thana quantitative analysis o f the old age insurance system in China, and the results should not be quoted in absolute terms without proper reference to the underlyng assumptions. Furthermore, due to the heterogeneous demographic and economic conditions in Liaoning, a macro-level projection o f its old age insurance system cannot be assumed to adequately reflect the vanability infinancial status. Data and assumptions used inprepanng estimates for Liaoning's OldAge Insurnace Systemwere based on information provided by the Liaoning Finance Bureau, the Liaoning Social Insurance Bureau andthe Liaoning StatisticalBureau. The publications "China's Population at the Turn o f the Centunes: Liaoning Chapter" and"Wanshan Chengzhen Shehui Baozhang Tixi Shidian" were also used as references. ProjectionMethodology PROST consists o f an input workbook and five output modules. The input filehemplate i s an Excel workbook with six embedded worksheets that contain a large number o f defined vanables. Eacho fthe five output modules contains vanous Excelworksheets with graphical presentations o f key results. PROST begins with a top-down approach, first projecting the population, then determining the size o f the labor force and the number o f employed persons. Depending on the method chosen, the number o f contributors can be denved either from the general population or AI-8  the labor force. Beneficianes are then determinedbased on the retirement patternandthe number o f contributors ineach age cohort. Revenues andexpenditures are calculatedby applyng average wage levels o f each age cohort against the number o f contributors and pensioners in the same cohort. Financial flows can thenbe calculated based on the resulting revenues and expenditures. Figure 6 provides a schematic illustration o fthe projection methodology Annex Figure 6: Projection Methodology Population ILaborForceand EmployedPersons I 7 IncreaseIDecrease InvestmentIncome The InputFile Datamustbe entered for the baseyear andthe endyear o fthe simulation honzon. Information for intervening years can be inserted if there are anticipated changes that may alter the linear relationship between the base year and the final year. The program i s equipped to handle both "stock"-based as well as "ji`ow"-based calculations. With the "stock" method, the current pools o f contributors and beneficianes are expressed as percentages o f either the overall population or the total number o f employed persons; from thereon, the numbers o f new retirees and the newly disabled are denved basedonthe stock o fcontributorsby age cohort, taking into consideration the assumed retirement age, length o f service and the histoncal retirement pattern. With the "$`ow" method, the increments (or decrements) o f different categones are computed first. These increments/decrements are then aggregated with the beginning stock o f contributors and beneficianes. Steps for entenng data requirements under the input templates are summanzed as follows: 0 General Inputinformation about the economy (inflation rate, realinterest rate, real GDP growth, etc.) as well as parameters o fthe pension system (current benefit expenditures, retirement age, accumulated reserve fund, etc.). Population Inputthe age structure o fthe population inthe base year alongwith age-specific fertility andmortalityrates as well as migration information for the simulation honzon. A1-9  Labor Inputthe labor force participation rates, unemployment rates, evasiodexemption rates and earningsprofiles for the simulation honzon. Pension Inputpension systeminformation for the base year including the age structure o f contributors andpensioners. Pension coverage rates and replacement rates for new beneficianes accordingto the relevant formula(s) will also berequiredfor the simulation honzon. Reform Inputparameters that are relevantto the reform o fthe system such as switching patterns, acquired nghts, replacement rates before and after the reform, andprovisions under bothnotional accounts and a funded defined contribution system. Profiles Inputprofile data for specific individuals includinggender, years o fservice, earnings and mortality profile. The Output File PROST generates five output modules. Eachmodule contains a number o f Excel worksheets and graphical summanes o fkey indicators. The output modules are arranged as follows: Population Key output includes populationpyramids, life tables (by age and Projection gender), life expectancy changes and demographic indicators such as population dependency rates. Demographic Projections o f labor force, employment, contributors and Structure beneficianes, demographic structure, lengtho f service at retirement, headcounts o fcontributors andbeneficianes, as well as calculated system dependency ratios. Financial Macroeconomic trends and wage distributions are shown, Flows contribution revenues andpensionpayments are calculated, and the financial balance and implicit pensiondebt are projected for each year. The total number o fpensioners i s divided into two groups -the existing stock andthe newly retired. Finances o f The financial balance, replacement rates from eachpillar and Multipillar aggregate results o f the reform are calculated. Individual These are illustratedusing individuals under a pay-as-you-go Accounts contributory scheme, with affordable replacement rates, indefined contribution andmultipillar schemes. A1-10  ProjectionAssumptions Simulation Horizon: To properly study a pension system and to generate meaningful projection results for sound policy analysis, the minimumnumber o fprojection years should spanmore than one generation or about 75 years - a penod deemed to be o f adequate duration to demonstrate emerging trends and implications for Liaoning's old age insurance system. Hence it was decided that the projection penod would run from 2001 through to 2075 Given the long term nature o f pensionarrangements, some experts even suggested that it wouldbewise to consider a simulation honzoninperpetuityto helppolicy makers understand the financial dynamics o fthe system more fully 2001 was chosenas the baseyear, the year for whichthemost complete documentationwas available for all the vanables needed for PROST More comprehensive data on the charactenstics o f contributors and pensioners under the old age insurance scheme came from on information provided inthe report preparedbythe Liaoning Social Insurance Bureau"Liaoning Enterpnse Old Age Insurance: Actuanal Report (2001-2075)". Populatzon Structure: Populationdata for 2000, including age structure and gender composition, was extracted from the 2000 Population Census, andthenprojectedto 2001 based on age-specific mortality rates inLiaoning. Mortality Rates: Age-specific mortality rates (in five-year age groups) for the years 2001-2075, includingprojectedimprovements inmortality (at five year intervals) were providedbythe World Bank's PopulationUnit. These mortality rates were usedto calculate the probability o fdyngby age cohort (age 0 to age 100) for both men and women for every year o f the simulation honzon. Both current and projected life expectancies at vanous ages were compared against other published sources as a means o f venfication for overall reasonableness. Key demographic indicators are shown inTable 1below 2001 2002 2003 2004 2005 2006 2010 2025 2050 2075 Male Life Expectancy:At Birth 71.8 71.9 72.0 72.1 72.2 72.3 72.7 74.2 76.3 78.7 At Age 20 53.6 53.6 53.7 53.8 53.8 53.9 54.2 55.3 57.0 59.0 At Age 60 17.6 17.6 17.7 17.7 17.8 17.8 18.0 18.8 20.1 21.7 At Age 65 13.9 13.9 13.9 14.0 14.0 14.1 14.2 14.9 16.1 17.5 Female Life Expectancy:At Birth 75.6 75.7 75.9 76.0 76.2 76.3 77.0 79.1 82.0 85.3 At Age 20 57.3 57.5 57.6 57.7 57.8 57.9 58.3 60.1 62.7 65.6 At Age 60 19.8 19.9 20.0 20.0 20.1 20.2 20.5 21.9 24.1 26.5 At Age 65 15.8 15.9 15.9 16.0 16.0 16.1 16.4 17.7 19.6 21.9 International expenence has shown that population groups that are covered by social secunty (usually associated with higher income, healthier living conditions and better medical care) generally exhibit lower mortality rates than the overall population. Furthermore, municipalities that are more developed tend to report longer life expectancies. Figure 7 compares the difference between the projected life expectancy for the national population and that projected for the Liaoningprovince. Under each o fthe ages shown, people inLiaoninglive longer than the average Al-11  in China. It is important to note that the prevalence o f early retirement in Liaoning is another source o ffiscal strainonthe old ageinsurance system -the average age o fretirement for women in Liaoning was estimated to be around 46, compared to the national average o f 50; for men the average age o f retirement was estimated to be the same as the national average. Annex Figure7: Comparisonof life expectancy derivedfromprobabilitiesof dyingbasedon nationalandLiaoningmortalitytables I1 1 INat'l Men I L N Men 0Nat'lWomen women 1 35 !! r a:Iif1:'1 30 15 0 50 55 65 ARA Note: ARA means average retirement age Source: World Bank staff calculations based on PROST Sex Ratio at Birth. The number o f males bornper 100 females was assumed to decrease slowly Ifrom the current level o f 115 to 112 by the year 2020 andbe hrther reducedto 111by 2075 Age-specific Fertility Rates: The age-specific distribution o ffertilityrates in2000 was assumedto be identical to those quoted in the publication "China's Pouplation at the Turn o f the Centunes: Liaoning Chapter" There was no consensus among demographers as to whether China's total fertility rate would continue to decline, remain constant at the current level, or increase to a minimumlevelo f2.1 to maintainzero population growth7 Under this projectionexercise, it was assumedthat total fertility rate will increase gradually from the current level o f 1.2 to 1.5 by 2015 and eventually returning to the maintenance level o f2.0 by2075 Net Migration. In2001, the net number o fmigrants (net flow o f immigrants and emigrants) i s estimated to be 560,000. Inthe absence o f information, it was assumed that these immigrants are evenly distributed between men andwomen betweenthe ages o f 16 and 55 The net inflow o f migrantsis expected to gradually dissipate by2010when there will be no matenalincrease or decrease inthe overall number o fmigrants. 7 Table 3.3, page 66. A1-12  Projected Populatzon 2001-2075 Based on the above adjustments, the population o f China is projected as follows: Pt =PO+B -D I-M + where, Pt i s the population in year t; POi s the population at the beginning o f the year; B is the number o fbirthsinyear t; D i s the number o fdeaths inyear t; Ii s the numbero fimmigrants inyear t; andMISthe number o femigrantsinyear t. Figure8 shows theprojectedshare o fpopulationby the three major age groups (0-14, 14-60 and 60+) and the corresponding old age dependency rate o f the population Annex Figure 8: PopulationandOldAge DependencyRateof the Population 100% - Population Old Age Share of working age 75% - 50% - Share of young 25% - 0% I , 2001 2011 2021 2031 2041 2051 2061 2071 Year Source: World Bank staff calculationsbased on PROST The Agzng Phenomenon. Results from the population projection (see Figure 7) show that Liaoning's working populationwill likely continue to increase through to 2009. Starting with the year 2010, its working age population i s expected to decrease until it reaches the lowest point around 2050. Onthe other hand, its old agepopulationi s projectedto grow steadily untilitreaches the highest point at aroundthe same time, 2050. The combined effect o fthis i s that the population old age dependency rate' will peak at 90% in 2050. To exacerbate this problem, this phenoemenon will grow at arapidpace. Figure 9 shows the growthrate o fthe working age andold age populations. The penodbetweennow and2009 is the only interval dunngwhich the working age population is going to expenence positive growth rates. Thereafter, the growth rate remains negative throughout the projection penod. Incontrast, the growth rate o f the old age population continues to expenence positive growth through to 2035, and at amuch faster pace than that o fthe working age population. 8 Populationoldage dependencyrate 1s definedas the rate denvedbythe numberof older persons (ages 60 and over) dividedbythe numberof workmgage mdividuals (ages 15-60). A1-13  In view of this emerging demographic phenomenon, many demographers and economists advocate the need to develop contingency plans now to forestall shortfalls associated with the impending aging cnsis. One strategy would be to capitalize on the demographic dividend when productivity is in abundance and to set aside reserves dunng times of surplus for the purpose o f supplementing available resources dunng times o f combined accelerated population aging and decreasing working age productivity -when the population burdenhits. Annex Figure9: Populationgrowthrate 8% 1 &\ A I\ # -4% ' Year Source:PROST output Contributors,BeneficiariesandCoverage Worworce: Age-specific labor force participation rates and unemployment rates for the total population were calculated from information published in the 2004 Labor Statistics Yearbook. Since the projections expressed old age insurance coverage as apercentage o fthe total population, the assumption that there would be no systemic changes in labor force participation rates throughout the simulation penod does not affect the overall dynamics o fthe projections. Coverage: Overthe 6 year penodfrom 1998 to 2003, the numbero fcontributors (from enterpnses and self employed) had increased marginally from 6.35 million to around 6.89 million, merely 8.5%. Incontrast, the number o fpensioners increased at a much faster pace-from 2.37 millionin 1998 to 2.96 million in2003, a 25% increase.' Another phenomenon that is equally alarming i s the surge inthe number o f contributors claiming the self-employed status, from slightlymore than 5% o f the contributors' base in 2000 to over 27% by 2003" It is envisaged that this ratio will remain high, at around 30% inthe foreseeable future. 9 ReportonActuanalProjections, LiaonmgProvinceEnterprise OldAge Insurance (Nov 2004). 10 H e Ping (2004). A1-14  Table 2 presents the evolution o f the enterpnse contributors and pensioners from 1998 to 2003. While some o f this increase could be explained by increased longevity and general aging o f the population, another problematic development i s the large proportion o f pensioners who are currently younger than the earliest statutory retirement age for both men and women (18% for male pensioners and 17% for female pensioners). Year Contributors Pensioners No. in 000 % changefrom1998 Pensioners % changefrom1998 1998 6,350 2,370 1999 6,600 3.9% 2,460 3.8% 2000 6,840 7.7% 2,640 11.4% 2001 6,670 5.0% 2,729 15.1% 2002 6,700 5.5% 2,830 19.4% 2003 6,890 8.5% 2,960 24.9% Currently, coverage i s estimated at around 25% when measured as the ratio o f contributors from enterpnses and self-employed to working age population. Since the Old Age Insurance System is largely an urban system, it might also be appropnate to measure coverage as percent o f urban population, which i s around 46%. Under the status quo, it was assumed that there would be no systemic coverage expansion, thereby allowing a more accurate assessment o f the overall sustainability o fthe system free from any exogenous influence. Bythe year 2000, the province o f Liaoning had already reached a level o furbanization (54.9% o fits population livinginurbanareas) just marginally second to the economically most developed province o f Guangdong (at 55 1%). However, there i s wide dispanty inthe level o f urbanization among the vanous municipalities - from as highas 70% inShenyang to as low 33% inHuludao. Inreality,itisverylikelythat coverageexpansionundertheoldageinsurancesystemhasalready been taking place by virtue o f the growth in size o f the urban population. In addition, many municipalities/provinces have begunto extendcoverage to pnvate, foreign-owned enterpnses and the self-employedinurbanareas as ameans o fcollecting more revenue to finance the deficits. A third factor that is expected to ultimately affect coverage is the emerging international trend whereby civil servants and employees o f State organizations and institutions are gradually being integrated into the national old age insurance system instead o f being provided with separate pension schemes. Although itwas assumed that there wouldbeno systemic coverage expansionunderthe status quo, given the above stated factors, three o f the scenanos included inthis projection exercise assumed that coverage in the old age insurance system would be expanded in accordance with the rates shown in Table 3. Given that rural workers have been participating in some form o f pension insurance scheme separate from the existingold age insurance system, the implicit assumption is that these rural workers will continue to do so, perhaps on an even larger scale. A1-15  Annex Table 3: Assumed CoverageRate under the coverage expansionscenarios Contributorsexprssed as: No expansion With expansion same as 2001 by 2020 by 2050 by 2075 --%Male of working age population -- Female 25.2% 34.6% 45.9% 47.9% 26.5% 34.9% 48.0% 53.8% % of estimated urban population ---- Male 45.0% 61.8% 81.9% 85.5% Female 47 4% 62.3% 85.8% 96.1% Retirement Pattern. Based on the collected statistics, it was estimated that menwould retire on average around age 56 with 30 years o f service, andwomen around age 47 with 25 years o f service. Inprojecting the impact of gradually increasing and unifyngretirement at age 65 (by the year 2022 for men and 2034 for women in all employment categones), further assumptions on behavioralpatterns were made: (i)Underthestatusquoscenano - the current retirement pattern pnor to the existing statutory retirement age would remain unchanged (approximately 15% o f the retirees); (ii)retirementageweretobepostponed,50%ofthosewhohavereachedtheassumedretirement If age would remain as contributors untilthey reach their statutory retirement age effective inthe year o fretirement; 25% would still elect to retire pnor to the statutory retirement age for health reasons or other exogenous factors beyond the system's control, while the remaining 25% would simply "disappear" into the informal sector. Macroeconomic assumptions: Economic indicators used for the years 2001-2004 are based on actual data. For 2005 and after, macroeconomic assumptions usedbyPROSTwere denved based on input from staff in the macro policy unit, taking into account histoncal indicators and projections from expert sources. Table 4 summanzes the key economic assumptions. GDP in 2001 was assumed to be RMB503,308 million. Real wage growth was assumed to be at panty with the growth in GDP per capita. These economic parameters were used in conjunction with assumptions concerning the population and the labor force for anticipated coverage and to simulate the fbture financial conditions o f the old age insurance system. Due to the high uncertainty associated with wage growth, the age earnings distribution o f contributors was assumed to observe the profile shown inFigure2. MacroeconomicTrends 2001 2002 2003 2004 2005 2006 2010 2025 2050 2075 Real GDP growth 7.0% 5.8% 12.1% 9.2% 6.6% 7.9% 7.4% 6.6% 4.1% 3.1% Real wage growth 12.1% 20.7% 12.7% 13.6% 6.9% 6.3% 5.8% 5.8% 4.0% 2.8% Inflation rate 0.0% -1.1% 1.7% 3.0% 3.0% 3.0% 3.0% 3.0% 3.0% 3.0% Real discount rate 4.0% 4.0% 4.0% 4.0% 4.0% 4.0% 4.0% 4.0% 4.0% 4.0% Evolution of contributors andpensioners. Basedon the coverage assumptions andthe retirement patterns described above, Table 5 illustratesthe demographics o fthe old age insurance system and its dependency rates inselected years dunngthe penod 2001-2075 under the status quo scenano. A1-16  2001 2002 2003 2004 2005 2006 2010 2025 2050 2075 Employed contributors (000) 6,673.5 7,124.1 7,308.5 7.365.2 7,451.3 7,421.9 7.586.0 5,859.1 3,764.3 3,006.0 Male 3,618.3 3,876.0 3,989.2 4,030.2 4,096.1 4,097.8 4,247.9 3,413.1 2,204.9 1,752.0 Female 3,055.2 3,248.1 3,319.3 3,335.0 3,355.3 3,324.0 3.338.1 2.446.1 1,559.4 1,254.1 Old Age Pensioners (in 000) 2,715.5 2,940.9 3,074.6 3.209.3 3,336.2 3,462.5 3.990.3 6.286.1 6,883.4 5,028.6 Male 1,089.2 1,175.6 1,211.2 1.236.8 1,261.3 1,288.8 1.437.6 2,331.5 2,613.1 1,872.6 Female 1,626.3 1,765.3 1,863.4 1,972.5 2,074.9 2,173.7 2,552.8 3,954.6 4,270.2 3,156.0 System Dependency Rate 40.7 41.3 42.1 43.6 44.8 46.7 52.6 107.3 182.9 167.3 SOU Table 6 illustrates the demographics o f the system under the expanded coverage scenano, where pension coverage will reach at least 85% o f its urbanworking populationby 2075 Annex Table 6: System Demographics and dependency rates of the old age 1 1 insurancesvstem - Coverage exDansion (2001-2075) 1 - 1 2001 20021 20031 20041 20051 20061 20101 20251 20501 207t EmployedContributors(in 000)II 6,673.5 II 7,124.1 I 7,308.5I 7,365.2I 7,451.3I 7,870.7 I 8,714.1 I 10.752.6I 9,319.2 I 8,032.9 3,618.3 3,876.0 3,989.2 4,030.2 4,096.1 4,348.6 4,841.8 5,830.2 4,740.1 3,958.5 Female 3,055.2 3,248.1 3,319.3 3,335.0 3,355.3 3,522.1 3,872.3 4,922.4 4,579.1 4,074.4 Old Age Pensioners(in 000) 2,715.5 2.940.9 3.074.6 3,209.3 3,336.2 3,097.4 3,520.0 4,829.2 6,919.3 5,437.4 1,089.2 1,175.6 1,211.2 1.236.8 1,261.3 1,009.9 1,074.4 1,647.7 2,640.7 1,979.4 Female 1.626.3 1,765.3 1,863.4 1,972.5 2,074.9 2,087.5 2,445.6 3,181.5 4,278.6 3,457.9 S stem Dependenc Rate 4 0 . 1 4 1 . 1 42.1 43.6 44.8 39.4 40 4 44.) 7 4 . 1 67.7 Source: World Bank staff calculations based onPROST Itshouldbenotedthat underbothscenanos, the dependency ratios continueto increasethroughout the projectionpenod. However, when the dependency ratios inTable 5 andTable 6 are compared, itisobvious that as aresult o fcoverage expansion (the combinedeffect oftheinclusiono fworkers from other sectors andthe urbanization process) which adds more contributors to the system, the burdeno f supportingthe pensioners is less onerous underthe Table 6 scenano. Pension System Financial Flows To project the financial flows of China's old age insurance system, revenues andexpenditures had to be determined first. Finances for the Basic Pension scheme (including transition pensions for the mzddle men) and the Individual Accounts were projected separately and outputs on key indicators were summed to give the final results. Revenues: The following chart gives a schematic view of the calculation o frevenues and sources. Contribution revenues were calculated based on the prescribed national average contribution rate (see descnption below), average earnings o f each age cohort, and the number o f contributors in each cohort. Government contributions (by local and centralgovernments) were provided only on an ad-hoc basis to supplement any current deficits. A1-17  Annex Figure11:Calculationof Revenues L Revenues Contributions Direct governmentcontributions .r Contributors in age group X Average wage of age group x Contribution rate PrescribedContribution Rates. Basic Pension-Annual contribution rates to the Basic Pensionvary amongmunicipalities -from as highas 25.5% inAnshan, FushunandDandongto as low as 19% inDalianandPanjin. Table 7 presentsthe contribution rates to the Basic Pensionbythe 14municipalities for the years 2001-2005, includingthe estimated annual average for the municipalities combined. Annex Table 7: Contributionratesbv m iniciDaliities (2001-2005) Municipality Shenyang 7 23.5% Dalian 19.0% 19.0% 19.0% 19.0% 19.0% Anshan 25.5% 24.6% 25.5% 25.5% 25.5% Fushun 16.0% 24.5% 25.5% 25.5% 25.5% Binqi 24.0% 24.0% 24.0% 24.0% 24.0% Dandong 25.5% 25.5% 25.5% 25.5% 25.5% Jinzhou 24.0% 24.0% 24.0% 24.0% 24.0% Yingkou 25.5% 25.5% 25.5% 25.5% 20.0% Fuxin 20.0% 20.0% 20.0% 20.0% 20.0% Liaoyang 25.5% 25.5% 25.5% 25.5% 20.0% Panjin 19.0% 19.0% 19.0% 19.0% 19.0% Tieling 24.0% 24.0% 24.0% 24.0% 20.0% Chaoyang 20.0% 20.0% 20.0% 20.0% 20.0% Huludao 20.0%1 20.0% 20.0%1 20.0%1 20.0% 22.0%1 21.0%1 17.0% * The estimated average takes into account the proportiona workers contributing at the rate of self emnlnved lIOO/,\ IndividualAccounts -Contribution rate to the IndividualAccounts was assumedto increasefrom 5% inJuly 2001 to 8% per year. For the purpose of this projection, no differentiation was made between employer and employee contributions. Only total contributions to the Basic Pension and IndividualAccounts were considered. A1-18  Collection Rates Statistics from the Liaoning SIA showed that collectionrate rangedfrom 84% in2001, to 76.1% in 2002 and 77 4% in2003. Segregated information on defaults andcontributions inarrears was not available. Since Liaoning has switched to tax collection (except for Dalian), it was envisaged that collection would likely improve over time. For this projection, the actual collection rates for 2001-2003 were used. It was further assumed that collection would increase to 80% in2004 and 2005, eventually reaching 85% by2030 andremaining steady at that levelthrough to the endo fthe projectionpenod. Expenditure. Figure 12gives a schematic view of the determination ofbenefit amounts andother sources of expenditure. Pension entitlements for the newly retired are determined based on the provisions described in the previous section. Pension entitlements for the existing stock o f pensioners are denvedbased on the extrapolated pensionprofile from the seven locations. Annex Figure12: Calculationof Expenditures I Expenditures I i Newly retired Pension formula Agelsex Length of service Stock Last year's pension I Indexation factor I Post-retirement Indexation All pensionpayments are assumedto be indexedto 60% ofnominalwage growth each year until 2025 From 2026 through to 2075, pensionpayments are assumed to be indexedto 100% o fpnce changes. Annuztypayments. Monthlypensionpayablefrom the IndividualAccount is calculatedbydividing the accumulated account balance at retirement by 120regardless of actual age at retirement. One o fthe measures underthe parametnc reform i s to use an actuanally fair annuity factor insteado f a factor o f 120 by taking into consideration: (i)the average remaining life expectancy at age o f retirement; (ii)post-retirement indexation policy; and (iii) anticipated interest rate dunng the the annuitant's life time. Figure 13 illustrates three sets o f annuity factors using three interest A1-19  assumptions for any male worker retinng at the ages indicated in the year 2040 and receiving a pension that will be indexed inline with pnce changes. 250 - - - - i! g - 5 200 - ; ----- 3% real ---- L 150- 4% real 1 I---- 120 factor -.I-- L -5 2, 100 - 5 i 50- O N , , , , , I I , , , , , , , , , , , , # 1 Fiscal Balance. From the revenue and expenditure payment streams, current balance and fund reserve were calculated. Currentbalance is definedto bethe difference betweenrevenues andexpenditures plusinvestment income, ifany. Fundreserve is normally determinedbased onthe following: A(t+I) =A(t) *(I+r(t)) +I(t) -E(t) where, A(t) is the amount o f fimdreserve at the beginningofthe year A(t+l) is the amount o ffundreserve at the endo fthe year r(t)isthe assumedportfolio returnfor the year t I(t) is the revenue for the year t E(t) is the expenditure for the year t Revenues under the Basic Pension and the Individual Accounts were accounted for separately, so contribution revenues to the Individual Accounts would not be available for subsidizing deficits in the Basic Pension. Furthermore, in any year when revenues from the Individual Accounts fall short o fexpenditures, the deficit will first be financedthrough the accumulatedfundreserve. Once the accumulated find reserve has been exhausted, any shortfall will then be provided through the A1-20  Basic Pension (social pooling budget)'' Inreality, it will be very difficult to implement such a policy Firstly, nationalpooling does not exist. Secondly, municipalities with significant reserves may not be the ones that are expenencing shortfalls in cash flow, while those with heavy cash deficits may not have sufficient reserves. It is important to keep inmindthat the purpose o f this exercise is to provide an estimation o f the overall fiscal condition o fthe Liaoning old age insurance system. As such, the methodology is not meant to be used as a budgeting tool for yearly estimates o f cash flows. Moreover, the only sure way to amve at an accurate picture o fthe national fiscal status o fthe old age insurance system is to perform a bottom-up simulation, which involves aggregating cash flow requirements from the lowest level o fpooling. Fund Reserve. No f h d reserve was assumed to exist as o f2001. Return on Investments. Histoncally, nominalinterest rates on deposits o ffinancial institutions had been quite low Table 8 presents the histoncal interest rates credited to the Individual Accounts since their inception in 1996. It is not uncommon that special bonus declarations were made from timeto time to provide ahighercreditedinterest than that paidbythe financial institutions for the deposits. Annex Table 8: Interest rates creditedto IndividualAccounts (1996-2003) Year I Credited InterestI Inflation I Real interest rate 19961 9.21%I 7.90%1 1.21% 1997 7 17% 3.10% 3.95% 1998 5.00% -0.70% 5.74% 1999 5.27% -1.40% 6.76% 2000 5.27% -0.10% 5.38% 2001 5.84% 0.00% 5.84% 2002 2.25% -1.10% 3.39% 20031 1.98% 1.70% 0.28% Source: World Bank staff calculation based on informationprovide by Liaoitung SIA staff Though much o f the reserves under the Individual Accounts are invested in negotiated bank deposits, the return on investments would likely remain low inthe next few years - not to exceed 2% p.a. in real terms. With the assumption that an asset allocation policy using a diversified portfolio management approach will be developed and fully implemented by 2010, the return on investments can be expected to increase to 3% p.a. inreal terms for the remainder o fthe projection penod. 11 Itwas reportedthat, mpractice, subsidy fromthe SocialPoolingbudget will be made available as soon as each mdividual's account balance hasbeendepleted despite the availability o ffundreserves ingeneral underthe Individual Accounts. A1-21  V. Summary of Financial Projections Development of Projection Scenarios. Projectionsunder atotal o f6 scenanos were prepared. Apart from the status quo (Scenano l), four scenanos were constructed to analyze the vanous policy implications, and one additional scenano was developedto highlightthe implications o f coverageexpansion. The scenanos were designed to help policymakers understand the composition o f the pension liabilities and hence find a financing solution as well as make informed decisions on how best to reform the system. Table 9 below summanzes these scenanos: Annex Table 9: Summary of Scenarios Scenarios Description 1 Liaoningprovisionsunder Document#42, includingearly retirementsand transition pension, with safeguard to contributionsto IndividualAccounts 2 Remove systemicflaws from IndividualAccounts (using actuariallyfair annuityfactors with no refundof principalupon death), increased and unifiedstatutory retirement age (65 for men by 2022 and 65 for women by 2034), post-retirementindexationchange from 60% wages to 100% prices in 2026, self employedto contributeat full rate 3 System with no systemicflaws, no early retirement assumed (past and future) 4 Removetransition pensionfor middle men, assume all middle men (active or retired) get the longterm replacementrate basedon length of service 5 Scenario4 using longterm requiredcontribution rate 6 Scenario6 with coverage expansion at long term requiredcontribution rate Scenario I representsthe status quo inline with the provisions under Document #42 as described inSection2. Scenario 2 includes a list of parametric reforms to address the critical flaws under Document #42 (Scenano 1). Although the design o f this Liaoning pilot was touted as a bluepnnt for improving benefit provisions and restonng fiscal sustainability, its implementation fell short o f addressing several cntical system flaws12 Therefore, the first step inthis analysis i s to examine what would be the implications ifthese systemic flaws hadbeen removed. These include: (i) the use o f actuanally fair annuity factors by age and gender inplace o fthe current factor o f 120; 12 WorldBankreport (May2005). AI-22  terminating the practice o f refunding accumulated contributions to deceased pensioners; gradually delayng andunifyngretirement age (age 65 by 2022 for menand age 65 by 2034 for women), although special early retirement would still be permitted; and workers employed by small establishments and who qualify under the current provisions o f "self employed" would be requiredto contribute at the hllrate, i.e., the provincial average o f 20% to the Basic Pension instead o f at the current rate o f IO%, and post-retirement indexation would be changed to 100%o fpnce changes beginning with the year 2026. Scenano 2 assumes that the above listedparametnc changes would be adopted under the Liaoning pilot. Scenario 3 removespast andfuture effects of early retirement. Overthe last decade or so, the old age insurance systemwas used as a venue to ease the social impact o f economic restructunng and SOE reform. Xia gang workers and others who could not find employment or were not "employable" were permitted to take advantage o f the Special Early Retirement provisions and retire early This created a stock o f some halfa millionworkers (roughly 15% o fthe total stock o f pensioners, see Figure 3) who became pensioners prematurely These were policy decisions inthe past which shifted the social costs o f economic reform onto the old age insurance system. The fiscal implication o f such actions needs to be quantified and corrected. Furthermore, with the Bingguz initiative and the introduction o f the Unemployment Insurance System, there is no justification for allowing the loosely defined Special Early Retirement to continue. Scenano 3 takes the parametncreform inScenano 2 a step further byremoving all existing early retirees from the system as well as prohibiting the old age insurance system from being used as a "dumping ground" to aid unemployment. Scenario 4 eliminates the entitlementsfrom the old system and the transitional arrangements and assumes the reformed system has reached maturation. Since 1996whenthe concepts o fold men, middle men andnew men were introduced, the average replacement rates beingpaid from the old age insurance system have declined slowly and steadily Pnor to the Liaoning pilot (2000), average replacement rate was 87.66%. After the introduction o f the pilot, the replacement rates were 88.86% in 2001, 85.59% in 2002, 82.99% in 2003 and 78.56% in 2004, an average annualized rate o f2.7%. At this pace, it willtake a longtime to lower the replacement rate downto the target level o f 58.6%. Furthermore, the middle men are entitled to supplements inthe form of transition pensions to make up for the "lost years" when they were not contributing to the Individual Accounts. Ina way, these are liabilities o f the past system and should be identified as such. Scenano 4 presents a case whereby no special provisions for transition or entitlements from the old system would continue. Instead, it was assumed that everyone who retires from the system (past and hture) would be covered by the same benefit formulas under the Basic Pension and Individual Accounts based on their anticipated years o f service, as if the system had already reached maturation. A1-23  Scenario 5 aims to reduce the Basic Pension contribution rate. The long term required contribution rate under Scenano 4 should be adopted as the statutory rate for the Basic Pension. This will make the system more attractive for workers, thus boost compliance, which inturn will create the combined desirable effects o fhigher coverage andmore contribution revenues, without jeopardizing the other fiscal sustainability benchmarks. Provisions for the Individual Accounts will remainthe same as under Scenano 4. Scenario 6presents the implications of expanding coverage under theprovisions of a matured system. Inthis scenano, coverage will be expanded, as shown inTable 3, to nearly all o fthe urban population gradually under the matured system described in Scenano 5 with a reduced contribution rate. Fiscal SustainabilityBenchmarks Fiscal sustainability refers to the financial soundness o f the system, now and inthe futureI3. It i s widely acceptedpracticeto measure the fiscal sustainability o fpensionsystems with vanous benchmarks. The following three commonly usedbenchmarks were used to test the financial soundness o f Liaoning's old age insurance system: Financing gap; Implicit pensiondebt (PD); and Required contribution rate The financing gap is a measure calculated by summing the net present values o f the current balances, i.e., revenues less expenditures, throughout the projection penod (in this case 2001-2075). This i s aparticularly useful indicatorwhen considenngthe financing requirements o f an ongoing pension system that has reached maturation. For systems that are still expanding, usingthe financing gap as afiscal sustainabilitybenchmarkcanbemisleading, since theprojection penod may include only contribution income from the newly covered workers without reflecting the total payout penod for these workers when they retire. As with all calculations using net present values, this measure i s highly sensitive to the discount rate used in the calculation and should be quotedwith pr~dence'~.However, as long as the macroeconomic assumptions adopted for the projection are dynamic and internally consistent, expressing the financing gap as a percentage o f the base year GDP may be a usefbl indicator. Implicit pension debt ( P O ) i s a measure o f the present value o f projected liabilities under the pension system and covers those who are already receiving payments from the system as well as those who have any acquired nghts at a particular date. This i s a useful indicator to measure the stock o f all outstanding liabilities. It i s also a common indicator used by the international community andfiscal affairs experts when comparing the maturation andfiscal burdeno fdifferent pension systems. As is the case with calculating the financing gap, this is an indicator that i s 13 Some experts believe that the honzon for the future should be to perpetuity. l4 Ifrealgrowthrates for wages were to mcreaseat a faster pacethanthe GDPgrowthrate, or ifinterest rates were to fall below the GDP growthrate, the slze o fthe future IPD would mcrease even more significantly. A1-24  highly sensitive to the discount rate used in the cal~ulation'~Hence, with the proviso that macroeconomic assumptions are on balance sound and internally consistent, expressing the implicit pension debt as a percentage o f the current year's GDP should be a reliable indicator o f fiscal burden. Nevertheless, it should stillbe usedwith prudence when making compmsons with other pension systems. It i s especially important to keep inmindwhen performing cross country cornpansons that factors such as coverage, dependency rate and generosity o f benefits can all affect the level o f IPD. Another useful indicator for measunng the fiscal sustainability o f a pay-as-you-go arrangement i s the required contribution rate that will allow the systemto manage on a cash flow basis whereby expenditures can be met solely through contributions, without any additional financing to supplement the shortfalls. For the system to remain inbalance inthe case where there are cash deficits and no other financing is possible, the required contribution will have to be set at a rate such that fiscal balance can be maintained. This will be the rate at which contributions must be made (regardless o fwhether they are leviedagainst the employee, employer or the Government) to forestall any negative current balance. Infact, the required contribution rate can be one o f many ways to fund the financing gap described above. This required contribution rate is often expressed as a percentage o f the wage bill and is calculated by amortizing the financing gap over the net present value o f the contributory wage bill and then adding or subtracting the existing statutory contribution rate. The usefulness o f this indicator i s also subject to the same limitations described under the financing gap, in terms o f its applicability to systems that have reached maturation or are close to maturation. For systems that have not reached maturation and where coverage is still being expanded, focus should be directed towards determining the long term actumally balanced contribution rate. Summaryof Projection Results It would be appropnate to emphasize again that the purpose o f these fiscal sustainability benchmarks i s to provide a compansono fthe relative magnitude o fthe effects o fdifferent pension policy measures under vmous scenanos. Therefore, as stated earlier, results from projections o f this nature should not be quoted in absolute terms without proper reference to the underlyng assumptions and limitations. Table 10 presents a summary o f the three sustainability benchmarks under each of the six scenanos. Table 11presents a summaryo fthe current balances for selected years, expressed as % o f GDP in the current year, for the penod 2001-2075. 15 Ibid. A1-25  Sustainability Status Quo Scenarios Benchmarks Scenario 1 2 3 4 5 6 IPD (as % of 2001 GDP) 153% 126% 120% 85% 85% 92% FinancingGap (as % of 2001 GDP) 172% 42% -13% -140% -80% -203% Required contribution rate 48% 32% 29% 27% 27% 27% (as % of wage bill) Annex Table 11: Summaryof CurrentBalance(2001-2075). As % f GDP inthe current year ( Source: World Bank staff calculations fromPR 3ST output Table 12presents a summary o f two key indicators that show the funded status o f the system for selected years under Scenano 6, expressedinterms o f (i) assets to liabilities; and (ii) o f GDP % Funded status 2002 2005 2010 2020 2030 2040 2050 2060 2075 Funded ratio (assetslliabilities) 0.7% 3.0% 6.8% 13.6% 20.3% 25.0% 25.6% 23.5% 23.0% Assets as % of GDP 0.7% 2.6% 5.8% 10.8% 13.9% 14.9% 13.4% 10.7% 9.6% AI-26  TECHNICALANNEX2 EVALUATION THELIAONING OF PROVINCESOCIAL SECURITY REFORM PILOT Fiscal and Economic Aspects of Public Pensions Reform' An evaluation o f the fiscal and economic aspects o f a pension system requires looking at most elements o f the system and its implementation. This note sketches an approach to evaluate the Liaomng pilot program ingeneral andthe fiscal and economic aspects inparticular. To the extent possible, it includes preliminary conclusions. Section A provides a background to the problems faced by Chinese pension systems; section B proposes an approach to the overall evaluation o f the Pilot; sections C and D discuss in more detail the evaluation o f fiscal and economic aspects. A. Background The Chinese pension system, including in Liaoning, seems to have senous problems. It covers only 12 percent o f the working age population, and -under the reformed system -first pillar pensions replace at most one-third o f previous wages. Yet, total pension fund contribution rates o f 28-35 percent form the bulk o f total social secunty contribution rates o f around 40-45 percent-very highfor a country as poor as China, and leading to significant distortions on the labor market. Moreover, the system shows a financing gap o f 95 percent o f 2001 GDP What i s going on, and what do policymakers need to know and do? A combination o f several factors is behind this situation. Underlyingproblems weigh on the system The system itself has deficiencies, with several parameters set inappropnately The individual account system i s ``acturnally unfair" the "amortization" factor that i s supposed to align benefits with contributions i s inconsistent with remaining life expectancy at retirement and the discounts and premiums applied in case o f early and late retirement are too small, actuanally - unduly promoting early retirement. Moreover, statutory pension ages are too low Economy-wide demographic factors are straining the system, especially when looking ahead. The old age dependency ratio -the number o f people over 65 per person o f working age -is expected to nse from around 32 percent now to 80 percent in 2050 (for China as a whole). Pension system transition costs have risen 1Ths techmcal annex was preparedby Louis Kuijs, Semor Economst, EASPR. 2World Bank East Asia Region, HDNSP, "Chma-Pension Liabilities and Reform Options for OldAge Insurance", May 2005", mbase scenano. A2- 1  0 I n the absence of a cut in pension payments, the 1997 move to pre-funding created a financial gap in thefirst pillar. The diversion o f some o f the contribution from the first to the second pillar left a financial gap in the first pillar (see Table 1). Inthe face o f relatively favorable pensionbenefits o f "old men" and "middle men", the gap was ~ubstantial.~ Inother countnes that introduced a funded pillar, a rule-based explicit government transfer compensated the first pillar find for this loss. The 1997 reform in China did not provide for such a transfer. Lacking a stnct separation between the first and second pillar, this also led to the usage o f individual account funds to finance the gap. I n addition, thepension system suffers from economic transition and restructuring 0 The retrenchment and restructuring of SOEs has resulted in unfavorable system demographics. Pension finds have traditionally relied on SOE employment for the bulk o f contributions. However, since the early 1 9 9 0 due ~ ~ to the restructunng o f SOEs, employment in SOEs has grown much less than overall urban employment - 13.5 percent versus 37 percent in 1993-2002. At the same time, a restructunng-relatedearly retirement wave buoyed an increase in the number o f pension recipients -significantly faster than expected on the basis o f demographic patterns. These trends will affect pension fund finances for a long time: while the economy wide dependency ratio is expected to nse from 30 percent in 2001 to 83 percent in 2075, the system dependency ratio is expected to increase from 34% to 133 percent dunngthis penod. 0 The unfavorable pension system demographics in China result in part from usage of the pension system as a social safety net, in the absence of a proper one. Dunngrestructunng inChina, thebasic pensionsystem hasbeenusedto serve the purpose ofmanyother forms of social benefits - unemployment support, social subsidies to workers in dangerous occupations, severance payments, and compensation for employees in bankrupt companies andsocial subsidies to disabled workers. Finally, thefinancial gaps are not as large as it might seem. Although the gap o f the Chinese pension system i s a large share o f expenditure, given the small size o f the system - with pension spending around 3 percent o f GDP in Liaoning-the absolute size i s modest, with deficits emerging after 2030. Considenng the penod until2075, the financing gap i s estimated by the World Bank at 95 percent o f 2001 GDP On current projections, which foresee strong growth inthis penod, the gap would be 30-35 percent o f 2038 GDP While political issues may need to be overcome to find an acceptable contribution of central and provincial, given China's current strong fiscal position, the fiscal implications appear economically quite manageable. I t is important to identijj the contribution of thefactors outlined above to thefinancing gap in order to help policymakers make decisions. Inparticular, it is important to separate the impact o f underlyng (parametnc) problems o f the system and the economy-wide demographic challenges on the one hand - which are "here to stay" and need to be resolved by changing Oldmen(that retiredbefore 1997) receive a pension under the old system, whichhada relatively hghreplacement rate. Middle menreceive pension benefits fromthe new pillars 1and 2 but receive additional transitory benefits from the social pooling part depending o n the length o ftheir employment pnor to 1997 World Bank, M a y 2005. A2-2  parameters -from the impact o f pension system andeconomic transition on the other hand. The latter are transitory, legacies o f the past, and not related to the parameters o f the system or economy-wide demographics. Their impact on the pension system should be offset by government transfers, based on the same arguments as inthe case o f the banking system. If, as in China, the pension system is not compensated for these transition costs, a combination o f unreasonably high contribution rates and financial turmoil o f the system anses. Too high contribution rates lead to compliance problems (deferral and evasion o f contributions), unwillingness o f the pnvate sector to join the system, unfavorable labor market effects (shifts into informal employment and hinng), and delay in SOE restructunng. Financial chaos leads to undesirable policies and undermines the credibility o f system. InChina, thelackofagovernmentsubsidytodealwiththe legaciesofthepasthadledto senous problems. Inthe words o f the Research Group o f the DRC "...the reason as to why pension system reform has not made substantial progress i s that the inhented liabilities from the old system have not been handled separately from the development o f the new ~ y s t e m " ~In the absence o f government transfers, pension systems increased already high contribution rates for the social pooling fund. The key results were: (i) distortions on the labor market, leading senous to evasion o f social secunty contributions and the promotion o f informal; (ii) lack of willingness o f pnvate sector employers and employees to join the system-which further affects the finances o f the system and depnves workers from their nghts on pensions; and (iii) highand unequal tax burdens on enterpnses, aggravating the financial problems o f many SOEs and their ability to compete with the pnvate sector. The increases in contribution rates were not enough to offset the transition effects. As a result, deficits inthe first pillar lead to depletion o f the individual accounts -further undermining the credibility o fthe system andthe willingness to join. B. Evaluating the Liaoningpilot-including lessons for the overall system I n the evaluation of the Liaoning pilot, an assessment of whether and how the pilot dealt better with the legacies of thepast shouldplay an important role. Indeciding what approach to take, one approach i s to stick to a fairly narrow evaluation o f the implementation o f the pilot. At the other extreme, another approach i s to try to draw lessons for the "big questions" o f fully- funded, pnvately managed versus PAYG.Nonetheless, the latter question should not be allowed to dominate the more urgent reforms. Given the root o f the problems faced by China's pension system, emphasis on questions as to what system is appropnate for China nsks missing the point. Ifthe fhdamental issue ofhow to dealwith the legacies ofthepas isnottackled, no systemwill work. The diversion o f some o f the contribution from the first to the second pillar due to the 1997 reform has been maintained inthe Liaoning pilot. Nevertheless, the Liaoning pilot was a step in the nght direction to de-linking the reformed system from the past by means o f government "Chna's Pension Secmty System Reform", Research Group o fDRC, in"Restructumg Chma's Social Secmty System" A2-3  transfers, combined with (the intention of) a reduction in contribution rates and some modest parameter adjustment to improve the underlyng financial sustainability TheLiaoningpilot set out to reform several key aspects. To strengthen the credibility of the system,widen coverage andimprove compliance and the willingness to join, by establishing truly finded individual accounts, clearly separated from the social pooling account system, (some) reduction in the contribution rate (apparently especially for "new individuals", with a total rate o f 18 percent), and a commitment by the government to support the systemwith transfers. To reduce the labor market distortions, by reducing the contribution rates and making the individual accounts "portable" inthe province. 0 To improve the underlyingfinancial sustainability of the system, by some improvement in the actuanal fairness (bybasing pensionbenefits on the length o fwork and contribution). 0 To strengthen the robustnessto shocks through the introduction o f aprovincial buffer find. What can be noted on these dimensions, with respect to implementation of the reforms and their effect? On credibility, funds have indeed been accumulated inthe individual accounts and there appears to be a clearer separation o f these accounts from the social pooling pillar, although the rule that "when individual accounts run up, they will be filled with funds from the social pooling" runs counter to the separation and has moral hazard implicatiom6 The individual accounts contribution rate has indeedbeen reduced, although the employers' social pooling rate has not, as planned, been decreased to 20 percent across the Province (see Table 1). Instead, it ranges from 19 to 25.5 percent. The government has indeed supported the system with transfers. But it is not clear what the terms and rules are for these transfers. Ifthere isneedfor anentityto undenvnte the system, it shouldbethe government/MOF A24  Table 1. Liaoning Province: Contribution Rates Pensionsystem Total social security contribution Total First Individual Emplo Employee Total Employer Employee pillar accounts Pre reform (1994) Liaoning 24.9 24.9 0 ... ... China Reform 1997 31 20 11 23 8 Liaoning pilot 2001 SOEs Objective 28 20 8 20 8 41 30 11 Average actual 30 22 8 22 8 43 32 11 "New individuals" 18 10 8 10 8 Selected other countries II Germany 33 12 21 Sweden us 24 17 7 14 6 8 Sources: "Old Age Security",World Bank (1995),and OECD Taxing Wages Database(2004). I / For 2003,for a 2 income,2 childrenfamily,with principle earner'swage 100 percentof APW and spouse earning 67 percentof APW Indications that the credibility has improved are (i) "organic coverage rateshare o f covered the employees making contributions-increased from 76.2 percent at end-2002 to 77 4 percent in July 2004; and (ii)the urban household saving rate has declined. While correlation with the pension reform pilot is difficult to prove, the decline in the household saving rate was at odds with the national development. On reducing labor market distortions, it i s very difficult to find convincing ernpincal evidence on the impact o f modest changes in contribution rates on labor supply due to the pilot, in particular in the absence o f micro-data (which had been requested, but not received). However, the international evidence on the impact i s sufficiently strong and robust to form the basis o f an evaluation o fthe level o f first pillarrates (see Box 1). Under the Liaoning pilot, first pillar rates are high, relative to the replacement rates, in international companson. Total social secunty rates are also high (41 percent; with the actual average even higher at 43 percent), even compared to west European countnes (see Table 1). Moreover, as mentioned, the vanation across municipalities creates additional distortions. The underlying financial sustainability has improved little, as the parameters o f the system remained largely unchanged. Key parameters o f the system that still need to change are: the "amortization" factor that i s supposed to align benefits with contributions is inconsistent with remaining life expectancy at retirement and the discounts andpremiums applied incase o f early and late retirement are at 0.6 percent per year too small, actuanally-unduly promoting early retirement. Moreover, statutory pension ages are too low The robustness to shocks has improved somewhat, including by the introduction o f the equalization scheme. A2-5  C. EVALUATIONFISCALIMPLICATIONS As part of the Liaoning pilot, the pension fund's expenditures are financed partlyby government transfers-largely from the central government. Transfers were 1.1 o f Liaoning GDP in 2004. The size o f these fiscal transfers is based in part on calculations. The fact that including the transfers the pension fund has been balanced in recent years (Table 2) would suggest that the government i s de facto fundingwhatever deficit emerges.' Table 2. Finances Liaoning Pension Fund (in percent of GDP) 2000 2001 2002 2003 2004 Revenues 3.5 3.5 4.0 4.0 4.0 Contributions 2.4 2.4 2.6 2.6 2.8 Employers 1.9 1.8 1.8 1.8 1.9 Employees 0.5 0.6 0.8 0.9 Individualaccounts I / 1.o ... 0.3 0.6 0.7 0.7 Budgetarytransfers 1.o 1.4 1.3 1.I From central government 0.9 0.9 1.o 1.o 0.9 From provincialgovernment ... ... ... ... ... Earmarketfor individualaccounts ... ... ... ... ... From central government ... 0.1 0.3 0.3 0.2 From provincial government ... 0.0 0.0 0.0 0.0 Interestincome individualaccounts 0.1 0.0 0.0 0.0 0.0 Other 0.0 0.1 0.1 0.0 0.0 Unidentified 0.0 0.0 0.0 0.0 0.0 Expenditures 3.4 3.1 3.3 3.2 3.2 Pensions 3.3 3.0 3.2 3.2 3.1 Other paymentsto individuals21 0.1 0.1 0.0 0.1 0.1 Administration ... ... ... ... ... Repaymentof arrears ... ... ... ... ... Other 0.0 0.0 0.0 0.0 0.0 Balance 0.1 0.3 0.8 0.7 0.8 Excludingindividual accounts accumulation ... 0.0 0.1 0.0 0.0 (principleand interest) Source: Liaoning authorities and staff estimates. I / World Bank staff estimate. 2/ Benefit paymentsother than pensions from the pensionfund. 7The Liaonmg Pilot government proposal hasno explicit reference to central government finance at all. It does however note as an objective "the matchof social secunty level with provincial situation and the affordability o f vanous parties." A2-6  Box 1. Pensionsystem contributions-incentives effects High social secunty contributions affect decisions to seek formal employment. Specifically, an increase inthe tax on labor increasesunemployment as long as there i s resistance among workers to a drop in post-tax wages. Ernpincal estimates of the impact o f the tax wedge on employment in OECD countnes are statistically significant and robust. Based on a survey of the literature, Nickel1 (2003) estimates that a 10 percentage point increase in the total tax on labor would decrease (formal) employment by 2-3 percent. While such results cannot be applied to China without qualification, the consensus inthe ernpincal literature leaves little doubt about the impact of labor taxes on (formal) employment. Guidance on the nght level of tax and social secunty contribution rates i s more difficult to obtain from suchcross country evidence, inpart because the willingness to pay taxes depends on the stage o f development and cultural factors, as well as on people's perception of the benefits of contributing to the system. Indeed, labor market distortions o f pension systems are minimized where contribution rates bear a full actuanal relationship to benefits, and are seen to do so (Barr, 2000). On the latter, as shown inthe figure below, there is generally a relationshipbetweenpension system contribution rates and replacement rates, with some countnes opting for relatively high replacement rates, which then require highcontribution rates, and other countnes opting for lower rates. The combination o f contribution and replacement rates of the first pillar in Liaoning i s relatively unfavorable, explaining inpartthe reluctance tojoin voluntarily Figure.IntemafionalCompansonofPublic PensionSystemContnbutionandReplacementRates 70 Y m:PublicPensionSystemReplacementrate (as share ofgross wage) 0 60 50 0 0 . 0 0 0 0 0 . 40 0 0 0 . 0 0 0 30 0 Llaonmg(iirstpillar, maxnmrm) 0 0 Llaonmg( h t pillar; afier 15 year) 20 0 10 X am: PublicPensionSystemContnbutionRate 0 0 5 10 15 20 25 30 35 40 45 50 Sources:PalaciosandPallares-Miralles (2000). OUSDcountnes,and countnes m LamAmnca and CentralandEastEurope. A2-7  Looking ahead, budget transfers should be based on a clearly identified rule that indicates the government commitment to finding. Such a rule would also limit the government's commitment, and thereby its liability Moreover, bolder parametnc reforms are needed to deal with the demographic issues. Evaluation of fiscal aspects The proposed approach consists o f the following. Based on the need to separate the inherent financial problems o f the pension system from the impact o f economic and system transition, estimate a "reasonable subsidy" that would offset the transition costs. To the extent that this fiscal transfer is economically and politically sustainable, and combined with changes in parameters on which there i s consensus, what would required contribution rates be7 The following discusses this approach in more detail. The actual work relies heavily on actuanal calculations and estimations. 1. Presentfiscal and economic implications of current Liaoning system. Budgetary transfers equivalent to between 1 and 1.4 percent o f Liaoning GDP broadly kept the overall budget balanced (Table 2). Data on the finances o f the Liaoning pension system provided by the authonties suggests that, including the budgetary transfers and counting the accumulation o f funds inthe individual accounts as revenues, the pension systemhad a surplus o f 0.7-0.8 percent o f GDP dunng 2002-04. Excluding these accumulating funds, the pension system was inbalance dunng 2001-04. As a ratio o f GDP, budgetary transfers peaked in 2002 at 1.4 percent o f GDP (RMB7.5 billion). In2004, transfers were 1.1percent o fGDP(RMB7.8 billion). Looking forward, present the fiscal implications o f the pension hnd projections on the basis of current policies for the penod 2001-2075 Two optional exercises would add important information. First, the same analysis for Liaoning under current national policies, to assess the financial impact o f the pilot, vis-a-vis the "status quo" Second, in order to inform the country- wide polic1 discussions, a preliminary estimate for China as a whole under Liaoning parameters. 2. Identijj economic transition costs. Estimate the impact on the pension system finances o fthe decline inSOE employment andusage o f the pension system for other social objectives. 3. Estimatepension system transition costs. Estimate the cost o fproviding pensions to the "old men" and"middle men", over and above the costs impliedunderthe policy for the "new men" 4. Decide on size of fiscal transfer. Policymakers should consider financing the estimated transition costs (from steps 2 and 3). But fiscal transfers have to be fiscally and politically sustainable in the medium and long run. The "economically justified" required transfers, as determined above, need to be assessed against (i) the size o f current transfers; (ii) fiscalthe capacity; and (iii) political considerations, including interprovincial issues. Key considerations are the following. 8 Ths has inpnnciple been done already. Care shouldjust be taken to do compansons using similar assumptions. 9 While the diversion of contributions from the first to the secondpillar would appear to add to the transition costs, t h s should not be the case ina forward loolung analysis. A2-8  Does the government have the fiscal capacity to support the transfer? Preliminary indications are that with sufficient central government involvement the required transfer would be feasible relative to the fiscal capacity o f government, especially considenng that the Liaoning pension system is among the most financially challenged. However, inthe case o f Liaoning, inthe absence o f central government support, the level o f government transfers would require 2.6 percent of total Liaoning Province GDP in2004, which would be difficult to shoulder. Would central government transfers that vary betweenprovinces be sustainablepolitically? Economically justified levels o f transfers as determined above would vary between provinces. They may invite political and inter-regional issues. Would long-term financing by central government that is much higher in relatively n c h Liaoning than in poor Western provinces be acceptable politically? What are the implications for the intergovernmental transfer system? If these considerations suggest constraints on central government transfers, the provinces need to be given leeway to fund the costs in other ways. In this connection, an increase in general income taxation would be less distortive thanhighpension contribution rates. Apart from interregional issues, decisions on how much to transfer are also questions of distribution over generations. These are among the reasons why policymakers may decide not to transfer the full economically justified transfer. That could be a perfectly reasonable outcome, as long as the decision is made explicitly andafter proper evaluation o f all considerations. 5. Define obvious parametric reforms. There i s virtually universal agreement on a few parametnc reforms. (1) Change the "amortization" factor that i s supposed to align benefits with contributions, since the current rate i s inconsistent with remaining life expectancy at retirement. (2) Increase the discounts and premiums applied in case o f early and late retirement; they are at 0.6 percent per year too small, actuanally-unduly promoting early retirement. 6. Economic implications of underlying Liaoning system with adjusted parameters and identified government transfers. With the actual subsidy policymakers are willing to finance as determined above, and the consensus parameter changes, what would contribution rates be, and what would the economic and labor market consequences o f such contribution? Inlight of the considerations highlighted inBox 1, it would be advisable to reduce rates from their current level. Ifthe contribution rates resulting from the above exercise are still deemed to have unacceptable labor market and economic consequences, policymakers should consider: (i) reducing benefits for those with transitionary arrangements (old and middle men); (ii)changing parameters, including increasing statutory retirement age; and (iii) financing by an increase in income taxes. On the latter, the base for income taxes i s significantly wider than that for pension contributions, which would allow a much lower increase intax rate. A2-9  D.(OTHER)ECONOMIC IMPLICATIONS Economic implications include macroeconomic viability, issues o f nsks and uncertainties, and consequences o f imperfect consumer information (Barr, 2000).lo additional implication stems An from special conditions inthe financial markets due to government intervention. Macroeconomic viability Key aspects o f macroeconomic viability are effects on the labor market and savings. Labor market distortions can be minimized if contributions and benefits bear a full actuanal relationship, and are seen to do so (Box 1). In the case o f China, were social secunty contributions are already too high, informal employment i s high, and enforcement o f compliance is weak, ensunng such an actuanal relationship i s particularly important, because correct incentives can to some extent substitute for enforcement inincreasing compliance. Saving i s generally assumed to increase as a result o f funding, although this is subject to some qualifications. In particular, a reduction in voluntary saving in response to the accumulation o f individual accounts would offset part o f the increase.l1If total saving is increased by pension system funding, it would allow for lower interest rates and, in pnnciple, higher investment and economic activity.12 The empincal evidence on the impact on saving across the world is less solid than previously supposed (Gale (1998), Holzman (1997), and Mackenzie, Gerson, and Cuevas (1997). The latter conclude that " ..it is not possible to generalize across countnes about the impact o f the public pension system on saving." Under certain circumstances, increasing total saving would support economic growth. In countnes with low savingrates (Latin Amencan ones inparticular) increasing saving has been an important additional motivation behindthe introduction o f a fully funded pension pillar. In the current context in China, saving and investment are very high and senior policymakers actually want to put more emphasis on consumption. Inother words, increasing saving may currently not be a sufficiently important consideration. This may be different in a decade, though, when demographic challenges are likely to put downwardpressure on household saving. What happened to saving in Liaoning? The overall gross domestic saving rate increased a bit in 2002 and a lot in 2003 (Figure 1). While it is tempting to associate this with the s h f t to real funded individual accounts, it is not possible to disentangle the impact o f the reform from other factors impacting on total saving. Moreover, a similar increase in the overall saving rate took place inthe rest o f China, inthe absence o f pensionreform. 10The latter aspectis not discussedhere; it is discussed mthe sectiononfinancial market issues. l1Inaddition, to the extent that saving mcreases, t h s is only so mthebuiltupphase.Insteady state, savmg by workers equals dissavmg bypensioners. And, it is important to know what happens to pensions o f "old men" and "mddle men" and thelr financmg. Inthe unlikely case that thesepensions are reduced, savmg will increase. If,as has beenmore common, they are financedby issuance o f government debt, t h s i s also an offset to any mcrease in pnvate savings. 12A concrete channel for the impact on mterest rates would be the buymg of government bonds bypension hnds as individual accounts accumulate. Conducting such transactions via the market would be preferable to bilateral arrangements, although the entry o fpension systems mthe government bond market could be a reasons for adjusting the composition o finstruments, for mstance interms o fmaturity. A2-10  The evidence on the urban household saving rate i s more striking. After peaking in 2001, the urbanhousehold saving rate declined significantly in 2002 and 2003 (Figure 1). This contrasts with the development in the rest o f China, where the urban saving rate remained unchanged in these years. While not providing robust evidence, the divergent trend inLiaoning would suggest that the credibility o f the pension system has been strengthened by the reform. Risksanduncertainties Pension systems face many types o f nsks and uncertainties, including from macroeconomic shocks, demographic shocks, and political nsks. These nsks are transferred differently in different systems. For instance, with individual accounts, persons are not insured against them. Inadditionto these generalnsks, a funded system also faces management nsk, investmentnsks, and annuities market nsks. Inthe Liaoning pilot, where the role o f the first pillar is modest - mainly one o f avoiding extreme poverty -a significant part o f the nsks associated is left inthe hands o fpeople. How nsks are transferred is also dependent on the level o f pooling. Indeed, conceptually there i s a strong case for pooling at a higher level. With pooling at a low level there are particularly large problems for migrants. Ifthey work formally, they contribute to the pension fund. However, it i s highlyunlikely that they will be able to meet the 15 years minimumcontribution requirement in order to be eligible for the social pooling component o fthe pensionprogram. Inorder to introduce some element ofnsk shanng across municipalities, Liaoning in2001 set up a Provincial Adjustment Fund. Each municipality i s required to contribute 5 percent o f its basic social secunty polling fund collected. Constraintson the rateof returndue to governmentpolicies The rate o f return on capital is high in China's fast growing economy However, the rate of return on official financial instruments i s significantly lower, due to vmous forms o f government intervention and policies (a "repressed financial systern").l3 As a result, the rate o f return on bonds and deposits i s significantly lower than wage growth. This causes a problem for funded accounts. Basically, the contributors and the system are short-changed by setting up a funded system in these circumstances-unless policymakers are willing to consider setting an administrative interest rate higherthan official interest rates. 13The mterest rate on the informal, or "pnvate" market is, however, at around 14 percent muchhgher than the official market rates. A2-11  Figure 1. SavmgRates, 1980-2003 Urbanhousehold savmg rate (inpercent ofhouseholddisposable mcom) 25 25 20 20 15 15 10 10 5 5 0 -5 0 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 Gross Domestic Savmg (inpercent o fGDP) -- 50 45 \ * - - * 40 V - - * - 4 - * * 4 35 - * I - * * * 30 Source: NBS,Liaonmg Bureau ofStatlstics, and staff estmtes. A2-12  TECHNICALANNEX3 EVALUATION THE LIAONING OF PROVINCESOCIAL SECURITY REFORMPILOT I s a 20 Percent Pension Insurance Contribution Affordable to Industrial Firms in Liaoning Province? I.Background Dunng 2001-2003, the State Council implemented a social secunty reform expenment in Liaoning province. As part o f an evaluation o f the Liaoning expenment, this paper addresses one issue, namely, the affordability o f the currently contemplated pension contribution rate to industnal firms inLiaoning province. Pension contribution o f firms, levied as a percentage o f payroll2, i s part o f their labor cost. It has been a senous concern to policy makers that a too high pension contribution rate may depnve Chinese firms o f adequate profit margin and weaken their competitiveness in the international market. This is particularly relevant to firms inLiaoning, China's rust-belt region, as they have been undergoing painful restructunng since mid-1990s, with senous unemployment pressure and social dislocation. An assessment o f the situation in Liaoning has important implication to decisions regarding the direction and key parameters o f pensionreform, not only inthe northeast provinces but also nation-wide. 11.Scope The effect o f a pension scheme on competitiveness o f an industnal firm is multi-folded. The direct effect is one that is easier to observe and measure: a firm's profit margindecreases with a pension payroll tax when total revenue and other parts o f the cost are given. With sufficiently highrate of payroll tax for pension, a firm may find itself depnved of any significant profit, or alternatively, it can keep a minimumprofit marginonly when its pnces of outputs (inputs) stay from falling (nsing). Ineither case, its competitiveness i s weakened. There i s also possible indirect effect. In a reasonably developed labor market, pension scheme should have its impact on wage rate, so that total labor cost o f a firm does not necessarily increase with pension contribution rate ina proportionate manner inthe long runwhen wage rate i s sufficiently flexible. That is, when the firm's contribution rate i s raised by P, a fraction o f P, say x, may be passed on to employees in forms o f lower wage rate or slower increase o f wage rate, so that total labor cost o f the firm increases only byP(I-x). Similarly, when the contribution rate o f employees i s raised by Q, a fraction o f Q, say y, may be passed on to employers informs 1 Thls techtllcal annex was preparedby ChunlinZhang, Semor Enterpnse Restructunng Specialist, EASFP 2 For the purpose o f thls paper, the concept of "contribution rate" and "payroll tax rate" are usedmterchangeably. While recogmzmg that mChina and some other countnes pension contributions are not considered formally as a tax and do not have the same legal basis, they are similar to the payroll tax levied mmany other countnes interms o f its ultimate purpose m financmg the social insurance programs. A3-1  o f higher wage rate or faster increase o f wage rate, so that total labor cost increases by ye. Therefore, the net effect o f P and Q on total labor cost o f the firm i s measured by P(I-x)+yQ The values o f x and y are determined by the bargaining power o f each employee vis-a-vis his employer, within the limit set by the flexibility o f labor market. To the extent that P(1-x)+yQ i s significantly larger (smaller) than P, the above-mentioned direct effect may under- (over-) estimate the real effect. There is evidence suggesting the existence o f this indirect effect inthe labor market inLiaoning, inparticular, for pnvate firms. Itwas reportedbypnvate employers andconfirmedby some SOE laid-off employees that pnvate employers tend to pay pension contribution for some key employees, such as senior managers, key engineers, technicians, in order to attract the needed talents. On the other hand, they are reluctant to do the same for low skill labor, such as workers laid off from SOEs, as they are over supplied. Workers o f this kind are also afraid o f bargaining with employers for pension insurance. Nonetheless, systematic data are lacking to enable an assessment o f the overall situation. In addition, given the ngidity o f the labor market and relatively short penod o f implementation o f pension reform in Liaoning, the indirect effect i s likely to be insignificant. This paper will therefore concentrate on the direct effect. There can be at least two ways to assess the direct effect o f pension contribution on competitiveness o f firms. One i s an international companson o f payroll tax rates, the other i s an analysis o fprofitability o f firms with a given payroll tax rate. Table 1 shows the level o f pension tax and social insurance tax o f some OECD countnes in the 1990s. It suggests that the target social insurance contribution rates to be born by employers in China, 20 percent for pension insurance, 6 percent for medical insurance, and 2 percent for unemployment insurance, are moderate. However, international companson has its limitation. Since "other things" are not equal, firms with the same pension payroll tax rate may differ greatly in terms o f profit margin. And it is the profit margino f the majonty o f firms o f a country that concerns policy makers and designers o f pension reform. Therefore, this paper will concentrate on analyzing the profit margin o f Liaoning industnal firms with a given pensiontax rate. A3-2  Table 1: Social InsuranceTaxes of High-IncomeOECD countries, mid-1990s (As Percentageof TotalLabor Cost) Country Pension Tax All Social Insurance Taxes Austria 17.8 35.2 Belgium 13.0 30.9 Canada 4.9 13.9 Finland 17.8 22.1 France 12.0 38.0 Germany 17.0 34.0 Greece 16.1 27.9 Iceland 7 1 17.8 Ireland 13.0 Italy 20.1 38.5 Japan 14.1 24.9 Luxembourg 13.9 25.0 Netherlands 28.9 50.5 Norway 19.3 Portugal 27 4 29.8 Spain 21 4 29.0 Sweden 15.9 20.0 Switzerland 8.7 17 1 United Kingdom 13.0 UnitedStates 10.4 18.5 Source: Robert Palacios andMontsen-at Pallares-Miralles, 2000, International Patterns of Pension Provision. Social Protection Discussion Paper Senes, No. 0009 The World Bank. 111.DataandMethodology Under the existing Chinese accounting system profitability data are generated and reported according the following formula3 1) Sales profit(SP) = Sales revenue o fproducts (SR) - Sales cost o fproducts (SC), including social insurance contributions ofproductionworkers - Sales taxes and surcharges (ST) 2) Operationalprofit(OP) = Sales profit +- Profit from other business operations (OBP) Selling expenses (SE), including social insurance contributions o f sales workers - Administrative expenses (AE), including social insurance contributions o fmanagenal workers MhstryofFinance, EnterpriseAccounting System. Effective since January 1,2001. ForChnese text, see http:llwww.mof.gov.cn/news/20050304~1900~5483 .htm A3-3  - Financial expenses (FE) 3) Totalprofit(TP) = +Operational revenues (IR) profit ++- Non-operational Investment Subsidies (SUBD) Non-operational revenues (NOR) expenses (NOE) 4) Netprofit(NP) = Total profit - Corporate tax (CT) A worth-noting difficulty for this study is that total social insurance contribution is a sum o f three elements, contribution for production workers, sales workers and managenal workers, which are mixed with other costs in three cost items, namely, sales cost, selling expenses, and administrative expenses. This study is based on a dataset drawn from an official database that covers all state owned and controlled industnal firms, as well as non-state industnal firms with annual sales revenue over 5 million yuan in Liaoning province. The dataset contains firm-level data o f 2001 and 2003 for indicators as shown inTable 2. The total numbers o f firms inthe dataset are 5847 for 2001, 6842 for 2003. There are some firms whose data are not adequate for analysis and have been omitted, either because all indicators are zero or one o fthe followings i s zero: employment, sales revenue, and payroll. These data are described in Table 3. The effective sample size o f this study is therefore 5504 for 2001 and 6655 for 2003. A number o f issues concerning data andmethodology deserve some hrther discussion. First, two key indicators, namely, payroll, labor and job-waiting insurance contributions, may not be as adequate as the analysis requires. The reported payrollmay not be exactly the same as the "social average payroll (shehuz pznglun gongzi)" used in calculating social insurance contributions in practice, and the exact meaning o f "labor and job-waiting insurance contribution" i s not clear. Without better alternative, this study uses the payroll data as a proxy for the social average payroll, and data for labor andjob-waiting insurance contributions as a proxy for unemployment insurance contributions. Second, how to measure pension and social insurance contributions for the purpose o f this analysis? An examination o f the dataset leads to the conclusion that one cannot simply look at the statutory or actual social insurance contributions for this analysis. As shown in Table 4, 50- 66 percent o f firms simply did not pay any social insurance contribution. Those who paid have a highly diverse range o f contribution rates, with no sign o f concentration around the legally obligated rates. A significant number of firms paid up to 100 percent o f payroll, and a small number o f firms paid even higher. Some investigation i s required to get a clear interpretation o f these data, but one possible reason i s that these firms paid arrears inthe reporting year. A3-4  Table 2 Indicators of the Dataset Administrative expenses Contributions ofpension Sumofcontributions for production workers, sales workers and and medical insurance managenal workers, picked out from three cost items Contribution of labor and Exact definition i s not clear, but contribution to unemployment job-waiting insurance insurance i s presumably included. Interest expenses ODerational Drofit Year I Numberoffirms I Employment as Sales Revenue as Percentageof Total Percentageof Total 2001 343 7 42 4.12 I 2003 I 187 I 8.87 I 3.56 I Table 4 Distribution of Effective Social Insurance Contribution Rates Among Firms (percent of payroll, 2001 and 2003) Indicators Range of actual 2001 2003 rates ofpayrolltax Numberof As % oftotal Number of As % of (T) firms f i r m s total Pension and T<O 3 0.1 0 0 medical T=O 2921 53.1 3291 49.5 insurance contributions Labor and job-waiting insurance contributions A3-5  In view of this difficulty, this analysis adopts an alternative approach, that is, calculate how much a firm would have paid in a situation where the rate o f total social insurance contributions was 28 percent o fpayroll, including 20 percent for pension, 6 percent for medical insurance, and 2 percent for unemployment insurance. This amounts to asking a question like this: What would have happened to profit margin o f a firm if it had paid a payroll tax o f 28 percent? Or equivalently, what would have happened if it had paid a pension payroll tax o f 20 percent, in addition to 6 percent o f medical insurance contribution and 2 percent o f unemployment insurance contribution? To make this approach work, it is necessary to deduct actual contributions from the total cost of each firm, or add them to total revenue, in calculating profit margin. Thus, a firm's "total revenue (TR)" is defined as the sum o f (i) sales revenue; (ii)actual social contributions; (iii) profit from other businesses; (iv) the difference resulted from a subtraction o f operational profit from total profit, reflecting profit generated from investment return, subsidies and non- operational revenues. For cost data, the dataset does not provide a complete picture, as some minor cost items are not covered, most notably, sales tax and surcharges, selling expenses, and loss caused by exchange rate movement (the difference between financial expenses and interest expenses). As a result, subtracting the sum o f three costs (sales cost, administrative expenses, interest expenses) from the sum o f two revenues (sales revenue, profit from other businesses) yelds a number that is significantly larger than operational profit. The discrepancy i s as large as 4.5 percent o f sales revenue in 2003 and 4.6 percent in 2001. It i s assumed in this analysis that the data for operational profit are reliable, andthis discrepancy is defined as "other costs" The sample firms are divided into 8 groups according to definitions shown inTable 5 Group 1-5 intend to identify the depth o f financial distress o f a firm, while group 6-8 are designed to measure its profitability Group 6 i s a marginal one in that firms in this group may be either slightly profitable or in slight loss. The 5 percent benchmark that divides Group 6 and 7 is arbitrarilychosen for simplicity, with a reference to average ratios o ftotal profit to sales revenue which were 3.2 percent in 2001 and 3.8 percent in 2003. The 10 percent benchmark dividing Group 7 and 8 i s also an arbitraryvalue. The 8 groups are firther put into 4 categones for ease o f analysis: --- Category Iincludes firms inGroup 1and 2, which are most likely commercially nonviable. Category I1puts together firms inGroup 3,4 and 5, which are probably infinancial distress. Category I11contains the marginal Group 6, and - Category N,including Groups 7 and 8, are those largelyprofitable. A3-6  Table 5 Definitionsof the 8 Groups of Enterprises IGroup IFirmswhose totalrevenue (TR) IV.PrimaryResults Table 6 and 7 presents pnmaryresults of grouping andcategonzation. Table 6 Eight GroupsandFour Categoriesof IndustrialFirmsinLiaoning,2001 (Category or Group Subtotalas % of TotalSample Firms) Groupand Number Employ- Total Total Sales Payroll Actual pensionand Category offirms ment assets liabilities revenue medical contribubons Category I 9.81 742 6.12 7.74 4.12 6.07 2.40 Category II 42.15 36.88 27.89 31.25 22.10 31.28 26.47 Category 111 26.49 36.77 42.91 42.11 47.66 41.01 47.01 Category IV 21.55 18.93 23.08 18.90 26.12 21.64 24.12 A3-7  Figure 1. irts of i ~ d w $ t ~ i ~ l its %t of Totals Figure 2, h-tGroupsand Four Categories of industrid Firms in Liaoning, 2003 AfI vrrrnr. 2001  Data analysis reveals the following main observations: 0 There were a significant number of firm in Liaoning that are commercially nonviable (Category I), accounting for 7 percent of total industnal employment and 8 percent of total liabilities o f industnal firms in2003. Little change took place dunng 2001 and 2003 interms of these two percentages. Financially distressedfirms (Category II)accountedfor 1/3 of total employment, % of total liabilities and sales revenue, and 1/5 of actual pension and medical insurance contributions in 2003, slightly lower than in 2001. The large shares of this Category highlight the formidable challenge o f enterpnse restructunng inLiaoningprovince. Putting together Category Iand 11, the implication i s that nearly 1/2 o f Liaoning industnal firms, accounting for 38 percent o f total industnal labor force and 31 percent o f total liabilities o f industnal firms, are in deep financial distress and many of them are probably nonviable. Without successful restructunng, they cannot afford any payroll tax. Firms at the margin (Category III)accountfor the largestportion of the social insurance contributions, but areprobably most sensitive to a change ofpayroll tax. In2003, firms in this Category produced % of the total sales revenue and made % of actual social insurance contributions. They are also employers o f more than 1/3 o f the industnal labor force. From 2001 to 2003, the share o f this Category increased slightly In 2003, 1942 firms o f this Category reported a total profit equivalent to 2.2 percent o f total sales revenue, which would have dropped down to 2.0 percent had they all paid a 28 payroll tax. However, this average rate masks the difficulties of most firms inthis Category There were 927 firms who did not pay pension and medical insurance at all in 2003, 661 firms paid less than 26 percent of payroll. Had a 26 percent payroll tax for pension and medical insurance been actually paid, 82 percent of firms inthis Category would have been negatively affected, although not have turned to red. Obviously, Liaoning cannot afford a worsening o f financial situation of firms inthis Category. Forprofit makingfirms (Category Iv), apayroll tax of 28percent isprobably affordable to them. However, they accountfor only 27percent of total employment and actual pension and medical contributions, and 32 percent of sales revenue. The pnmary challenge for enterpnse restructunng inLiaoning inthe coming years i s to move firms from Category I11to Category IVYwhich has the potential to increase the number for firms who can afford a 28 percent payroll tax. Before this i s achieved successfully, it seems difficult for this Category alone to sustain Liaoning industnal sector and the pension system. In this regard, it is encouraging to note that the shares o f this Category in total enjoyed a significant increase from 2001 to 2003. V. FurtherAnalysis It is usehl to further analyze the categonzation results with 2003 data, considenng factors of ownership, size and sector. A3-9  Ownership and size The 2003 dataset contains 1164 SOEs and 5491 non-SOEs. Table 8 describes their shares inthe sample total. SOEs are on average 4 times larger than non-SOEs in terms o f employment. Employng % o f the industnal labor force in Liaoning, SOEs control 2/3 o f industnal assets, produce 60 percent o f industnal output and generate '/z o f industnalprofit. Table 9 divides SOEs into 4 categones and 8 groups. It shows that about 1/3 o f employees o f industnal SOEs are still found innonviable and distressed firms, which account for 28 percent o f total liabilities o f industnal SOEs. These firms tend to be much smaller than those in Category I11and IV,accounting for 60 percent o f total innumber. However, their low share (14 percent together) insales revenue suggeststhat manyo fthem are not active anymore. SOEs in Category I11are about the same size as those in Category IV They employ over 40 percent o f the SOE labor force, control 46 percent o f SOE assets, and produce nearly 60 percent o f output. In addition, nearly 60 percent o f SOE pension and medical contribution comes from them. Again, the hture o f SOEs in this category determines the overall picture o f SOEs in Liaoning. In companson, the best performing SOEs in Category IV represent about 1/3 o f the SOE sector interms o f employment, assets and output. Turningto non-SOEs as showninTable 10, it is strikingto see that the deptho ffinancial distress o f the non-SOE sector i s rather greater than the SOE sector. Employees o f Category Iand Category I1 firms account for 44 percent o f non-SOE total. In terms o f liabilities and sales revenue, these firms represent shares as high as 38 and 24 percent, all significantly higher than their SOE counterparts. More investigationi s required to explain why financially distressed non- state firms have such a high share in the total o f non-SOE output, which should not have been the case if they stopped or reduced their production, as is probably the case for financially distressed SOEs. One explanation may lie in their size. Firms in these two categones are about the same size as those in other two better performing categones, a clearly different feature from SOEs. Compared with SOEs, better performing non-SOEs have a significantly lower concentration in Category 111.They distribute inthe two categones ina half-half pattern. This suggests that better performing non-SOEs as a whole may have a stronger ability to accommodate a 28 percent o f payroll tax than their SOE counterparts. A3-10  Table 9. Eight Groups and Four C rimof IndustrialSQEs in Liaoaing,2 ory OF Group su as % of Total of sample SQEs) Figure3. Four Categories of Industri~lSOEs in [ , ~ a o f l j 2003 ~ ~ , Cateriratian of lndus~ri~l SOEs in tiaoning, 2003 A.34 1  cnrp oyer per I I I i 1--7 I I I A3-12  Dividing firms in each sector into the 4 categones yelds a result as shown in Table 12. The sector o f oil and natural gas extraction i s a special case, despite a 4 percent share intotal output, the sector has only 4 firms, and dominated by one large and well-performing firm, which is in Category IV The rest o f 9 sectors display the following patterns. A. The Ferro metallurgy and electncity sectors have the best financial situation, where Category IV firms enjoy large shares o f over 40 percent and the shares o f Category Iand I1firms are insignificant. B. Three sectors are dominatedby CategoryI11firms, while not facing very heavyburdenof restructunng, as shares o f firms in Category Iand I1 are relatively small. They are transportation equipment, coal mining, and oil processing. These three sectors produce about a quarter o f total industnal output o f the province. C. Other three sectors, namely, non-metal mining products, chemical and garments, face heavy burden o f restructunng, reflected by high shares o f Category Iand 11, particularly inemployment. However, interms ofoutput, their CategoryIVfirms represent very large shares, andtogether with Category I11firms, dominate the sector. D The textile sector seems to have the most difficult situation. While Category I11and IV firms account for over half of the sector's output, over 70 percent o f its employment is trapped inCategory IandI1firms. A3-13  Table 12. Four Categories and Eight Groups of Industrial Firms in Ten Selected Sectors (Category or Group Subtotal as YOof Sector Total) A3-14  Figure 5. Four ~ ~ ~of Ie strial Firms~in Oil and~ ~ o i ~ Sector 7 Sector I 7  Sector 32 Sector 44  Sector 37 , . ........ nIV nlll II ,I. , Soclot OS  e 1I,Four Categoriesof ~ ~ d # s ~ Firms i~Oil Process r i a l Sactar 25 I c,IV Dlll .II . I. Categories of Industr Firms in Xon-metal ing Products Sector Sactor 31  Sector 28 Figure I&FourC a t ~ ~ of Ind~ ~ s o ~ ms in Garments Sector Sector 1s  VI. PreliminaryConclusions Liaoning industnal firms as they were in 2003 would have found great difficulty had a 28 percent social insurance tax, (or equivalently, a 20 percent payroll tax for pensions on the top o f 6 percent payroll tax for medical insurance and 2 percent for unemployment insurance) been stnctly imposed andpaid. Apart from 1/4o f well-performing firms which account for 1/3 o f total output, all others would have found it difficult to cope with such a level o f payroll tax for social insurance programs. Nearly 1/2 o f the firms, accounting for 38 percent o f total industnal labor force and 31 percent o f total liabilities o f industnal firms, were in deep financial distress and many o f them were probably nonviable in 2003, who could not afford any social insurance tax. Inbetweenlies the mainstay of Liaoning industry, about 1/3 innumber and accounting for ?4 of the total output, ?4o f actual social insurance contributions, and employng 1/3 o f the industnal labor force. These firms would have either made slight loss or a moderate profit margin less than 5 percent o f sales revenue. This would have made them particularly vulnerable to significant market fluctuation. Overall, this category o f firms i s the representative case for Liaoning industnal firms. This has been confirmed by further analysis from perspectives o f ownership and size, and 10 selected key sectors. The future o f Liaoning industnal sector depends on to what extent firms inthis category will achieve improvement intheir performance, and how successful financially distressed firms can be restructured. Before these are achieved, a 28 percent social insurance tax would appear difficult for most Liaoning industnalfirms to accommodate. A3-20  TECHNICALANNEX4 EVALUATION THELIAONING OF PROVINCE SOCIAL SECURITYREFORMPILOT PensionAdministration' I.Background According to State Council Document 42, the general objective o f the Liaoning pilot was: ..to establish a ... standardized social secunty system and socialized public administration and " service." The main tasks were to include: " .to socialize the administration and service delivery o f social secunty; to strengthen financing and management o f the social secunty fund [and] to speedup the process o f social secunty legislation." The document went on to list a number o f specific changes that would be introducedinorder to achieve these broad objectives. These included: Covenng free lance professionals, the self-employed, and contracted rural immigrant workers; Shifting to the local tax authonties the responsibility for collecting all social insurance contributions; Improving financial management, including o Institutingthe two-track systemto allbranches o f social insurance, o Separating the management o f the individual account funds from the management o f the social pooling funds, and assigning the responsibility for the former to provincial- level officials, o Establishinga "robust systemo finternal controls" at social insurance agencies, o Regularlypublishinginformation onrevenue and expenditure, and o Improving statistics onwages and contributionbases; Improving program administration, including o Maintaining pension, unemployment and health insurance records on both in-service workers and retirees at local social insurance offices, and o Formulating local regulations on contribution collection, subsidies and benefit delivery; Improving public understanding through strengthened education programs for workers, enterpnses, andthe general public; and 1Ths techcal annex was preparedby Lawrence Thompson, Consultant. A4-1  0 Creating a social secunty information network to link labor and social secunty, civil affairs, finance, statistics, and taxation offices and to include information on contributions, worker records, andbenefit payments. This section reviews the administrative reforms instituted as a part o f the pilot project and suggests some additional steps to improve further the efficiency and effectiveness o f the social insurance system. It begins by noting the significant progress that has already been made under the pilot program. When fully implemented, the reforms outlined in Document 42 will substantially strengthen both the financing and the management o f these programs. The section then outlines a number o f additional steps that international expenence suggests would further strengthen program management. Some can be implemented fairly easily in the current administrative structure while others may require testing ina subsequent reform pilot. 11.Accomplishments The changes introduced in the pilot project have improved both the quality o f the services delivered to program participants and the financial integnty o f the province's social insurance programs. A number o f the reforms outlined in Document 42 have already been implemented fully, and substantial progress has been made in implementing others. The onginal reform agenda was ambitious, and it is not surpnsing that it could not be achieved completely in the relatively short time frame o f the pilot. As this agenda i s completed, further improvements in service quality and financial integnty canbe expected. 1. Separationof individualaccountandsocialpoolingfunds A pnmary objective o f the pilot was to prevent the diversion o f contributions intended for individual accounts to finance current benefit payments from the social pool. This objective has been achieved through new contribution collection procedures and accounting systems. Now, pension contributions from individual workers are segregated at the point o f their initial collection, deposited in a special treasury account in the central bank, and transferred directly andimmediately to an account controlledbythe provincial-level finance bureau. The rest o fthe social insurance contributions are deposited in another set o f special accounts located in commercial banks and controlled by the financial bureaus at the same pooling level. 2. Design and construction of information system linking social insurance agencies at provincial,municipalandcounty/districtlevel Working with a contractor, the provincial labor and social secunty bureau information center designed a new social secunty information system capable o f maintaining information on social secunty contributions, work histones, individual account balances, and benefit payments and automating most o f the steps in calculating social secunty benefit entitlements. The provincial government financed the installation o f this system in its own offices. The governments at the provincial, municipal and county/distnct level cofinanced its installation in 10 o f the province's 14municipalities. A4-2  The new system consists o f a provincial level database and a set o f municipal level databases. The municipal databases contain the detailed information on contribution collections, individual account balances, work histones, and pension payments that i s used to support program operations. Each o f the municipal systems covers all o f the county and distnct level governments in the municipality The provincial database contains general administrative and statistical information as well as a copy o f the detailed infomation on the individual pension accounts. The information on each worker's individual account balance i s transmitted from the municipal to the provincial levels once a month. The remaining four municipalities - Dalian, Shenyang, Fushun and Panjin -- had already installed computer systems pnor to the start o f this pilot. In order to conserve provincial resources, these four were not converted to the new system as part o f this initiative. Their systems already contain information on contributions, work histones, individual account balances, andbenefit payments, and these municipalities are able to transmit the requireddata to the provincial system. In order to gain the full functionality of the new system, these four municipalities will have to be converted. Under current policies, such conversions would apparently have to be financed out o f their own resources. 3. Automation of social insurancebenefit calculations All but 10 to 20 o f the county/distnct level units have finished entenng the information they collect regularly on individual work histones into the municipal databases. Where this task has been completed, local social insurance agencies are able to automate most o f the steps in the calculation o f pensions. The databases do not contain complete information on service pnor to 1992, however, or on service in those arduous occupations that entitle workers to earlier retirement. Thus, for the foreseeable future, pension calculations will still rely, at least in part, on manual checking through the paper records contained in each individual's personal archive. This function is now performed at eachjunsdiction's labor and social secunty bureau. 4. Socialization of social insurance record keeping and benefit payment by shifting responsibility from enterprises to social insurance agencies Essentially all o f the record keeping and benefit payment responsibilities in the pension and unemployment insurance programs have now been shifted to the social insurance agencies. A minonty o f enterpnses have not yet joined the health insurance program. 5. Shift of responsibility for collecting contributions from social insurance agencies to localtax agencies. The responsibility for collecting contributions has been shifted from the local social insurance agency to the local tax agency in all but one o f the Province's municipalities.2 From the data currently available, it i s not possible to know the effect this change has had on collection rates. Although enterpnses can now make social insurance contribution and local tax payments in one stop at the tax agency, in most municipalities they must still go to their local social 2The excephon is mDaliancity, which for b s purpose is outside ofthejurisdiction o fthe LiaomngProvmcial Government. A4-3  insurance bureau each month to record any changes in individual account information and have their payment forms venfied. 6. Improvements ininternal controls through extension of two track accountingsystem to all branches of social insurance. The two-track system o f financial management has been extended to all o f the five branches o f social insurance. Together with the separation o f the individual pension account contributions from the rest o f the contributions, this has improved the internal controls over social insurance contributions. Under the two-track system, contribution collections are immediately transferred to special accounts controlled by the respective financial bureaus. Funds not needed immediately for benefit payments are deposited ininterest-beanng accounts. When funds are needed for benefit payments, they are first transferred to accounts controlled by the social insurance agency and then transferred to the social insurance beneficianes (or other payment recipients). Only ina few cities (no more than 5), the fbnds used to pay basic pensions from social pooling are transferred to beneficianes without transit via social insurance agency (This procedure has not been endorsed by MOLSS, however.) 7. Unification of Social Insurance Finance and Management Social insurance management has already been integrated vertically in four o f the province's municipalities - Shenyang, Fushun, Benxi and Dandong. The remaining ten provinces are to convert to this model by the end o f this year. The vertical integration means that county and distnct social insurance agencies are converted from independent agencies attached to their respective local general government to departments o f the municipal social insurance agency. Their employees become employees o fthe municipal government. Similarly, pension finances are now pooled at the municipal level in four o f the fourteen municipalities and are scheduled to be pooled at the municipal level inthe other ten at the end o f this year. For this purpose, municipal pooling means that enterpnse contribution rates are equalized throughout the municipality and, for the most part, financial flows are adjusted to equalize the burden among the country anddistnct government^.^ The vertical integration o f the administrative units, equalization of the burden across the entire municipality, and the automation o f many o f the administrative tasks are three important steps toward the goal o f a standardized social secunty system. They help to assure the implementation o f the kind o f uniform costs, benefits and administrative procedures that will assure equal treatment o f all enterpnses and individuals inthe municipality 3 For example, FushunMumcipality currently has both vertical management and mumcipal pooling. Untilnow, the mumcipal government has covered pension deficits remaimng after transfers from the national andprovmcial governments and the equalization fund. However, the mumcipal government i s considermg shifting a portion o fthat responsibility to its county and distnct governments. Absent a clear formula for distributing the localburden accordingto the relative resource bases ofthe different local governmental units, this would represent a step backward from mumcipalpooling A 4 4  8. ExtendCoverage Inpnnciple, coverage has been extendedto free lance professionals, the self-employed andrural immigrantworkers, although there are few measures to assesshow effective the change has been implemented. 111.NextSteps Additional steps that built on the progress already made in the Liaoning pilot are needed to further strengthen financial control and overall program management. Some involve competing initiatives that have already been started or that were part o f the onginal vision but have not yet been completed. Others involve adapting to the Chinese environment processes and procedures that are now employed in other countnes and that have come to represent international best practice. 1. Completeimplementationof the reformagendacontainedinDocument42 (1) CompleteDevelopment of theInformation System As recognized in Document 42, a modem information system i s one o f the most important requirements for efficient social insurance program administration. The system now being implemented in Liaoning seems to hold the promise o f substantially improving program operations, once it i s completed. Several important steps remain in the initial implementation phase, however. These include: Completing the entry o f the work histones o f current contributors. As noted previously, some 10to 20 counties and distncts have not been able yet to fully automate their records. Converting the remaining four municipalities to the new system. Automating data transfers between the tax bureaus and social insurance agencies. At least inShenyang Municipality, these transfers are now done manually Finishing the clean up o f the provincial data base on individual accounts to allow generation o f improved program statistics, particularly on earnings. Data edits show that this data base contains numerous errors, affecting perhaps as many as half o f all the records. Many o f the errors appear to be instances where the data are internally inconsistent, such as when the reported age and sex are inconsistent with the reported national identification number. Most o f these are errors that will be corrected before the data is used to calculate benefit entitlements. Although they are unlikely to cause erroneous benefit payments, they do compromise the ability to generate reliable program statistics and to venfy the transfer o f the deposits into individual accounts that are flowing from the municipal level to the provincial level o f government. A4-5  Completion o f these steps will give the province a powerful tool for improving program management. Provincial officials should also be examining ways to improve further the h c t i o n i n g o f the information system, including: i)Reviewingtheneedforacommondatasystemacrosseachofthedifferentagencies. The vision outlined in Document 42 was o f a single system with uniform software and hardware connecting each o f the vanous agencies involved in administenng the social insurance system. At least with respect to the tax bureaus, this has not happened. The tax bureaus already had a data system that integrated their operations throughout the province andhave not participated inthe development o f the new social insurance system. Provincial officials need to review the relationship between the tax agencies and the social insurance agencies to see whether improved data exchange protocols can achieve most o fthe gains that a fully integratedsystem would bnng. ii)Captmngalloftheinformationthatwillsubsequentlybeneededtocalculatebenefitsso that, over time, manual reviews o f the information inthe personal archives will no longer be necessary. The personal archives are currently usedto establish service pnor to 1992, service in arduous occupations, and other factors that may tngger special retirement provisions. It may not be cost effective to input information on service pnor to 1992, but to the extent possible, any other information required to establish benefit entitlement and calculate benefit amounts should be captured contemporaneously and entered into the work records maintained at the social insurance agency The objective is eventually to eliminate the need to use the personal archive in the administration o f social insurance programs and rely instead exclusively on data maintained in the agency's data bases or submitted bythe retinng worker at the time a claim i s filed. iii)Venfyngthecompletenessandaccuracyoftheinformationsystem. Whenthesystemis stabilized, it would be advisable to have an outside organization conduct a thorough review of its operations to assure that the software i s doing all calculations correctly and i s moving and stonng data as intended. A pnvate accounting firm that audits organizations with major computer systems should be qualified to conduct such a review. iv) Creating a provincial data base covenng pension recipients. After the individual account data base has been cleaned up, a provincial level data base covenng pension recipients should be constructed. Such a data base would greatly improve the quality of program statistics and, together with the individual account data base, provide a sound basis for projecting future program costs. It would also be the first step in eventually moving to provincial level pooling and management o f the pension system. Initially, the provincial database covenng pension payments might consist o fjust a sample of the municipal data bases, containing, for example, one tenth o f the records o f each municipal data base selected using a process that produces a random sample. A ten percent sample would provide ample basis for generating program statistics, tracking program operations, and basing cost projections and it allow the information technology specialists to straighten out any new data problems before attempting to create a database covenng the entire pensionrecipient population. A4-6  (2) Regularly Publish Information on Revenue and Expenditure The provincial agency should improve its regular publication of information on the revenue and expenses o f the social insurance system and the composition of its investments, as provided in Document 42. As soon as practical, the province should arrange for regular audits o f the financial statements by outside auditors. Outside audits have several benefits. They enhance the credibility o f the financial statements and increase public confidence in the government's management o f social insurance finds. The audit process should also involve a review o f the internal controls that have been established to assure accurate accounting o f all financial flows, thereby helping to fblfill the promise contained in Document 42. Annex 4-1 contains some samples o f financial reports issued inother countnes. (3) Improve the Legislative Basisfor the Program The tax authonties report several areas where weaknesses inthe legislative basis of the program undermine compliance. One is the absence o f clear consequences for noncompliance. Another is uncertainty about the obligation to contribute among enterpnses that may have negotiated tax concessions in return for locating in specific junsdictions. There i s also confbsion about the status o f people working in many enterpnses that are registered in rural areas and people who may not havewntten labor contracts. In September 2004, Shenyang Municipality adopted legislation that establishes penalties for noncompliance with social insurance contribution requirements that are close to those for noncompliance with tax requirements. Fushun Municipality is now developing similar legislation. At the very least, these efforts need to be expanded to the rest o f the municipalities inLiaoning Province. Legislationat the provincial level would probably be more effective that the efforts o f individual municipalities. The most effective approach would be legislation at the national level. The legislation needs to establish clear consequences for noncompliance. It should also clanfy the vanous coverage issues, particularly the status o f pnvately-owned enterpnses registered(and/or located) inrural areas, the status o f the township andvillage-owned enterpnses, the status of workers registered in urban areas but working in rural areas, and the status o f enterpnses that may have negotiated agreements containing tax concessions from local governments inreturnfor their locating ina particularjunsdiction. The authonties should also consider breaking the link between that portion o fthe labor code that specifies the conditions for labor contracts and the coverage o f the social insurance system. Social insurance coverage should be based on a simple test that there was work performed for pay Enterpnses and workers should not be allowed to avoid social insurance contributions by fiddling with the provisions o f labor contracts (or not having one at all) and the tax authonties should not be expected to enforce other portions o f the labor code in order to gain compliance with the social insurance contribution requirements. (4) Move to ProvincialLevel Pooling There are currently five contribution rates: 25.5%, 24%, 23.5%, 20% and 19%. The provincial government plans to lower employer pension contribution rates first to 22.5%, 22%, 20% and  19% by the end o f this year, then to 20 percent in all municipalities (except Dalian o f 19%) by the end o f 2006. Once rates have been equalized and all municipalities have established vertically integrated social insurance administrations, it is a relatively easy step to move to provincial level pooling, at least for the pension program. The province should also develop plans to complete the vertical integration o f the social insurance agencies by merging the municipal level social insurance agencies with the provincial level agency 2. Instituteadditionaladministrativereforms Several relatively modest additional reforms could have a major impact on the quality o f both financial and general management. (1) Financial Management Improvements Financial management can be improved by streamlining several cash management processes and beginning to reconcile the accounting records covenng individual accounts with the balances being camed inthe provincial financial accounts. Specific steps include: Converting the current two track accounting system into a full treasury cash management system. The treasury cash management systems found in most advanced countnes separate the process o f approving transactions fkom the process o f managing cash flow Cash transactions are processed through the treasury system in response to payment orders executed by program administrators. The treasury managers do not have the authonty to initiate transactions without a payment order from the program manager and program managers never exercise direct control over the fiscal resources. The two track accounting system has some o f the attributes o f the treasury system inthat contributions are deposited directly into accounts controlled by the financial bureau. At present, however, the financial bureaureturns the funds to the social insurance agency for the purpose o f making benefit payments, merging the cash management and payment authonzation function in the same agency Internal controls would be improved if the financial bureau retained control over the finds and transferred them directly to the bank accounts o f the program beneficianes, in response to payment orders drawn up by the social insurance agen~ies.~ Accelerating the movement o f finds from enterpnse deposits to interest beanng accounts. It is reportedthat as much as 30 days elapses betweenthe time that an enterpnse deposits its social insurance contributions in the special accounts controlled by the local financial bureau and the time those funds are invested. In contrast, the employee pension contributions are channeled through a different special account managed by the central bank in a matter o f hours. The financial departments need adjust the structure o f the special accounts used for those social insurance contributions other than the individual pension account contribution to reduce significantly the current processing time. This may involve using the central bank as the fiscal transfer agent or consolidating the Inaddition, wthn the financial bureau, the authonty to set andexecutemvestment policies shouldbeassignedto a different group than the group responsible for managing the fiscal flows. A4-8  vanous program and county level accounts into one municipal level account containing all social insurance contributions (except those for individual pension accounts) but maintaining separate accountingbyprogram. c) More effective pooling o f financial resources. At present, separate accounts are maintained in each county and distnct into which contributions are deposited and from which transfers are made to cover benefit payments, even inmunicipalities that officially pool at the municipal level. The multiplicity o f accounts may be one o f the factors underlyng the money management delays just noted. Municipal authonties should evaluate the relative efficiency o f merging all o f the financial operations into one single municipal level account. d) Reconciling provincial level individual accounting records with the provincial level financial flows. Although accounting records and bank records are matched at the local level at the time that contributions are collected, the individual account information at the provincial level i s not currently matchedwith the provincial level fiscal balances. This is a potentially senous internal control weakness. Matches will not be possible until the cleaning up o f the provincial level data base is complete. At that time, regular matches between account records and financial balances should be instituted. (2) Create Performance Standards,Measure Performance, and Report Results Provincial and national program administrators need to create a set o f performance standards to measure current service quality and facilitate service quality improvements. Among national social insurance agencies, best international practice is to create a set o f performance standards, establish goals for each standard, and report regularly to the legislature and the general public about progress inachieving the goals. Common performance standards for social insurance agencies include: the time required to process applications for benefits or reports o f changes in beneficiary status, 0 the fraction o f new benefit calculations or continuing benefit payments that have errors, the effectiveness o frecovery o f debt owed to the social insurance agency, and 0 the level o fpublic satisfaction with the service deliveredbythe agency. h e x 4-2 contains examples o f performance standards taken from public reports posted on the internet for several different pension agencies. (3) Establish an Internal Performance Review Group An integral part o f the process o f setting performance standards and measunng performance is the establishment o f a separate group within the social insurance agency whose pnmary hnction i s to review internal operations. Such a group would track performance to assure the accuracy o f reports o f processing times, debt recovery, and similar management indicators. It would also select a small sample o f case files and review them intensively in order to construct measures o f A49  the accuracy o f administrative processes. Where performance measures include the level of public satisfaction, the internal review staff would develop performance measures through household surveys or other appropnate means. An internal review staff also plays an important role in assunng fiscal integnty They are responsible for evaluating the structure o f internal control systems and monitonng the effectiveness o f their implementation. They also provide the resources to investigate any allegations o f fraudulent behavior or mismanagement among agency staff. (4) Standardize Operating Procedures Standardizing procedures involves developing detailed, step by step instructions for handlingall important transactions, prepanng wntten documents outlining these procedures in a senes o f operations manuals, distributing the manuals throughout the organization and monitonng operations to assure that the approved procedures are being followed. To achieve uniformity o f operations, standardized procedures need to be executed by staff operating under uniform organizational structures occupyng jobs with common job descnptions and standard skill requirements. Like the operating policies, the organizational structures, job descnptions and skillrequirements needto bewnttenandcommunicated throughout the organization. Standardized operating procedures are both a reform objective and a necessary precondition for efficient administration. They are a reform objective because they assure equal treatment o f all program participants. They are a necessary precondition because they allow objective analysis o f operating costs and results and the structured development and implementation of operational improvements. When an internal review finds a problem in a particular program operation, such as erroneous payments or improper financial transactions, those conducting the review can assess whether the problem was caused by a failure to follow standard procedures or whether standard procedures need to be modified in order to eliminate the problem. In either case, the standardization o f internal processes allows similar corrective actions to be implemented throughout the organization. 3. Pilot Test More ComplexReforms Several important administrative reform areas would be good candidates for further pilots. In these cases, although the reform objectives are clear, the most effective approach inthe Chinese context is not as clear. Alternatives can be developed and piloted in different locations with the results evaluated to determine which alternative i s the most promising. Three such areas are streamlining current administrative process, particularly those involving data collection and data processing, improving collections, andimproving the dissemination o fprogram information. (1) Streamlining Processes Many o f the current social insurance administrative processes are cumbersome and impose significant and unnecessary costs on enterpnses, potential beneficianes and the administrative agencies themselves. They appear to have been designed inan era inwhich the costs incurred by enterpnses and government agencies was not a particularly important concern and when A4-10  information technologies were pnmitive. These processes are no longer appropnate in a market economy where much o f the economic activity i s camed out in profit-making concerns, where state enterpnses must keep operating costs low in order to compete, and where modem information technologies are widely available. Cumbersome administrative procedures increase the cost to all participants o f operating the social insurance programs and are a bamer to compliance, particularly among small enterpnses andthe self-employed. One example is the monthly procedure enterpnses are required to follow to pay their social insurance contributions. Typically in Liaoning Province this requires a representative o f each enterpnse to fill out multiple copies o f a set o f forms, personally visit the social insurance office, personally visit the tax office, and personally visit the banks5The stop at the social insurance office is to venfy the information on the form filled out at the enterpnse, the stop at the tax office is to drop o f f one set o f forms and pick up another form, the stop at the bank i s to make the actual payment. The bank certifies that contributions have been paid on the form received at the tax office and six copies o f the properly certified payment form are then distributed among different entities. Other cumbersome procedures include the process for registenng a new enterpnse or renewing a business license, for transfer a worker's account from one pooling area to another, and for applymg for a pensionbenefit. Most countnes have streamlined their business registration, contribution collection and benefit applicationprocesses inorder to minimize both employer burden and government administrative cost. In the United States, a business can register with both the tax and social insurance authonties and obtain a tax identification number (which is also the social insurance identification number) in one ten minute telephone call; social insurance contributions are paid, usually monthly, either by mailing a check to the bank or through electronic funds transfers executed via the internet from the enterpnse's office; and the vast majonty o f new retirees never visit their local social insurance office, applymg for benefits instead either on the telephone or over the internet. Other OECD countnes follow similarly streamlined procedures; some are even more automated that the United States. Contribution payment procedures in several OECD countnes are described inAnnex 4-3. Many o f the steps in the current processes are designed either to make sure that information i s transferred between different government agencies, that the reported data i s correct and that correct amounts are paid. These are all valid objectives. The problem with the current processes i s that the objectives are being achieved through paper-intensive routines that place a disproportionate amount o f the burden on the citizen and, in fact, may not be as effective at achieving these objectives as alternative approaches would be.6 Reform will require developing new processes that reduce the reliance on paper forms and that automate data transfers but that do not sacnfice (and may improve) the controls built into the current processes. Procedures The actual process may differ m certam details from one mwcipality to another. The process outlined here is that used mFushun. Currently, enterpnses report each month all changes mthelr employment patterns, including all workers who are hred, all who leave, and all who have a change inthelr monthly mcome. Ths approach probably works well for enterpnses that have little turnover and few changes inemployment relationships. As pnvately runsmall and medium sued enterpnses become more important inthe economy, however, the strategy is likely to become mcreasingly difficult to implement. Employee turnover will mcrease and wages will become more variable, mcreasing the time it takes each month to report andvalidate every change. A4-11  employed in other countnes will provide useful input in developing alternatives for China, but can never be completely replicated. Each country must develop its own procedures taking into consideration the division o f responsibilities among governmental units, the level o f technology, the legal environment, andbusiness practices employed elsewhere inthe community (2) Improving Collections Once the legal basis for collecting contributions has been improved, alternative contribution collection techniques should be pilot tested. Collection enforcement involves addressing three kinds o f challenges: (1) making sure all covered enterpnses are registered, (2) making sure all registered enterpnses accurately report their contribution liabilities, and (3) collecting the liabilities reported. An improved legal basis will help deal with each o f these challenges by clanfyng coverage issues. It will be particularly important in facilitating actual collection o f enterpnse liabilities. The tax bureau in Fushun Municipality has already begun testing several potentially powerful enforcement techniques. It has matched its records with those of the Industry and Commerce Bureau to make sure that all enterpnses registered with the ICB are also registered with the tax bureau and compared information from the statistics bureau with its own data to determine the scope o f its compliance problem. It has sent teams o f auditors into selected enterpnses that it suspected o f misreporting their social insurance contribution liabilities, finding substantial underreporting in24 o f the 25 enterpnses it examined. Once its municipal compliance ordinance has been adopted, it plans additional steps to improve compliance, including publicity campaigns, publication o f lists o f delinquent enterpnses, and denial o f social benefits or honors to delinquent enterpnse mangers. The impact o f each o f the vanous steps being taken in Fushunshould be monitored carefully so that information on effective practices can be shared with other municipalities and provinces. Similar techniques should be piloted inother localities. One promising change may require the assistance o f the central government. Inpnnciple, local tax bureaus should be able to compare enterpnse reports filed inconnection with their profits tax payments with those filed inconnection with social insurance contribution payments to check on the accuracy o f enterpnse reports o f their total payroll. These compansons can not be made currently, however, because the profits tax returns do not ask that comparable information be reported. The national tax authonties should work with local authonties to revise the profits tax reports to allow them to be used to improve enforcement o f social insurance contribution requirements. Finally, the feasibility o f usingthe national tax authonty to collect social insurance contributions should be tested in one or more locations as part o f a pilot. Eventually, China should further centralize the responsibility for operating its pension program through national pooling and national databases. At that point, it may also want to shift collection responsibility to the national tax authonties, ifthat option proves to be feasible inpilot testing. (3) Issuing Annual Statements A4-12  Internationally, an increasing number o f national pension systems are sending each worker an annual statement showing the worker's current account balance, informing the worker o f the implications for his future retirement income, and telling the worker how to report potential errors in his account records to the pension agency. These statements are a form o f internal control to assure the accuracy o f employer reports and agency records, increase public confidence that their contributions are being used in the manner intended, and better inform workers about how to plan for retirement. Annual statements were tned in 1998 inat least some o f the municipalities o f Liaoning province, but it was decided at that time that their issuance was too expensive. Since that time, however, the individual accounts have become a much more important component o f the pension package, making them potentially more valuable to contributors, and the new information technology system has been installed, which should have made it much easier to issue the statements. Once the data bases have been fully computenzed, the province should once again issue annual statements and assess both the cost andthe benefits o f the issuance. With modem computers, the cost o f issuing such statements should not be much more than the cost o f the postage needed to mail them plus the cost o f answenng any questions that the statements generate or o f correcting any errors they uncover. 4. The LongerTerm Goal: A NationalPensionSystem The longer term goal should be the gradual evolution to single national pension system operating on the basis o f a single national data system and financed through a single national pool. National pooling i s the only effective way o f evening out the financial consequences o f regional demographic differences on a long-term and sustainable basis. To be implemented effectively, national pooling will require the construction andmaintenance o f a national database containing information on collections, individual account balances, individual work histones and individual pension payment records. Such a database will provide the accounting controls needed to protect the integnty o f the national financial pool. Itwill also automatically eliminate the problems associated with worker migration, including the administrative burden o f moving records from one pooling region to another and the impact on regional fiscal balances o f transfemng either the account balances or the pension payment liabilities from one region to another. National pooling i s probably easier to accomplish if the responsibility for collecting social insurance contributions is shiftedto the national tax authonties, although local authonties might continue to collect, particularly ifthey were integratedwith the provincial level institution, as the tax bureaus in Liaoning Province currently are. As the collection processes are streamlined, either the tax authonty or the social insurance authonty will eventually assume responsibility for collecting and processing the information on individual work expenence and deposits to individual accounts. In either case, as long as the accounting records are carefully reconciled with the financial flows, these data can then be fed into the national data system. Eventually, pension benefits will likely also be administered through a national benefit agency For an extended intenm penod, however, national pooling could be combined with provincial A4-13  level administration o f benefit payments. Before such a system i s implemented, however, many o f the administrative reforms recommendedhere would need to be developed and implemented, including the development o f efficient data transfer protocols, the standardization o f administrative processes, the development o f performance standards, and the creation o f internal review units. These steps would be needed to assure the efficient and effective exchange o f information andto allow effective central monitonng o f regional and local administrative actions. Inthe longer run, the operations of social insurance, financial, and tax agencies should also be subjected to a formal process o f nsk analysis. Such a process involves conducting a structure review o f all o f the administrative operations, assessing the nsk of loss associated with each possibleproblem together with the cost o f addressing the nsk through additional internal controls, and developing a list o fthe relativepnonties o f implementing different types o f internal control. A4-14  Annex 4-1 Examplesof PublishedFinancialStatementsOfNationalPension Schemes 1. Canada StatementofNetAssets at March31 (Millions ofDollars) - Assets ~ ~~~ Investments: Provmcial bonds 24,890 26,188 Canadabonds 3,3 86 3,403 CPP InvestmentBoard- at fax value Canadian equities 10,114 5,024 Non-Canadian equities 4,175 2,131 2002 2001 NetAssets, beginningof year 45,688 41,261 Increaseinassets Contribuhon mcome 23.533 21,407 Investmentmcome 3,565 2,849 Subtotal 27098 24,256 Source: Annual Report o fthe Canadian Pension Plan, 2001-2002, available at: http:llwww.sdc.gc.c~e~isp/pub/cpplrepo~2OO3lannual.sh~ A4- 15  2. UnitedKingdom BalanceSheet. as at 31March2004 &'OOO~ Source: The Pension Service Annual Report andAccounts, 2003104 h~:11www.theuensionservice.aov.u~udf/a~ualreuo~a~ualreuo~artl0304.udf 2003104 2002103 Net cash (outflow) from operating activities (note 1) (458,494) (276,485) Capital expenditure (2,569) (1,075) Net financmg from the Consolidated Fund (note 2) 262,604 133,278 Financing from the Nabonal Insurance Fund (note 2) 198,562 144,285 Increase m cash inthe penod 3 3 2003104 2002103 Net operating cost (760,809) (513,429) Adjustment for non-cash transactions 296,192 227,554 Adjustment for changes inworlung capital 6,052 9,390 Amount o fprovisionutilized inyear (29) 0 Net cash outflow from operating activities (458,594) (276,485) 2003104 2002103 Financing from the ConsolidatedFund 262,604 133,278 Financing from the National Insurance Fund 198,562 144,285 Financing inflow 461,166 277,563 (Increase m cash) (3) (3) Net cash requlrement 461,163 277,560 A4-16  Annex 4-2 SelectedPerformanceIndicators 1. UnitedStates SocialSecurityAdministration (1) Issuingsocial securitvnumbers Semice measurements % o f applicants notified o fnumber within 24 hours 91.2% % o f cards issued within 5 days 97.0% Average processing time 0.6 days % o fnumbers issued accurately 99.8% EfJici ency measurements Numbero fcards issued 16.8 million Staff years used 2,560 Outputper staff year 6,575 Cost per card issued $12.53 (2) Postingearningsto individualrecords: Service measurements % o fitemspostedcorrectly 98.8% % o f employer reports postedwithin 6 months 83.9% % o fself-employed reportspostedwithin 9 months 84.9% EfJicien cy measurements Items processed 221.9 million Staff years used 1,556 Output per staff year 142,641 Cost per itemposted $0.57 (3) New benefitawards Service measurements Average time to process retirement or survivor benefit 14.0 days Average time to process disability benefit 83.8 days Average time to process social assistance benefit for aged 16.4 days %o fretirement benefitsprocessed within 14days or before first payment due (iflater) 83.O% % o f social assistancebenefits for agedprocessedwithin 14 days65.4% % o fcontinuing payments made on scheduled day 99.9% % o finitialretirement benefit paymentswith an error 5.3% % ofinitial social assistancepaymentswith an error 6.3% % o flifetime retirement benefits paid incorrectly 0.2% % oflifetime social assistance benefitspaidincorrectly 5.2% A4-17  EfJici ency measurements Total claims received 7,242,802 Staff years used 20,7 17 Claims processed per staff year Retirement insurance 618.7 Disability insurance 236.9 Social assistance for aged 502.1 Social assistance for disabled 277.2 Overall average 350.0 Cost per case processed Retirement insurance $219.97 Disability insurance 574.58 Social assistance- aged 271.12 Social assistance- disabled 491.05 Overall average 388.87 (4) Service to the public Service measurements Visitors to local office with appointments: - - average waiting time inlocal office 6.8 minutes % seenwithin 10minutes 83.8% Visitors to local office without appointment -- average waiting time inlocal office 25.2 minutes % seenwithin 30 minutes 72.7% Persons contacting agency by telephone -- %reaching agency within 5 minutes 73.5% % o f contacts handledcorrectly 96.7% Ofcontacts involvingan issuedaffectingbenefit payments, - %with error leadingto incorrect payment 5.6% Satisfaction measures %rating local office as satisfactory -- Office location 89% Waiting time 71% - Waiting area comfort 76% -- Staff courtesy 87% Clanty o f information 82% % o fclaimants rating other agency services as satisfactory - telephone service 90% - informationreceivedby mail 75% Source: Social Secunty Adrmmstration, Accountability Reportfor Fiscal 1998, available at http://www.ssa.gov/finance/index.html A4-18  2. UnitedKingdomPensionsAgency Indicator Goal Actual Telephone semce: Percent of calls answered 88% 91.1% Percent of callers receivlng busy signal 1% 7.3% Accuracy ofbenefit calculations: Retirementpensions 2% 1.2% Minimumincome guarantee benefit 6% 6.1% Percent of incorrectbenefit payments 4% 4.0% Timeliness of actions: Percent of pensions awarded within 60 days 95% 95.6% Average time to settle minimumincome claim (days) 10 12.1 Source: UnitedKingdom, The Pension Service, Annual Report and Accounts, 2003104 Presented to the House of Commons pursuant to Sec 7(2) of the Government Resources and Accounts Act of 2000. Available at: http://www.thepensionservice.gov,uklaboutus/who_we_are.asp#annualreport A4-19  Annex 4-3 ContributionPaymentandReportingProcedures' This annex outlines the procedures usedinthe United States, Canada andthe UnitedKingdomto collect pension contributions and report individual earnings. Social insurance contributions are collected by the tax authonties inall three countnes. 1. Payments In the United States, pension and health insurance contributions (both the employer and employee shares) and individual income taxes withheld by employers are collected simultaneously. Employers remit their payments on a schedule determined by aggregate payments made in a previous year. Employers with liabilities in excess o f $50,000 per year deposit their payments roughly 3 days after each payday, or one day after their undeposited liability has reached $100,000. Employers with more than about 15 employees earning the national average wage would fall under this provision. Employers whose annual payments are less than $50,000 but more than $10,000 deposit once a month and those whose annual payments are less than $10,000 deposit quarterly Large employers are required to make their payments through electronic funds transfer via the internet. Smaller employers may mail their payment to the national tax agency or mail or bnngit to a commercial bank that has agreed to accept tax deposits. At the time that payment i s made, the employer submits a form (either on paper or electronically) listing: (1) Employer identification number, (2) Employer name and address, (3) Calendar quarter to which the payment applies and (4) Total amount being paid. At the end o f each calendar quarter, the employer submits a two page form summanzing the payments made the previous quarter. The form includes information on the employer's name and ID number, his total wage payments, his total payments subject to pension and health insurance, the total amount he has remitted in income tax withholding, pension contributions, andhealth insurance contributions. H e also reports adjustments ansing due to any payments o f subsidiesto low wage workers, sickness payments, and so forth. InCanada, payments for pension contributions are handled separately from other tax payments. As inthe US.,the payment schedule depends on the aggregate paymentsmade inprevious years, andranges from four times a monthfor the largest employers to once per quarter for the smallest. Payments can be made electronically via the internet or via ATM machines or through banks. Each payment must be accompanied by a payment from listing the employer's name, identification number, aggregate payroll, number o f employees, aggregate payment and ending date o f the payment penod. Inthe UnitedKingdom, small employers pay quarterly and all others pay monthly Payments can be made at a post office, by mail to the tax agency, through any bank that processes 'Thedormation presentedhere was taken from employer guides availableonthe internet at www.11s.gov(the United States), http://www.cra-arc.gc,ca/menu-e.html (Canada) andhttp://www,hmrc.gov.uk/ (United Kingdom) A4-20  electronic payments or through fund transfer directly over the internet. U.K.employers making an electronic payment need only transmit their employer identification number with a prefix to show whether they are making a national insurance payment or remittingincome taxes withheld and a suffix to show the month for which the payment is being made (ifnot the current month). 2. Employeeinformation In each of the three countnes, employers are required to file detailed reports once each year showing the earnings, insurance contributions, and income taxes withheld for each o f their employees. Inthe United States and Canada, the report covers a calendar year and must be filed with the government by the end o f February (although inthe U.S.copies must be given to each employee by the end o f January and the report to the government can be delayed untilthe end o f March if it i s submitted electronically ) In the U.K.,the reports cover the year ending the first week in April and are due to the government by May 19 Inthe U.S.,the reports are filed with and processed by the Social Secunty Administration. Inthe other two countnes, the reports are filed with andprocessed by the tax agency All three countnes allow reports to be filed over the internet or bymail, although employers with more than a certain number o f employees (250 in the U.S., 500 in Canada) are required to file electronically Small employers may file their reports on line. Medium sized employers are either required or encouraged to download software which will allow them to fill in the forms off-line and then upload the data. Alternatively, they are given data protocols that allow them to submit electronic files containingmultiplerecords. In all countnes, the annual form contains general information about the employer, including name, address, identification number, total payroll, industry, and so forth. This i s followed by individual information on each employee, including name, social insurance number, annual wages, social insurance contributions withheld, and income taxes withheld. The countnes also collect information other individualized information that each uses either in enforcing its tax laws or inmonitonng other payments or collections, such as subsidies for low wage workers or collections o f student loans. A4-21