Report No. 25174-IR Islamic Republic of Iran The Pension System in Iran: Challenges and Opportunities (In Two Volumes) Volume I: Main Report September 2003 Middle East and North Africa Social and Human Development Group (MNSHD) Document of the World Bank Currency Equivalents Unit of Currency=Iranian Rhials (IR) Average Exchange Rate (IRper US Dollar) 1Dollar =7,500 Rhials Basic Indicators 2001 GDP =IR 666,165billion (USD 88.9billion) GDP per capita =IR 10.4million (USD 1,396million) Population=63.9 million Labor Force= 17.1million ACRONYMS AND ABBREVIATIONS cs Contractual Savings CSRO Civil ServantsRetirement Organization DB Defined Benefit DC Defined Contributions FF Fully Funded FFYP First Five-Year Plan GDP Gross DomesticProduct IA Individual Accounts MPO Management and Planning Organization MENA MiddleEast and North Africa Region MNSHD MiddleEast and North Africa Human Development NDC Notional Defined Contributions NPF NationalPension Fund OECD Organization for Economic Cooperation and Development PAYG Pay-As-You-Go SSIC Social SecurityInvestment Company sso Social SecurityOrganization USD United StatesDollars Vice President: Jean Louis Sarbib CountryDirector: Joseph Saba SectorDirector: Jacques Baudouy SectorManager: George Schieber ACKNOWLEDGEMENTS This report has been preparedby a team ledby David Robalino and comprising Albert0 Musalem and Tatyana Bogomolova. It has been prepared in close collaboration with the Managing and Planning Organization of Iran, and the technical teams o f the Social Security Organization and the Civil Servants Retirement Organization. The report has benefited greatly from extensive comments providedby the peer reviewers Robert Palacios and Michal Rutkowski, as well as from Dr. Mehdi Karbasian, First Vice Minister of Finance. Remaining errors and omissions are the authors' responsibility. Data for international comparisons were compiled by Massimo Sabbatini and Cheikh Fall. The report i s part of a series of studies on Social Protection requested by the Government of the Islamic Republic of Iran. TABLEOF CONTENTS EXECUTIVESUMMARY ................................................................................................................ i 1 BACKGROUND . ..................................................................................................................... 1 2. GENERAL OVERVIEW OF THE IRANIANPENSIONSYSTEM .............................................. 3 3.3.1. THEInstitutional SOCIAL SECURITY ............................................................. Issues ..................................................................................................... ORGANIZATION (SSO) 11 11 3.2. Contributions and Labor Force Coverage................................................................. 12 3.3. Benefits and Rates ofReturn on Worker Savings....................................................... 14 3.4. Financing Mechanisms and Sustainability................................................................. 19 3.5. Management Policiesfor Fund Reserves................................................................... 25 4.4.1. THEInstitutional CIVIL SERVANTS RETIREMENT ..................................... Issues ..................................................................................................... 31 ORGANIZATION (CSRO) 31 4.2. Coverage and Contributions ...................................................................................... 32 4.3. Benefits and Rates ofReturn ...................................................................................... 33 4.4. Financing Mechanisms and Sustainability................................................................. 37 4.5. Management Policiesfor Fund Reserves................................................................... 40 5. TAX TREATMENT WORKER SAVINGS OF ........................................................................ 45 6. A FRAMEWORK 6.1. Objectives of the Public Pension System. Costs. and Implementation Mechanisms..47 FORPENSIONREFORM ......................................................................... 47 6.2. International Experiences: One Size Does Not Fit All.............................................. 51 7. REFORMINGTHE IRANIANPENSIONSYSTEM ................................................................. 55 7.1. Dealing with the Finances of Current Systems WhileImproving Equity. Incentives. Reallocating Risks. and Creating a Better Balance between Mandatory versus Improving the Management of Fund Reserves: The Needfor Better Governance... 56 Voluntary Savings ...................................................................................................... 7.2. 69 7.3. Institutional Issues to Consider over the Short and Medium Terms: Strengthening Institutional Capacity and Merging Pension Funds .................................................. 74 7.4. Expanding Coverage.................................................................................................. 77 7.5. Promoting Voluntary Savings and Developing CapitalMarkets............................... 80 8. DESIGNINGANDIMPLEMENTINGAREFORMPROGRAM ............................................... 83 8.1. Triggeringa Successful Reform Program.................................................................. 83 8.2. Next Stepsfor Iran ..................................................................................................... 84 9. REFERENCES ..................................................................................................................... 85 LISTOFTABLES Table 1 Total Mandatory PensionBenefits as Percent o f Individual Earnings inHigh-Income Countries5 Table 2 Vesting Periods. Accrual Rates. and Retirement Ages inMENA Countries ................................ 6 Table 3 Pension Expenditures and Determinants in Selected MENA Countries........................................ 8 Table 4 SSO. Evolution o fthe Number o f Contributors by Type o f Establishment (thousands) ..........14 Table 5 Benefits Provided by the SSO ..................................................................................................... 15 Table 6 16 Revenues and Expenditures inthe SSO (billions of rhials) ........................................................ Evolutiono f the Total Number o f Pensioners inthe SSO........................................................... Table 7 21 Table 8 SSO. Basic Indicators o f Financial Sustainability ....................................................................... 22 Table 9 28 Table 10 SSO. Profitability o f Selected Companies. 2000 and 2001 (billion of rhials) ............................ Sources o f Revenues from Investments inthe SSO (billions o f rhials in2001) ......................... 28 Table 11 Evolutiono f Contributors inthe CSRO ...................................................................................... 32 Table 13 Evolutiono f Total Beneficiaries inthe CSRO ............................................................................ Table 12 Benefits Providedby the CSRO .................................................................................................. 34 35 Table 14 Revenues and Expenditures inthe CSRO (billions o f rhials) ..................................................... Table 15 CSRO, Structure o f Reserves, 2001 (millions o f dollars) ........................................................... 39 42 Table 16 Pension Reforms inEasternEurope and Central Asia ................................................................ 52 Table 17 Pension Reforms inLatin America ............................................................................................. 53 Table 18 Proposed Reforms for the SSO and the CSRO............................................................................ 63 LISTOFFIGURES Figure 1 IranianPension System................................................................................................................. 3 Figure 2 Coverage o f Contributory Pension Systems inIran and the MENA Region................................. 4 Figure 3 4 Rates o f Return inOECD and Selected MENA Countries............................................................ Statutory Payroll Contributions to Pension Systems inMENA Countries .................................... Figure 4 7 Figure 5 Organizational Chart o fthe SSO................................................................................................. 12 Figure 6 SSO. Economy-Wide and Individual Replacement Rates by Income Level ............................... 16 Figure 7 Internal Rates o f Returninthe SSO ............................................................................................ 19 Figure 8 SSO. Sources o f Income.............................................................................................................. 20 Figure 9 Figure 10 Summary Results o fFinancialProjections inthe SSO ............................................................... Functional Compositiono f Expenditures inthe SSO.................................................................. 20 Figure 11 SSO, Investment Portfolio as a Share o f GDP ............................................................................ 24 26 26 Figure 13 Shasta's Companies: Ownership and Economic Sectors o f Activity, 2000 ............................... Figure 12 SSO, Structure ofthe Portfolio o fInvestments ........................................................................... 27 Figure 14 Organizational Chart o fthe CSRO.............................................................................................. 31 Figure 15 CSRO, Economy-Wide and Individual Replacement Rates by Income Level............................ 35 Figure 16 Evolution o f the Real Average Old-Age Pension........................................................................ 36 Figure 17 Internal Rates o f Returnfor Males and Females inthe CSRO .................................................... 37 Figure 18 CSRO, Composition o f Total Expenditures ................................................................................ 38 Figure 19 Summary o f Financial Projections inthe CSRO ......................................................................... Figure 20 CSRO Investment Company ....................................................................................................... . 40 41 Figure 21 Individual Replacement Rates for the SSO, the CSRO, and Selected Countries ........................ Figure 22 Effect o f Increasing Vesting Periods on Rates o f Return............................................................ 56 58 Figure 23 Marginal Financial Impact o f Alternative Reforms inthe SSO .................................................. 64 Figure 24 Financial Impacts o f ReformType I(Parametric) ....................................................................... Figure 25 Financial Impacts o f ReformType I1(Downsize DB-PAYG) ................................................... 66 Figure 26 Financial Impacts o f ReformType I11(NDCs) ........................................................................... 66 66 Figure 27 Fiscal Impacts o f Alternative Reforms........................................................................................ 67 Figure 28 Replacement Rates and Rates o f Return under ReformType I Figure 29 Replacement Rates and Rates o f Return under ReformType I1.................................................. ................................................... 68 68 Figure 30 Replacement Rates and Rates o f Return under ReformType I11................................................ 68 Figure 31 Estimated Cost o f a USD 200 Demogrant inIran....................................................................... 79 EXECUTIVE SUMMARY The present report evaluates the Iranian pension system and proposes a series o f policy interventions to improve outreach, financing mechanisms, incentives, equity, and management. It has been prepared at the request o f the Management and Planning Organization. Given data constraints, the analysis concentrates primarily on the Social Security Organization and the Civil Servants Retirement Organization, the two main pension funds. Occupational funds and non- contributory regimes are surveyed only briefly. The need to look more closely at these schemes inthe near future is emphasized. The report is organized in eight sections. Section 1presents the background o f the study. Section 2 provides a general overview o f the Iranian pension system. Sections 3 and 4 present detailed assessments of the Social Security Organization and the Civil Servants Retirement Organization. Section 5 concentrates on the tax treatment o f retirement savings. Section 6 proposes a framework for guiding pension reform, presents a typology o f pension mechanisms, and reviews intemational experiences. On the basis o f this framework, Section 7 outlines strategic directions for reforming the Iranian pension system and analyzes the financial and fiscal implications. Finally, Section 8 discusses the political economy o f pension reform and recommends necessary steps for the design and implementation o f a successful reformprogram. A. MAINRESULTS THEANALYSIS FROM The Iranian Constitution mandates the government to protect the elderly and women; Iran has thus developed contributory and non-contributory pension schemes that cover close to 50percent of the labor force and 40 percent of the population older than 60. Contributory systems in Iran-including occupational funds-regroup close to 9.5 million employees. The resulting coverage rate i s high compared to the 30-40 percent observed in other countries with similar levels o f income. Coverage, however, i s concentrated in urban areas (close to 80 percent o f the labor force). Thus, large segments o f the rural labor force remain uncovered. Non-contributory schemes seem well developed, reaching 40 percent o f the population older than 60. Unfortunately, inthe absence o f data about beneficiaries, it i s not possible to assess whether these schemes are reaching the most vulnerable. As a result of the "baby boom" that Iran experienced during the mid-80s, there is potentialfor a rapid increase in the number of contributors over the medium term, but raising unemployment rates are likely to dampen the effect. During the late 80s and early 90s, Iran experienced exceptional growth in the total number o f contributors to the Social Security Organization (10 percent per year), which allowed dependency ratios to remain low (at 20 beneficiaries per 100 contributors). The causes behind this rapid expansion are not well understood. High economic growth during the First Five-Year Development Plan as well as the opening o f the system to voluntary contributors provide only partial explanations. The growth rate o f total contributors has now slowed to less than 5 percent per year. Driven by the baby boomers who will attain working age inthe next 5 years as well as by risingfemale participation rates, the labor force i s expected to grow at 5 percent per year over the next 10 years. Unfortunately, even under optimistic assumptions about GDP growth (6 percent per year for the period 2002-2012), labor markets are unlikely to absorb all the newcomers. Unemployment rates are expected to reach 20 percent by year 2012. High contribution rates are also likely to discourage enrollment in the pension system. Moreover, the current Social Security Organization benefit formulas have built- in incentives to leave the system after 8-10 years and resume contributions only 2 years prior to retirement. Hence, there are no good reasons to expect that the coverage rates o f the employed population will increase significantly. If these coverage rates remain at current levels over the next 10years, thenthe share o f the labor force coveredby the system i s likely to stagnate. Both the Social Security Organization (SSO) and the Civil Sewants Retirement Organization (CSRO) target high replacement rates' and generous rates of return. Statutory replacement rates for full-career workers (40 years) can reach 75 percent in the CSRO and 100 percent in the SSO at all levels o f income. This contrasts with most OECD countries where statutory replacement rates for the average full-career worker approximate 60 percent o f gross earnings. Moreover, in these countries, replacement rates decrease with the level o f income (see Figure I). the SSO In and the CSRO, with current contribution rates, implicit real rates o f return range from 5 percent to over 20 percent per year. Unfortunately, implicit rates o f return above the growth rate o f the economy are unlikely to be sustainable over the long run. Figure I:Statutory ReplacementRatesfor the SSO, the CSRO, and Selected Countries -France - Korea --)lt - - - -Sweden CSRO 0 1 2 3 4 5 Individuals earnings (proportion of average) Source: Whitehouse (2002) and mission calculations for the SSO and the CSRO Pension expenditures have increased rapidly, doubling their share in GDP within the last decade. Pension expenditures (old age, disability, and survivor pensions) grew inreal terms at an average o f 11percent per year duringthe last decade. Today pension expenditures capture 1.5 percent o f GDP. Higher expenditures reflect the rapid increase in the total number o f beneficiaries (9 percent per year) fueled, inpart, by low minimumretirement ages and vesting periods, benefit formulas that encourage individuals to retire as soon as possible, and rising life expectancies. At various points in time the CSRO has provided additional incentives for early retirement as a mechanism to reduce the size o f the civil service. In 1988 and 1992, the SSO also implemented two generous early retirement programs. On an accrual basis, both the SSO and the CSRO are insolvent. Financial projections under various economic scenarios show that the operational balances o f the CSRO and the SSO will continue to deteriorate, becoming negative within the next 3-10 years respectively (see Figure11). This is not only the result o f the gradual aging o f the population but probably, more importantly, a reflection o f the generosity o f the system. Ifthe systemremains open to new entrants, unfunded pension liabilities for the period 2002-2070 could attain 140 percent o f today's GDP in the SSO and 35 percent in the CSRO. If the systems were closed to new entrants, unfunded pension liabilities would be lower but still in the order o f 20 percent o f GDP for the CSRO and 100 percent for the SSO. These large, unfunded pension liabilities threaten the credibility o f fiscal policy and the welfare o f future generations. The replacement rate is defined as the old-agepensiondividedby last gross wage. 11 Years I -baseline - - --low case-closed -baseIlne - - - Years .low case -closed In addition to problems of financial sustainability, important challenges remain to improve the management of reservesfor bothfunds. Current investment policies are complex, risky, and not necessarily in the best interest o f plan members. Lately, the funds have been converted defacto into agencies responsible for restructuring public companies. Indeed, in 2001, the government transferred assets worth Rhials 4,000 billion to the SSO and Rhials 3,750 billion to the CSRO to cover part o f its debt (mostly arrears). In the case o f the CSRO, these transfers have caused losses amounting to Rhials 132 billion. Further investment in these companies to improve management and profitability i s likely to bring additional losses to plan members. The pension funds are over-expanding their mandate and now have considerable market power in several economic sectors-including the stock market. This interferes with corporate governance and hampers the development o f the financial sector. Inpart, inappropriate investment policies reflect weak governance structures, as processes to select and structure the governing bodies, define fiduciary responsibilities, and enforce accountability do not follow best international practices. The analysis has also raised concerns about equity, as the system is prone to adverse inter- and intragenerational transfers. There are various sources o f inequities in the current arrangements. First, part o f the resources that the government transfers to the SSO (3 percent o f covered wages) and the CSRO (resources for administration) mostly benefits well-off workers in the formal sector. Second, as in other defined benefit systems, rates o f return on contributions vary as a function o f wage histories, life expectancies, and retirement strategies. Workers with fast- growing salaries and who live longer obtain higher rates o f return. Often they are well educated and healthy individuals. Workers who join the system late inlife or who evade the system during mid-career also obtain higher rates o f return. This heterogeneity in rates of return creates intragenerational transfers that do not always favor low-income individuals. Third, the fact that only the last two years o f salaries are included in the calculation o f the pension penalizes low- income workers (blue-collar workers), who usually have their peak income relative to the average during mid-career. Finally, if reforms are delayed, future bailouts and/or cuts in benefits necessary to balance the finances o f the system will imply a massive intergenerational transfer from tomorrow's poor to the current well-off. While very scant information is currently available regarding the structure, size, andfinances of occupationalfunds, it is likely that the majority is accumulating unfunded pension liabilities. The occupational funds seem to be heterogeneous in terms o f size and benefits offered. Actuaries from the SSO Research Institute consider that while a few o f the funds are properly managed and seem to be financially sustainable, the majority i s likely to generate an operational deficit within the next few years. Since the funds are linked to public companies, their implicit pension debt i s part o f the contingent liabilities o f the government. Another issue o f concern is the fact that in 111 ... transferring from one fundto another or from any fund to the SSO, the current rules are complex, give room to discretion, are not always fair from the workers point o f view, and thus constrain labor mobility. B. PROPOSEDPOLICYINTER VE"S The report identifies various types o f interventions that the government could consider to strengthen the social function o f the Iranian pension system while improving financial sustainability and reducing economic distortions. The recommendations are classified in five groups: (i) interventions to improve the financial situation o f the current systems and to improve incentives and equity; (ii) interventions to improve the management o f reserves; (iii) interventions to expand coverage; (iv) interventions to promote voluntary savings; and (v) other interventions to improve institutional capacity and management. The main policies are summarized below. Dealing with the Finances of Current Systems While Improving Equity and Reducing DistortionsinLabor Supply andRetirementDecisions The government and civil society need to make a decision as to what is an adequate and affordable level of income replacement for retirement and how this level of income should be generated by a combination of mandatory and voluntary savings. In OECD countries, for instance, the public system focuses on replacing income for low-income individuals. Middle- income and high-income individuals are expected to complement the savings in the mandatory public system through voluntary savings. In Iran, current highreplacement rates at all levels o f income are neither affordable nor sustainable. This section presents three reform options for reviewingthe pattern o f income replacement in order to improve financial sustainability, while improvingincentives and equity. Reform TypeI: Maintain the current mandate to save while introducing parametric reforms in the SSO and the CSRO. This reform maintains the level o f the current mandate to save (i.e., contribution rates are kept at current levels) through the defined benefit-pay-as-you-go (DB-PAYG) systems. Implicitly, the reform considers the DB-PAYG system as the only source o f savings for retirement and preserves flat replacement rates across levels o f income. In order to improve financial sustainability, however, the targeted replacement rate for a full-career worker (40 years) would need to be reduced to at least 80 percent (close to Spain). This implies a 2 percent accrual rate that could be reached over a period o f 10years. Other interventions include: Eliminate multiple retirement conditions and benefit formulas and target a uniform rule for all workers, inboththe SSO and the CSRO. Target a retirement age o f 63-65 years for both males and females2 with no maximum retirement age. The new age could be reached ina period o f 6-10 years for males and 16- 20 years for females. Afterward, the retirement age would be indexed with life expectancy. Allow for early retirement with actuarially fair reductions in replacement rates and eliminate special incentives inthe CSRO as well as current lump-sumpayments. Eliminate incentives to strategically manipulate wages, reduce heterogeneity in rates o f return, and improve equity by each fiscal year including an additional year o f salary in the benefit formula. Wages included inthe calculation o f the pension are indexedby the growth rate o f system-average wage. * This intervention takes into account that, today, a male worker retiring at age 60 can expect to leave untilage 76 and that a female retiring at age 55 can expect to live untilage 77. iv 0 Reduce incentives for retirement over work, by reducing the minimumpension to at least 70-80 percent o f the minimumwage. The minimumpension should be granted only to individuals who have contributed for a period o f 30 years. 0 Modify current articles governing the indexation o f pensions in such a way that the responsibility to implement the adjustment i s transferred directly to the department in charge o f processingpayments. Use consumer prices as the index. Thispolicy, however, should only be implemented after the other reforms have been adopted. Reform Type 11: Downsize the current DB system while giving a more prominent role to voluntary savings. This reform considers that the current DB-PAYG system will not be the only system replacing income for retirement. Therefore, it proposes reducing the size o f the mandate to save (i.e., the contribution rate), reducing the targeted replacement rate, and also imposing a maximum covered wage o f three times the economy-wide average. This would generate patterns o f income replacement similar to those observed in OECD countries. The targeted replacement rate for the average full-career worker (40 years) could be set initially at 60 percent, implying an accrual rate o f 1.5 percent per year. This reform would not affect current retirees or workers close to retirement. The implementation o f the reform couldproceedas follows: New CSRO and SSO workers would enter a new DB scheme (that could have some degree o f pre-funding) where the contribution rate i s lower (below 15 percent), where benefits are lower (accrual rate i s set at 1.5 percent), retirement conditions are tighter (a minimum retirement age of 63 years indexed with life expectancy is enforced), and pensions are computed on the basis o f lifetime earnings. Wages are indexed by the growth rate o f the system average wage, and pensions are indexed by prices. The system guarantees a minimumpension o f 70-80percent o f the minimumwage. For those workers who are in the current system, benefits and retirement conditions are adjusted according to the recommendations o f ReformType I. In parallel, necessary incentives and an appropriate regulatory and supervisory framework are put in place to promote voluntary savings in the form o f contractual savings. Eventually, a mandated defined contribution (DC) fully funded (FF) pillar could be introduced, while the DB-PAYG system is downsized further. Reform Type111: Downsize the current DB-PA YGsystem while introducing Notional Dejlned Contributions (NDC) andpromoting voluntary savings. This reform has the potential to reduce the contingent liabilities intrinsic to a traditional PAYG system, by legally enforcing that the rate o f return paid on contributions approximates the sustainable rate o f return. Although the system remains unfunded, it could have the following advantages: allows for an easy and rapid harmonization o f retirement schemes across professions; allows for more fundamental reforms o f disability and survivor benefits; eliminates perverse redistributive features o f the traditional PAYG system; reduces distortions in individual labor supply and retirement decisions; and generates a better distribution o f risks between current and future generations. The reform could proceed as follows: 0 New workers enter a new system where the total contribution rate i s lower (below 15 percent), thus allowing for a better balance between voluntary savings and mandatory savings. 0 Contributions are accumulatedin "virtual" individual accounts, earning a notional interest rate that i s a function of the growth rate o f covered wages. Stabilization mechanisms are V also incorporated to reduce volatility in rates o f return and to ensure that the system liabilities do not surpass the system's assets. At retirement, the accumulated capital in the virtual accounts i s transformed into a lifetime pension on the basis o f estimates about life expectancy. 0 Current contributors are allowed to switch to the new system. For them, an initial capital i s accredited to the "virtual" account, which i s calculated on the basis o f past contributions plus notional interest. 0 Workers who remain in the current system face the type o f adjustments described in the case o f ReformType I. 0 Inparallel, necessary incentives and an appropriateregulatory framework are put inplace to promote voluntary savings inthe form o f contractual savings. These three types o f reform would considerably enhance the financial situation o f the funds while improving incentives. Each has pros and cons that would be amplified or dampened by the particularities o f the Iranian economic and social systems. The first reform could be easier to implement at the political level, but it maintains highcontribution rates that may not be affordable by the Iranian economy and that would continue to distort labor markets. In addition, it leaves the finances o f the system vulnerable to changes in macroeconomic conditions; it does not fully resolve distributional problems; and it does not allow for the development o f other forms o f saving. Thus, financial risks remain concentrated in public hands and hture generations. The second type o f reform i s subject to the same criticisms, except that it does create room for other forms o f savings and does reduce labor market distortions. Finally the third reform, while having the potential to reduce risk, distributional problems, and improve incentives to enroll and contribute to the system, could fail if choices in terms o f rates o f return on notional accounts and formulas to compute pensions are not technically sound. This reform may also be more demanding in terms o f institutional capacity. Improving the Management of FundReserves A review o f international experiences shows that, in general, public pension funds have failed to generate appropriate rates o f return on investments, in part, given weak governance structures. The report discusses best practices to improve governance, which involve reviewing how duties and obligations are specified; how the governing body i s structured and selected; how the management o f the pension fund i s structured and selected; and how accountability i s enforced. Policy options are also given for refinancing the public debt and reducing risks in current investment policies. The main recommendations can be summarized as follows: Governing Body. To improve the effectiveness o f current governing bodies, it i s necessary to clarify responsibilities and objectives, as well as to modify their composition and their selection process. In particular: substitute direct government appointments to the High Councils by recommendations from a high-level Selection Committee with representatives from civil society; focus the mandate o f the High Councils on managing the pension funds in the interest o f plan members, while eliminating other objectives such as pursuing social or economic development policies; give the High Councils the freedom to select and remove the Managing Director o f the pension fund(s) and to decide on compensation modalities; lastly, separate governance responsibilities from management/executive responsibilities. Public Debt. It i s recommended to eliminate the current practice o f transferring public companies as payments on the public debt. Instead, external audits should be conducted to establish the correct value o f current government arrears. Then, a plan to collect these arrears at fair value should be developed, which i s based on the use o f government bonds and revenues vi from privatization. This implies that the restructuring and selling o f public enterprises would be carried out by specialized institutions within the government and not by the pension funds. Investment Policies. The report advises against investing fund resources in the restructuring o f companies that have already been transferred. Rather, it i s recommended to find strategic investors with controlling ownership. Other recommendations to improve investment policies include: adopting a program to reduce controlling stakes in corporations by holding minority participation; encouraging the participation o f other fund managers in addition to the funds' own investment companies (Shasta in the case o f the SSO and the Investment Company inthe case o f the CSRO); adopting exposure limits to ensure maximum diversification on investments in shares (a maximum o f 5 percent o f the capital o f any company should be owned by the fund, and a maximum o f 5 percent o f fund assets should be invested in any company); allowing for investments outside the country; and eliminating taxes on investmentincome. Accountability. To improve accountability it is recommended that a mark-to-market valuation o f current assets i s conducted by an independent auditing company. In addition, it i s necessary to improve current reporting mechanisms and make Annual Reports available to plan members; publish the balance sheets of the companies owned by the pension funds; introduce periodic external audits (if possible by international consulting firms); and make use of external custodians. Expanding Coverage In the case of Iran, reducing the contribution rate and implementing a credible reform program that ensures a better balance between mandatory and voluntary schemes could contribute to expand coverage. If a DB system i s preserved, computing pensions on the basis o f full-career salaries will also reduce incentives to evade. Structural reforms leading to faster economic growth and better employment opportunities in the formal private sector will contribute too. Nonetheless, neither a reformed pension system nor a more dynamic economy i s likely to suffice, at least over the medium-term. This implies that social assistance programs targeted to the elderly will remain an important component of the government strategy to ensure an adequate level o f income during old age. The following recommendations are made: 0 Conduct appropriate surveys to estimate the coverage gap and its causes. The objective o f this activity i s to identify population groups not covered by the system as well as their geographic and socioeconomic characteristics. 0 Conduct a review o f current social assistance programs for the elderly-including estimates o f costs, benefits, number of beneficiaries, and their socioeconomic characteristics. The review should assess management, targeting, and monitoring mechanisms and present recommendations in terms o f the need to expand or eliminate existing programs and/or design additional ones. 0 Consider the introduction o f a demogrant. The demogrant i s a special form o f cash transfer that i s not limited to the elderly poor, but to all elderly. While in the case o f the nonpoor the transfer brings negligible benefits, for the elderly poor it can represent a sizable share o f total income. Although the grant i s universal and, therefore, does not require the establishment o f targeting mechanisms, administrative issues should not be underestimated, particularly in a large country like Iran. On the other hand, no funds needto be invested, and estimating the present and future costs o f the grant i s a relatively straightforward task. Today, providing USD 200 per year (roughly 20 percent o f the minimum wage) to the entire populace above age 65 (2.9 million individuals in year 2001) would cost between 0.4 percent and 0.7 percent o f GDP. vii Promoting Voluntary Savings An appropriate balance between the mandatory and voluntary components of apension system is an importantfeature of its design. Among voluntary savings schemes, contractual savings (CSs) are promising alternatives. Contractual savings are savings accounts created to promote long- term savings and manage social risks. These savings can be usedto finance funded pension plans (accumulation period), annuities (payout period), life insurance, unemployment, and other contingencies. Thus, they constitute an instrument to improve the management o f social risks. Contractual savings can have six important effects on financial markets: increase depth and liquidity by increasing the demand for shares and bonds, market capitalization, and volume traded; increase the demand for long-term bonds and the supply o f long-term loans; create incentives to improve regulations and transparency; foster financial innovation, competition, and efficiency; improve corporate govemance; and, contribute to the reduction o f financial risks and, therefore, o f output volatility. The report recommends that over the medium term the government put in place the necessary regulatory and supervisory infrastructure to stimulate the development of contractual savings in Iran. A variety o f international experiences can guide this strategy. The development o f CSs could become part o f the agenda that the government i s currently putting in place in the area o f private sector development and financial sector reform. An important element that will need to be evaluated i s the tax treatment o f different types o f savings. To this end, the necessary studies should be conducted. C. NEXT STEPS The requiredfollow-up activities to move the reformprocess forward are summarized inTable I. First,it is necessary that the government create the body that will be responsiblefor managingthe reform process, the Pension Reform Commission. One of the first activities o f the Commission will be to disseminate the current report within the government and among representatives o f civil society. To this end, various seminars and workshops can be organized. On the basis o f these discussions, the Pension Reform Commission will need to prepare a White Paper that presents the key elements o f the government's pension reform strategy. A detailed multi-year reform program can then be prepared and executed. Pension reform i s taking place at a time when unemployment rates remain highand the impact o f ongoing structural reform on faster growth and higher labor productivity still needs to materialize. Some have argued that in this context increasing retirement ages could be counterproductive. Or that reducing replacement rates will imply an even lower purchasing power for future retirees. There are also concerns regarding the readiness o f the private sector to provide supplementary retirement plans. All o f these are valid points. It i s important, however, that policymakers and civil society recognize the following: 0 Different problems require different policy instruments. Keeping retirement ages at current level will not be enough to solve the unemployment problem and it will only compromise the financial sustainability o f the pension system. Reducing unemployment rates in a sustainable way requires other policy instrumentshtrategies to promote investment and growth and reduce distortions in labor markets. Appropriate social protection instruments, such as well-designed and monitored active and passive labor market programs and a solvent unemployment insurance system, should accompany these. Pension reform should thus be part o f a larger program o f structural reforms. Interventions in the various sectors should be implemented in sync. That is why the recommendation i s to implement reductions inthe retirement age ina phased manner. ... Vlll 0 The level o f the pensions that the contributory systemcan afford i s ultimately constrained by economic and labor productivity growth. Ifcurrent benefits are not gradually brought down to sustainable and affordable levels, future generations will be penalized either through abrupt cuts in benefits or higher taxation. The reforms being proposed will not affect current retirees or those who are close to retirement. A minimum pension will continue to be guaranteed. Also, the gradual implementation o f the reform should allow the economy to catch-up. Newcomers while facing lower accrual rates (and probably lower contribution rates) will also face better economic prospects. Moreover, it i s important to recognize that the contributory system regroups relatively prosperous individuals. Indeed, the poorest and more vulnerable population groups in Iran cannot afford to enter the contributory scheme. Thus, public resources allocated to the contributory scheme to cover unfunded liabilities are resources implicitly taken away from programs to assist the poor and vulnerable. This is an important source of inequality. 0 Itis clear that the private sector cannot start providingvoluntary complementary pensions overnight. Several activities will need to be conducted to set up the necessary regulatory and supervisory framework and to assist the industryin developing new products. These activities should all be part o f the medium-term reform program (see Table 1). Changes in the mandate of the public pension system should thus be timed with reforms in the private sector. Table I:Next Steps in the Reform Process Activity Description Unit Responsible and SuggestedTime Frame Creationof the The role of the Pension Reform Commission is to act as MPO, by August Pension coordinator and manger of the reform process. Hence, it is the 2003. Reform Pension Reform Commission that i s given the responsibility of Commission. studying, consulting, and proposing a reform program to the govemment. Dissemination The current report identifying problemsand options for reform will Pension Reform of the current be discussed within the govemment and among representatives of Commission, by report. civil society to create consensus on a reform strategy. To this end September 2003. various seminarsand workshops will be organized. PrepareWhite On the basis of various discussions the Pension Reform Pension Reform Paperon Commission will prepare a White Paper describing the key Commission, by pension components of the reform strategy that better reflect the December2003. reform. preferencesof the Iranian society. Preparemulti- Once a final reform strategyicourseof action has been established, Pension Reform year reform the various activities necessaryto implement the strategy and their Commission, by program. distribution over time can be outlined. A first set of activities i s February2003. related to the preparation of additional studies to define the final structure of the pension system, the level of different parameters, and the most appropriate transition mechanism. A second set of activities concentrates on the preparationof the new legislation. In parallel, it is necessary to develop activities to reinforce the institutional capacity of the pension funds (e.g., training, review of management and information systems). If the new legislation is approved, then the final implementation phase can be initiated. Source: Mission signon the basis o fdiscussions with govemment officials. ix 1. BACKGROUND The Government o f the Islamic Republic o f Iran, through the Management and Planning Organization (MPO), has engaged in a medium-term program to assess the effectiveness and efficiency of its social protection system and to introduce necessary reforms. As part of this program, MPO has requested Bank technical assistance to evaluate social assistance programs, active labor market programs, and insurance programs. The present report concentrates on the Iranian pension system and has two major objectives. First, to identify the strengths and weaknesses of the system, particularly with respect to financing mechanisms, incentives, institutional capacity, management, and outreach. Second, to propose strategic lines for eventual policy interventions that could strengthen the system. The bulk o f the analysis concentrates on the Social Security Organization (SSO) and the Civil Servants Retirement Organization (CSRO), the two most important pension funds in Iran. Given data constraints, the various occupational funds and the non-contributory schemes are only surveyedbriefly. The report has beenjointly preparedwith the Government o f the Islamic Republic o f Iran. It i s the result of an extensive analysis o f demographic, economic, and financial data prepared by the different technical departments and a series o f discussions and seminars held with concerned authorities and technical staff during five field visits: January 11-24, April 20-30, and June 21-July 4,2002; March 7-13 and June 9-13,2003. The report i s comprised of eight sections. Section 2 provides a general overview o f the Iranian pension system and summarizes major challenges and opportunities. Sections 3 and 4 present detailed assessments of the Social Security Organization (SSO) and the Civil Service Retirement Organization (CSRO). Section 5 briefly discusses the tax treatment o f retirement savings. Section 6 proposes a framework for guiding pension reform, presents a typology o f pension mechanisms, and reviews international experiences. On the basis o f this framework, Section 7 outlines strategic directions for reforming the Iranian pension system and analyzes their financial and fiscal implications. Finally, Section 8 discusses the political economy o f pension reform and recommends necessary steps for the design and implementation o f a successful reformprogram. 1 2. GENERAL OVERVIEW OF THE IRANIANPENSIONSYSTEM The Iranian Constitution mandates the government to protect all elderly; Iran has thus developed an extensive pension system that i s composed o f contributory and non-contributory schemes, which together cover 50 percent o f the labor force and close to 60 percent of the elderly population. Contributory systems are defined-benefit with pay-as-you-go finan~ing.~ The Social Security Organization (SSO) is the largest fund, mainly covering workers in the formal private sector and workers retaining government contracts (see Figure 1). It currently has 6 million contributors and 1.14 million individuals receiving old-age, disability, or survivor pensions. The Civil Service Retirement Organization (CSRO) covers approximately 1.5 millioncivil servants and pays benefits to 662,786 pensioners. Inaddition to these funds, close to 18 public companies in sectors such as telecommunications, transports, oil, steel, copper, finances-including the Central Bank-and the military have put inplace defined-benefit systems for their employees. It i s not clear at this stage how many employees are covered through these schemes, but unofficial estimates put the figure at around 1.8-2 million. Hence, contributory regimes would cover close to 50 percent o f the labor force, among the highest in the MENA region (see Figure 2). Finally, there are several non-contributory schemes. The largest i s the Shahid Rajaee program managed by the Emam Khomeini Relief Committee, which supports 1.5 million elderly, or 40 percent of the elderly population. These non-contributory regimes have been crucial inexpanding coverage among low-income population groups. Figure 1: Iranian Pension System I -- I I I 1 Contributory Non-Contributory DefinedBenefits 1 1 1 Social WelfareOrganization SocialSecurity Civil Servants Occupational EmamKhomeiniRelief Organization Retirement Committee Organization IslamicRevolutionMartyr's Foundation 6 Million PrivateSector Workers 2 Million Civil Servants Low-IncomePensioners 1.14 MillionPensioners 662,786 Pensioners I I Source: MPO Statistical Annex, SSO Annual Report, and various interviews. While the funds have some reserves, they are not beingmanagedas partially-fundedschemes. Figure 2: Coverageof ContributoryPensionSystems in Iran and the MENA Region 60% 1 Coverage as Percent of the Labor Force 50% I 40% 30% 20% ~ i 10% o% i Algeria Bahrain Egypt, Iran, Jordan Lebanon Morocco Saudi Tunisia I Arab Islamic Arabia I Rep. Rep Source: World Bank. MNSHD Pension Database The SSO, the CSRO, and the occupationalfunds impose large mandates to save that are likely distorting labor supply and savings decisions. The total contribution rate for pensions ranges from 17 percent in the case o f the Ministry o f Jihad-Sazandegy to 27 percent inthe case o f the Central Insurance. These contribution rates are the highest in the region, only surpassed by Egypt, Kuwait, and the United Arab Emirates (see Figure 3). In the case o f the SSO and the CSRO, total contributions paid represent 1.8 percent o f GDP. The concern i s that this high mandate is crowding-out other forms o f savings that could contribute to the development o f financial markets. In addition, there i s some evidence in the case o f Latin American countries that highcontribution rates to DB-PAYG schemes can reduce the probability o f enrollment and encourage work in the informal sector. Thus, high contribution rates may in fact be reducing the potential revenues o f the pension system. Source: World Bank, MNSHD Pension Database. The individual replacement rates offered by the SSO and the CSR04are high (equal or above 100percent in the case of the SSO) at all levels of income. This contrasts with most OECD countries where the public pension systemfocuses on replacing income at retirement for low- income individuals. As an illustration, Germany replaces 80 percent o f the income o f a full- career worker earning 50 percent o f average earnings (see Table 1). For workers earning 5 times average earnings the replacement rate drops to 30 percent. Similarly, in Canada, the replacement rate for a full-career worker goes from 74 percent to 6 percent. Even in Spain where the system tends to be more generous, the replacement rate drops from 88 percent to No information is availableregardingreplacementrates for occupationalfunds. 4 60 percent. Inthe SSO, with an accrual rate o f 3.3 percent per year and only the last two years used to compute the base salary for the pension, statutory replacement rates for full-career workers are equal to 116 percent. In the CSRO because contributions and the pension are computed on a base o f 75 percent o f total income, effective replacement rates are lower (75 percent). In the SSO the ceiling to compute contributions and benefits is a little over 8 times the average wage; most plan members are not affected. In the CSRO there are no ceilings. Hence, highreplacement rates are received by the quasi-totality o f workers, regardless o f income level. Today, the average replacement rate for the SSO and the CSRO old-age pension (the average old-age pension divided by the average wage) i s 112 percent and 61percent respectively. Table 1: Total Mandatory Pension Benefits as Percent of Individual Earnings in High-Income Countries I- Individual Earnings as a Proportion of Economy- WideAverage 50 Percent of S Times Average Average Average Iran SSO 133 99 99 Iran the CSRO 133 75 75 Australia 71 43 31 Canada 74 43 6 Finland 50 38 13 France 77 60 60 Germany 80 72 30 Italy 58 58 42 Japan 72 53 15 Korea 108 78 54 Netherlands 70 70 70 Norway 62 52 15 Spain 88 88 30 Sweden 93 69 55 Switzerland 63 58 14 United Kingdom 51 35 9 United States 57 45 15 Source: Whitehouse (2001). Numbers for Iran are mission calculations. Benefit formulas and eligibility conditions provide incentives for underdeclaration, evasion, and retirement over work, and may penalize low-income individuals. Statutory retirement ages ranging between no minimum (males with 30 contribution years or females with 20 contribution years inthe CSRO) and 60 years (males with 10 contribution years inthe SSO or females with no minimumnumber o f years inthe CSRO) are among the lowest inthe region (see Table 2). These retirement ages seem unsustainable taking into account that the life expectancies for males i s 76 at 60, and for females 77 at 55. Individuals enrolled in the SSO system have incentives to evade the system after 10 contribution years and return to the system two years prior to retirement. As discussed in the next section, this strategy considerably increases the rates o f return. In the CSRO, the government has created special rules as a mechanism to reduce the size o f the public administration. Inaddition, the fact that the pension i s computed only on the basis o f the last two years o f wage history penalizes blue-collar workers (low-income workers). Indeed, for these workers the peak in earnings relative to average earnings in economy tends to occur in mid-career. Moreover, this formula encourages the underdeclaration o f wages in early and mid-career. Finally, generous minimumpensions (equal to the minimumwage) provide incentives for retirement over work. 5 Table 2: Vesting Periods,AccrualRates, and RetirementAges in MENA Countries VestingPeriod Minimum Accrual Rate RetirementAge (Percent) Reference Wage Algeria ~ CNR 15-M 10-F 60-M, 55-F 2.50% AMW last 5 yrs CASNOS 15 60-M, 55-F 2.50% AMW last 5 yrs Djibouti OPS 15 55-M, 50-F 2.00% AMW last 10 yrs CNWFunctionaries 25 50 3% (first 10yrs) last monthly wage 2% (after I O yrs) CNR/Po1ice 26 40 3% (first 10 yrs) last monthly wage 2% (after I O yrs) CNRiMinisters 10 55 3% (first 10 yrs) last monthly wage C M R 15 __ 3% last monthly wage Egypt Law 79/1975 (Main Scheme, Basic Pension) I O 60 2.22% AMW last 2 yrs Law 79/1975 (Main Scheme, Variable Pension) 2.22% all years Law 108/1976 (Self Employed) 10 65 2.22% all years Law 50/1978 (Egyptians working abroad) I O 60 2.22% all years Law 112/1980 (Casual employment) 10 65 flat contribution all years Iran CSRO 25-M, 20-F 50-M, none-F 3.30% AMW last 2 yrs (04 0 60 (or) 30 none I sso I(or) 10 60-M, 55-F 3.30% AMW last 2 yrs 30-M, 25-F 50-M, 42-F I I 2%.(yrS >IO) 5 years RNS 10 65 3% (yrs <=lo) AMwlastf3 Out O O f 2% (yrs >IO) last 5 years RTTE 10 65 3% (yrs <=IO) Yemen Private Sector Scheme 15 60-M, 55-F 2.50% AMW last 2 yrs Civil Service and Public Enterprises Scheme 15 60-M, 55-F 2.86% AMW last year West Bank and Gaza West Bank 15 60 2.00% basic salary (y before 1987) (or) 40 none last basic salary and allowances (years after 1987) Gaza Pension and Insurance COT. 15 60 2.50% last monthly wage (Civil servants) (or) 20 none Security Forces 15 60 5.33% basic wage (or) 20 none (ifyrs <15) 6 The rates of return offered by the SSO and the CSRO seem average when compared to other MENA and OECD countries. The analysis reveals, however, that large intragenerational transfers are taking place. Comparisons o f rates o f return at the international level are complicated by the many differences inthe benefit formula, eligibility conditions, and existence o f minimum and maximum pensions. When only looking at statutory contribution rates and accrual rates in a sample o f OECD and MENA countries (see Figure 4), rates o f return for an individual entering the system at age 30 and retiring at age 60 with a growth rate in wages o f 2 percent per year, range between 1 percent (Egypt) and 9 percent (Tunisia). In Iran, the internal rate o f return (IRR) for this individual i s close to 5 percent in the SSO (similar to Germany) and 4 percent in the CSRO (similar to Spain). Nonetheless, there i s high heterogeneity in rates o f return. First, rates o f return are higher for individuals with fast- growing wages and who live longer (this i s a feature o f defined-benefit systems). Second, rates o f return are higher (over 6 percent), for individuals entering the system later inlife. Finally, in the case o f the SSO, rates o f return for individuals who contribute for 10 years early in life, leave the system, and return to the system two years prior to retirement can be above 10percent per year. This heterogeneity inthe rates of return gives room to intragenerational transfers that do not always favor low-income workers. Figure 4: Rates of Return in OECD and Selected MENA Countries I I 1 I 1 10% 1 9% - ~ 8% ~ 7% - 6% - 5% - ~ Source: Actuaria (2002). Note: For comparison purposes calculations are limited to an individual who enters the system at age 30 and retires at age 60. Only old-age pension benefits are considered and only statutory contribution rates and accrual rates are taken into account. No early retirement is allowed. Life expectancies used inthe calculations are country-specific. Total pension expenditures through the SSO and the CSRO have doubled as a share of GDP within the last 7 years and now stand at 1.5percent of GDP. This level o f expenditures is lower than that observed in countries o f similar demographic structure like Morocco or Algeria (see Table 3 - military expenditures are not included for these three countries). In part, this i s explained by a low average-covered wage. The high growth in expenditures i s linked to the increase inthe number o f beneficiaries, who grew at an average o f 9 percent per year duringthe last decade inboth the SSO and the CSRO.' Inthe case o f the SSO, the financial impacts were neutralized, inpart, by a fast expansion in the number o f contributors between 1987 and 1995 (an average increase o f 10 percent per year). This expansion followed the end o f the war with Iraq and can be partially explained by highrates o f economic growth duringthe First Five-Year Development Plan and by extending the system to voluntary contributors. I t i s unclear whether In SSO growth rates were higher between 1986 and 1996 (12 percent average per year). The reasons are unclear at this stage. Inthe CSRO high growthrates are explained inpart by the provision of incentives for early retirement to reduce the size of the civil service. 7 other structural changes within the SSO andor the economy-at-large took place during that time. Regardless, growth rates for the total number of contributors have now slowed down. Table 3: Pension Expenditures and Determinants in Selected MENA Countries Old-Age, Disability, and Old-Age Only Survivor Year Pension Pension Average Share of the Average Share of the Expenditure Expenditure Replacement 60+ Years Covered Population (Share of (Share of Rate Olds Wageover Older than 60 GDP) GDP) Covered by GDPper the Svstem Cavita ~ Algeria 1999 2.84% 5.90% Kingdom of Bahrain' 2000 1.39% 4.90% Islamic Republic of Iran 2000 1.12% 0.74% 98.97% 19.85% 0.65 6.50% Kingdom of Jordan' 2001 5.72% 4.51% Kingdom of Morocco 1998 2.05% 6.70% Tunisia 2001 3.31% 2.66% 67.29% 30.46% 1.73 7.50% Djiboutia 2000 3.26% 2.67% 51.09% 14.40% 6.97 5.20% Saudi Arabiab 1998 0.19% 4.50% EgyptC 2001 2.69% 6.10% Yemen' 1999 0.91% 3.80% Lebanon' 1999 3.89% 8.30% West Bank and Gaza 2000 0.88% 4.59% Source: World Bank, MNSHD PensionDatabase. Note: (a) Includes the Military Pension System. (b) No information on the civil and military sector is available. Data contains information on the private sector only. (c) Pension expenditure data for Egypt regard only pension schemes regulated by Law 79/1975 (main scheme), Law 10811976 (self-employed scheme), Law 50/1978 (scheme for Egyptians working abroad). Pension expenditure regulated by Law 11211980 follows a contributoy scheme, which is not directly comparable with those inspiringpension funds inthe rest ofthe MENA region. As a result of the "baby boom" that Iran experienced during the mid-80s, there ispotentialfor a rapid increase in the number of contributors over the medium term, though rising unemployment rates are likely to dampen the effect. The Iranian population i s essentially young. Fertility rates picked up duringthe early 80s. Hence, the 1996 census displays a large proportion o f the population to be 8-12 years old. These individuals will start to enter the labor force in the next five years. If at the same time the trend o f increasing female participation rates continues, the labor force could grow at over 4-5 percent for the next 10 years. If a majority of these individuals were to work in the formal sector, the finances o f the SSO could be improved considerably over the short and medium terms6. Unfortunately, with expected GDP growth rates of 6 percent' and labor and capital productivity growth rates of 2-3 percent, total employment i s forecasted to grow at less than 4 percent per year. To keep the share o f the labor force covered by the system constant or growing, coverage rates o f the employed population will need to increase. Given the incentive problems discussed above, this will not necessarily be the case. Even if an optimistic stance were taken-with total beneficiaries growing at 9 percent per year-the system's dependency ratio would rapidly deteriorate. Under the current structure of contributions andbenefits eachindividual is a liability to the system. Expanding coverage wouldincrease the financinggap over the long-run. World Bank, Country Office Estimate. 8 Looking forward, both the SSO and the CSRO are on an unsustainable path with unfunded pension liabilities for the period 2002-2070 reaching 140 percent and 35 percent of GDP respectively. Intwo to five years, an operational deficit couldbe observed inthe CSRO, and an operational deficit in the SSO could be observed within the next 10 years. Reacting to this imbalance in the future through bailouts and cuts in benefits will impose an adverse intergenerationaltransfer from tomorrow's poor to today's relatively well-off workers. While very scant information is currently available regarding the structure, size, andfinances of occupationalfunds, it is likely that the majority is accumulating unfunded pension liabilities. The occupational funds seem to be heterogeneous in terms o f size and benefits offered.* Actuaries from the SSO Research Institute consider that while a few o f the funds are properly managed and seem to be financially sustainable, the majority i s likely to generate an operational deficit within the next few years. Since the funds are linkedto public companies, their implicit pension debt i s part o f the contingent liabilities of the government. Another issue o f concern i s the fact that intransferring from one fund to another or from any fundto the SSO, the current rules are complex, give room to discretion, are not always fair from the workers point o f view, and thus constrain labor mobility. Non-contributory schemesfor the elderly are well developed in Iran relative to other countries in the region, but there are concerns in terms of the level of eflciency of these schemes and whether targeting mechanisms are appropriate. Non-contributory regimes ought to concentrate on the elderly poor. The fact that these programs cover 40 percent o f the elderly population raises questions about targeting. Indeed, it may be the case that the programs are reaching individuals who could have been part o f the contributory regimes, or individuals who are indeed also receiving benefits from a contributory scheme. On the other hand, it is possible that individuals who should be covered by non-contributory schemes are being left out. Inthe absence o f data about the characteristics o f the beneficiaries, it i s not possible to conduct a rigorous evaluation o f these programs. This i s an area where more work is required. While the SSO and the CSRO still own reserves amounting to 5 percent of GDP, these are mostly invested infixed assets and government debt, generating low or negative rates of return. The funds were designed with some degree o f pre-funding. However, no funding targets are defined explicitly. Inpractice, the funds are managed as pay-as-you-go systems with a stock o f savings that i s used to smooth adjustments to contributions and benefits. Today's portfolios are illiquid and risky. Nonetheless, even if higher rates o f return could be achieved, current reserves would not be sufficient to cover pension liabilities over the next decade. Thus, maintaining the current level o f benefits implies accumulating a debt that future generations will have to finance, either by diverting resources from other sectors (e.g., education and health) or by increasing the tax burden. the SSO and the CSRO have over-expanded their mandate by directly managing companies that operate in most economic sectors and lately have become de facto agencies to restructure public enterprises. This gives the pension funds excessive influence in financial markets and corporate governance. The strategy o f transferring public companies as payments on government arrears imposes unnecessary risks on worker savings. Looking forward, important challenges remain to improve the management o f these savings. For instance, improving the process for structuring goveming bodies; creating appropriate incentives for managers; and increasing accountability-including better disclosure, auditing, and custody. Fiduciary responsibilities also need to be reviewed to ensure that the pension fund i s managed inthe best interest o f the plan members. These are preconditions for increasing rates o f return on fund reserves. * Meeting with representatives fromthe OccupationalFunds. MPO. January 2002. 9 Although the problems outlined are complex, policymakers have a unique opportunity to initiate reforms. First, the pension funds are not yet facing an operational deficit and demographics remain favorable, allowing for a larger margin o f maneuver. Second, the government has improved economic management, leading to a more stable macroeconomic environment. Third, Iran i s initiating structural reforms in the financial sector, particularly the banking ~ y s t e m . ~In addition, the government has made considerable progress in the privatization o f insurance companies. In the last two years, eight insurance companies were privatized. The government i s also creating the conditions for the development o f private auditing companies. These reforms are likely to reduce the risks facing pension funds, to improve accountability and to expand investment opportunities. At the same time, the reforms open the way for the development o f pensionplans that complement the public pension system. Finally, the rationalization o f the system o f indirect subsidies i s expected to liberate considerable public resources that could be used to restructure the current government debt with the pension funds and absorb transition costs. Looking forward, policymakers need to create awareness among the population regarding the problems facing the pension system while generating commitment for reform. This requires framing discussions within a long-term horizon to make explicit the tradeoffs between the potential social and economic costs o f today's policies with the future benefits. The analysis presented inthis report will, it i s expected, contribute to the discussions. See Fetini (2002a, 2002b, and 2002~). 10 3. THESOCIAL SECURITY ORGANIZATION (SSO) 3.1. Institutional Issues The Social Security Organization was established in I952 to provide pensions, unemployment insurance, and health insurance to workers in the private sector-including the self-employed and voluntary contributors-and to contractual workersfrom thepublic sector. The SSO i s an autonomous institution attached to the Ministry o f Hygiene, Health, and Medical Education. Today the binding law for the institution i s the 1975 Social Security Law, which has been amended several times since 1994." The Social Security Organization is governed by the High Council of Social Security, which is constituted by 7 government representatives, 5 employer representatives, and 3 employee representatives. The Minister o f Hygiene, Health, and Medical Education chairs the council. The Social Security Organization is administered by a Board of Directors, o f which three members are selected on the basis o f recommendations from the Minister o f Health. The Managing Director o f the SSO i s the Chair o f the Board. A Supervising Committee monitors activities and comments on balance sheets before they are referred to the High Council. The Supervising Committee i s composed o f a government representative who i s proposed by the Ministry of Economic Affairs and Finance, an employer representative, and a employee representative (see Figure5). The SSO is a large organization employing 30,533 employees distributed between its headquarters in Teheran, 29 regional ofices, and 364 local branches. Eighty percent of employees are full-time and the majority, or 83 percent, works in the regional offices. Their distribution by region i s closely correlated with the size o f the population. Employees are allocated half-and-half between the health branch and the insurance branch. I n the case of thepension branch, expenditures in administration are more or less in-line with international standards. Wages paid amount to Rhials 538,400 billion (USD 71 million) or 79 percent o f operational expenditures. In-line with levels observed in other countries, these expenditures represent 6 percent o f total benefits paid. The SSO is a pension fund where institutional capacity can be characterized as average; considerable challenges remain to upgrade management and information systems. At present, it is difficult to generate data regarding the finances o f the funds, investments, or simply demographic and economic data regarding contributors and beneficiaries. While information technologies are widely available, these do not operate in the context o f an integrated information system. Electronic mail i s still not used routinely and much o f the exchange o f information continues to take place through traditional means (mostly paper). Furthermore, there i s highheterogeneity inthe level o f access that the different regional offices have to these technologies. While the SSO i s working on the implementation o f a new MIS, through one o f its own IT companies, the proposed design has serious flaws. First o f fall, the software and operating systems being used are not appropriate for this type o f MIS. Second, the module to track contributors and payments to the pension plan i s not being designed as an integrated system. This implies that each local branch will have its own system. Information can be routinely transferred to the Center, yet the Center has no easy mechanism to merge the databases. A t this stage, the information that i s transferred from the local branches i s not disaggregated by individual but only by employer. In the case o f beneficiaries, local branches receive requests to process pension liquidations and compute the value o f the pension. The request to pay the claim i s then sent to the Center. The Center's database o f current ~ lo For a briefhistory see SSO (2000). 11 beneficiaries i s compiled with the information from the branches. The central office, however, cannot verify the calculations since individualrecords for the contributors are not available. Figure 5: Organizational Chart of the SSO High Council I I Supervision Committee Board of Directors I I I Managing Director Public Relations Office I I I I I I Assistance Office Provincial Offices I I Source: SSO. 3.2. Contributions andLabor Force Coverage The total payroll contribution to the SSO is set at 33 percent to finance cash benefits-including unemployment, pensions, and health services. It is divided between employers (23 percent), the government (3 percent), and workers (7 percent). There i s a maximum contribution of Rhials 2.5 million per month, implying a maximum covered wage o f Rhials 7.5 million per month (USD 1,000 per month or 8 times the minimum wage"). No contribution can be paid on a wage that i s below the minimumwage set at Rhials 850,000 per month in 2003 (Rhials 650,000 or USD 80 in 2001 or 66 percent o f the average wage. Family allowances are required to be paid directly by the employer, which raises the effective contribution rate above 23 percent. As discussed in the next section, while contributions are formally dividedby type o fbenefit, there are no separated accounts. The Social Security Organization currently covers 753,435 establishments in theprivate sector, 16,867 establishments in the public sector, and voluntary contributors (self-employed, non- economically active individuals, or the unemployed). The majority pays a contribution equivalent to 30 percent o f wages. Some o f the public establishments (1,070, or 6 percent o f the total) operate under contractual arrangements where the SSO reimburses part o f the contributions received inexchange for services that are directly provided by the employer to its workers. Inthe case o f small enterprises in the private sector (9 percent o f the total registered establishments), workers and employers are allowed to pay lower contribution rates, totaling 18 percent (only 2 percent o f workers belong to establishments that benefit from the lower contribution rate). Voluntary contributors also pay an 18 percent contribution rate. The contributionfor pensions and short-term benefits (e.g., cash transfers) is set at 21percent, which is high and may be promoting the underdeclaration of wages and evasion. Indeed, the average wage necessary to generate the level o f total contributions accrued to the SSO in 2001 o f Rhials 13,527 billion (USD 1.8 billion) with a 30 percent payroll contribution (excluding I'In2003 SSO staffreportsamaximum coveredwage of40,000,000 permonth (see Ghorbanali, 2003). 12 contributions from the government) i s estimated at Rhials 7.5 millionper year (USD 1,000). A t the time, however, the average wage o f the covered population was estimated at 1.5 times the minimum wage or Rhials 10.8 million (USD 1,440) per year.12 Given the high ceiling on contributions, the 30 percent difference between the covered and average wage i s most likely explained by the underdeclaration o f wages and moratoria. The incentive to under-declare wages i s particularly strong given that the pension paid i s not linked to lifetime earnings. For instance, consider the case o f an individualretiring after 25 years o f work. Ifhisher pension i s a function o f the total number o f contribution years and the average wage received during the last 2 years o f work, then there i s no incentive to declare the full wage duringthe first 20 years o f work, since those years do not contribute anything to the final pension. Moreover, a payroll contribution acts as a tax on labor that distorts labor supply and demand decisions. Recent research in the case o f Latin American Countries suggests that high payroll contributions to PAYG systems discourage enrollment (see Sections 6.1 and 7.4). Between 1986 and 1997 the total number of contributors grew on average by an impressive 10 percent per year; this exceptional growth rate is expected to be halved during the next decade. The total number o f contributors in the SSO increased from two million in the mid-80s to six million today. This allowed the SSO to maintain a low dependency ratio (7 percent for old age). The expansion mainly involved private sector and voluntary contributors (see Table 4). The causes behind this expansion are unclear. The opening o f the SSO to voluntary contributors in 198713may be part o f the reason. Rapid economic growth duringthe First Five- Year Development Plan following the Iran-Iraq war i s also a plausible explanation. Still, some other structural change i s likely to have taken place either within the SSO or inthe economy-at- large.14 Technical staff at the SSO suggest that the highgrowth rates reflect an underestimation of the number o f contributors in the mid-80s, as a result o f weak information systems. The important conclusion, however, i s that growth rates o f 10 percent per year are not likely to continue. In the future, the mission estimates that the total number o f contributors could grow at best at 5 percent per year.15 A rate higher than the current 2 percent i s possible given the "baby boom" that Iran experienced inthe early 80s'. Individuals born inthe early 80s will soon enter the labor force. At the same time, female participation rates are expected to increase. Hence, the labor force could grow at 5 percent per year. With an estimated GDP growth rate o f 6 percent per year, and factor productivity expected to grow at 2 percent per year, employment could expand at close to 4 percent per year on average. Hence, if the share o f the employed population that i s covered by the system increases, the share o f covered labor force could remain constant. The growth rate would be lower, however, ifthose individuals who joined the system 10years ago leave only to return to the systemclose to retirement. Calculations by SSO's Department o f Economics Insurance and Planning. Takes into account the seasonality o f income for workers in the agricultural sector. l3The mission has not received the specific regulations. l4This is an issue that will needto be explored further. l5 See Technical Appendix for a detailed explanation o f assumptions for demographic, economic, and financial projections. 13 Government Private Sector Voluntary Total Growth Rate 1986 537 1,416 7 1,960 1991 893 2,226 185 3,304 11.01% 1996 943 3,433 744 5,120 9.16% 1997 984 3,837 855 5,676 10.9% 1998 946 3,988 945 5,879 3.6% 1999 927 4,121 926 5,974 1.6% 2000 872 4,263 960 6,095 2.0% 3.3. Benefits and Rates of Return on WorkerSavings The SSO provides a wide array of benefits that can be grouped into three categories: cash assistance and compensation, pensions, and health services. Cash assistance and compensation include maternity, sickness, and family allowances; grants for marriage and for funeral expenses; lump-sumtransfers for physical disability; and unemployment benefits.16 Pension benefits include old-age, disability, and survivor pensions. In addition, the SSO offers outpatient and inpatient health services for the insured and the family o f the insured (see Table 5 for a summary o f the eligibility conditions and the formulas used to compute different benefits). The discussionhere focuses on pension benefits. l6SSO staff have the right to various additional transfer. 14 Table 5: Benefits Providedby the SSO Pensions I Condition Benefit I Financing Old-Age Today, minimum I O years o f For eachcontribution year, the insureris entitledto Financedby 21 percentage contributions.A recentchange in 3.3% of hisiher average wage duringthe lasttwo years points out ofthe 33% legislation graduallywill increase of work. The pensioncannot representmorethan 116% contribution. the vestingperiodto 20 years. The o fthis average. It cannot bebelowthe minimumwage minimumretirementage is 55 years (set at Rhials 600,000 per month, or USD 80, in2001- for females and60 years for males. today the minimumwage is Rhials 850,000 per month). Males with 30 years o f work can retire at 50.. Females who have For individualsinhazardousjobs, each contribution year contributedcontinuouslyfor 20 i s worth 1.5years. years can retire at age 42. Individualswith 35 years in hazardousjobs can retire at any age. Disability Certifiedover 66% disabledby Receives 3.3%of averagewage over the lasttwo years Financedby 21 percentage Total work- MedicalCommittee due to a work for each contributionyear (minimumof 50% and points out ofthe 33% related injury. (No informationabout maximumof loo%). Ifthere aredependentsandthe contribution. recertification ). replacement rate is below 60% allowances are givento attainthe 60%. Totalnon- Same calculation (see Note 2 o fArticle 72 of SSL). work-related I Partial work- Certifiedbetween33%and66% Same calculation,but the resultingpensionis multiplied related disabledby MedicalCommittee due bythe degree o fdisability(see Note2 of Article 72 o f to a work-related injury (no SSL). informationabout recertification). Survivor Dependentsof retirees or insurers Spouse receives 50% of the pension; children 25% each; Financedby 21percentage who have contributedat least one andparents20%. The totalcannotbe above 100%. points out ofthe 33% year duringthe last I O years anda contribution. positive amount duringthe last two For insurersthe pensionis calculatedat the time o f years are allowedto receive a death with the same rules as the normalretirement pension. Dependentso ftotal pension. disabilitypensioners ifcontributed 90 davs durinethe last vear. Cash Transfers Marriage Insureris marrying for the first time, Receivesaveragesalary o fthe last two years. Ifboth Financedby 21percentage i s working, andhas contributed at spouses are insured, each receivesa separate grant. points out o fthe 33% least 720 days duringthe last two contribution. years. Maternity The insuredwomen needsto have 60 Receives213 o fher last wages duringthe 12weeks Financedby 2 1percentage days of contributionsduringthe last precedingandfollowing the delivery for breast-feeding points out o fthe 33% year precedingthe pregnancy. mothers. contribution. Sickness Insuredunder treatment or needs rest Duringthe periodoftreatment an insurerwith Financedby 21percentage as a result o f work-relatedor non- dependentsreceives 75% of salary. An insurerwithout points out ofthe 33% work-related injuries or diseases. dependentsreceives 66%. contribution. Lump-sumfor Certified 10%-33%disabledby the 36 times the applicabledisabilitypension(see disability Financedby 21 percentage physical MedicalCommittee. benefit above) times the percentagedisability. points out ofthe 33% disability contribution. Family Insurerhaspaidcontributionsfor a An allowanceis paidfor eachchildren (upto a Financeddirectly fromthe Allowances minimumof 730 working days. maximum of 2). This allowance is equalto 2-3 times employees(outsidethe 30% the daily wage o f anunskilledworker. contribution). Unemployment Beenunintentionallyunemployedand 55% of the wage. Ifthe individualhas dependentshe/she Financedby 3% points ofthe liable to the UnemploymentInsurance will receive 10%of the min. wage for each(max.4). 33%contribution. Law. Totalbenefits cannot be morethan 80% o fthe wage or less than the min. wage. Benefitspaidupto 36 months for singlesand 50 months for marriedindividuals. Health Services Be insuredwith the SSO. Minimum Insureranddependentsare allowedinpatientand Financedby 9% o fthe 33% contributions not specified. outpatient services. Direct method: insurerand contribution. dependentsuse the SSO facilities for free. Indirect methodinsurer uses private facilities andpaysbetween 10% and25% ofthe costs. Source: SSO (2000), Social SecurityResearchInstitute (1997), andvarious interviews. 15 Today there are close to 1.2 million pensioners in the SSO receiving pension payments equivalent to 5,875 billion or I percent of GDP. Sixty percent of these expenditures are related to old-age pensions, 33 percent to survivor pensions, and only 7 percent to disability. Pensioner growth averaged 7 percent per year during the last decade. The fastest growing group, at an average of 9 percent per year, is old-age pensioners. They represent 34 percent of pensioners while the survivor and disabled pensioners represent 60 percent and 6 percent respectively (see Table 6). Table 6: Evolution of the Total Number of Pensionersin the SSO Year Old-Age Disability Survivor Total 1986 95,228 42,338 258,103 395,669 1991 170,455 55,981 383,316 609,752 1996 310,005 67,541 572,069 949,615 1997 323,192 68,789 592,906 984,887 1998 344,762 69,153 627,135 1,041,050 1999 369,784 69,112 660,567 1,099,463 2000 387,534 67,067 689,421 1,144,022 2001 419,078 69,009 720,540 1,208,627 Latest Estimate' 1,245,000 Average Growth 1996-2001 6.21% 0.43% 4.72% 4.94% Source: Mission calculations on the basis o f MPO (2002) and SSO staff numbers. Nore: (a) Ghorbanali (2003) The system offers high replacement rates across all income levels. Take the case of individuals who enter the system today and contribute for 30 years. If current minimum pensions and ceilings on contributions are assumed to be constant in real terms, then the majority o f individuals, regardless of their income, would receive a replacement rate o f 100 percent (see Figure 6). Only individuals with incomes above the current ceiling of 8 times the average wage (likely to be a handful) would receive lower replacement rates. Individuals withincomesbelow the current minimumwage (66 percent o f average earnings) would receive replacement rates above 100 percent, but legally nobody i s allowed to enroll with a wage below the minimum. If one assumes that the minimumpension and the ceiling grow inrealterms, then individuals with incomes above 2.5 times average earnings would receive lower replacement rates. Figure6: SSO, Economy-Wideand IndividualReplacementRatesby Income Level -2 1.4 Ceiling grows 1.2 .L P .-n l k 0 e e P 0.8 Z 2 0.6 0 '2 8' 0.4 .t e I 0.2 0 0.5 2.5 4.5 6.5 8.5 10.5 0.5 2.5 4.5 6.5 8.5 10.5 Individual earnings (proportion average wage) Individual earnings (proportion average wage) ' Source: Mission calculations. 16 The systems offer a generous minimum pension equal to the minimum wage (Rhials 600,000 per month or USD 80 in 2001), or 66 percent of average earnings. This relatively high minimum pension is likely to create negative incentives to work. Moreover its lax eligibility conditions can create adverse redistributive effects. Take the case o f a male worker entering the system late in life (age 50) with the average wage. Further, assume that all wages grow at the same rate. At age 60 this individual can retire, as he would have fulfilled 10 contribution years. H i s final wage will still be equal to the average wage. With a 33 percent replacement rate, however, his pension would be 50 percent below the minimum (set at 66 percent o f average earnings). This relatively well-off individual will be eligible for the minimumpension, thus receiving considerably higher rates of return than low-income full-career workers. In general, the minimum pension i s likely to primarily benefit workers who have entered the system relatively late in life, regardless o f initial earnings. Full-career workers who experience positive growth rates in wages will tend to retire with pensions above the minimum. As a consequence o f generous minimum pension, SSO appears to be largely a flat system, with an average old-age pension o f Rhials 8.4 million per year (USD 1,121) compared to a minimum pension o f Rhials 7.2 million per year (USD l,000)'7. Disability and survivor benejits appear to be generous as well, but the mission has had only limited information about implementation details. The same accrual rate that applies to old-age pensions applies to disability benefits (see Table 5). However, beyond generosity in benefit formulas, the usual problem in this type o f system i s that benefits are granted without appropriate certification procedures. Disability pensions often substitute for unemployment insurance. For the SSO it i s unclear how the certification process operates and whether recertification is required. In the case o f survivor pensions, there seems to be no limit on the age o f dependent children. The treatment o f divorced spouses i s also unclear. The SSO does not provide formal early retirement but has implemented generous exceptions to the statutory rule. Thus, while men and women are expected to retire respectively at age 60 and 55 with 10 contribution years, in the following cases retirement ages can be lower. Male employees who have contributed for 30 years can apply for retirement at age 50. Female employees who have contributed continuously for 20 years can apply for retirement at age 42 (however, inthis case, the minimumpension does not apply). Workers inhazardous jobs who have worked at least 35 years can retire at any age without penalties. Duringthe late eighties and early nineties, the SSO also retired 91,000 workers before eligibility conditions were met; however, the regulations giving the optiodincentive for early retirement have been eliminated.18 However, the short vesting period and the fact that only the last two years o f salaries are included in the pension calculation reward individuals who leave the system after 10 years and return to the system two years prior to retirement (see calculations on internal rates o f return below). Rulesfor transferring to otherfunds are complex and constrain the mobility of the labor force. When transferring to a different fund (e.g., an occupational fund), the new fund sets the conditions for the transfer. When workers transfer from an occupational fundto the SSO, they are required to pay a lump-sum equal to "late" contributions. "Late" contributions are computed on the basis o f an 18 percent contribution rate applied to the last wage o f the employee. The occupational fund gives the employee back hidher contributions without interest, and the employee i s responsible for completing the required amount. These rules constraint the mobility o f the labor force and reduce incentives to join the system. '*Thesemission l7 numbers are for 2001. The has not received information about the regulation(s) that allowed these retirements. 17 While the legal provisions to protect pensions from increases in the cost of living exist, an automatic indexation mechanism is not in place. In practice, adjustments are subject to discretion by government officials. Data on adjustments were not available at the time of writing this report, but according to the SSO staff, on average, pensions have been allowed to fall in real terms by a few percentage points per year. Hence, plan members are affected by uncertainty interms o f the evolution o f prices. Rates of return for individuals vary widely and depend on wage histories, life expectancies, time of enrollment, and strategies for retirement, giving place to potentially large intergenerational transfers. Figure 7 summarizes these rates o f return. The first two panels present rates o f return for males and females for different combinations o f wage histories (0 percent, 2 percent, and 4 percent growth rates per year) and enrollment ages, under the assumption that retirement takes place as soon as eligibility conditions are met. A feature of DB systems is that rates of return increase with the growth rate of wages and life expectancy. Inthe case of the SSO, growth rates are also higher for workers joining the system late in life (this is particularly true for individuals who benefit from the minimum pension). Women receive higher rates o f return than men. Among men, as compared to those enrolling between ages 25 and 35, men who join before age 25 enjoy higher rates o f return because they have 35 contribution years and are allowed to retire before age 60. Workers receiving the highest rates o f return, however, are those who leave the system after completing 10 contribution years (the minimumvesting period) and return to the system to receive their pensions two years prior to retirement (see the last two panels o f Figure 7). Rates o f return in this case can be higher than 10 percent real and even reach 20 percent real in the case o f workers with fast-growing wages. This analysis suggests that rates o f return within the SSO vary widely across individuals of the same generation. Individuals with low rates o f return are implicitly transferring resources to individuals with high rates o f return. It i s not clear that these transfers favor low-income individuals. As discussed in the case o f the minimumpension, these transfers may operate in the opposite direction. For instance, individuals with fast-growing wages and long life expectancies are likely to be healthy and well-educated workers. 18 Figure 7: Internal Rates of Return in the SSO Females(NormalRetirement) 0 - _ _ gW=O% gW=2% gW-a% z . 5 0 B 8 ! 0 20 25 30 35 40 45 Age when individuol ioin3 lhe system Females(Gaming) - 1 8 " 5 E -- I 1 0 I a 20 25 30 35 ' 0 45 so Age when individuol join3 t h e system Source: Missioncalculations. Note: Calculations are for individuals earning the average wage when enrollinginthe system. These calculations take into account effects of the minimumpension(66 percent o f average wage) andthe maximumcoveredwage (8 times average wage), which are assumed constant in real terms. Life expectancies are calculated at the time of joining the system (i.e., vary by age). The first two panels, normal retirement, assume that individuals retire as soon as eligibility conditions are met (see Table 5). Specialrules for workers operating in "hazardous" sectors are not considered. The last two panels (gaming) assumethat individualsleave the system after 10 contribution years andreturnto the systemtwo years beforeretirement. All pensions are assumedto be indexedby inflation. 3.4. FinancingMechanisms and Sustainability Contributions from workers and employers are the major source of income for the SSO, accountingfor over 90percent of total revenues and representing 3 percent of GDP. Second are revenues resulting from investments; duringthe last decade these accounted, on average, for a mere 7 percent of the total. Other earnings, such as those resulting from late fees are marginal (see Figure 8). 19 Figure8: SSO, Sourcesof Income 1W ? 80% Premium 40% 20% 60% 0% ~ 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 2000 2001 Source: Various interviews. Out of the 33 percent payroll contribution, 9 percentage points are usedformally tofinance health expenditures and 3 percent to finance unemployment insurance, thus leaving 21 percentage points tofinance pensions and cash transfers. Out of the 21 percent for pensions and cash transfers, 18 percent supposedly is allocated to "long-term benefits," which according to the SSO classification include old-age, permanent disability, and survivorship pensions (family allowances are also included among long-term benefits, but these are paid directly by employers). The remaining 3 percent finances cash transfers (short-term expenditures). Expenditures in year 2000-including investment jlows-approximated Rials 15,075 billion (US0 2 billion), or the equivalent of 2.5percent of GDP. The share of investment expenditures has declined over time from 59 percent in 1988 to 10 percent in 2001. The share of health and pension expenditures, on the other hand, has increased considerably from 17 percent to 30 percent and from 25 percent to 60 percent respectively (see Figure 9). Figure9: FunctionalCompositionof Expendituresinthe SSO I 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 2000 2001 Source: Data providedby SSO staff. I n practice, none of the branches has an explicitJinancing mechanism linking contributions to expected expenditures. There is no transparent allocation of total contributions across benefits; neither is there an explicit accounting of cross-subsidies. Little is known about the unit costs of the interventions and the factors that affect their dynamics; therefore, it is difficult to define appropriate financing arrangements. While the 9 percent contribution (out of the 33 percent) allocated to the health branch is at present sufficient to cover expenditures (close to Rhials 3 trillion in2000), it is not clear that current unit costs and levels of spending are efficient. Inthe case of unemployment benefits, the 3 percent contribution rate in 2001 generated revenues in 20 excess o f 16 percent o f expenditures, yet no fund has been established to accumulate the surplus (see Table 7). The pension branch is still displaying a surplus due to still-favorable dependency ratios and high contribution rates. The current average replacement rate for old age, as discussed in the previous section, i s equal to 112 percent; yet, for old-age pensioners the dependency ratio i s only 7 percent. This means that for each retiree there are still 14 workers contributing to the system. Thus, the equilibrium contribution rate for the old-age pension branch would be 7.8 percent (see Table 8) below the current implicit contribution rate o f roughly 10 percent (since old-age pension expenditures represent close to 47 percent o f total expenditures excluding health, unemployment and administration and these expenditures are financed by a 21 percentage points contribution rate ). For the system as a whole, the average replacement rate i s relatively low (64.8 percent), due to a lower average pension for survivors (Rhials 4.8 million, or USD 650 per year), and the dependency ratio i s high (20 percent), leading to a higher equilibrium contribution rate (13 percent); yet, the contribution rate remains well below the current 17 percent implicitly allocated to pensions. Hence, the pension branch i s still generating a surplus that i s equivalent to 37.8 percent of the covered wage bill, giving the illusion o f a healthy financial position. Table 7: Revenues and Expendituresinthe SSO (billions of rhials) Source: Dataprovidedby SSO staff. Note: Numbers havebeenrecently revisedby SSO FinancialDepartmentas follows. Short-term= Rhials 185 billion; Long- term (includingpensions) = Rhials 7,539 billion; Medicalcare = Rhials 4,319 billion; Administrative = Rhials 1,215 billion; Total= Rhials 13,259 billion (see Ghorbanali,2003). 21 Table 8: SSO, Basic Indicators of Financial Sustainability Average Covered Yearly Wage (USD) 1,000 Current. Imdicit. Contribution Rate 17%a Average Yearly Pension(USD) 1,121 Average RedacementRate 112% II DeDendencvRatio 6.98% EauilibriumContribution II 7.83% II Average Yearly Pension(USD) 648 Average RedacementRate 64.8% Dependency Ratio 20.1% EquilibriumContribution 13.1% Source: Missioncalculations. Note: (a): In practice there is nor formal division o f the 21 percent contribution rate for pensionand short-term benefits. Since pension expenditures represent close to 80 percent of total expenditures, excluding health unemploymentandadministration,it is assumedthat 17percentagepointsare beingallocatedto pensions. Over the medium and long term, the SSO's pension branch is unsustainable with unfunded pension liabilities reaching 140percent of today's GDPfor the period 2002-2070. Financial projections were conducted for three scenario^.'^ The scenarios differ in the assumptions about economic growth, total factor productivity growth, labor productivity growth, and whether the system i s open to new entrants or not (see notes to Figure 10). The base-case scenario assumes GDP growth rates o f 6 percent over the next ten years, converging to 3 percent over the long run. The low-case scenario assumes a lower growth rate of 4 percent for the next 10 years. Given assumptions about total factor productivity and labor force participation, unemployment rates and the share o f the labor force coveredby the system are computed endogenously. Inthe case o f the baseline scenario, this share remains more or less constant at 35 percent during the next few years and then increases gradually, reaching 50 percent by the end o f the simulation period. In the low-case scenario, coverage drops slightly-to 33 percent-during the next 10 years but rises to 50 percent by the end o f the simulation period. The third scenario (closed- case) uses the same macroeconomic assumptions as the base-case scenario but closes the system to new entrants. The first two scenarios show that the system dependency ratio increases continuously from 10 percent today to 30 percent by year 2030, and close to 70 percent in year 2070. Required contributions to keep the system in balance would need to increase from 18 percent today to over 70-80 percent by the end o f the simulation period. If contributions remain unchanged, then a growing deficit would be observed starting in year 2012. This deficit could reach 8 percent o f GDP by year 2070. The deficit will be drivenby a dramatic increase in total expenditures from 1percent o f GDP today to 4 percent in year 2030 and 10 percent by year 2050. Revenues will also grow as a share o f GDP, but at a much slower pace, without ever reaching 3.5 percent o f GDP. As a consequence, the SSO is accumulating considerable unfunded pension liabilities, in the order o f 115 percent o f today's GDP (low- case) to 140 percent o f today's GDP (base-case) for the period 2002-2070. If the SSO was closed to new entrants, the value o f unfunded pension liabilities would be equivalent to 100 percent o f today GDP. This i s the amount o f resources that the government would need to transfer to the SSO today to guarantee the promises to current retirees and contributors. l 9The financial projections need to be interpretedwith caution given the limitations of the baseline data. Several assumptions have been required. These have been extensively discussed with SSO technical staff and have been constrainedby the experiences in other countries (see Technical Appendix). 22 Effectively, it i s a transfer from future generations-including poor workers-to those currently well-off and operating inthe formal sector o f the economy.20 2o These projections ignore the outcome of the current vesting period remaining constant at 10 years and those individuals who joined the system during the early 90s starting to leave it with the hopes of returning close to retirement. It is shown that this strategy increases the rates of retum on savings. One of the implications is that coverage rates would be lower as new entrants are neutralized by those leaving the system. The other implication is that the lengtho f service at retirementwould bereduced. 23 Figure 10: Summary Resultsof FinancialProjectionsinthe SSO PensionSystemDependencyRate ContributionRate Requiredfor Zen, Balance (total beneficiaries/contributors) E]] 60% *__..-- 50% 30% 20% 10% 0% 200120062011 20162021 2026203120362041 2046 205120562061 2066 Years I Years I -base line- - --low case 1 '-base line- - - .low case PensionFundTotal Revenueas % of GDP PensionFundTotal Expenditure as % of GDP 14 0% c- ______-__ 3.0% - 120% 2.5% 1 100% 2.0% - 8 0% 1.5% - 6.0% - I 1.0% - 4.0% ~ 0.5% i .~ I '-base line- - - -low case -closed - - - 1 I 1-baseline .lowcase -dosed ~ ~ PensionFundCurrentBalance I PensionFundDeficit:Present Value from 2001 I as % of GDP to 2070 as % of 2001 GDP 2 0% 160% I w ~ 00% 140% 1 .2 120% -40% 100% 4 0% 80% .a 0% 60% 1 .loo% 40% .12 0% I 20% I -baseline - - -Yearscase 0% .low -closed l l Dase line IOW case ClOSea Source: Missioncalculationson the basisofthe PROSTmodel. Scenario 1(base-case). This scenario is basedon the projectionsof the Iran Country Office (World Bank, 2002). GDP grows at 6 percent for the period 2001-2010, droppingto 4.5 percent by 2015, and to 3 percent by year 2025. During the same periods the growth rate o f real wages is set respectively at 4 percent, 3.5 percent, and 2.5 percent, and total factor productivityat 2.4 percent, 1.8 percent, and 1percent. The growthrate of employmentis a function ofthe growth rate of GDP and the growth rate of total factor productivity. Over the period 2002-2012 it grows at an average o f 3.6 percent per year. The male labor force participation rate is assumedto be constant. The female participationincreases from 10percenttoday to 20 percentby year 2012. Underthese assumptions, unemploymentrates go from 15 percentto 21 percent by year 2012. The system coverage of the employedpopulation is assumed to grow slightly so that labor force coverage does not drop. Scenario 2 (low-case): RealGDP grows at 4 percent duringthe next 10 years, drops to 3.5 percent in 2015, and then converges to 3 percent. Wages and total factor productivity are adjusted downward. Assumptions about labor force participationrates are the same, thus unemploymentrates are higher andcoveragelower. Scenario 3 (closed): Same as Scenario 1, but assumes the system is closed to new entrants. The discount rateis set at 5 percent. A 10 percent discount rate produces unfunded pension liabilities to 130 percent, 70 percent, and 65 percent respectively. 24 3.5. Management Policiesfor Fund Reserves The SSO investment policy is in principle proposed by the High Council of the SSO and executed by the Managing Director of the SSO in coordination with its Board of Directors. The High Council is responsible for setting the general framework o f the investment policy (e.g., limits o f investment risks, classes o f assets where the SSO can invest) and for definingthe annual operations budget. Investment policies need to take into consideration the SSO's triple mandate: i)to provide health, unemployment, andpensionbenefits to its members; ii)to support social development (e.g., through investments inhousing); and iii)to support economic development (e.g., through the financial support o f national projects). The Managing Director and the Board o f Directors execute these policies through the Deputy o f Economic and Investment Affairs o f the SSO and the managers o f the different companies directly owned by the SSO-including the Social Security Investment Company (Shasta). The Social Security Investment Company (Shasta or SSIC) was created in I984 to manage the Social Security Organization's investments in the productive sector. Shasta is governed by a General Assembly comprised of Shasta Board members, SSO Board members, and SSO Economic and Investment Department representatives. Shasta's mandate i s to maintain and increase the value of the SSO funds through: (i) investments in industrial, commercial, and mining sectors that impact economic development and yield a reasonable rate of return; (ii) short- and medium-term investments in manufacturing firms, such as those registered with the Teheran Stock Exchange; and (iii)management o f industrial, construction, and commercial firms whose shares predominantly are owned by the SSO or Shasta.21 Shasta has a Managing Director who is selected by the SSO Managing Director in coordination with the SSO Board and High Council. The General Assembly meets once a year to review Shasta's performance and to approve new plans and programs: other meetings follow the mandates set by the Iranian Business Act. The General Assembly is responsible for defining the investment policy-including deposits) . types o f investments and minimum rates o f return (usually not below bank The other companies owned by the Social Security Organization report directly to its Managing Director and Board of Directors: while in theory autonomous, the companies are under close supervision. The following companies are SSO-affiliated institutions: the Iran Housing Construction Company (1986), the Social Security Consulting and Computer Company (1992), the Social Security Research Institute (1992), the Labor and Production Company (1992), Labor and Security Services (1992), the Social Security Auditing Company (1993)' the Social Security Real State Agency (1993), and the Worker's Welfare Bank (1961).22 Each has its own Managing Director and Board o f Directors. The preparation o f nominees usually rests with the SSO Department o fEconomic and Investment Affairs; the selection of candidates i s retainedby the respective company's Managing Director and Board o f Directors. The respective operating budgets, however, must be approved by the SSO Board o f Directors and the High Council, and large financial operations require the approval o f the HighCouncil. The size of the SSO portfolio has been declining as a share of GDP while the structure has changed, givingpriority to direct and indirect investments over lending and loans (see Figures II and 12). Upto 1976, the SSO investedits cashreserve fundinthe Worker's Welfare Bank inthe form of fixed deposits. After the creation of the Social Security Fundin 1976, the SSO started to diversify its investment activities, but by 1989 still close to 80 percent o f the portfolio was composed o f long-term deposits. By year 2000, however, the share of deposits had dropped to 10percent, while the shares o f indirect and direct investment increased to 27 percent 2'Social Security Research Institute (2002a). 22SSO owns 34 hospitals, 174 clinics, 31polyclinics, and 3 day clinics. 25 and 43 percent respectively, or 70 percent of the total portfolio (see Figure 12). The companies owned by the SSO (directly and through Shasta) produce 43 percent of the pharmaceutical and hygienic products, 36 percent o f the cement, 35 percent o f the televisions, 25 percent o f the fireproof products, 31 percent o f the refrigerators and freezers, and 35 percent o f the rubber. The SSO is one o fthe most active investors inthe stock exchange, with 11.3 percent ofthe total portfolio belonging to the SSO in2001, Inaddition, the SSO invests ingovernment securities; provides financial support for public construction projects; and accords liquidityto the banking system; and gives financial assistance to pensioners and contributors (e.g., housing loans, marriage aid, etc.) at subsidized interest rates. In 2000, the SSO investment portfolio (excluding government debt) was valued at Rhials 7,000 billion (USD 933 million). Figure 11: SSO, Investment Portfolio as a Share of GDP we 3.00% el GD 2.50% P 2.00% - 1.50% I.00% 0.50% 0.00% , ~~ Source: Missioncalculations. Figure 12: SSO, Structure of the Portfolio of Investments T i 1100% 80% 7196 1 60% 40% 20% a% 1 0% 10% 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 I Source: Dataprovidedby SSO EconomicandInvestmentAffairs Department. Note: Indirect investment refers to investments through the Investment Company Shasta. Directinvestmentrefersto investmentsincompaniesdirectly ownedbythe SSO. An important part of the fund reserves are invested through Shasta which by year 2001 had accumulated assets equivalent to Rhials 1,537 billion23(USD 200 million). Today, Shasta owns some 86 companies, o f which 65 percent are majority owned, in six major economic sectors (commercial services, non-metallic minerals, metal industry and household appliances, pharmaceuticals, agriculture, cellulose and chemicals, and commercial services) (see Figure 13). 23 See SSO(2002b). 26 Figure 13: Shasta's Companies: Ownership and Economic Sectors ofActivity, 2000 1 35.0% I Commercial and Services Non-Metallic 25.0% 15.0% b; I 0 ~ 2 5 % 25.49% 50-59% 60.90% More 90% Shrrrsf the Compmkr'Arrrtr 16% ~ Source: Based on Social SecurityInvestment Company (2002). Recently the Social Security Organization has added to its stock of companies those transferred by the government to cover part of its debt with thefund; the book value of the transferred assets is estimated roughly at Rhials 4,000 billion. By 2001,the government had accumulated a debt with the SSO, in large from unpaid contributions, estimated at Rhials 8,000-14,0000 billion. To cover its debt with the fund, the government transferred assets from the following public companies to the SSO: 25.5 percent of the petrochemical companies in Khark, 8.5 percent in Arak, and 17 percent in Esfahan; 25.5 percent o f the Ahwaz pipe-making company; 50 percent of national gas industries; 50 percent o f Persi gas company; and 51 percent o f cultural and scientific publishinghouses. The SSO plans to keep some of these companies and sell others after restructuring and recapitalization. Revenuesfrom investments in year 2001 reachedRhials 1,383 billion (USD 184million), which relative to total assets of Rhials 7,000 billion (excluding government debt) implies a real rate of return of -0.6 percent.24 The majority o f revenues, or 80 percent, are generatedby Shasta and other dependent companies (see Table 9). The assets invested inthese companies approximate Rials 3,500 billion (USD 460 million). According to official figures, Shasta appears to be a profitable holding, generating in year 2001 a 3.5 percent real rate o f return after taxes. The same is not true of the other dependent companies; in aggregate they generateda negative real rate o freturn after taxes (see Table 10). ~ ~ 24Inflationinyear 2001 was reportedat 20.4percent. 27 Table 9: Sources of Revenuesfrom Investments in the SSO (billions of rhialsin 2001) Total Revenuesfrom Investmentsa 1,383 Shasta 505 36.5% Other Dependent Companies 596 43.1% DepositsinWWB 24 1.7% Banking Sector 148 10.7% GovernmentSecurities 29 2.1% PersonalLoans 9 0.6% CorporateLoans 25 1.8% Management Dues 15 1.1% Late Payments 15 1.1% Real State 10 0.7% Source: Dataprovidedby SSO staff. Note: (a) Original numbersdo not add-up. Table 10: SSO, Profitability of SelectedCompanies, 2000 and 2001 (billionof rhials) Year Capital Profits Profits Rate of Rate of Injlatio Real Real before after Return Return n Rate Rateof Rateof Tax Tax before after Return Return Tax Tax before after Taxes Taxes Shasta 2000 1,330 435 323 32.7% 24.3% 20% 10.58% 3.58% 2001 1,537 505 379 32.9% 24.6% 20.4% 10.38% 3.49% SSO-Dependent Companies 1998 1,602 330 216 20.6% 13.5% 20% 0.50% -5.42% 1999 2,069 468 353 22.6% 17.1% 20.4% 1.83% -2.74% Source: SSO (200 All these numbers, however, need to be interpreted with caution, as currentfinancial reporting practices are weak. An official evaluation o f the SSO investment policies suggests that current reporting practices are not adequate. To date, a system that provides updated information about investments, revenues, and profits i s lacking. There i s high variability in the way that different companies calculate and report operational results, and the problem seems to be particularly serious among construction projects. Managers have also indicated inconsistencies between the budgeting system and the accounting system. The former, for instance, includes total revenues from investments as part o f the total budget o f the SSO. As far as the companies are concerned, only net profits matter. Based on this review, the main problems facing the SSO interms of its investment policies can be summarizedas follows: The governance structure and the practices in terms of disclosure, auditing, and custody are unlikely to promote accountability and incentives for a prudent management of reserves. The SSO has opted for a tripartite governing body (the High Council) with representatives from the government, plan members, and employers. As suggested by international experience, this type o f governing body i s rarely conducive to a prudent and 28 efficient management o f reserves (see Section 7.2). In fact, the poor financial performance of the fund i s to be explained, at least in part, by the current governance structure. First, the fact that the majority o f the Board members are selected by high-ranking officials creates a governing body that i s overly dependent on the government. This leaves room for the emergence o f conflicts o f interest betweenthe government and the plan members. The decision to pay part o f the government debt by transferring companies at prices above market value i s an example o f a conflict o f interest. Moreover, there i s little or no accountability regarding these financial decisions. For instance, the losses created by the transfer o f the public assets are simply assumedbythe membersofthe plan. As previously discussed, there are problems interms of the quality o f the financial information flowing from the managers o f the different companies owned by the SSO and Shasta to the Department o f Investment, Economic, and Planning Affairs, to the SSO Boardo f Directors, and to the High Council. These problems are amplified when looking at the flow o f information between the SSO and the plan members. The various companies, for instance, are not required to publish their financial statements. While the SSO publishes an Annual Report with a one- year delay, the information disclosed i s insufficient to give to the plan members an idea of how their savings are being invested. Several brochures have been produced about the SSO and Shasta. However, these focus on publicizing contributions to social and economic development and on the impressive range o f activities where the SSO i s involved without providing any hint o f how successful these activities are interms o f protecting and increasing the value o f worker savings. While all the companies conduct periodic audits, they are carried out by the Supreme Auditing Organization, which i s a government body. This practice can also create conflicts o f interest between the government and the plan members. Finally, where best international practices emphasize the need to have external, independent custodians, the SSO and its holding companies do not comply. At the same time, the SSO over-expanded its mandate rendering it difficult now for managers to assess the performance of investment policies and implement corrective measures when necessary. Beyondproblems o f governance and disclosure, it i s unclear if the SSO currently has the resources (human and physical) to effectively monitor the operations o f the large number o f companies it owns directly or through Shasta. Since the Managing Directors o f the companies know the limitations o f the monitoring system, they are more willing to take risks. Some SSO staffhave observed, for instance, that the cost-benefit analysis presented by the Managing Directors o f the companies to justify additional investments often lack realism. During implementation, results are usually far-off from predictions. Yet, these are approved during the meetings o f the General Assembly. Another problem with the direct management o f several corporations i s that considerable resources need to be allocated to deal with administrative problems. These are resources that need to be taken away from the design and implementation o f an effective investment policy. Other SSO activities such as housing projects, lending to the corporate sector-including its own companies-and subsidized lending to beneficiaries, also bring administrative problems that divert the attention o f the Board25and create conflicts o f interest. The Department o f Economic, Investment, and Planning Affairs i s currently proposing to break Shasta into smaller holdings that are more specialized but that would respond to the same governance structure. This is unlikely to bring sizable improvements in management, as incentives would remain unchanged. Other alternatives are discussed in Section 7.2. 25Example, Dealing with the complaints of beneficiaries in terms of the quantity and quality of housing services; pervasivedelays observedinconstructionsprojectsor the difficulties involved in selling ahouse (see SSO, 2002). 29 The SSO has considerable market power in several economic sectors: this may impede private-sector development and interfere with corporate governance. As previously discussed, the SSO controls large shares o f the market in sectors such as pharmaceutical, domestic appliances, and food. This gives a government institution considerable influence over the corporate sector. Conflicts o f interest may arise, for instance, if protective policies are applied to keep inefficient companies afloat. The SSO also owns 11 percent o f the assets traded in the stock market. The implication i s that changes in the financial position o f the fund may create large fluctuations in stock prices, which discourages the participation of private investors. Current investment policies do not follow best practices in terms of exposure limits. Best international practices suggest that the pension fund not own more than 5 percent o f the capital of any company, nor the investment capture more than 5 percent o f the pension fund's portfolio. In the SSO's case there are no exposure limits, leading to a risky and illiquid investment portfolio. The SSO is also negatively affected by an inappropriate regulatory environment. Besides the problems with its own regulatory structure, the SSO faces several institutional and legal constraints impeding a more efficient management o f reserves. These are not only related to an over-regulated stock market26but also to restrictions on foreign investment (the exact nature o f which remains unclear), to an underdeveloped Iranian financial sector, and to a poorly diversified economy. These constraints limit investment opportunities. 26 While part o f these assets in principle can be traded in the stock market, transactions seem to be suspended when stock prices drop below a given threshold. 30 4. THECIVIL SERVANTS RETIREMENTORGANIZATION (CSRO) 4.1. Institutional Issues The Civil Servants Retirement Organization (CSRO) was created in 1922 to provide old-age, disability, and survivor pensions to government employees. Initially, the CSRO was part o f the Ministry of Finance. In 1975 the fund became an independent institution, but its accounts remained under the Treasury's control. Authorizations from the Ministry o f Finance and Managing and Planning Organization (MPO) were required for financial transactions. Not until 1998 didthe CSRO assume its own financial management. The CSRO is governed by a Supreme Council where sits the Head of the Managing and Planning Organization (MPO), the Head of the SSO as a Deputy of the Ministry of Health, a Deputy of MPO, a Deputyfrom the Central Bank, and a Deputyfrom the Ministry of Finance. The last three are selected by the Heads of their respective organizations for a period o f four years. The Director o f the CSRO i s named by the Head o f MPO and reports to a four-members Supreme Council (see Figure 14). A new piece o f legislation i s currently being considered by the Parliament to include employers and plan members on the HighCouncil. The CSRO has 1,400 employees distributed in 28 regional offices covering all provinces in the country. The regional offices link contributions from close to 4,000 public institutionshnits to the CSRO. In 2001 the CSRO's total regular operational expenditures represented a modest 0.65 percent o f total benefits paid. This low number is explained by the fact that the government (i.e., the central budget) directly finances part o f operational expenditures. Inyear 2000 and 2001, these government transfers accounted for roughly 7 percent o f total benefits paid. Wages account for 50 percent o f operational expenditures. Nonwage expenditures are related to general administration (e.g., maintenance, banking, and legal fees) and account for roughly 47 percent o f the total. Finally, expenditures related to corrective transactions (e.g., excess contributions returned) represent 3 percent ofthe total. Figure 14: Organizational Chart of the CSRO LegalAffairs ChairmanofCSRO Public Relations Administration h Administrationand Teclnucal Economicand FinancialAssistance Assistance InvestmentAffairs Planningand InformtionOffice I FinancialBureau Incomand ExpmesBureau Invesmt and I I Affiliated Corporations IAdministrationBureau I I I FProvincialOflices(28) I burce: CSRO. 31 During the last three years the CSRO has made considerable progress in strengthening institutional capacity by updating management and information systems, training stafi and streamlining the administrative process. The new information system, developed in-house, uses state-of-the-art technology and reflects the high quality o f its technical staff.27 The system processes liquidations for new beneficiaries, emits payrolls, keeps track o f bank accounts, and provides managers with detailed information about current pensioners. According to management, it has increased employee productivity threefold. However, given the lack o f an updated database o f contributors and employers (MPO i s currently updating this database), data still need to be entered manually to process liquidations. These data are submitted electronically or in hard copy from the regional offices. Eventually, the wage and work history for all contributors should be loaded into the system and accessible from all regional offices. The CSRO also has computerized the accounting system. Iranian accounting law, however, does not follow international best practices and current regulations make it difficult to assess in a transparent and clear manner the financial situation o f the pension fund. For instance, even when government contributions have not been transferred to the CSRO, they are registered in the income and profit statement. 4.2. Coverage and Contributions The CSRO covers 1,572,825public sewants that represent 8.7percent of the labor force; this number is expected to remain more or less constant over time. Despite the implementation o f policies to reduce the size o f the civil service, between 1990 and 2000 the average growth rate o f the stock o f public servants approximated 1.8 percent. There have been large fluctuations around this trend, partially reflecting discretion in the hiring process (see Table 11). Looking forward, however, the government seems to be committed to restricting the expansion o f the public sector. The implication for the CSRO i s that inevitably, over time, there will be a sharp increase inthe number o fpensioners per contributor. The total payroll contribution to the system is set at 22.5 percent, but it is applied to the base salary, which represents roughly 75 percent of total earnings. Hence, effectively, the contribution rate i s 16.8 percent o f total earnings. From the 22.5 percent, workers pay 9 percentage points while the government pays 13.5 percentage points. The employee contribution rate was increased from 8.5 percent to 9 percent in 2000. Employees with more than 30 years o f service do not pay their share o f the contribution. 27 Puttingthis system in placerequiredanalyzingand streamliningthe different administrative processescarriedout by the CSRO. Detaileddocumentation ofthese processesis available. 28 Numbersobtainedfrom CSRO do not seemto beobserveddatabutrather some type o fmovingaverage. 32 4.3. Benefits and Rates of Return The system offers old-age, disabiliv, and survivor pensions as well as contingency loans and family allowance^^^; the system also covers health insurance contributionsfor pensioners (see Table 12). Over the last ten years, there has been afast increase in the number of benejciaries, attributed in part to the government's efforts to control wage expenditures by reducing the size of the civil service. Not only has this lead to limitedhiring, but also to the provision o f incentives for early retirement (see below). Hence, during the last decade, pensioners grew at an average o f 10 percent per year (see Table 13). As a consequence, the ratio between total old-age pensioners and contributors has deteriorated sharply, from 13.7 percent in 1990 to 33.4 percent in 2001. In a way, budgetary pressures in the general budget were implicitly transferred to the CSRO. Today, there are 662,786 pensioners inthe CSRO-including old-age, disability, and survivors. Inthe absence of new policies that change retirement incentives, the number of pensioners is expectedto grow at 4 percent per year. 29Theseare paiddirectly fromthe generalbudget. 33 Table 12: Benefits Provided by the CSRO Benefit Rules To qualify for a pension, the individual needs to be 60 years old with no minimum number of years of contribution, or 50 years old with at least 25 years of service; if male, or no minimum age with 20 years of contribution if female. Males with 30 years of contribution do not have aminimum retirement age. The pensioni s computedon the basis of a 3.3% accrual rate. The replacement rate i s applied to the average salary during the last two years." The salaries used inthe calculations, however, represent only 75% of the total compensation. There is a maximum replacement rate of 100%. Old-Age There is a minimum pension, which i s adjusted yearly by Parliament, and set equal to the minimum wage Pension (Rhials 600,000 (USD 80) per monthin2001; andRhials 850,000 per monthin2003). Professions considered of high-risk allow individuals who have contributed for 20 years to retire with the same benefits as individuals who have contributed 25 years. Similarly, those who have contributed 25 years can retire withbenefits equivalent to 30 contribution years. Individuals retiring before verifying retirement conditions obtain a lump-sum, which is financed by the govemment, equal to 3 months of salary for eachyear of service for unskilledworkers, or 45 days of salary per eachyear of service for skilledworkers. The CSRO gives backthe contributions. Non-work-related: Individuals receive a pension equal to 0.033 *A * average wage for the last two years, Disability where A =number of years of services + 3. A cannot be less than 15. Pension Work-related: The pension i s equal to the wage received by an individual with two more ranks. A rank is equal to 4 years of service. The wage of an individual i s proportional to the rank. A disabled individual receivestwo additional ranks of salary. Category 1: When an old-age retiree dies. The pension i s divided by the number of dependents. Each dependent receives hisiher share. Category 2: Contributor dies from work-related cause. The pensioni s calculated as inthe case of a disability Survivor caused by a work-related injury. The pension i s divided by each dependent. [Restrictions regarding Pension dependencynot specified.] Category 3: Contributor dies from non-work-related cause. The pension i s calculated as in the case of disability caused by a non-work-related injury but with a 3 years bonus. The pension i s divided by each dependent. Family There i s an amount received by the spouse and by each child (up to a maximum of 3). The exact formula i s Allowances cumbersome,but the allowance i s basically apercentageofthe basic wage, indexedwith the growth rate ofthis wage). Family allowances are paid directly by the govemment andnot by the CSRO. Low- Pensionersand contributors have the right to receive loans at a subsidized interest rate. This rate varies from Interest case to case. In2001 the average loan was equal to Rhials 4,000,000. Loans represent only a marginal fraction Loans of the investment portfolio. Source: Van0 interviews with the CSROtechnical staff andPublic Employee Act. 30This i s a recent modification. Before 2001 the replacementrate was applied to the average salary o f the last three years. 34 Table 13: Evolution of Total Beneficiariesin the CSRO Old-Age Disabled Survivor Beneficiaries Pensioners (Families) Total 30.00% 1 1995 236,348 5,949 90,144 332,44 1 25.00% - I 1996 255,305 6,858 96,507 35,867 1997 273,355 7,744 102,835 383,934 1998 294,572 8,545 109,618 412,735 1999 320,051 9,336 116,764 446,15 1 2000 349,479 122,927 482,335 I2001 1401,549 I 9,929 I 125,395 I 10,82 537,766 I ' I m 0.00% I L I I 1990 1992 1994 1996 1998 2000 2002 Source: Missioncalculationson the basisof computer reportsprovidedby the CSRO IT Department As in the SSO's case, the eligibility conditions and benefits that the system attempts to provide are generous by international standards. With regard to old-age pensions, if the individual i s 60 years old or above, there i s no minimumnumber of contribution years. Females can retire at any age as long as they have contributed for 20 years, and males can retire at age 50 if they have contributed for 25 years. In all cases the accrual rate o f the system i s set equal to 3.3 percentage points per year, but it also applies to 75 percent o f wages. Hence, the effective accrual rate i s 2.47 percent, still considerably high by international standards. The pension i s computed on the basis of the last two salaries. Hence, an individual retiring after 25 years o f contributions will obtain a pension that represents roughly 62 percent o f hidher last total compensation. After 30 years o f contributions the replacement rate will attain 74 percent. The maximum replacement rate has been set at 100 percent. Because the CSRO does not have a ceiling on covered wages, all employees regardless o f income level receive the same replacement rate (see Figure 15)31. Figure15: CSRO, Economy-WideandIndividualReplacementRatesby IncomeLevel I I I I 3h '- 10 iI Minimum pension i s K e' o'204 I 0 6 6 2 6 6 4 6 6 6 6 6 8 6 6 1066 0.66 2.66 4.66 6.66 8.66 10.66 Individual earnings (proportion average wage) Individual earnings (proportion average wage) Source: Missioncalculations. Note: Calculationsassume 30 contributionyears. Minimumpensiongrows inreal terms. While initial pensions are generous relative to contributions, like in the case of the SSO, pensioners do not count with an automatic indexation mechanism to protect them from increases in the cost of living. Article 4 of the Law on Coordinated Pay for Civil Servants 3' Recently a law hasbeen approved to provide complementary plans to civil servants. The missionhas not received information about the type of scheme nor the mechanisms used to manage these schemes. 35 states that the government should adjust yearly civil service employee wages and pensioner pensions by the consumer price index. The adjustment occurs after consensus i s reached between the Managing and Planning Organization, the Minister o f Finance, and the Central Bank. Discretioninthe adjustments explains, inpart, the erratic fluctuations inthe real average pension observedbetween 1995 and 2001 (see Figure 16). Figure 16: Evolutionof the RealAverage Old-Age Pension 4,000 3,500 29 2,500 I 1 1 1995 1996 1997 1998 1999 2000 2001 , Source: Mission calculations on the basis of CSRO individual records. Like the Social Security Organization, the CSRO system offers a minimum pension, which is equal to the minimum wage of Rhials 850,000per month ;only afew of the new retirees have had the need to apply. Inyear 2000, when the minimum wage was equal to Rhials 400,000, only 2.5 percent o f total retirees received a pension equal to the minimum. However, 50 percent received a pension below Rhials 500,000 and often close to the minimum. In2001,2.7 percent o f retirees received the minimumpension. Adjustments to the minimumpension have overcompensated for increases in the cost o f living. For instance, between 1995 and 2000 the minimumpension increased 3.4 times (from Rhials 117,000 per month to Rhials 400,000 per month), while prices increased 2.3 times. Between 2000 and 2001 the minimumpension was increased by 25 percent, while the inflation rate was 15 percent. Disability and survivor benefts are disconnected from riskfactors andfrom considerations of financial sustainability. These pensions represent 23.5 percent o f total pension expenditures. The same accrual rates used to compute old-age pensions apply, but workers are arbitrarily accredited with additional contribution years (see Table12). Inthe case o f disability pensions, the mission has not been able to assess the appropriateness o f the current accreditation process. The system, however, does not appear to be prone to abuses, as these expenditures currently represent only 2 percent o f total pension expenditures. The share of survivor pensions is considerably higher (21.5 percent). Eligibility conditions for dependency are not known at this point. Despite high contribution rates, the system offers rates of return that are not sustainable. As previously discussed, over the long-term, the sustainable rate o f return in a pay-as-you-go system i s equal to the growth rate o f the wage bill. Inthe CSRO, given that the size o f the civil service is expected to decline as a share o f the total population, the sustainable growth rate i s given by the growth rate of the average wage plus at best 1 percentage point. Calculations under various assumptions about the growth rate o f this average wage show that implicit rates o f return are considerably higher and therefore unsustainable. For instance, under the assumption o f a 2 percent average growth o f real wages, internal rates o f return for males would be inthe 5-7 percent range (depending on the age o f the individual when enrolling in the system), and for females inthe 5-10 percent range (see Figure 17). 36 Figure 17: Internal Rates of Return for Males and Females in the CSRO I Males Females i;;- r 0 -2D: z. - 0 E _ _ - - - -_ c u l -- _ - = P - D 8 While statutory early retirement is not particularly attractive, special programs have been implemented by the government over theyears as a way of reducing the size of the civil service. Individuals retiring early receive, in principle, a lump-sumequal to three months o f salary for each contribution year in the case o f unskilled workers, and 1.5 months in the case o f skilled workers (see Table 12)32. This lump-sumi s financed directly by the government. In addition, the CSRO gives back the contributions, without interest. These statutory rules, nevertheless, have often been overruled. For instance, in 1980, all employees with 15 years o f services were given 15 additional years with no restrictions. In 1999, an act was approved that provided a lump-sumequal to two weeks of salary per contribution year to those individuals who agreedto retire before meeting requisite retirement conditions. In2001, the act was modified to offer one month o f salary per contribution year, up to a maximum o f 30 years. In 1980, all employees with 15years of services were given 15 additional years with no restrictions. 4.4. FinancingMechanisms andSustainability The CSRO is expected to mobilize revenues amounting to Rhials 5,170 billion (US0 689 million), or 0.9percent of GDP, mainly from contributions and investment^.^^ Revenues from government and employee contributions in principle account for over 80 percent o f the total. Revenues from investments in years 2000 and 2001 amounted to Rhials 528 billion (USD 70.4 million) and Rhials 600 billion (USD 80 million), or 12 percent and 11 percent o f total revenues respectively. An implicit revenue i s also generated by the fact that the government pays directly part o f administrative expenditures. These payments inyear 2001 were equivalent to Rhials 311billion (USD 41 million), or 5.7 percent o f total revenues. Actual revenues, however, are lower as the government has not been paying in-full its contributions. The government suspended the payment o f contributions in 1995. In 2001, for instance, unpaid contributions totaled Rhials 1,000 billion (USD 180 million), excluding penalties that amounted to Rhials 765 billion (USD 80 million). As a result o f the moratoria, the government has accumulated arrears with the CSRO amounting to Rhials 17 billion (USD 2.3 billion). Recently, the government settled USD 500 billion o f these arrears by transferring 32These lump-sumis financed from the general budget. 33 These revenues exclude non-regular entrances such as those related to the payment of late fees from the govemment, late contributions, or unaccounted revenues. In year 2000 and 2001 for instance, these unaccounted revenues reached Rhials 32 billion. 37 public companies to the CSRO (see Sub-section on Investment Policies inthe CSRO below). It i s expected that starting inyear 2002, no new arrears will be accumulated. The share in GDP of total expenditures in pension benefits has doubled during the last five years; today expenditures amount to Rhials 4,308 billion (US0 609 million), or 0.75percent of GDP. The largest shares of expenditures are absorbed by old-age pensions (76.7 percent) and survivor pensions (21.4 percent). Disability pensions represent only 1.9 percent o f the total. These shares have been roughly constant over time (see Figure 18). In terms o f non-pension benefits, expenditures on health insurance contributions are very modest (0.45 percent o f total pension expenditures). Family allowances are paid directly from the general budget and therefore are not included in the CSRO's profit and losses report. Information on the total amount i s not available. Figure18: CSRO, Compositionof Total Expenditures 0.70% I---- 1 100% 90% 2 0.60% - 2E 80% -4 0.50% - L l 70% a 0.40% 60% --d n ;// 2 m 50% 0.30% - .-E 40% 0.20% - i 30% 2 a 0.10% - 20% 0.00% I 10% 0% 1996 1997 1998 1999 2000 2001 I 1996 1997 1998 1999 2000 2001 Source: Missioncalculationsbasedon CSRO individualrecords. The 2001 ofJicial profit and losses reportfor the CSRO, which counts as revenues the arrears accumulated by the government for its contributions and latefees, displays a positive balance of Rhials I,764 billion (US0 235 million).34 When these fictitious contributions and late fees are removed, the operational balance generates a deficit o f Rhials 186 billion (USD 24.8 million). Similarly, the official pension balance (total revenues from contributions minus total pension expenditures) i s equal to Rhials 262 million. When government contributions are removed, pensions expenditures exceed revenues by Rhials 1,088 billion (USD 145 million). Since government contributions are being normalized, a better indicator of the financial situation of the fund i s the operational balance excluding late fees and the losses for the valuation o f assets. This balance equals Rhials 814 billion (USD 108 million). Thus, if government contributions are paid regularly, over the short term the CSRO could generate a surplus and accumulate reserves (see Table 14). 34 This includes the extraordinary expenditures related to the losses imposed by the transfer of govemment companies to the CSRO to settle part of the public debt (see Section 4.5). The losses result from the difference betweenthe book value of the company, which was usedto compute the value of the transfer, and the market value. For 2000 and 2001 these losses were estimatedat Rhials 232 billion (USD 31million) and Rhials 311 billion (USD 41 million) respectively. 38 Table 14: Revenues and Expendituresin the CSRO (billionsof rhials) 2000 Percent Share of 2001 Share of GDP Percent GDp TOTAL EXPENDITURES' 3,099 0.53% 4,356 0.65% PensionBenefits 3,048 98.37% 0.52% 4,308 98.90% 0.65% Old-Age 2,320 74.87% 0.40% 3,306 75.90% 0.50% Disability 60 1.94% 0.01% 80 1.84% 0.01% Survivor 668 21.56% 0.11% 922 21.17% 0.14% HealthInsurance Contributions 20 0.65% 0.00% 20 0.46% 0.00% Operational Expendituresb 31 0.99% 0.01% 28 0.64% 0.00% General Administration 13 0.42% 0.00% 13 0.30% 0.00% Wages 12 0.39% 0.00% 14 0.32% 0.00% Other 6 0.18% 0.00% 1 0.02% 0.00% TOTAL REVENUES 4,136 0.71% 5,170 0.78% Contributions' 3,608 82.60% 0.62% 4,570 83.38% 0.69% Investments 528 12.09% 0.09% 600 10.95% 0.09% BALANCE 1,037 0.18% 814 0.12% Source: Interviewswith the Director ofthe Financial Department. Note: (a) The table excludes non-regular expenditures (e.g., losses imposedby the transfer of govemment companies) and revenues(e.g., unaccountedrevenues, late fees from the govemment). (b) Operationalexpenditures do not take into account payments that are directly incurred by the govemment's central budget. (c) Contributions reported are legal contributions (Le., do not take into account government arrears). Lookingforward, however, theJinancia1situation of the system is expected to deteriorate as the number of retirees increases relative to the number of contributor^.^' Under the assumption that the size o f the civil service grows at half the growth rate o f the population, actuarial projections show that the dependency ratio would double by year 2030, reaching 80 beneficiaries per 100 contributors (see Figure 19). To keep the system inbalance, contribution rates would need to increase from 22 percent today to close to 60 percent in year 2075. If contribution rates and current benefits remain unchanged, then the balance o f the CSRO will deteriorate, generating a deficit o f 0.8-1 percent o f GDP by year 2030. While the share in GDP o f this deficit i s likely to drop afterward, as the economy expands faster than the pension system, the absolute value o f the deficit will continue to grow. The present value o f unfunded pension liabilities i s estimated at 30-35 percent o f today's GDP, depending on assumptions about economic growth and labor productivity growth. Ifthe CSRO i s closed to new entrants, the additional resources necessary to honor the pensions o f today and o f hture retirees approximate 20 percent o f GDP. 35 As in the case o f SSO, the financial projections need to be interpreted with caution given the limitations of the baseline data. Several assumptions have been required. These have been extensively discussed with the CSRO technical staff and have been constrained by the experiences in other countries (see Technical Appendix). 39 Figure 19: Summary of Financial Projections in the CSRO I 120%, PensionSystemDependencyRate ContributionRate Requiredfor Zero Balance (totalbeneficiarieskontributors) .. I 1 70% 3 /-- 40% 30% I -base line - - - .low case 1 1-baseline- - -.lowcase PensionFundTotalRevenue as YOofGDP PensionFundTotal Expenditureas 'YOofGDP I I -base line- - - -low case -closed 1-baseline - - -.lowcase -dosed PensionFundCumnt Balance I I PensionFundDeficit:PresentValue from 2001 as YOofGDP to 2070 as % of 2001 GDP I I 40% ... I 1 .1. -base line - - - .low case -closed I "'" 1; base line low case closed I ~~~ Source: Missioncalculations. Note: Size o f the civil service is assumed to grow at halfthe rate of the total population. Macroeconomic assumptions for Scenarios 1(base-case), 2 (low-case), and 3 (closed system) are the same as those o fthe SSO (see Figure 10). 4.5. ManagementPoliciesfor Fund Reserves Investment policies are dictated by the Supreme Council of the CSRO and executed by the Director of the CSRO with the technical advice of an Investment Committee. Policies include the budget of the CSRO, the classes o f assets where the CSRO can invest, and maximum allocations. While the law creating the CSRO imposes restrictions on the shares o f new investments that can be allocated to loans (50 percent), bank deposits (25 percent), and stocks (25 percent), these do not seem to be respected in practice. The Supreme Council does not have the authority to remove the Director of the CSRO, who i s selected by the Head of MF'O. The Investment Committee only has an advisory role. The Committee is composed o f the Director of the CSRO, an Economic and Investment Assistant (nominated by the Director o f the CSRO), and the Managing Director of the CSRO's Investment Company (also nominated 40 by the Director of the CSRO). Each o f the Committee members has investment advisors-including affiliated stockbrokers. There are two main types of investment activities: direct investments, which are executed by the Department of Investments and Economic Affairs of the CSRO, and indirect investments, which are executed by the newly created Investment Company. Direct investments refer to investments in companies (i.e., equity positions) that do not take place through the stock market. The level of CSRO ownership inthese companies varies from 100percent o f assets, as in the case o f the construction company, to less than 13 percent. Companies where CSRO ownership exceeds 50 percent o f assets are managed directly; in companies where the CSRO owns between 13 percent and 50 percent o f the assets, there are representatives on the Board o f Directors; in companies where ownership i s below 13 percent, there is no representation. Direct investment activities have intensified with the recent transfer o f public companies to cover part o f the government debt with the fund. Looking forward, however, the strategy i s to prioritize indirect investment activities through the Investment Company. The Investment Company was created to manage investments in the stock market; it currently has assets equivalent to USD 125 million, representing close to 5percent of total reserves (see Table 15). To give the company some degree o f autonomy from the CSRO, the Investment Company was created with its own Supreme Council. The Supreme Council o f the Investment Company, however, i s composed o f the same members as the Supreme Council o f the CSRO. The Board o f Directors is constituted by the Headofthe CSRO, the Director o f the Department o f Investment and Economic Affairs o f the CSRO, the Managing Director o f Behvar (a company owned by the CSRO), and the Managing Director o f the Auditing Company (also owned by the CSRO). The Manager o f the company i s appointed by the Board o f Directors. The company has three departments: the Stocks Department that i s in charge o f trading stock, the Financial Department that acts as a custodian, and the Research Department that provides recommendations in terms o f investment (see Figure 20). There i s no information about company-specific investment policies or strategies, except that investments cannot be above 15 percent o f the capital o f a given company (to avoid involvement in decisionmaking) and that above a given threshold approval i s required from the Board o f Directors. Figure20: CSRO Investment Company Supreme Council ~ 1 a1 Boardof Directors Director Manager Shares Department 1 1 Financial Department (Custodian) ResearchDepartment Stock Market 1 ~ ;ource: CSRO. 41 The CSRO has accumulated reserves equivalent to USD 2.5 billion, or 3.3 percent of GDP, although 70 percent assumes the form of government arrears. This government debt, even when interest and penalties are applied, generates negative real rates o f return o f close to -2 percent. An additional 20 percent o f the reserves i s captured by pubic companies recently transferred by the government to cover part o f its arrears with the fund. Indeed, in 2001 the government transferred assets in 50 companies with a book-value o f USD 500 million to the CSR0.36 The remainder o f the CSRO reserves is allocated to the Investment Company (5 percent) and to government bonds, bank deposits, and loans to members and the CSRO companies (5 percent). Hence, in practice, fund managers have a very narrow margin o f maneuver to reallocate assets and improve rates o f return. Companies Transferredby the Government 500 19.60% I InvestmentCompany I 125 I 4.90% I ConstructionCompany 37 1.45% Long-Term Deposits 25 0.98% GovernmentBonds 63 2.45% Loans 1 0.04% GovernmentArrears- Including Interestand Penalties 1,80037 70.57% I Total I 2.551 I I Share of GDP 3.29% The Director of the CSRO with the support of the Investment Committee has decided to use the CSROfunds to recapitalize the companies transferred by the government and to restructure their management andproduction processes so that the companies achieve a minimum level of projitability and can then be traded on the stock market (through the Investment Company). When an agreement was reachedbetweenthe CSRO and the government to pay part o f the debt through the transfer o f public assets, a list o f companies operating in various sectors (e.g., telecommunications, transport, and energy) was proposed to the CSRO. Companies were then selected on the basis o f six criteria: i)the volume o f shares issued by the company in domestic and foreign markets; ii)the consumption patterns for the goods produced by the company; iii)the book value o f the company in comparison with its market value; iv) the quality o f its human resources; v) the level o f technology; and vi) the financial performance o f the company in recent years. The companies were appraised by qualified experts from the stock market andjudicial specialists. Despite this selection process, however, only a few o f the transferred companies can be traded in the stock market. The majority o f the companies do not meet the necessary criteria. The CSRO has assumed control o f these companies (through the Department o f Investments and Economic Affairs) andplans to reformthe management process in order to make the companies profitable. These reforms are being conducted by specialized groups o f legal, actuarial, financial, and auditing experts. 36 Conservative estimates by stock-market experts reflected in the state of profits and losses suggest that the revaluation of the assets (from book value to market value) cost the CSRO Rhials 132 billion (USD 17.6 million). Additional transfers equivalentto USD 1billion are expectedto take placein 2003. 37Total arrearsin 2003 are estimated at over USD2,300 million (Rhials 19,000 billion). 42 The main weaknesses in current investment policies, similar to those o f the SSO, can be summarized as follows. The governance structure is not conducive to prudent management of reserves. As in the case of the SSO, the current governance structure of the CSRO does not follow best international practices in making fund governors and fund managers accountable to plan members. Investment decisions are made by a closed group o f senior civil servants appointed by high-ranking officials. The fund therefore remains overly dependent on the government. While a proposal i s currently being considered by the Council o f Ministers to include workers and beneficiaries representatives on the Supreme Council, even if approved, the policy i s likely to bring little or no improvement in the fund performance. Indeed, as shown by the SSO and the experience o f other countries, tripartite governing bodies are not conducive to better governance (see Section 7.2). There is no clear separation of governing functions from managerial functions, and there are no incentives in place for the Managing Director to optimize the performance of the fund. First, it i s unclear whether the governing body has the technical capacity to design suitable investment policies and monitor performance. The Investment Committee i s the technical body that bringstogether experts inthe areas o f economics and finance, but it plays a purely advisory role and has little influence on investment decisions. Second, the Supreme Council does not have the power to select and remove the ManagingDirector o f the CSRO (who i s appointed by the Head o f MPO). The Managing Director i s therefore not really accountable to the governing body. Finally, there i s no mechanism linkingthe performance o f the fund to the compensation o f the ManagingDirector and the management team. While the Investment Company was created with the objective of increasing the level of independence of investment policies, the current governance structure is unlikely to achieve this goal. The two bodies that share governing responsibilities, the Supreme Council o f the Investment Company and its Board of Directors, are composed o f top-level officials from the CSRO and the government. Again, it i s not clear whether either o f the two bodies has the technical capacity to design investment policies and evaluate the performance o f the Managing Director. The Board o f Directors of the Investment Company does have the power to select and remove the Managing Director, but there are no policies linkinghidher remuneration to the performance o f the fund. Disclosure, auditing, and custody practices do not promote accountability. Managers have inprinciple full access to the financial informationofthe different companies andto the results o f the various investment activities. This information, however, i s not processed and reported ina simple way for the consumptionbyplanmembers. Infact, the CSRO does not produce the equivalent o f an Annual Report for its members, and no formal process exists for a worker or beneficiary to access financial information. Financial audits o f the CSRO and the CSRO-owned companies are conducted periodically by the Supreme Auditing Organization, which i s a government body. The fact that there are no audits by completely independent firms i s a negative factor interms o f accountability. Like the SSO case, the CSRO does not use an independent custodian. The custodian for the Investment Company resides in its own financial department. The custodian for the CSRO i s also part o f its financial department. The same is true o f the Housing Company. 43 There is no clear statement of investment policies and investment strategies do not follow basic principles in terms of asset exposure limits. As in the SSO, the practice o f limiting investment to 5 percent o f the capital o f any given company i s not followed by the CSRO. For the Investment Company, the only restriction i s that investments remain below the level giving the CSRO representationon the Board o f Directors. There are no limits interms o f the share o f a given company inthe CSRO portfolio. The current strategy to recapitalize and restructure the transferred public companies to improve profitability is highly risky. Due in part to an inappropriate governance structure, the CSRO's activities are starting to deviate fkom the original mandate, imposing unnecessary risk on workers savings. As compared to the SSO, the CSRO is not mandated by law to implement social or economic development policies. The basic mandate i s simply to collect contributions, manage reserves to the benefit o f the plan members, and pay pensions. CSRO activities that relate to the direct management o f companies are increasing, inpart as a result o f the transfer o f the public companies. If some o f these companies really have the potential to become profitable, it is not clear why the CSRO should assume all the risks involved in the restructuring process. Indeed, these companies could attract other investmentpartners who are willing to share the risks inexchange for higher profits inthe future. As inthe case of the SSO, directly managing companies can create serious conflicts of interest. For instance, credits to the companies are likely to be approved without the proper assessment of risks. Inan effort to improve the financial sustainability of the companies, anticompetitive practices may be put in place (e.g., subsidies, price floors). These practices can jeopardize the development o f other companies operating inthe sector. The mechanisms that are being used to refinance public debt are costly and non- transparent. The government still has accumulated arrears with the CSRO amounting to USD 1,800 billion. To pay part o f this debt, a new transfer o f companies i s being considered. This transfer could bringadditional losses to the CSRO as the assets are not valued at market prices. Even if the companies are not directly managed by the CSRO, resources will need to be allocated to outsource management and/or find investment partners to restructure part o f these companies. Hence, the CSRO i s implicitly being converted into a government restructuring agency. As discussed in Section 7.2, other alternatives could be considered to refinance the public debt that do not put excessive risks on worker savings. 44 5. TAXTREATMENT OF WORKER SAVINGS The tax treatment of pensions has important implications for fiscal policy, individual savinghonsumption decisions, and equity. A generous tax system can encourage pension savings but also bring costs in terms o f forgone revenue. Adverse redistributive effects may also take place if high-income individuals benefit more from tax exemptions than low-income individuals. This short section benchmarks the Iranian pension system with best international practice^.^' The tax treatment o f non-pension income i s not discussed. I n Iran, contributions and income generated by the investments of reserves in the SSO and the CSRO are taxed alike. Pensions are tax exempted. This i s known as the tax, tax, exempted scheme (TTE). It i s a comprehensive income tax, as consumptiodexpenditures and savings are both taxed. This is a rare form o f taxation for retirement income. Among OECD countries only New Zealand seems to apply this scheme. Best international practices suggest exempting interest income from retirement savings and taxing either contributions or benejts. In other words, consumption-type taxation is applied to mandatory retirement income. This i s the mechanism used by most OECD countries as a way to induce long-term savings. The issue then becomes whether to tax contributions (the tax is front-loaded) or benefits (the tax i s back-loaded). The majority o f OECD countries exempt payroll contributions from taxes and tax pensions. This i s known as the EET (exempt, exempt, taxed) treatment. It i s also possible to have TEE schemes. While both schemes are likely to mobilize the same amount o f resources,39governments may prefer one or the other depending on liquidity constraints. For instance, some governments may not be able to afford EET schemes as tax revenues are delayed. The Government of Iran could consider exempting the SSO and CSRO's investment income from taxes. Nevertheless, it is important to notice that this does not mean introducing tax exemptions for the companies where the funds are invested. Indeed, this would create undesirable economic distortions. Companies owned by the CSRO or the SSO should have the same tax treatment as companies not owned by the funds. It i s the after-tax profit from the activity o f the companies that i s distributed to the SSO or the CSRO that should be exempted from the tax (see Section 7.2). ~~ ~ 38 For a review o f altemative practices in terms o f tax treatment of retirement savings see Whitehouse (2001) and Lindeman (1999). 39 This assumes that the tax rate is constant andthat the discount factor is equal to the interest rate on savings. 45 6. A FRAMEWORK FOR PENSIONREFORM Prior to presenting reform options for the pension system in Iran, this section outlines a framework to guide policy discussions. It starts by identifying the rationale for government involvement in the design and implementation o f a public pension system and its social objectives. Then a typology of pension mechanisms that can achieve these objectives is proposed. Finally, a short review o f international experiences regarding the implementation o f these various mechanisms i s provided. 6.1. Objectivesof the Public Pension System, Costs, andImplementation Mechanisms The question of what is the rationale for government involvement in the area ofpensions is central to discussions about pension reform. Prior to deciding reform options, policymakers and civil society ought to have a clear idea of the social objectives that they are trying to achieve through the public pension system. In other words, why i s it better for a given society to put in place a public pension system instead o f letting individuals plan for their retirement through formal mechanisms (e.g., savings accounts) and informal mechanisms (family or community support)? There are at least three reasons. First, some individuals simply cannot save enough for retirement due to low-income levels and/or frequent periods o f unemployment; these individuals may not always be able to count on their families for support. Second, individuals may have difficulties thinking about the future and as consequence consume excessively when young. Finally, even those individuals who have the means to save and are willing to save may face problems when managing the risks associated with the investment o f their capital. Here it is argued thatpension reforms should be conducted with three main social objectives in mind for the pension system:40 i)guaranteeing a minimum pension at retirement to those individuals who did not have the means to save enough; ii)encouraging individuals to save for their retirement; and iii)assisting individuals to manage investmentrisks. Pension mechanisms that can achieve these objectives, however, also bring costs. First, guaranteeing a minimumpension for low-income individuals involves transfers that are usually financed through taxation and that add distortions to the economy. Guarantees can also create negative incentives for work and impose large contingent liabilities on governments. Second, mandating savings can have adverse effects ifthe contribution rate i s set too high(e.g., evasion, lower social welfare, excessive labor market distortions). Third, if the government assumes excessive financial risks, macroeconomic stability and the welfare o f future generations can be compromised. Hence, when analyzing pension reform options, decisions need to be made in terms o f the following: How large should the mandate to save and the targeted replacement rate be? How much redistribution is affordable? How should jnancial risks be distributed between the government, individuals, and private insurers? Who should manage and how to regulate investment policies? The objective o f a pension reform program should be to maximize social objectives while minimizing micro- and macroeconomic distortions. A large number o f pension mechanisms could be considered to achieve required social objectives; they are characterized by four elements: i)how the benefits are computed; ii)how the system i s financed; iii)who manages the system and how; and iv) who bears the financial risks. Benefits can be computed on the basis o f work and wage history (the case o f defined benefits schemes-DB); or on the basis o f the total contributions accumulated, rates o f return received on investments, and life expectancy (defined contribution systems-DC); or 40For discussions about the rationalof the public pensionsystem see for instanceWorld Bank (1994). 47 combinations o f both. In terms o f financing, pensions can be paid from current contributions (the case o f the pay-as-you-go system-PAYG); from accumulated savings (fully funded system-FF); or a combination o f both. Management can be public or private with different proportions o f mandatory versus voluntary savings, different types o f regulations for the investment o f fund reserves (e.g., centralized versus decentralized management), and different mechanisms for collecting contributions, keeping records, and reporting. Finally, financial risks can be distributed differently between the government, private insurers, and individuals. Hence, there i s a large menu o f pension mechanisms that could be put in place by selecting alternative options within each o f the four dimensions outlined here. How efficient some o f these choices are inachieving social objectives while minimizing distortions i s next discussed. Social Obiective 1: Protecting the Elderlv Poor It is useful to distinguish between redistribution toward the core elderly poor, who are usually outside the contributory system, and redistribution within the contributory scheme, ideally from high-income workers to low-income workers. Redistribution toward the elderly poor involves social assistance programs, in-kind or in-cash transfers, that should befinanced directlyfrom the central budget. The elderly poor, and those individuals likely to become poor during old age, are usually outside the contributory system. These individuals face liquidity constraints and short-term risks that outweigh the risk of longevity. For them joining a contributory scheme i s usually welfare decreasing. Non- contributory schemes need to be in place to protect these individuals. These are discussed in more length in Section 7.4, which addresses the issue o f how to expand coverage. Here it i s only emphasized that these programs need to be implemented in a way that does not create negative incentives for working and for joining the contributory system. Most countries pursue anotherform of redistribution, from high- to low-income workers within the contributory system (usually a DB-PAYG scheme). The costs of this redistribution mechanism, however, can outweigh the benefits. Whether redistribution from high- to low- income workers inthe contributory system i s desirable or not i s a matter o f social preferences. Ifit is, however, policy makers ought to consider the less-distorting redistribution mechanism. Achieving the redistribution through the pension system41 involves taxing labor and thus distorting labor markets. An alternative is to finance the transfers (e.g., guarantees on minimum pensions) directly from the central budget. Countries following this approach include Australia, Chile, andNew Zeeland. Moreover, in the case of a DB-PAYG system, transfers can be regressive, meaning that low- income individuals receive lower rates of return than high-income individuals; furthermore, future generations usually receive lower rates of return than current generations. As exemplified inthe discussions about the SSO and the CSRO, a given individual's rate o f return from the DB-PAYG system i s determined by two main factors: the average growth rate o f the real wage; and life expectancy. Individuals whose wages grow faster and who live longer, thus receiving pensions for longer periods o f time, have higher rates o f return. These individuals are often educated and healthy workers, more likely to belong to middle- and high-income households. In principle, the system can be designed so that low-income individuals receive higher rates o f return than high-income individuals (for instance by introducing minimum pensions), but progressivity is not an inherent feature of the DB-PAYG system. Another characteristic o f the DB-PAYG system i s that the sustainable rate o f return that the system can 4'Transferswithin the pension system are not a unique characteristicof defined benefit schemes with pay-as-you-go financing (DB-PAYG). Transfers can also be implemented in a filly funded defined contribution system with individual accounts(DC-FF-IA),for instance, by discounting a few percentagepoints from the contributions of high- incomeworkers to guarantee aminimumpensionto low-income workers. 48 pay decreases over time. The implication i s that new generations receive lower rates o f return than old generations. Thus, there i s an implicit transfer o f resources from the young to the old. Even if the system i s designed with some degree o f progressivity within generations, high- income members of the "old" generation can still receive implicit transfers from low-income members of the ``young" generation. Social Obiective 2: EncouragingIndividuals to Save The challengefor policymakers is to design a pension system with the right balance between mandatoy and voluntay savings. Any type of public pension mechanism attempts to promote savings by mandating a minimum contribution rate. However, if the contribution rate and the rates of return are not set appropriately, the mandate can have adverse effects. When contributions are set too high, that i s when individuals are being asked to save beyond their optimal rates (assumed to be unknown to them), the mandate will reduce and not increase social welfare. High contribution rates and low expected rates o f return (resultingfor instance from political uncertainty) can also promote evasion and underdeclaration o f wages, thus undermining coverage and financial sustainability. Underdeclaring wages i s another mechanisms to escape the mandate in order to obtain higher expected rates o f return on savings or a better allocation o f income between consumption and assets. Underdeclaration i s more likely when part o f the benefits received are not linkedto the contributions. For instance, ina DB system when the pension i s computed only on the basis o f the most recently receivedwages, declaring in-fullthe wages received at the outset o f the career contributes nothing to the final pension. Ingeneral, finding a mandate that proves desirable for a majority o f individuals (at least when old) i s a difficult ~ndertaking.~' An alternative mechanism to mandatoy savings is toprovide incentives that induce individuals to save. Popular instruments to encourage long-term savings include tax credits on interest income, preferential interest rates, and matching grants. If properly used, these instruments increase the relative rate o f return that individuals receive on long-term savings (see Section 7.5). Nonetheless, by themselves these instrumentsare not sufficient. A necessary condition to promote voluntary savings i s that households perceive a stable macroeconomic environment and trust the financial system. Voluntary savings are unlikely to develop in economies with high inflation rates, volatile exchange rates, restricted investment opportunities, and poorly regulated and supervised banks. Hence, prior to considering instruments that add points to the expected market rate o f return on long-term savings, governments need to create a macroeconomic environment and regulatory institutions that raise the expected market rate o f return. Social Obiective 3: Helping Individuals to Manape Investment Risks The allocation o f risks between individuals, private insurers, and the government depends on how benefits are computed and the type o f guarantees providedby the sponsor.43 When financial markets are underdeveloped, the rationale for government provision of pensions is the strongest. In this case, individuals may lack the necessary instruments to transfer income into the future or to manage their risk o f longevity (e.g., annuity markets). Even if the instruments exist, an inappropriate regulatory and supervisory framework can make these instruments too risky. Individuals close to retirement can be extremely vulnerable to a sudden downturn in financial markets. Governments can mitigate negative impacts, for instance, by guaranteeing a minimum pension like in Chile, but the costs o f these guarantees 42 See Robalino and Sabatini (2003). 43 See Lindeman and Galer (2000). 49 can be ~onsiderable.~~ DB-PAYG appears to be a mechanism to help individuals smooth The consumption over time while shielding them from financial risks and the risks o f longevity. Still, DB-PAYG systems can transfer excessive risks to the government and therefore tofuture generations. In DB-PAYG systems the rate of return that individuals receive is not pre- determined, since it depends on the growth rate o f wages and life expectancy. Inthe absence o f appropriate indexation, individuals also face the risk o f inflationthat can erode the real value o f pensions. The bulk of the financial risks, however, i s assumed by the government. If the contingent liabilities o f the government are too high, fiscal and macroeconomic stability can be compromised and with it the welfare o f future generations. A more equilibrated allocation of risks between the government and individuals can be achieved by combining defined benefit systems with defined contribution systems. By relating part o f the benefits to the contributions made by the individual and macroeconomic and demographic developments, the government implicitly reduces its share o f the financial risk. For instance, in the case o f the NDC system (see Section 7.1), pensions received can be a function of contributions accumulated that earn an interest rate equal to the growth rate o f the economy. Government guarantees can still be implemented and, thus, assure a minimum pension upon retirement after a full-career. Secondarv Obiectives: Enabling Sustainable Economic Growth Increasing the level of funding of a pension system while improving the management of reserves can contribute to the accumulation of long-term savings,financial-sector development and through this channel economic growth and lower output volatility. The traditional view regarding how a mandatory-fundedpension system can contribute to economic growth focused on the level o f savings o f an economy. The argument was that mandatory-funded systems could increase national savings, thus promotingthe accumulation o f capital and growth. While there i s some evidence that in financial markets where plan members have restricted access to credit, mandatory systems increase the national saving there i s in general scant support for the proposition. Moreover, even if savings increased, the effect on growth would depend on how the savings are allocated. If higher saving rates are simply accompanied by higher surpluses inthe capital account, growth effects would be slimor nil. On the other hand, there i s growing evidence that mandatory-funded schemes can alter the composition o f savings (favoring long-term versus short-term savings) and can promote the development o f securities markets, especially in countries with closed capital accounts making them more liquid and deeper as well as more sophisticated and i n n ~ v a t i v e .At ~ ~the same time, there i s evidence that financial-sector development contributes to higher economic growth47 and lower output volatility. Since pension benefits are paid out o f current GDP, regardless o f how the system i s financed, a pension system that promotes economic growth is at the same time promoting its financial sustainability and allowing its members to enjoy higher rates o f return on their savings. 44See Lachance and Mitchell (2002). 45Bailliu and Reisen (2000). 46Impavido, Musalem, and Tressel (2002); Walker and Lefort (2000). 47Levine (1999); Levine, Loayza, and Beck (1999). 50 6.2. International Experiences: OneSizeDoes Not Fit All The purpose o f this section is not to provide a complete overview of pension reforms around the world; other references are suggested for that purpose.48 Rather it i s simply to give the reader an idea o f the heterogeneity o f reform efforts and the need to tailor reform programs to the initial conditions ofindividualcountries. I n practice, the reform of public pension systems generally takes one of two paths: non- systemic reforms and systemic reforms. Non-systemic reforms refer to reforms involving parametric changes to key parameters, such as retirement ages, vesting periods, accrual rates, and so forth. Systemic reforms are more complex and typically affect the underlying management and financing o f a pension system or its method o f determining benefits. Three basic types o f systemic reforms have been observed worldwide. These include (i) shifting- either in part or entirely-from a pay-as-you-go defined benefit scheme to a funded defined contribution scheme where funds are held in individual investment accounts and managed by private parties, (ii) shiftingfrom a fully funded defined contribution scheme to a pay-as-you-go defined benefit scheme managed by the public sector (the exact opposite o f the former approach), and, more recently, (iii) introducingnotional defined contribution accounts (NDCs). What are the regional trends? Ofthe countries that have reformedtheir public pension systems over the past few decades, most (roughly 70 percent) have pursued parametric reforms.49 The remainder have pursued systemic reforms. Funded individual accounts were introduced in most o f the Latin American countries that undertook reforms (e.g., Argentina, Bolivia, Chile, Colombia, Costa Rica, El Salvador, Mexico, Peru, and Uruguay). Australia and the United Kingdom have also followed this approach. In funded individual accounts, contributions accumulate in individual investment accounts and earn market interest rates. At retirement, the capital in the account i s transformed into an annuity (i.e., a pension paid until the beneficiary dies). The reverse approach-moving from a fully funded defined contribution scheme to a pay-as-you-go defined benefit scheme-is less common but has beenpursued insome countries in Sub-Saharan Africa such as Nigeria, often to address the disappointing performance of state- runprovident funds.50The third approach to systemic reforms-NDCs-is relatively new, at least in its current form.5' This approach to reform changes the method used to compute pensions (from a classic defined benefit formula to a defined-contribution-type formula) without actually changing the scheme's underlying pay-as-you-go financing. It is being pursuedby Latvia, Sweden, Poland, and Mongolia. Inthe case o f the three European countries, NDCs are being augmented by funded individual accounts to encourage capital-market development and to increase total replacement rates at retirement. Within these general approaches to reform, however, considerable variation exists. For instance, in terms o f implementation schedules, methods o f financing the transition costs, if there are any (transition costs arise, for example, when shifting from pay-as-you-go financing to a funded system, because invested contributions are no longer available to pay benefits to current beneficiaries), and whether the reform o f the public system i s done in isolation or as part o f a multipillar approach which involves the introduction o f mandatory or voluntary private pension schemes. Tables 16 and 17 illustrate the variations in terms o f systemic reforms introduced inEasternEurope and Central Asia, and Latin America. 48 See for instance Schwarzand Demirguc-Kunt(1999). 49 Ibid. 50 For a discussionofpensionreformstrategies beingpursuedinAfrica see Bonnerjee (2001). 5' Valdes-Prieto(1999) points out that a scheme resemblingNDCs was proposed in the UnitedStates in 1968 by Buchanan. 51 Table 16: PensionReforms in EasternEuropeand CentralAsia Projected Second Sizeof Pension Fund Vorkforce Country Status start Date First Pillar Pillar as Shareof Assets in 2020 Pillar Strategy (2003) (% GDP) Mandatory New Hungary i i 1 1 Operating Jan 1998 DB-PAYG 6% 31% 45% Entrants, Voluntary Others I I Kazakhstan Legislated, Guaranteed I I Operating Jan 1998 Minimum 10% 30% 100% Mandatory Poland Legislated, mc Mandatory <30, Overating Jan 1999 7.2% 33% 70% Voluntary 30-50 July 2001 2% I 1 I I Latvia Legislated, Mandatory <30, Operating (NDC NDC growing 20% I Jan 1996) to 9% DB-PAYG Croatia Legislated Jan2003 (point 5% 25-30% Mandatory <40, system) 60-70% Voluntary 40-50 2% Bulgaria Legislated, Operating Jan 2002 DB-PAYG growing Mandatory <42 to 5% 1 Estonia Legislated lJuly2002 I DB-PAYG 1 6% I 20% 60% (ovt-out +2%) ~ Partially DB-PAYG Mandatory >20 Romania Legislated, Then Jan 2003 (point 8Yo 30% 75% Years from Questioned system) Retirement Macedonia 26% 15% New Entrants Partially 2% (<35) Russia to 6% Mandatory <50 Operating (36-50) 2% Ukraine Partially Mandatory Legislated Jan2003 1 PAYG 1growing to7% 1 NewEntrants I ~ Partially Kosovo Legislated, Jan 2002 Minimum 10% Mandatory Operating- I I Source: Wo I Bank, Human Development Network, Social Protection, For a discussion on pensions systems in Eastern Europe, see Lindeman et al. (2001). 52 Table 17: PensionReformsinLatinAmerica Nicaragua 2001 Full Optional? Phased-out Venezuela 2001 Full Optional? Phased-out Old New Special Brazil 200x? UF/DB Mixed Decentralization Costa Rica 200x? UF/DB Mixed Good Prospects Ecuador 200x? UF/DB Mixed Argentina 2 Guatemala 200x? UF/DB Mixed Myriad Paraguay 200x? UF/DB FF/DC High Costs Which reforms have been the most successful? Pension reform necessarily involves both political and economic costs and benefits. The troublesome reality for policymakers is that costs tend to be felt inthe short term, while benefits tendto manifest over the long term. 0 Parametric reforms typically generate relatively less political resistance and have fewer fiscal implications in the short term, particularly in cases where, as an alternative to parametric reform, officials are considering moving from a pay-as-you-go system to a funded system and must find the resources to fund the transition costs. Parametric reforms, however, usually leave some structural problems unresolved. These reforms typically fail to (i)reduce distortions in the labor supply and in individual decisions regarding savings and retirement timing; (ii) eliminate heterogeneity in rates o f return; and, (iii)fully assure the long-term fiscal sustainability o f a system, as there are no automatic mechanisms to adjust the parameters o f the system in response to changes in economic and demographic conditions. In addition, these reforms may lack credibility among young generations which harbor unfairly pessimistic expectations about the capacity o f the system to deliver on its promises (which may encourage evasion and, thus, fulfill expectations). Finally, parametric reforms fail to exploit the spillover effects that a funded system may generate for the economy by helping to promote financial-sector development. 0 Introducing fbnded individual accounts, on the other hand, i s complex and logistically and administratively challenging. I t typically encounters more political resistance, invokes higher short-term fiscal costs, and imposes new risks on participants. It should be pursued only if key preconditions have been met. First, countries need to be in a favorable fiscal position to finance the tran~ition.~~Second, countries must have sufficient regulatory capacity and a solvent banking system. A limited supply of financial instruments (e.g., corporate bonds and stocks), the absence o f a secondary 52See Holmann 1998. 53 market for government debt, and limited human capital in pension fund management are all factors that should discourage the introduction o f mandatory-funded individual accounts. 0 Replacing defined benefit systems with Notional Defined Contributions (NDCs). While N D C systems generally remain unfunded(i.e., they continue to operate on a pay- as-you-go basis, perhaps with an investment fund to cushion long-term demographic changes), they have the potential to offer some o f the benefits o f funded individual accounts. By linking contributions to benefits in a more transparent way than can be done with a defined benefit (DB) formula, NDCs could reduce incentives to under- report income or evade the pension system altogether. NDCs can also improve financial sustainability by bringing the rate o f return paid by the system towards its sustainable level, and by introducing automatic mechanisms to account for unexpected changes in life expectancy, macroeconomic shocks, andor changes in preference for leisure and consumption. They can also facilitate the portability o f funds, enable more substantial reforms to disability and survivor pensions, and, for countries seeking to do so, do not preclude the eventual introduction o f mandatory-funded individual accounts. Intheory, DB formulas can be set up in a way that exactly replicates NDC outcomes, thus generating homogenous and sustainable rates of return. The DB formula that achieves this, however, i s considerably more complex.53 For the average worker the linkbetweenthe benefit and the contributionmaybe less transparent than inthe case of the N D C formula. If different funds within or across countries have different DB formulas, transferring benefits from one fund to another i s less straightforward than in the N D C case. This brief review of international experiences elucidates the heterogeneity o f pensions systems and therefore the complexity o f designing a reform program. The main message i s that policymakers ought to keep in mindhow different choices-in terms o f the size of the mandate to save and the targeted replacement rate, the way benefits are calculated and financed, the mechanisms used to managed and regulate investments, and the distribution o f financial risks-affect the social objectives that the public pension system i s supposed to achieve and the economic efficiency. 53See Robalino and Sabbatini (2003). 54 7. REFORMINGTHE IRANIANPENSIONSYSTEM The main results from the analysis o f the Iranian pension systemcan be summarized as follows: 1. The system targets very generous replacement rates at all levels of income that are neither affordable nor sustainable. Pension expenditures are growing fast and despite high contribution rates and still-favorable demographic conditions the system is heading toward a financial collapse. The unfunded pension liabilities o f the SSO and CSRO for the period 2002-2070 surpass 170 percent o f today's GDP. Thus, the government i s assuming excessive risks that can ultimately compromise fiscal and macroeconomic stability and the well-being o f future generations. 2. The situation is complicated by a structure of payroll contributions, benefit formulas, and eligibility conditions that distort labor supply, retirement, and savings decisions. High contributions may discourage enrollment, promote the informalization of the economy, and crowd-out other forms o f savings. Benefit formulas and elegilibility conditions provide incentives to underdeclare wages and to evade and game the system. A generous minimumpensionrewards retirement over work. 3. The system is prone to generating adverse inter- and intragenerational transfers. First, given a benefit formula that penalizes low-income workers (blue-collar workers) by considering only the last two years o f salaries in the calculation o f pensions. Second, through a very heterogeneous structure o f rates o f return. Third, through eventual bailouts that favor workers inthe formal sector, and cuts inbenefits and/or higher taxes for future generations-including low-income workers. 4. Current investment policies are risky, complex, and not necessarily in the benefit of plan members. The pension funds are being transformed de facto into agencies to restructure public companies. The funds over-expanded their mandate and now have considerable market power in several economic sectors-including the stock market. This interferes with corporate governance and hampers the development o f the financial sector. In part, inappropriate investment policies reflect weak governance structures, as processes to select and structure the governing bodies, define fiduciary responsibilities, and enforce accountability do not follow best international practices. The objective o f the reform o f the Iranian pension system i s to strengthen its social function (protecting the elderly poor, encouraging individuals to save, and assisting individuals to manage risks), while guaranteeing financial sustainability and reducing economic distortions. This section examines the type o f policy interventions that could be considered. I t starts by analyzing what could be done to provide sufficient, affordable, and sustainable income replacement at retirement,to improve equity, to create a better balance between mandatory and voluntary savings, and a more prudent distribution o f risks. Then the section outlines a set o f interventions to improve the management o f the funds as well as mechanisms to promote the development o f voluntary savings. The final section discusses how reducing distortions in labor markets could contribute to expand coverage and what options could be considered to cover vulnerable population groups operating outside o f the formal economy. 55 7.1. Dealing with the Finances of Current Systems while Improving Equity, Incentives, Reallocating Risks, and Creating a Better Balance between Mandatory versus Voluntary Savings The government and civil society need to make a decision as to what is an adequate and affordable level of income replacementfor retirement and how this level of income should be generated by a combination of mandatoy and voluntary savings. It has been evinced that the SSO and the CSRO currently provide high replacement rates to the majority o f workers regardless o f income levels (see Figure 21). In the sample o f high-income countries studied, replacement rates for the average full-career worker range between 35 percent and 88 percent, the average being 57 percent. Hence, Iran needs to consider reducing current replacement rates of the mandatoy system,particularly for high-income workers while designating a more prominent role to voluntay savings as a source of incomefor retirement. At the same time, the benejt formulas and eligibility conditions need to be reviewed to improve equity and incentives. Figure21: IndividualReplacementRatesfor the SSO, the CSRO, and SelectedCountries 140 , .I120 1 \, I ~ -----t-Australia - ~ -Canada 3B -France 100 P -Korea .e -0 80 .S Sweden g 60 - - -CSRO %n 40 --Iran SSO Iran .E z ~ 20 e E o 0 1 2 3 4 5 I Indinduals earnings (proportion of average) I Source: Whitehouse (2002) and missioncalculations for the SSO and the CSRO Inthis sectionthree types ofreformare proposed: 1. A first type o f reform maintains the level o f the current mandate to save (Le., contribution rates are constant) through the DB system, considering it the unique source o f savings for retirement. However, the targeted replacement rate for a full- career worker i s reduced to 80 percent (close to Spain), in order to reduce long-term deficits. Additional parametric adjustments are introduced to improve financial sustainability and equity and to reduce negative incentives for working. 2. A second type o f reform reduces the mandate o f the mandatory system for new contributors, targeting a replacement rate o f 60 percent for a full-career worker (40 years). This allows a reduction inthe contribution rate. The reform provides room for the development o f funded voluntary savings and eventually the introduction o f a mandatory funded pillar. 3. A third type o f reform reduces the mandate o f the current DB-PAYG system for new contributors, but changes the benefit formula to an NDC type in order to reduce contingent liabilities, while generating a more transparent link between contributions and benefits and better incentives for work over retirement. This reform also targets the development o f voluntary private funded pensions. 56 Reform Type I:current mandate is preserved and the system remains a DB-PAYG Under this reform option, addressing the financial problems of the system is a more or less mechanical process that involves the following steps: 0 Given the contribution rate, evaluate the unfunded pension liability, which in the case o f a pay-as-you-go system i s equal to the present value o f future pensions, minus the present value o f future contributions, minus current reserves. 0 Depending on the severity o f the unfunded pension liability and the non-pension deficit o f the government, reduce the rate o f return paidto individuals by reviewingthe pattern o f income replacement across levels o f income, and tightening retirement conditions. In addition, it is desirable to review a number of mechanisms that affect equity as well as individual incentives for evasion, underdeclaration, early retirement, or abuse. These include the number o f years used in the calculation o f the final wage, pension indexation mechanisms, rules for early retirement, and eligibility conditions for disability and survivor benefits. Next specific recommendations in each o f these areas are discussed. Reviewing pattern of income replacement. There are three key parameters that influence the pattern o f income replacement: (i) the accrual rate; (ii) minimum pension; and (iii) the the maximum covered wage. Accrual rates should be set in reference to a full-career worker (40 years) in order to reach the desired replacement rate. Inthe case o f this reform, if the targeted replacement rate for a full- career worker i s 80 percent, the accrual rate should be set at 2 percent per year. The adjustment o f the accrual rate should also be gradual and should only affect new contributions. The minimum pension should be set in a way that does create incentives for retirement over work. Allocating this minimum pension without a minimum number o f years o f contribution negatively affects the finances o f the system and can be inequitable. Indeed, some workers will choose to enroll the system close to retirement. Because these individuals accumulate a low replacement rate, they are often eligible for the minimumpension (independently o f their level o f income). Their rates o f return can increase considerably. Introducing a maximum covered wage ensures that the public system limits pension expenditures on high-income individuals. Individuals with salaries above the ceilings effectively receive lower replacement rates, even if in absolute terms, their pensions are equal to those o f individuals with salaries equal to the ceiling. This policy also allows high-income individuals to diversify their sources o f savings. Inthe case o f Reform Type I, however, there would be no changes in ceilings. To summarize, inthe case o f ReformType I, the following recommendations are made: 0 Gradually reduce the accrual rate to 2 percent per year. The new accrual rate can be reached over a period o f 10years. 0 On the basis o f the latest household survey and estimates o f the poverty line, conduct a study to determine the appropriate level o f the minimum pension. The minimum pension should be set below the minimum salary. A maximum level o f 70-75 percent o f this minimum salary i s suggested. At the international level minimum pensions represent 30-40 percent o f the average wage. The study should provide estimates o f the number of individuals who are likely to be candidates for a minimumpension and the cost ofthe policy intervention. 0 Pay minimum pensions only to workers who have contributed for a minimum o f 30 years. 57 Rationalizing retirement conditions. Further reductions in rates o f return are achieved by increasing the period during which individuals make contributions, reducing the period during which individuals receive pensions, and reducing the pension o f individuals who retire early. Three sets o f parameters are adjusted in this case: the vesting period, the minimumretirement age, and the age-dependent penalties for early retirement. Each i s discussed inturn. Vestina ueriods. Three issues need to be kept in mind. First, adjustment to vesting periods should be gradual. Vesting periods should not increase faster than half a year each calendar year. Otherwise, individuals close to retirement would have to continue working until the new target i s reached. As an example, if the current proposal in the SSO to increase the vesting period from 10 years to 20 years i s implemented in a period o f 10 years, individuals who are one year away from retirement will have to wait ten additional years. Second, when increasing vesting periods there i s a danger o f introducing incentives for evasion. For instance, workers aged 45 years or more are unlikely to join a systemwhere the minimumvesting period i s equal to 30 years. Third, changes investing periods can induce abrupt drops inthe rates o f return for older workers (see Figure 22). On the other hand, allowing workers to retire with a few contribution years without adjusting replacement rates can generate disproportionably high rates o f returnfor workers joining the system relatively late inlife. :1L/,,/I Figure22: Effect of Increasing Vesting Periods on Rates of Return ;ol I- -gw=07. qw=2z gW-4% 5 u - u;;i - E - - _ _ - - - _ - - - - - - - - - & m - _ - - -.- \ 0 - - - _---' & m - - \ 8 0 20 25 30 35 40 ' 0 50 0 20 2 s 30 35 40 4 5 50 Age when individual joins the system ~ g when 'naividual joins the system e Source: Missioncalculations on the basis of SSO rules. Note: Individuals are assumed to retire as soon as retirement conditions are met. First panel considers a 10-year vesting period. Second panel considers a 20-year vesting period. Retirement age and early retirement rules. A male worker retiringat age 60 today can expect to live until age 76. A female worker retiring at age 55 can expect to live until age 77. On average, pensions will be received for 16 and 22 years respectively. Ten years from now, a male worker retiring at age 60 will expect to receive a pension for 18 years and a female retiring at 55 for 25 years. As people live longer it i s reasonable to expect them to work longer. Otherwise, the finances o f the system cannot be sustained. InIran, both the SSO and the CSRO could initially target retirement ages o f 63 for both men and women. As in the case o f the vesting period, retirement ages should not increase faster than 0.5 years per calendar year. Hence, the new retirement age for males could be attained in a period o f 10 years, while for women it would require twenty years. Over the long run, it i s desirable to have an automatic indexation mechanism that periodically adjusts retirement ages in-line with changes in life expectancies. Clearly, for various reasons, individuals may wish to retire before reaching the minimumage. This should be allowed with actuarially fair adjustments to pensions. Inother words, ifa pension i s going to be received over a longer period o f time, then the pension should be lower. Actuarial reductions inreplacement rates increase with the numbers o f years that the individual advances retirement. 58 In summary, the following recommendations are made to reduce benefits and rationalize retirement conditions: 0 Eliminate multipleretirement conditions and benefit formulas and target a uniform rule for all workers, inboth the SSO and the CSRO. 0 Target a retirement age o f 63-65 years for both males and females with no maximum retirement age. The new age could be reached in a period o f 6-10 years for males and in a period of 16-20 years for females. Once the new retirement age is reached, it should be indexed by life expectancy. 0 Gradually increase the vestingperiod to 20 years for both males and females. The new vesting period should be attained in a period o f 10 years for males and 20 years for females. 0 Allow for early retirement with actuarially fair reductions inreplacement rates. Reducing the strategic manipulation of wages and improving equity by gradually considering all wages in the calculation of the pension. As discussed in the diagnostic sections, underdeclaration o f wages i s in part explained by the fact that the final pension i s not linked to the predominant career-wide wage level. Because blue-collar workers would benefit from an increased number o f years included in the average wage used to calculate the pension, because they usually have peak earnings relative to the average inmid-career. Hence, for both funds the following recommendationsare made: 0 Update individualrecords o f current contributors to include informationo f past wages. 0 Change the benefit in such a way that each fiscal year an additional year o f wages i s included inthe calculation o f the pension. 0 Index the wages included in the calculation by the growth rate o f the system's average wage. Protecting pensions from inflation. The laws regulating the SSO and CSRO operations mandate adjustments in nominal pensions to protect beneficiaries from increases in the cost o f living. The way the current legislation is written, however, discretion in adjustments occurs. To ensure that pensions are not eroded by inflation an automatic indexation mechanism i s needed. I t i s therefore recommendedto: 0 Modify current articles governing the indexation o f pensions in a way that the responsibility to implement the adjustment i s directly transferred to the department in charge o fprocessingpayments. 0 The new article(s) should define the index that will be used to adjust the pension and the periodicity o f the adjustment. It is recommended to use the growth rate ofprices as the index. This policy, however, should only be adopted ifthe other reforms have beenimplemented. Adjustments in disability and survivor pensions. There i s an ongoing debate over the best mechanism to provide disability and survivor benefits. Traditionally, DB systems are prone to abuses. Moreover, the way survivor benefits are implemented tend to provide little incentive for survivor spouses to enter the labor market; and, in the case o f death o f the latter, surviving children can be left unprotected. Questions to consider include whether the SSO and the CSRO could outsource the provision o f these benefits, and/or whether i s it desirable to have independentpension rights for both spouses with accumulated pensionrights split after divorce or death. In the short run the new accrual rates used to compute old-age pensions should also 59 apply to disability and survivor pensions. The SSO and the CSRO should review the appropriateness o f the current accreditation process and reinforce eligibility conditions. A more detailed study o f current financing mechanisms should be conducted in the future. Over the medium term it i s desirable that disability and survivor pensions be financed independently o f old-age pensions. The downside of this strategy. Introducing severe cuts in benefits is difficult politically. Even after introducing adjustments, a sizable unfunded pension liability is likely to persist. Another problem with the current strategy i s that, even with demographic indexation o f the retirement age, the system remains vulnerable to changes inthe sustainable rate o f return due to the vagaries o f the economy. Indeed, there are no automatic mechanisms to adjust the system's parameters to reflect changes in macroeconomic performance. This creates a credibility problem over fiscal policy, as the contingent liability for the government (and future generations) remains large. The reform also creates a credibility problem among beneficiaries who may generate pessimistic expectations about the capacity o f the system to deliver its promises. This may encourage evasion thus fulfilling these expectations. In addition, the strategy leaves room for adverse redistributive effects.54 Moreover, contribution rates remain high, which discourages enrollment inthe system, impedes the diversification of savings, and negatively affects labor markets. Hence, this strategy i s considered only a temporary fix to the problems o f the current DB-PAYG. Reform TvDe 11: current DB svstem is downsized givinp a more prominent role to voluntary savings and an eventual mandatorv-funded svstem This reformconsiders that the current DB-PAYG will not be the only systemreplacing income for retirement. Therefore, it proposes reducing the size o f the mandate to save (i.e., the contribution rate) and the targeted replacement rate, as well as imposing a cap on the covered wage o f 3 times the economy-wide average, and a minimum pension o f 70-75 percent o f the minimumwage. The targeted replacement rate could be set initially at 60 percent for a full- career worker. This implies an accrual rate o f 1.5 percent per year. The remainder o f the replacement rate would be financed by capitalization, through a combination o f voluntary and possibly mandatory schemes. This reform would not affect current retirees or workers who are close to retirement. The reformcouldproceed as follows: 0 New workers inthe SSO and the CSRO enter a new DB scheme with some degree o f pre-hnding where the contribution rate i s lower (below 15 percent), where benefits are lower (an average accrual rate in the 1-1.5 percent range), retirement conditions are tighter (a minimum retirement age o f 63 years for both males and females subject to demographic indexation), and pensions are computed on the basis o f lifetime earnings. Wages are indexed by average earnings, and pensions are indexedby prices. 0 For those workers in the current system, benefits and retirement conditions are adjusted accordingly to the recommendations o f the previous section. For these workers, the unfunded pension liability i s financed out o f three sources: i)current reserves, ii) general revenues, and iii)probably part o f the contributions o f new workers. 0 The remaining contributions could be accumulated in a separate fund that depends on the SSO and the CSRO but that is subject to different governance rules and investment policies (see Section 7.2). 54This issue could be resolved ifall wages are introduced in the calculation of the pension and a cohort-dependent accrualrate, which is a function of the contribution rate and the life expectancy at retirement of the members of the cohort, is introduced. No country, to our knowledge, uses this type of formula, probably in part due to its complexity. 60 In parallel, necessary incentives and an appropriate regulatory and supervisory framework are put in place to promote voluntary savings in the form o f contractual savings (see Section 7.5). 0 Eventually, a DC-FF pillar could be introduced while the DB-PAYG is further downsized. The downside of this strategy. Under this reform the financial sustainability o f the DB scheme remains frail. While parameters could always be adjusted to reflect changes in demographic and economic conditions, in practice these adjustments can be subject to discretionary decisions. Another potential problem facing this strategy i s related to the transition costs. Since the new contribution rate would be lower and part o f the revenues could be accumulated, current pensions would need to be financed from general revenues. This imposes a fiscal burden over the short and medium terms that requires issuingdebt, increasing taxation, or reducing non-pension expenditures. None o f these options i s neutral from a macroeconomic point o f view. Reform Tvpe 111: Downsizing; the current DB-PAYG while introducing Notional Defined Contributions and womoting; voluntarv savinps In recent years Notional Defined Contribution Systems (NDCs) have evolved into a new alternative for reforming DB-PAYG pension systems. NDCs remain unfunded mechanisms, but attempt to emulate a defined contribution scheme. Indeed, the plan guarantees to its members a rate o f return on contributions that approximates the sustainable rate o f return o f a pay-as-you system (a function o f the growth rate o f the covered wage bill). At retirement, the pension i s calculated by dividing the sum o f indexed contributions by a so called "Gfactor", that takes into account the average life expectancy o f the individual. Advocates o f this mechanism emphasize the following advantage^:^^ 1. Overcomes issues o fthe political economy relatedto parametric reforms 2. Legally binds the financial sustainability of the system by linking rates o f return to macroeconomic performance and demographic changes 3. Allows for an easy and rapidharmonization o f retirement schemes across professions 4. Allows for more fundamental reforms o f disability and survivor benefits 5. Eliminates perverse redistributive features o f the traditional pay-as-you-go system 6. Provides incentives for work over retirement and may reduce incentives to evade. Since the model i s not a complete D C scheme, contingent liabilities persist, but part o f the risks are transferred to plan members. Hence, there i s a better distribution o f these risks between current and future generations. Relative to a DC-FF scheme, however, risks are lower because the volatility o f the return on Notional Accounts, the real wage bill growth, i s lower than the volatility o f the return on the funded system scheme, which depends on portfolio choice and the capital market). I t is emphasized that the NDCs outcomes could be approximated in a standard DB setting, if all wages are included in the calculation o f the pension and a cohort-related accrual rate i s introduced. For a cohort expected to retire at age a and time t, the accrual rate i s defined as the contribution rate divided by the G factor at age a and time t. Clearly, those individuals in the cohort who for some reason retire before or after time t will receive higher or lower rates o f return (since the accrual rate i s defined ex-ante and i s the same for all individual inthe cohort). 55 See Disney (1998) for a reviewofthe pros and cons ofNDCs. 61 The complexity o f the modified DB formula, however, suggests that if N D C outcomes are targeted, then the NDC formula shouldbe used. InIran, the implementation ofthisreformcouldproceedalongthe followinglines: 0 New workers enter a new system where the total contribution rate i s lower (below 15 percent), thus allowing for a better balance between voluntary savings and mandatory savings. 0 Contributions accumulate in "virtual" accounts, earning an interest rate that i s a function o f growth rate o f the covered wage bill. Stabilization mechanisms are also incorporated to reduce volatility in rates o f return and ensure financial sustainability. At retirement, the accumulated capital is transformed into a lifetime pension on the basis o f estimates for life expectancy (alternative mechanism can be used to index the pension).s6 Current contributors are allowed to switch to the new system. For them, an initial capital i s accredited to the "virtual" account, which i s calculated on the basis o f past contributions giventhe appropriate notional interest rate. 0 Workers who remain inthe current system face the type o f adjustments described inthe case of ReformType I. 0 In parallel, necessary incentives and an appropriate regulatory framework are put in place to promote voluntary savings inthe form o f contractual savings (see Section 7.5). The administration o f this scheme will necessitate improvements in management and information systems. Indeed, as in the case o f a DB-PAYG, adequate management and information systems are required to track contributions, register life events, and compute and pay benefits. To fully realize the benefits o f the new schemes, reporting mechanisms need also to be updated so that plan members are periodically informed on the value o f the funds "accumulated" intheir accounts. The downside of this strategy. This type o f reform has been introduced successfully in Sweden, Poland, and Latvia. The success o f the NDCs system depends on the credibility o f the rules determining the valorization o f accounts and the rules determining the calculation o f the G factor. If these rules allow for discretionary decisions, then the financial sustainability o f the system i s compromised. Even if there are no discretionary decisions, definingrules that allow for an accurate estimation o f the valorization factor and the G factor can be problematic. The other limitation of this strategy is that the NDCs remains an unfunded system. This being the case, it fails to exploit the positive effects that a mandatory-funded system may have on the economy by contributing to developing financial markets and fostering economic growth. Potential financial and fiscal impacts of the DroDosed reforms For illustrative purposes, this sub-section summarizes the results o f simulations for the period 2002-2070 o f the financial and fiscal impacts o f Reforms Type I(parametric reform o f the current systems), Type I1(downsizing o f the DB-PAYG), and Type I11(downsizing the PAYG scheme and introducing NDCs). The various reforms are summarized inTable 18. 56 See Technical Appendix for a summary o f methodsto computepensions andreplacementsratesinthe NDC system. 62 Table 18: ProposedReformsfor the SSO andthe CSRO Parameter(,,) Reform TypeI Reform Type11 Reform Type111 Simple parametric New workers enter a New workers enter an reform. new, downsized, DB- N D C Scheme. Current PAYG. Current systems systems are closed and are closed and subject to subject to Reform Type ReformType I. I. Contribution Remains unchanged at Reduced to 14%b Reduced to 14% Rate 18% for SSO (only for long-termbenefits) and 22% for the CSRO Minimum Increases to 65 years. Set at 63 for new Set at 63 for new Retirement Age For male the change workers (male and workers (male and takes 10 years; for female). female) women 20 years Accrual Rate Targets 80% Targets a replacement N.A. replacement for FC rate o f 60% for new worker. workers. Number of Years Increases by one year Usedto Compute each year untilfull- indexed with average BaseWage career. Wages indexed earnings with average earnings Penaltiesfor Actuarially fair Actuarially fair N.A. Early Retirement reductions to reductions to replacement rates replacement rates adjusted on a yearly adjusted on a yearly basis basis Rateof Return N.A. N.A. Growth rate o f average on Contributions earnings' Source: Mission design. Note: (a) Other reforms, such as changes in minimumpensions andceilings on coveredwages, have not beensimulateddue to lack of the necessarydata (Le., wage andpensiondistributions). (b)The contributionrate could be lower if accumulated funds are not used to cover the deficits of the closed systems. (c) This mechanism is used simply for illustrative purposes and computationalconvenience. Other mechanisms are available, such as using the growth rate of the wage bill that also takes into account change inthe size of the coveredpopulation. This index, however, is morevolatyle. All simulations have been applied only to the baseline demographic and economic scenarios. Inthe case of ReformType Isimulations are conducted by introducing one policy at a time, in order to demonstrate their marginal impact. For Reforms I1and I11only the aggregate financial impacts o f the policies are presented. Illustration of the financial impacts of individual parametric reforms in the SSO. The simulation considers the following policies: (i) increasing the retirement age; (ii) allowing for early retirement with appropriate penalties,57 (iii)gradually introducing all wages in the calculation o f the pension; and (iv) reducing the accrual rate. In Figure 23 it i s observed that each policy on its own has non-negligible impacts on the finances o f the system. The increase in the retirement age alone can reduce the present value of the accumulated deficits for the period 2002-2070 from 140 percent o f current GDP in the status quo to 120 percent. 57 While in principle early retirement i s not available, under special conditions individuals can retire before ages 55 (female) and 60 (males). Since retirement conditions for all workers are being normalized, early retirement with actuarially fair penaltiesi s allowed. It is assumed that given the penalties, current probabilities of early retirement are halved. 63 Introducing penalties for early retirement can gain additional 20 percentage points. Moving to the full-career average can reduce the present value of accumulated deficit from 100percent to 80 percent. Under this last policy, the effects on average replacement rates (pension dividedby the average wage) differ by type of worker. Indeed, because wages are indexed by average earnings, those workers who had their highest salary (relative to average earnings) in mid- career (usually blue-collar workers) benefit from the reform. This reform also provides incentives for workers to declare in-full their salary and to contribute continuously. Thus, in the calculations it i s assumed that the underdeclaration o f wages i s reduced by 10 percentage points. Finally, the reduction in accrual rates i s simulated, which i s the policy that has the largest impact on the finances o f the system. This policy reduces benefits while providing incentives tojoin the system early rather than late, thus increasing the average length o f service at retirement. When age-specific accrual rates are set as to generate a 2.4 percent average accrual rate, unfunded pension liabilities are reduced below 20 percent. Ifthe average accrual rate i s set to 2 percent, the unfunded pension liabilities for the period 2002-2070 are virtually eliminated. Nonetheless, the problem o f financial sustainability over the long term i s not resolved, as deficits above 2 percent o f GDP persist. Figure 23: Marginal Financial Impact of Alternative Reforms in the SSO PensionFund Current Balance PensionFund Deficit: Present Value from 2001 as Percent of GDP 20% ; 160% T- to 2070 as Percent of 2001 GDP ~- -1 0.0% 140% 120% 100% 4.0% 80% 60% 40% .10.0%- - 20% ....., Years -+penalties 0% - ...+luiIcareer baseline , rel.age -+accr.rate2.4% - - - - - - - - +accr.rate 2% base line relirementage tpenalties +lullcareer +accr.rale2.4% +accr.rale2% Source: Mission calculations usingPROSTmodel. Financial impacts by type of reform. The results of the simulations show that the three proposed reforms improve thepnancial situation of thefunds signi$cantly. In all cases, the present value o f unfundedpension liabilities for the period 2002-2070 could be reduced below 15 percent. The results are sensitive to two key policy choices. The level o f the accrual rates in the SSO and the CSRO (2 percent average inthis case), and the contribution rate to the new DB or NDC systems (14 percent inthis case). A Type I reform could virtually eliminate unfunded pension liabilities in the SSOfor theperiod 2002-2070 while reducing them below 11percent of GDP in the CSR0.58In the SSO deficits would be observed starting in year 2045 that could reach 3 percent o f GDP by the end o f the simulation period. The present value o f fund balances for the period (2002-2070) i s negative but could be covered by current reserves. In the CSRO deficits would persist for most o f the periodranging between 0.1 percent and 0.3 percent o f GDP. As discussed below, these deficits will require fiscal support (see Figure24). A Type 11reform could shrink unfundedpension liabilities from I75 percent of GDP today for both the SSO and the CSRO) to less than 10percent, while reducing contribution rates and giving roomfor alternativeforms of long-term savings. Inthe calculations of unfunded pension liabilities it assumes that the surpluses accumulated in the new system are used to finance the 58This considers current reserves of close to 3.3 percent of GDP. 64 deficits o f the reformed systems that were closed. If this is not the case and the unfunded pension liabilities of the closed systems are financed through the central budget, then the new system could accumulate resources equivalent to 45 percent o f today's GDP (see Figure 25). Contribution rates could also be lower in this case. Over the long-run, as the dependency ratio increases, the new system would generate a deficit of close to 1percent of GDP. This implies that contributions would needto be adjusted upwardor accrual rates d~wnward.~' The Type111reform produces similarfinancial results as TypeII. This simulation does not take into account potential changes in behavior, such as longer contributory periods. Still, the results show that if the surpluses accumulatedby the new NDC system are not usedto finance the closed systems, the present value of balances for the period2002-2070could also attain 45 percent o f GDP (see Figure 26). In the projections for Reform Types I1and 111, it i s assumed identical trends in coverage and length of services at retirement. In the latter, however, given the change in the benefit formula, coverage rates, length of service, and therefore reserves could be higher. Over the long term, the system could initially generate a deficit that reaches 1 percent o f GDP by the end o f the simulation period. This deficit, however, would be auto- corrected as rates of return (and therefore individualreplacement rates) are reducedwhen either the growthrate of average wages or total contributors slows down. 59Deficitsare likely to be lower if demographic indexationof the retirement age, which has not being incorporated in the simulation, is considered. 65 Figure 24: Financial Impacts of Reform Type I(Parametric) PensionFundCurrentBalance PensionFund Balance: PresentValue from 2001 as % of GDP to 2070 as % of 2001 GDP 04% -20% -40% -60% -60% .loo% -120% .140% ... .. baselineSSO - t a s e k CSRO -rebrmedSSO - - - .rebrmedCSRO .160°% Source: Mission calculationsusingPROSTsimulation model Figure 25: Financial Impacts of Reform Type I1(Downsize DB-PAYG) PensionFundCurrentBalance ' PensionFund Balance:PresentValue from2001 as % of GDP to 2070 as % of 2001GDP 4 0% 100% 50% 0% 1-4 :E 0% - .50% .._.._ - l i 1. -100% ('0 0',) -- ~ (!2O">' 1 i .......SSOiCSRO ~ - (baseline)-SSO dosed CSRO closed - - ~ -newPAYGDB Source: Missioncalculationsusing PROST simulationmc Figure 26: Financial Impacts of Reform Type I11(NDCs) I PensionFundCurrentBalance I PensionFund Balance:PresentValue from 2001 as % ofGDP 1 to 2070 as % of 2001 GDP I -150% .......SSOtCSRO(baseline)-SSOClosed-CSRO Closed- - - - I W NDC Source: Missioncalculationsusing PROST simulation model Fiscal impacts. In the absence o f reforms, the fiscal balance o f the government could be seriously compromised, with transfers to cover deficits in the SSO and the CSRO starting at 1 percent o f GDP within the next couple o f years and surpassing 8 percent o f GDP by year 2050. The calculations assume that reserves are liquid and that surpluses in the SSO can be used to finance deficits in the CSRO. The reforms discussed here can reduce considerably the demand for fiscal support over the next 60 years (see Figure 27 first panel). In the case o f Reform Types I1and 111,this assumes that the surpluses o f the new systems (DB or NDC) will be used 66 to finance the deficits o f the closed systems. If this i s not the case, then the resources required to cover the aggregate deficits o f the closed the SSO and the CSRO, after reserves have been depleted, would amount to 2 percent o f GDP during the period 2020-2050, declining to zero afterward (see Figure 27 second panel). Figure 27: Fiscal Impacts of Alternative Reforms ~ 1200 t Parametnc tParemetnc ofthe closedSSOandCSRO 140% 8 00 % , I 100%t G 600 DP 4 00 2 00 0.00 .lil.-...-.....X.n..."..~.~,.-"..-,-.-.,..~~ 2001 2011 2021 2031 2041 2051 2061 - Source: Mission calculations. Nore: Assumes that current reserves are liquidand fungible across funds. Discount rate set at 5 percent (real terms). Welfare implications. Clearly, the proposed reforms improve financial sustainability by reducing the rates o f return on contributions. The targeted rates o f return, however, are affordable while still generating replacement rates above 50 percent o f the last salary. Moreover, Reforms I1and I11open the door to additional forms o f long-term savings. With the current reforms, the redistributive power of the PAYG system would concentrate on low-income individuals. To illustrate some o f the welfare implications, replacement rates and rates o f return for individuals at different points o f the income distribution are reviewed. It i s assumed that the individuals in question enter the system at age 30, retire at age 65, and live until age 73 (the current life expectancy at birth for males), and that wages grow at 3 percent per year. For illustration purposes, the minimumpension i s set at 60 percent o f the minimum wage and the maximum coveredwage at 2.5 times the average wage. Ifminimumpensions and ceilings are allowed to grow in real terms, replacement rates across reforms would range from 80 percent in the case o f low-income individuals to 26-36 percent for individuals earning 5 times the average wage. If the minimum pension and the ceilings are kept constant in real terms, then replacement rates would be lower, ranging between 50-70 percent in for low- income workers and 8-11 percent for high-income individuals (see first panels in Figures 28, 29, and 30). In terms o f rates o f return, these could range between 3-4 percent for low-income individuals and 2.5-3 percent for middle- and high-income individuals. I n the absence of the reforms, unfunded pension liabilities will need to be financed either through abrupt reductions in benefits for future generations, budgetay reallocations, or higher taxation that will be welfare decreasing. This will create adverse intergenerational transfers from the future poor to the present wellhff. It can be argued that given low wages, reductions in replacement rates will not allow some retirees to satisfi basic needs. This could be the case for a given segment o f the population, but unfortunately it i s not a problem that can be solved in a sustainable way through the contributory pension system, which in any case is not covering the long-term poor. Higher standards o f living for the whole population can only result from higher economic growth. Clearly,to minimizethe social impacts of the proposedreforms, all adjustmentsto system parametersneedto be implementedin a phasedmanner. 67 Figure 28: ReplacementRatesand Ratesof Return under ReformType I 0.90 ;; 0.80 5s.I e 4.00% - 0.70 a E 3.50% - 0.60 h:p0.50 B$ 'E 0.40 : .-Keg2 -E 2.00% - 0.30 0.20 1.50%- e 0.10 1.00% - e 0.00 0.50% r 0.50 1.50 2.50 3.50 4.50 Individual earnings (proportion average wage) Source: Missioncalculations. Note: Case of individuals entering the system at age 30, retiring at age 65, and living until age 73. All wages are assumedto grow at 3 percent inrealterms. Graphsare for SSO. Results for the CSRO are similar. Figure 29: ReplacementRates and Ratesof Return under ReformType I1 0.9 4.50% 1 I -e ;; 0.8 a 4.00% 2 2 0.7 .a e 3.50% .I 0.6 *:p 2 3.00% 1 1 - 1 CI ._.-50.5 0 2.50% 0.4 49 0.3 ;:::: 2.00% I 'I L e 0.2 P 0.1 e 0 0.50% 0.50 1.50 2.50 3.50 4.50 0.50 0.70 0.90 1.10 1.30 Individual earnings (proportion average wage) Individual earnings (proportion average wage) Source: Missioncalculations. Note: Case o f individuals entering the system at age 30, retiring at age 65, and living until age 73. All wages are assumedto grow at 3 percent inrealterms. Figure30: ReplacementRates and Rates of Return under ReformType I11 1 0.9 aaini-n-wnd . . _ " 1 " 4.00% - __ 3.50% - 3.00% - s 2.50% ~_____ 1 7 2.00% ~ ~ _ _ _ ~ .- 1.50% - E 0.2 0.1 maximum covered wage 1.00% - constant in real terms i 0 ' 0.50 1.50 2.50 3.50 4.50 0.50 0.70 0.90 1.10 1.30 1.50 Individual earnings (proportion average wage) Individual earnings (proportion average wage) Source: Missioncalculations. Note: Case of individualsentering the system at age 30, retiring at age 65, and living until age 73. All wages are assumed to grow at 3 percent in real terms. Notional interest fixed at 3 percent. Interest rate usedto compute annuity set at 5 percent. 68 7.2. Improving theManagement of Fund Reserves: TheNeedfor Better Governance During the last few years, the study of the determinants of pension fund investment performance, and in particular public pensions funds, has generated considerable interest. This i s inpart due to the practical relevance o f the question. Indeed, today, 49 percent of the world's labor force (close to 800 million people) i s covered by mandatory, publicly managed, DB- PAYG systems that have accumulatedreserves. The issue of management is relevant to some 62 public pension schemes inthe world.60 The main message from the various studies is that while the macroeconomic environment places constraints on thepotential performance of a pensionfund, the governance structure of thepension fund is the keyfactor in achieving this potential. Governance refers to the manner in which authority or power is exercised to fulfill duties and obligations to a constituency of stakeholders. A review of international experiences shows that, in general, public pension funds havefailed to generate appropriate rates of return on investment for theplan members, in part given weak governance structures. A recent statistical analysis o f cross-country data suggests that there i s a strong correlation between the average real rate of return on pension assets and a governance index6' Furthermore, the positive impact that improved governance appears to have had on rates o f return i s higher when the initial level o f governance i s low. A corollary is that changes in governance structures can have particularly high payoffs when initial structures are very weak. The public sector i s not intrinsically a bad manager o f pension funds. Ifthe majority o f public pension funds have displayed weak performances, it i s because they have all followed similar badgovernance practices. This section starts by defining measures o f governance and reviewing best international practices. It then presents policy recommendations specific to Iran that try to take into account local institutional constraints. Basic principles of good governance Governance i s a multidimensional concept that is difficult to define and measure. Here the focus is on four dimensions that appear to be critical for the performance o f a pension fund: i) how duties and obligations are specijied; ii) how the governing body is structured and selected; iii) how the management of the pension fund is structured and selected; and iv) how accountability is enforced. Duties and obligations. Duties and obligations should be specijied by law. The objective or mission o f the pension fund should be clearly stated in order to ensure the adoption o f measurable goals against which the performance o f the plan, its governors, and administrators can be evaluated. The objective of the governing body of thepension fund is to ensure that resources are used to serve the interest of pension plan members. The governing body should not target social and development goals. Basic activities should be limited to collecting contributions, investing reserves, and paying benefits. Unfortunately, this last principle i s too often not observed, particularly in developing countries. For instance, the fiduciary responsibility o f SSO Governors toward beneficiaries i s compromised, because the SSO has mandates to support the development o f national projects such as housing and to intervene in areas such as corporate lending. In its 2000 Annual Report, the SSO even mentions the promotion and diversification o f exports as one o f its goals. 6o This section relies heavily on Impavido (2002) and Palacios (2002). 61 See Iglesias and Palacios (2000). 69 Pension funds should not be considered as lenders of last resort for the government. In this regard, an example o f best practice i s the Canadian Pension Plan Investment Board (CPPIB). Unlikefunds inJapan, Korea, and the United states, the CPPIB is only required to make funds available to the government ifit is consistent with investment targets. The governing body. The governing body i s the entity with the highest level o f governance authority. InWestern common law, the governing body usually i s called the Boardo f Directors or Board o f Trustees. The SSO equivalent i s probably the High Council for Social Security, although it also has a Board o f Directors. In the CSRO, the highest governing body i s the Supreme Council. International experiences suggest that in general tripartite governance bodies do not prudently and efficiently manage reserves. Inthe majority o f developing countries, the governing body i s composed o f representatives o f interested parties. These usually include the government (as plan sponsor), employers, and plan members. This i s the case o f the SSO HighCouncil: it has seven government representatives, five employer representatives, and three worker representatives. In general, tripartite Boards end up composed of fiduciaries who lack the necessary technical expertise implicit to the job. Because in most cases the appointees are selected by high-ranking government officials, the resulting governing body lacks the necessary independence from the government. The lack o f independence creates conflicts of interest between government objectives (for instance reducing the government debt with the pension funds) and plan members (receiving the best rate o f return on this debt). This problem i s pervasive even inmore developed countries. A recent study o f the United States shows that the weak performance o f public funds relative to private funds i s explained, in part, by the composition o f the governing bodies. Funds in the private sector are governed by qualified professionals who have a clear economic mandate, while public sector governors respond to economic as well as political pressures. Two questions that need to be addressed in order to improve the composition o f the governing body are: i) what are the appropriate qualijkations; and ii) how governors should be selected to ensure independence. I n terms of qualifications, governors should understandfinancial markets, risk management, and actuarial principles. They should be prepared to study and understand the promises and policies o f the pension fund. Governors should understand the conflicts o f interest and commit to resolve them in favor o f the plan's beneficiaries. Canada i s again an example o f best practices. The twelve members o f the Board o f the CPPIB are selected for their investment and business expertise in areas such as economics, accounting, actuarial science, finance, investment, banking, and business in general. The requirement for relevant expertise and experience is set out in the legislation. I n terms of the selection urocess, the goal is to ensure independencefrom the government; this requires an open and transparent process. One option is to have the selection made by an independent Selection Committee. Members o f this committee can be nominated by representatives of the government in different provinces as well as employers and workers. The role o f the Selection Committee i s to identify prospective candidates for the Board o f Directors or High Council from across the country to produce a shortlist. The shortlist i s then presented to the relevant minister (usually the Minister o f Finance or Planning), who in consultation with other high-level government officials, and with directors already appointed, selects the members o f the Board from the shortlist. Selected members o f the Board can serve for periods o f three to four years, renewable for a maximum o f three to four times. During a term, members cannot be removed except for illegal or immoral misconduct. 70 I n terms of the size of the governing bodv it is important to consider the trade-offs between independence and effectiveness. As the number o f Board members increases, independence increases; and the probability o f collusion and moral hazard i s reduced. On the other hand, effectiveness i s reduced, first due to coordination problems, but also because incentives are given for free riding on the work o f others. The size o f the Board should be selected to maximize its overall performance. The resuonsibilities of governors should be defined by the law of fiduciary duty; governors should act as the owners of the assets (to ensure accountability) and on behalf of the beneficiaries. Fiduciary obligations include: i)complying with legislative requirements; ii) communicating to members their rights and entitlements, iii)ensuring that actuarial valuations are performed routinely; iv) ensuringthe required contributions are submittedto the plan on a timely basis; v) ensuring that funds are prudently invested; vi) ensuring that payments o f benefits are accurate and timely; and vii) ensuring that the level o f funding i s appropriate. The obiectives of the povernors should be stated in the relevant sets of regulations and should include: i) managing funds in the best interest of contributors and beneficiaries; and ii) maximizing investment returns without incurring undue risk of loss. Governors should not conduct any business that i s inconsistent with these two objectives. To meet their fiduciary obligations, governors need to define the main uolicies for the fund. These include: i)determine acceptable levels o f balance sheet risk and targeted returns for a given level o f risk; ii)approve the businessplan; iii)monitor outcomes versus expected results and develop a system o f compensation linking economic performance to the remuneration of management; and iv) hire and fire the Chief Executive OfJicer of the organization (see next section on Management). The governing body can be divided into committees with separate responsibilities. A process should be in place to evaluate the performance o f the committees and to report the outcome to the stakeholders. One o f the critical committees i s the Investment Committee. This Committee should establish both an investment policy and an implementation policy covering issues such as asset allocation and active versus passive management. Management. Governing functions and responsibilities should be clearly separated from managingfunctions and responsibilities. Different individuals should belong to each o f these groups. Management fiduciaries should be responsible for day-to-day operations and the execution o f the policies established by the governing body. Members o f the management team should be constituted as independent fiom the Board but report to the Board through the Chief Executive Officer. I n meeting its fiduciary obligations, operational management should assemble the necessary human and operational resources. I t should define a strategy to meet the investment targets and to identify tactics for implementing the strategy. It should also develop a system to measure the performance o f the strategy. I t should report outcomes to the governing body. Accountability. Accountability is fundamental to good governance structure. Accountability depends on the process by which management reports to and i s evaluated by the Governing Board, and the process by which the Governing Board reports to and i s evaluated by stakeholders. A strict system o f internal controls is required to regulate the activities of fiduciaries and ensure the commitment to public transparency and reporting. I n private pension funds the legal basis for accountability is personal liability. Insurance for personal liability can hrther ensure the ability o f the pension fund to recover losses in case o f mismanagement. For public systems, implementingpersonal liability can be dijficult, because the government assumes contingent liabilities and, therefore, becomes an important 71 stakeholder. Hence, the performance of the pension fund, which i s the ultimate criteria to evaluate the Governors, often depends on policies that are beyond the Governor's control. For instance, investment regulation and contribution rates are often established by law and may be inconsistent with the fiduciary responsibilities of the Governors. Ifpersonal liability cannot be established, the members of the governing body should be made accountable to an independent body representing all stakeholders. A natural choice is the Parliament. Since legal accountability is limited in public funds, transparency and information disclosure become the main mechanisms to ensureprudent management. All members o f the pension plan should receive information on the objectives o f the fund, its agenda to achieve objectives, and plan member rights. This information should be stated in a simple and easy-to-understand format. The governing body should report on investment strategies, implementation mechanisms, and the results of the operations- including rates of return on the different investments. Summary balance sheets should be regularly published. Plan members should receive periodic statements reporting accrued benefits and the overall performance o f the pension plan. Two additional mechanisms to improve accountability include external audits and independent custodians. The financial activities o f the fund should be audited on a regular basis by an external auditor. The auditor should be required to verify that fund activities comply with all relevant regulations. The auditor should also conduct a periodic actuarial evaluation of liabilities and provide an analysis o f hnding levels. Custody should be provided externally only by an independent financial institution. The assignment o f custody responsibility to an entity other than the asset manager i s an efficient way to ensure the physical and legal integrity of the assets and to oversee the transactions o f the assets manager. For apublic pension fund, an ideal custodian would be a truly independent central bank. Other internal controls that enforce accountability include requiring directors and employees to clear personal trades before execution and to report on their personal investment activities on a regular basis. The CPPIB has implemented these controls. In-line with the previous recommendations other measures adopted include: i)the appointment o f an independent, external accounting firm to review the operations o f CPPIB and to record and report directly to the Audit Committee o f the Board of Directors; ii)a procurement policy covering the selection o f outside organizations and suppliers; iii)an external custodian selected through a rigorous process o f due diligence; and iv) the determination o f signing authorities and limits to protect cash and portfolio assets. These are policies that have allowed CPPIB to earn public trust. Recommendationsfor the CSRO and the SSO The general principles discussed in the previous section are highly demanding and cannot be achieved overnight. However, policymakers and plan members should be aware that a large number of case studies at the international level indicate that unless these principles are adopted, the management of fund reserves is unlikely to improve in a sustainable way. Recommendations that could be gradually adopted by the SSO and the CSRO to improve investment policies are presentednext. Strengthen the role of current governing bodies by clarifying responsibilities and objectives and by modifying their composition and selection process. Today, the High Councils o f the SSO and the CSRO, the Boards o f Directors, and the Advisory Committees have overlapping functions that dilute accountability. Moreover, the technical competence and 72 independence o f the High Councils can be improved. To this end are the following recommendations:62 0 Create a new governing body or HighCouncilthat unifies the current HighCouncil and Board o f Directors while modifying its size and composition. Within the new High Council, different subcommittees could be created with specific functional responsibilities (e.g., administrative issues, investment policies). 0 Change the selection process for members o f the HighCouncil. The Head o f MPO, the Ministry of Finance, the Ministry o f Health (inthe case o f the SSO), and the Central Bank could continue to have representatives on the High Council, but the other members would be nominated by a Selection Committee appointed by the Head o f MPO on the basis o f suggestions from other members o f the cabinet. The Selection Committee's mandate is to identify individuals to serve on the HighCouncils who have technical expertise in the areas o f finance, economics, and actuarial analysis, and who have demonstrated through their professional achievements that they can contribute to prudent and efficient management o f the pension funds. 0 Focus the mandate o f the HighCouncils on the management o f the pension funds inthe interest o f its plan members. Eliminate other objectives such as pursuing social or economic development policies. 0 Give to the High Councils the freedom to select and remove the Managing Director o f the pension fund(s) and decide on compensation modalities. 0 Make the HighCouncils accountable to Parliament. Modify the current strategy used to refinance the public debt with the pension funds. The current strategy i s risky and it i s creating losses for plan members. The following recommendations are made: 0 Stop the practice o f accepting public companies as payments on the government debt. 0 Conduct an external audit o f current government arrears and agree on a total amount. 0 Adopt a plan to collect arrears at fair value. The plan would involve issuing tradable bonds (both coupon and zero coupon bonds). The use of government bonds improves transparency and assures that even if the debt is not paid in cash to thepension funds, it continues to earn appropriate real rates of return. Clear clauses need to be put in place to protect the pension funds in the case o f default. This policy i s less risky than accepting public companies, often insolvent, and then investing resources in their restructuring. Introduce changes in current investment policies to reduce risk, promote capital-market development, improve monitoring, and ultimately increase rates of return. Discard the current policy o f restructuring the enterprises received in payment for cancellation o f arrears. Instead find strategic investors with controlling ownership. Adopt a program to reduce the controlling stakes in corporations by holding minority participation. It is recommended that the SSO and the CSRO get out o f the construction business. 62 If pension funds are merged as discussed in Section 7.3, these recommendations would still apply to the new integratedpensionfund. 73 Encourage the participation o f other fund managers in addition to the fund's own investment companies (Shasta- SSO and the Investment Company for CSRO) and have the different managers compete. The process for outsourcing i s delicate; therefore, appropriate competition rules among prospective managers need to be put in place. International experiences suggest, however, that this strategy i s likely to produce higher rates o f return. Moreover, it reduces the problems associated with managing a large public fund in a small financial market. In particular, market performance becomes less sensitive to changes in the plan's strategy, as other managers are allowed to compete andthe market power o f the public pension fund i s reduced. Forbid the extension o f credits to firms owned by the pension fund. Adopt a program to gradually reduce member loans and charge market interest rates. Adopt exposure limits to ensure maximum diversification on investments in shares: maximum 5 percent o f the capital o f any company should be owned by the fund; and a maximum of 5 percent of the fund assets should be invested in any company. Allow for investments outside the country. Eliminate taxes on profits from the investment income o f fundreserves.63 Prohibit the SSO and the CSRO from having representation on corporate Boards and from exercising shareholder rights. Instead, authorize the pension funds to delegate the representationo f shareholder rightsto outside fundmanagers. Improve accountability Conduct a mark-to-market valuation o f current assets by an independent auditing company. 0 Improve current reporting mechanisms. In particular, review the structure, scope, and outreach o f current Annual Reports. These reports should be freely available to stakeholders and provide information, organized in a concise and clear way, about the finances and performance o f the fund. Beyond standard accounting reports and the notes from the auditors, the Annual Report needs to provide information about the allocation o f the portfolio of assets and the rates o f return on different investments. Publish the balance sheets o f the companies owned by the pension funds. Introduce periodic external audits, ifpossible by international consulting firms. Introduce external custodians. 7.3. Institutional Issues to Consider over the Short and Medium Terms: Strengthening Institutional Capacity and Merging Pension Funds Information systems and human resources The operation o f a pension fund involves several administrative processes-including registration o f new members and their employers, follow-up o f life events, tracking and collection o f contributions, processing o f claims, payment o f benefits, monitoring o f investments, book keeping, and reporting. These processes needto be supportedby appropriate management and information systems and a human resource base that i s coherent in size and 63Notice that this recommendationdoesnot imply that the profits of companies in which SSO has investedshould not be taxed(see Section5). 74 skills mix. This section presents recommendations for the SSO and the CSRO to improve institutional capacity. The CSRO has developed a state of the art information system, based on a detailed analysis of the different administrative process: the software problem is resolved. The challenge now is to define and implement procedures so that information can be collectedhpdated into the system on a routine basis, particularly in the case of contributors. The following activities need to be completed over the short run: 0 Update databases o f current contributors-including the creation o f a unique identifier number and when possible wage histories. 0 Develop a mechanism for routinely collecting and updating information about new contributors and life-events. This could be done at the central level on the basis o f manually filled forms, printouts from special software, or electronic support sent by the local branches. The best alternative, however, i s to provide the regional offices access to a centralized database. 0 Develop a mechanism to routinely collect information on contributions paid by current and new beneficiaries. Today, the collection o f contributions takes place at the local offices o f the CSRO. Then contributions are deposited in the local branches o f the banks where the CSRO has accounts. A more efficient mechanism is to have contributions directly deposited in the bank's local branches, these can send electronic records to the central office on the basis o f the unique identifier. The information can then be easily updated at the central database. 0 Optimize the processing o f claims and the transfer o f payments. For new beneficiaries this process should be automatic. Claims could be submitted electronically or in manually filled forms to the central office where the information systems will perform the calculations usingthe information already collected inthe databases. Payments can then be transferred to the bank's local branches along with a list o f identifiers for individuals who should receive the payment. For current contributors, especially those close to retirement, processing the claims will first require manually updating informationabout wage and work history and life-events. 0 Conduct an analysis o f the needs interms o f information technologies (computers, fax, e-mailhnternet) at the central and local levels and estimate implementation costs. The SSO lags behind in the development of its management and information system and the definition of a coherent human resourcepolicy. The design and implementation o f the M I S i s being conducted by a software company owned by SSC. To date, several technical problems have been identified in the design (see Section 2.1). Furthermore, since the SSO i s involved in a large number o f activities that go beyond those proper to a pension hnd, it has been difficult to define a coherent strategy in terms o f information needs, human resources needs, and the best organizational structure for the institution. Looking forward the following interventions are considered priorities: 0 Explore the possibility o f transferring the current MIS o f the CSRO to the SSO. 0 Develop administrative guidelines for routinely collecting information about new contributors and their employers in the local branches and assign a unique identifier number. Today, data on contributors are aggregated by employer. 0 Review the current system used to record the information about new beneficiaries to include the work histories used by the local branches to compute benefits. 75 0 Conduct an audit o f the current databases o f beneficiaries. 0 Develop a plan to update the databases o f current contributors-including a unique identifier number. 0 Develop a plan to update information technologies at the central and local levels. Merging the pension funds Over the medium term it is desirable to move to an integrated National Pension Fund. This fund would merge the CSRO and the SSO, while also extending to individuals currently covered by occupational plans. Occupational plans would then become complements to and not substitutes for the National Pension Fund. The integration o f funds would generate economies o f scale while facilitating labor mobility and improving financial sustainability. Indeed, the financial sustainability o f a PAYG system needs to be evaluated at the aggregate level. Pension funds need not be sustainable by sector or subsector. Beyond improving aggregate financial sustainability, having a single system allows for reductions in management costs (personnel, infrastructure). Indeed, in a pension fund a large part o f operation costs are fixed. Inthe case ofIranthe integrationcouldproceedalong the following lines: 0 Eitherthe pensionbranchofthe SSO or the CSRO would become the basis for the new system; in this discussion it i s called the National Pension Fund (NPF). This implies that all new contributors (from the private sector and public sector) will join the NPF. The fund will have two types o f contributors: i)current contributors, who are subject to a parametric reform, Type I;and ii)new contributors, who mightjoin a Type I1or Type I11system (see previous section). 0 Ifreforms Type I1or Type I11are implemented, current contributors from the CSRO and the SSO could be given the choice to switch to the "new" system within the NPF. Workers who transfer are subject to the same rules as the new plan members. This often involves lower benefits. If the government finds it appropriate to honor worker- accrued rights, the necessary resources to finance the additional benefits need to be transferred to the new pension fund or guaranteed by the central budget through recognition bonds that mature when workers retire. Ifthe "new" system in the NPF i s an N D C (Reform Type 111), capital equal to past contributions plus appropriate interest earnings is transferredto worker notional accounts. 0 The accounts o f the plan members who remain outside the NPF could also be managed by the new NPF. Another alternative would be to leave the accounts underthe pension fund that is beingphased out. The latter is an easier option to implement, but it means that reforms to improve management and the information system need to be coordinated separately with two funds. Jordan, for instance, has recently adopted this strategy. Indeed, the Civil Servants and Military pension funds are being phased out. New civil servants and military recruits now enroll inthe Social Security Corporation, which previously was open only to workers from the private sector. The accounts o f current civil servants, military personnel, and pensioners continue to be managedby the Civil Servants and the Military pension funds. 76 There are three necessary conditions for the merger: i)the newly integrated pension fund needs to be financially sustainable (the reforms discussed in the previous sections would need to have been adopted); ii)it has to have the capacity in terms o f information systems and human resources to manage the merger; and iii)transfers o f the current plan members to the new system should not involvetransfers o fpensionliabilities. The merger in Iran ought to be accompanied by reforms mandating that new employees in Jirms that currently have occupational plans join the new system. Appropriate financing mechanisms could also to be put inplace to allow current plan members to join the new system. As previously observed, occupational plans would be maintained, but as complementary regimes not substitutes. A detailed analysis o f the financial sustainability o f these plans i s necessary as well as a review o f the current regulatory and supervisory framework. A merger is a complex process that needs to be carefully planned and orchestrated, as it involves changes to current administrative structures, management and information systems, and accounting systems. To implement the merger inIran, a multi-year programwould be need to developed. 7.4. Expanding Coverage The Iranian contributory systems covers approximately 50 percent o f the labor force, which i s a relatively highpercentage given the country's level o f income. As in other countries, however, workers in the informal sector and some populations living in rural areas remain uncovered. Ideally, Iran would like to achieve universal coverage through the contributory pension system. As discussed below, pension reform can contribute to expand coverage. Structural reforms leading to faster economic growth and better opportunities o f employment in the formal private sector can contribute as well. Nonetheless, neither a reformed pension system nor a more dynamic economy are likely to suffice, at least over the medium term. Policymakers should keep in mindthat there i s no automatic increase in coverage with economic development. This implies that social assistance programs targeted to the elderly will remain an important component of the government strategy to ensure an adequate level of income during old age. Expanding; the coverage of the contributorv systems Recent research has shown that two aspects of the payroll contribution affect coverage: its level and its allocation between DB and DC schemes.64 Highcontribution rates, in any type o f pension system, are detrimental to expand coverage. This occurs not only because expected rates o f return decrease, but also because when the mandate to save is too high relative to individual preferences, these individuals are worse off if joining the pension fund. Finally, some individuals may face liquidity constraints and therefore simply cannot afford the payroll contribution. There i s also evidence that for a given contribution rate and a given rate o f return, moving from DB to D C schemes increases the probability o fjoining the system. This suggests that individual expectations about rates o f return in a D C system are higher, maybe as a result o f a more transparent linkbetween contributions and benefits and less political uncertainty. I n the case of Iran, reducing the contribution rate and implementing a credible reform program that ensures a better balance between mandatory and voluntary schemes could contribute to expand coverage. If a DB system i s preserved, computing pensions on the basis o f full-career wages will also reduce incentives to evade. A key challenge will be to reduce unemployment rates and the size of the informal sector. Labor force growth will soon accelerate as a result o f the "baby-boom'' in the mid-80s and 64Packard(1999 and 2001). 77 increasing female participation rates. Even under the hypothesis o f a 6 percent GDP growth rate, an insufficient number o fjobs are likely to be created and unemployment rates could be driven to 20 percent within 10 years. Better employment opportunities will require focusing efforts onjob creation (through private sector participation) rather than onj o b protection. As part o f the review o f the current social protection systems, Iran should also carefully scrutinize current passive and active labor market policies. Policymakers should avoid developing sector-specificpublic pension programs, such as those covering housewives or agricultural workers, with the objective of expanding coverage. In many countries, different professions are covered by their own public pension system (e.g., teachers, miners). Iran has also allowed the emergence o f multiple occupational funds that have implicit government guarantees. The proliferation o f "personalized" pension systems reduces transparency, can bring equity problems, and complicates management and financing. Ideally, over the medium term, Iran will have a single pension system applying to all individuals the same rules. It i s often argued that housewives who are dependent on the income o f their spouse would be ina vulnerable situation in the case o f a divorce and that therefore the government ought to put in place special pension programs to protect them. This i s a serious problem that requires government consideration. One alternative would be to separate the rights of both spouses. Thus, in the case o f a divorce, part o f the retirement pension can be vested to the wife inthe same way that a share o f current earnings i s vested. Nonetheless, the government should encourage the emergence of complementaryholuntary programs that are designed,financed, and managed by private associations for-profit or non- profit). One mechanismto do this i s to provide preferential tax treatment for long-term savings (other options are discussed in Section 7.5). Another mechanism to support the emergence o f these programs i s to provide information and training. Hence, the government could help finance seminars and workshops for interested organizations and disseminate experiences from other countries. The government, however, should not be directly involved in the design and financing o f these complementary programs. Instead, it should focus on developing adequate regulation and supervision. Expanding coverage throuph social assistance programs and demogrants It is unlikely that contributory schemes will reach vulnerable groups such as the lifetime poor. They usually have to focus on addressing short-term risks rather than planning for the future. For them, joining a contributory system would be welfare-decreasing. The implication i s that in-line with the reform of the pension system, social assistance programs that target the elderly- poor need to be carefully assessed, both from the demand and supply side. In terms o f the demand, policymakers need to have a clear understanding to the size and socioeconomic characteristics and geographic distribution o f population groups not covered by contributory regimes. From the supply side, it i s necessary to assess whether current programs are meeting demands. Of particular import is to benchmark current targeting mechanisms and monitoring systems with best international practice^.^^ By now there i s some evidence regarding the costs o f the various systems, their strengths and weaknesses. Self-targeting mechanisms (provision o f goods/services that are mostly demanded by the poor) are less costly to administer but unlikely to be effective for large transfers. Moreover, this system imposes costs on the beneficiaries, which reduces the net benefit o f the transfer. Categorical transfers are also easier to administer, yet these can be subject to substantial leakages when the goal i s to reachthe poor. For instance, one could think about a system that provides a transfer to all the elderly not covered by the 65See Subbarao et al. (1997). 78 contributory system. Some o f the elderly, however, are nonpoor. Geographic targeting i s demanding in terms o f the necessary baseline data. For large geographic units considerable leakages can be observed. Effectiveness depends on the institutional capacity o f local governments. Proxy means tests use a synthetic score calculated on easily observed characteristics-household structure, location and quality o f housing, ownership of durable goods, etc. They are increasingly popular although implementation and administrative costs are higher. Finally, some countries have considered community-based target systems, which use existing local actors (teachers, nurses, clergyman) or a new civic committee to decide who receives the transfers. Local actors may have better information. However, local institutions (formal and informal) may affect negatively the actors' performance. There i s the risk that elites capture most o f the benefits. To date, there i s still little empirical evidence o f the effectiveness o f these programs. Another option that the Government of Iran could consider to protect the elderly poor is to provide a universal grant or demogrant.66The demogrant i s a special form o f cash transfer that i s restricted to the elderly, not solely the elderly poor. While inthe case o f the nonpoor elderly the transfer brings negligible benefits, for the elderly poor it can represent a sizable share of their total income. While the grant i s universal and therefore there i s no need to establish targeting mechanisms, administration issues should not be underestimated, particularly in the case o f a large country like Iran. On the other hand, there are no funds to invest and the estimation o f the present and future cost o f the grant i s a relatively straightforward task. Today, providing USD 200 per year (roughly 20 percent o f the minimumwage) to all o f the population older than 65 (2.9 million individuals in year 2001) would cost between 0.4 percent and 0.7 percent o f GDP (see Figure 31). Figure31: Estimated Cost of a USD 200 Demogrunt in Iran 0.90% 0.80% - 0.10% +- -4I 0.00% ~ 2001 2011 2021 2031 2041 2051 2061 Source. Missioncalculations. Note: Calculations basedon base-line demographic and macroeconomic scenario When designing any type of non-contributory system to protect the elderly poor, basic to the concept is evaluating the incentive effects on the contributory regime. If benefits under the non-contributory system are too high, they will induce some individuals to drop out o f the contributory system. Hence, benefits have to be designed in a way that only the core poor are benefited. In Chile, for instance, the minimumpension o f the contributory regime i s set at 75 percent o f the minimumwage, while in the non-contributory regime it only reaches 25 percent o f the minimumwage. 66 See Dulitzsky et al. (2000). 79 Summarv of recommendationsto expand coverage 0 Adopt a pension reform program that allows for a lower contribution rate and a better balance between DB and DC schemes. 0 Conduct appropriate surveys to estimate the coverage gap and its causes. The objective o f this activity i s to identify population groups that are not covered by the system, as well as their geographic and socioeconomic characteristics. Only on the basis o f this information it i s possible to define specific activities to expand contributory and non- contributory schemes. Conduct a review o f current social assistance programs for the elderly-including estimates o f costs, benefits provided, number o f beneficiaries, and their socioeconomic characteristics. The review will assess management, targeting, and monitoring mechanisms and will present recommendations in terms o f the need to expand, eliminate, andor design additional programs. The review should include detailed estimates o f financing needs. 0 Conduct a viability study for the implementation o f a demogrant inIran. 7.5. Promoting VoluntarySavings and Developing CapitalMarkets As discussed in Section 7 o f this report, one o f the objectives o f the public pension fund is to encourage individuals to save for their retirement (given "myopia"). A mandate that is too high, however, can discourage enrollment or reduce individuals' welfare. An appropriate balance between the mandatory and the voluntary components o f a pension system i s therefore an important feature o f its design. Among voluntary schemes, contractual savings (CS) appear as a promising mechanism for developing countries. Contractual savings are savings accounts created to promote long-term saving and to manage social risks. These savings can be used to finance funded pensions plans (accumulation period), annuities (pension pay-out periods), life insurance, funded unemployment benefits, end o f service indemnity, and other contingencies such as the down payment for a house, education, weddings, or funerals. Thus, they constitute a mechanism to improve the management o f social risks. m i l e the impact that CSs have on the savings rate of the economy remains controversial, particularly if voluntay, there is evidence that they can contribute to the development of capital markets and, through this channel, economic eflciency and Indeed, CSs can have the following effects on financial markets: 1. Increase depth and liquidity by increasing the demand for shares and bonds, market capitalization, and volume traded 2. Increase the demand for long-term bonds and the supply o f long-term loans 3. Create incentives to improve regulations and transparency 4. Foster financial innovation, competition, and efficiency 5. Improve corporate governance. CSs also contribute to reduce financial risks and therefore output volatility. This occurs through three channels. First CSs reduce the debtor-including the government-refinancing risks by lengtheningthe maturity o f debts. Second, CSs reduce the pressure on banks to engage in excessive term transformation risks. That is borrowing short-term (from savings deposits) 67See Impavido et al. (2001, 2002a, 2002b). 80 while lending long-term. Finally, CSs increase the demand for the equity o f firms and induce an increase in the equity/debt ratio. A higher equity/debt ratio implies lower risks to fluctuations in interest rates and demand shocks. Indeed, while the value o f a company's debt remains constant during a contraction, the value o f its equity (the debt with stakeholders) decreases. Over the mediumterm, the government could consider putting inplace the necessary regulatory infrastructure to stimulate the development o f contractual savings in Iran. A variety of international experiences can guide this strategy. The development o f CSs could become part o f the agenda that the government i s currently putting in place in the area of private sector development and financial sector reform. It i s particularly timely today when the government i s in the process o f privatizing insurance companies. Indeed, life insurance companies are natural providers o f private pension products. An important element that will need to be evaluated i s the tax treatment o f different types o f savings. There i s no robust evidence supporting the view that tax exemptions can increase the aggregate level o f savings. At worse, the preferential tax treatment can create income effects that reduce savings. Nonetheless, ifthe preferential treatment focuses on long-term savings, it i s likely that it will change the composition o f financial intermediation favoring long-term funds. Insummary, to develop voluntary savings and contractual savings inparticular the government should consider the following interventions: 0 Initiate studies to develop an appropriate regulatory and supervisory framework for the providers and managers o f contractual savings. The lessons from several international experiences are available to guide this task. 0 Review the regulations in the Tax Law related to income taxation and evaluate the fiscal viability o f alternative forms o f tax exemptions for long-term savings. 81 8. DESIGNING AND IMPLEMENTINGA REFORM PROGRAM This section addresses questions related to the political economy of pension reform. In particular, what are the triggers o f a reform program? What are the necessary conditions for a successful reform? It then outlines a series o f short-term actions to initiate the process. 8.1. Triggering a Successful Reform Program Several of the more substantial reforms in the world, unfortunately, have been initiated as a result of a severe crisis. Reforms in East European countries are the best example. While the financial problems o f a pension fund system can be detected long before the most undesirable symptoms appear, governments have shown a tendency to delay the cure until the patient i s seriously ill. This tendency can be explained, in part, by a disconnect between the planning horizon of policymakers (usually short-term) and the relevant planning horizon for analyzing the finances o f a pension fund (necessarily long-term). The economic and political costs o f a reform program manifest in the short term and therefore affect the government in power, while the bulko fbenefits materialize over the longterm. Reforms can also take place as a result of demonstration effects; hence, if a given country undertakes a successful reform, other neighboring countries are likely tofollow. This i s more or less the experience o f Latin America during the 90s. The reform o f the Chilean system was crucial for inspiringand guiding subsequent reforms. Inthe MENA region, however, structural reform o f the pension system has yet to occur inany country. The region is still waiting for the leader to appear. d In Iran the conditions to initiate a comprehensive reform program are in place; the risk is that the relatively healthy financial position of the SSO gives room for complacency. Among policymakers and some groups within civil society awareness about the problems o f the pension system i s rising. The government has taken initiatives to increase institutional capacity within the pension funds, and technical groups are now evaluating options for reform. Moreover, the actions that the government i s undertaking to improve economic management, rationalize public expenditures, and strengthen the regulatory framework for the banking and the financial sector, in general, create an environment that increases the likelihood o f success o f a structural reformprogram. Ifthe momentum i s preserved, Iran could become the first country inthe region to solve the problems of the pension system ina sustainable way. The challenge for policymakers and civil society i s not to take the operational surpluses that the SSO i s still generating as a sign that only minor adjustments are necessary. There are two necessary conditions for the initiation and successful completion of a pension reform program: adopt a comprehensive approach and have a long-term commitment. First, pension reform should be part o f a comprehensive reform o f the public and private sectors. Indeed, there are several synergies that needto be taken into consideration. As an illustration, the reform o f the civil service will affect (and has already affected) CSRO finances. Similarly, realistic reforms in the area o f pensions depend on the types o f reforms that are introduced to develop capital markets and implement appropriate regulatory institutions (e.g., Stock Exchange Commission, Insurance and Pension Commission). More importantly, financial reforms inthe pension system are closely linked to expectations about the non-pension balance o f the government. The second condition i s to a have a long-term commitment for pension reform. Reforms do not occur overnight and often extend over a decade. To create continuity, it is important to establish a Reform Commission with a structure that is independent of changes in the government. Clearly, a successful reform program also requires some luck. 83 Unexpected internal or external shocks can jeopardize reform efforts by switching resources and attention to restoring the equilibrium. 8.2. Next Stepsfor Iran The Management and Planning Organization has initiated the reform process by producing an analysis o f the major challenges facing the Iranian pension system and identifying reform alternatives. Looking forward, the following steps are proposed: Creation of a Pension Reform Commission. To date, the various organizations involved in pension reform are moving in their own direction with little coordination. The role of the Pension Reform Commission i s to act as coordinator and manger o f the reform process. Hence, it i s the Pension Reform Commission that i s given the responsibility o f studying, consulting, andproposing a reformprogram to the government. Dissemination of the current report. One o f the first activities o f the Pension Reform Commission would be to disseminate the current report within the government and among representatives o f civil society. The goal i s to create awareness about the challenges facing the pension system and the options for reform. Preparation of a white Paper on pension reform. On the basis o f the various discussions and the feedback received from stakeholders, the Pension Reform Commission should prepare a short White Paper that presents the key element o f the government strategy to reform the pension system. Preparation of detailed integrated multi-year reform program. Once a final reform strategy/course o f action has been established, the various activities necessary to implement the strategy and their distribution over time can be outlined. A first set o f activities basically relates to the preparation o f additional studies to define the final structure o f the pension system, the level o f different parameters, and the most appropriate transition mechanism. A second set o f activities relates to the preparation o f the new legislation. In parallel, it i s necessary to develop activities to reinforce the institutional capacity o f the pension funds (e.g., training, review o f management and information systems, investment policies). Ifthe new legislation i s approved, then the final phase o f implementation can be carried out. 84 9. REFERENCES Actuaria. 2002. Pension ReforminTunisia. Draft Document. Paris, France. Bailliu, J. and Helmut Reisen. 2000. Do FundedPensions Contribute to Higher Aggregate Savings? A Cross-Country Analvsis. InHelmut Reisen, ed., Pensions, Savings, and Capital Flows: FromAgeing to Emerging Markets, pp. 113-31. OECD, Paris. Bonnerjee, Aniruddha. 2001. Snapshot o f Pension Systems inAfrica. Workshop on DefiningPension ReformPriorities and Approaches for Africa. World Bank, Washington DC. CSRO. 2002. The Analysis o f Financial Crisis inthe Iran Retirement System Fundo f Civil Servants and the Ways to Overcome It. Teheran, Iran. Disney, Richard. 1998. Notional Accounts as a Pension Reform Strategy: An Evaluation. World Bank, Washington DC. Dulitzky, Daniel, Kerry Flanagan, and Zafiris Tzannatos. 2000. Expandim Coverage o f Pensions in MENA: Is The Zero Pillar anOption? Unpublishedmanuscript. World Bank, Washington DC. Fetini, Habib. 2002a. Energy Subsidies: Deadweight Loss and National Savings. World Bank, Washington DC. -. 2002b. Financial Services for the Evolving Iranian Economy. World Bank, Washington DC. -. 2002c. Islamic Republic o f Iran Off-Budget Revenues and Expenditures. World Bank, Washington DC. Ghorbanali, K. June 2003. The Social Security Organization o f Iran. Paper presentedat the Workshop on Management o f Public Pension Funds. Malta. Holzmann, Robert. 1998. Financingthe Transition to Multipillar. World Bank, Washington DC. Iglesias, August0 and Robert Palacios. 2000. Managing Public Pension Reserves. Part I:Evidence from International Experience. Social ProtectionDiscussion Paper 3. World Bank, Washington, DC. Impavido, Gregorio. 2002. Governance Issues inPublic Pension FundManagement: Preliminary Considerations. Mimeo. World Bank, Washington DC. Impavido, Gregorio, Albert0 Musalem, and Thierry Tressel. 2001. Contractual Savings Institutions and Banks' Stability and Efficiency. Policy ResearchWorking Paper 2751. Financial Sector Development Department ,World Bank, Washington DC. -. 2002a. Contractual Savings and Firms' Financing Choices. InS. Devarajan and F.H.Rogers eds. World Bank Economists' Forum, Vol. 2, pp. 179-222. Washington D C -, 2002b. Contractual Savings and Impacts on Securities Markets. UnpublishedManuscript. World Bank, Washington DC. Lachance, Marie-Eve and O h i a S. Mitchell. 2002. Understanding Individual Account Guarantees. Wharton School. Philadelphia, PA, USA. Levine, Ross. 1999. Law, Finance, and Economic Growth. Journal o f Financial Intermediation. 8( 112): 8-35. Levine, Ross, NormanLoayza, and Thorsten Beck. 1999. Financial Intermediationand Growth: Causality and Causes. UnpublishedManuscript. Development Research Group, World Bank, Washington. DC. 85 Lindeman, David. 1999. Tax Treatment o f Retirement Savings. HumanDevelopmentNetwork, World Bank, Washington DC. Lindeman, David and Russell Galer. 2000. Shifting Risks: The GrowingResponsibility o f the Individual inPension Provision. OECD, Paris, France. Lindeman, David, Michal Rutkowsky, and Oleksiy Sluchynskyy. 2001. The Evolution o f Pension Systems inEastern EuroDe and Central Asia: Otmortunities. Constraints, Dilemmas, and Emerging Practices. OECD Financial Market Trends. 80. Packard, Truman G. 1999. Extending Coverage o f Multipillar Pensions Systems: Hwotheses and Implications. World Bank, Washington DC. -. 2001. I s there a Positive Effect from Privatizing Social Security? Evidence from Latin America. UnpublishedManuscript. University o f Oxford. Department o f Economics. Palacios, Robert. 2002. Managing Public Pension Reserves: Part 11. Lessons from Five Recent OECD Initiatives. DiscussionPaper. No. 0219. Social Protection Network. World Bank, Washington. DC. Robalino, David and Massimo Sabbatini. 2002. Welfare Impact o f Alternative Pension Mechanisms in MENA Countries: The Case o fIranand Jordan. Mimeo. MNSHD, World Bank, Washington DC. Schwarz, Anita and Asli Demirguc-Kunt. 1999. Taking Stock o fPension Reforms Around the World. World Bank, Washington. DC. Social Security Investment Company. 2002. A Flourishing Trade is Infinite. SSIC. Teheran, Iran. Social Security Organization. 2000. The Social Security Organization inIslamic Republic o f Iran. SSO, Public Relations and International Affairs. Teheran, Iran. Social Security Organization. 2002a. A Guide to the Social Security InvestmentCompany and Its Subsidiary Companies. Teheran, Iran. -. 2002b. SSO Investment Policies. Translated Chapter from Annual Report. Teheran, Iran. Social Security Research Institute. 1997. Social Security inIran. SSO. Teheran, Iran. Subbarao, K.,A. Bonnejee, J. Braithwaite, S. Carvalho, K.Ezemenari, C. Graham, and A. Thompson. 1997. Safety Net Programs and Poverty Reduction: Lesson from Cross-country Experience. World Bank, Washington, DC. Valdes-Prieto. 1999. The Financial Stability o f Notional Accounts Pensions. Mimeo. Pontificia UniversidadCatolica de Chile, Santiago, Chle. Walker Eduardo and Fernando Lefort. 2000. Pension Reform and Capital Markets: Are There Any [Hard) Link? School o f Business Administration, Pontificia Universidad Catolica de Chile, Santiago, Chile. World Bank. 1994. Averting the Old Age Crisis: Policies to Protect the Old and Promote Growth. N e w Y o r k OxfordUniversity Press. Whtehouse, Edward. 2001. The Tax Treatment o f Funded Pensions. Unpublished manuscript. London: Axia Economics. Whitehouse, Edward. 2002. Pension Systems in 15 Countries Compared: The Value o f Entitlements. UNSW. Discussion Paper. 02104, Centre for Pensions and Superannuation. 86