33381 World Bank PROST model Modeling pension reform The World Bank’s pension reform options simulation toolkit T oday’s pension policies can affect retirement incomes and the public finances for decades to come. Retirement income systems that may be af- programs can be politically popular yet the full financial impact is only realized over long periods of time, often generations. Pay-as-you-go schemes fordable today will often prove unsustainable in when first introduced, can generate net income for the future, given the twin pressures of demo- the government, because payroll tax revenues graphic aging and the maturing of pension exceed (initially modest) benefit expenditures. The schemes. true costs of the pension scheme only become The World Bank’s pension reform options simula- apparent in the medium- to long-term as the sys- tion toolkit (PROST) models pension tem matures. People often consider these future contributions, entitlements, system revenues and benefits as strict entitlements, making it politically system expenditures over a long timeframe into very difficult to adjust them in line with demo- the future. The model is designed to promote graphic and economic pressures. Pension benefits evidence-based policy-making, bridging the gap are sometimes even protected by the constitution. between quantitative and qualitative analysis of pension regimes. It is a flexible, computer-based toolkit, easily adapted to a wide range of countries’ circumstances. Policy makers and other key stakeholders have used PROST to evaluate existing schemes and reform options in more than 90 countries around the world (Figure 1). Why model pensions? Pay-as-you-go pension systems were often intro- duced with little or no analysis of the medium- and long-term effects on the elderly, on workers and PROST used in more than 90 countries 1 on the public finances. Once schemes are intro- duced, there remains a need for regular scrutiny of the impact on the pension system of factors such Countries such as Brazil—where three-quarters of as changing life expectancy and patterns of labor the fiscal deficit had been attributable to social force participation. security—are not isolated cases. The cost of paying for pensions crowds out spending on other Pensions policy-making can be dangerous without deserving programs: health, education etc. And long-term projections of system finances. Pension when payments exceed contribution revenues, For more information, please contact Social Protection, Human Development Network, World Bank, 1818 H Street NW, Washington, D.C. 20433; telephone +1 202 458 5267; fax +1 202 614 0471; e-mail prost@worldbank.org or visit the web site www.worldbank.org/pensions 2 Modeling pension reform cross-subsidies are required from broadly based closed population civil servant schemes or open taxes, such as value-added taxes. Pension benefits population national schemes. tend to go to a privileged minority that works in the covered, formal sector so that subsidies A special section in PROST analyses the impact of required to support such benefits can be highly pensions at an individual level. The user can regressive. explore the impact of the system and reforms on individual workers with different income levels, Pension modeling assesses schemes’ cash balances mortality rates, earnings profiles, job entry ages, over the long-term and thus determines system retirement patterns etc. sustainability. Modeling pensions can also rigorously evaluate different policy reforms, Population pyramids 2 informing both policy-makers and the public about the financial consequences of change. age Modeling pensions with PROST PROST is designed to answer the following kinds 1999 of question:  What will be the contributions, benefits, earnings and potential need for subsidies be over the long-term? Will the system be viable and sustainable in the long-run?  What kind of benefits can people expect to receive in the future? Will the pension system provide a decent retirement income to age different categories of people?  How large are the government’s implicit pen- 2050 sion liabilities?  How would broadening coverage, changing eligibility, changing benefits, or adjusting contribution rates affect the system? How will costs, expenditures and liabilities change under number of men number of women various reforms? The model utilizes country-specific data provided Population pyramids graphically illustrate by the user. It generates population projections changing demographics. In this example, (Figure 2), which, combined with economic the 1999 population is relatively young, with assumptions, are used to forecast future numbers a steep decline in the number of people by of contributors and beneficiaries. These in turn age. By 2050, the projected ‘pyramid’ is generate flows of revenues and expenditures. The much less steep, with fewer children and model then projects fiscal balances, taking account youth and many more elderly. Bulges in the of any partial pre-funding of liabilities. 2050 pyramid also show baby ‘booms’ and ‘busts’. The model can use either a ‘stock’ or a ‘flow’ ap- The model can accommodate a distribution of proach. In the stock concept, parameters such as wages per cohort which allows users to determine retirement are expressed as total retirees as a the effects of changes in floors and ceilings of percentage of population rather than as income subject to contribution and the effects of probabilities of retirement, since the stocks can be changes in the minimum and maximum pension more stable predictors of the future. Also, levels. projections can be based on either population or employment. The model can also be used for Modeling pension reform 3 The model can assess anything from ‘parametric’ A growing pension deficit… 3 reforms of initial pay-as-you-go systems— changing pensionable ages, contribution rates, expenditure benefits, indexation etc.—to structural reforms, such as the introduction of individual, funded retirement savings accounts or notional accounts deficit can also be modeled. PROST can handle provident fund schemes as well as pay-as-you-go systems as the starting point, before reform. revenue The model also allows for different transition paths to a new system, including the age cohorts (generations) covered by the new system (such as 2000 2010 2020 2030 2040 2050 2060 2070 applying reforms only to younger workers) and the treatment of pension rights accrued before the reform. Accrued rights can be paid in multiple ways, including as recognition bonds and as …means higher contributions… 4 proportional wages. On-going funded defined contribution schemes and notional accounts can contribution rate be modeled in PROST as well. needed for Model output balance The PROST program produces five output mod- ules, comprising Microsoft Excel tables with graphical summaries. The modules are: current  Population projections, including life tables, contribution population pyramids, population dependency rate (30%) ratios etc. (Figure 2).  Demographic structure: labor force and em- ployment, numbers of contributors and 2000 2010 2020 2030 2040 2050 2060 2070 beneficiaries, system dependency ratio.  Financial flows: projections of wages, bene- fits, revenues and expenditures of the pension …or lower benefits 5 system (Figure 3), pension scheme balance and pension replacement rate the implicit pension debt. The financial flows current system module also calculates the adjustments—to benefit levels or contribution rates—that would ‘balance’ the system, i.e. bring revenues and expenditures into line (Figures 4 and 5).  Fundamental, systemic reform: this module replacement rate illustrates the effect of a shift to a ‘multipillar’ needed for balance regime, incorporating both a pay-as-you-go, defined-benefit pension and a funded, defined contribution scheme or exclusively one or the other. Again, it measures the impact both on 2000 2010 2020 2030 2040 2050 2060 2070 the system finances and on individuals’ pension entitlements, including measurement of transition costs. The total pension benefit  Effects on example individuals: the model and the value of each of the pillars are works out contributions and benefits for dif- provided separately. ferent example individuals, specified by age, sex, age of labor market entry, retirement age, 4 Modeling pension reform earnings profile, mortality etc. With results for The model also relies on informed assumptions of six different example people per age cohort, future patterns including: the distributional effects of the current system  Wage growth and potential reforms can be assessed both  Real returns on pensions assets within and between generations.  Economic growth  Growth in coverage of a contributory pension All of the modules allow for analysis of the scheme sensitivity of results to key demographic and economic parameters, such as fertility, PROST users longevity, wage growth and interest rates. PROST is used by World Bank staff, country policy makers and other counterparts and Technical requirements development partners. PROST is policy oriented, allowing easy operation  PROST is generally easy to use, with training for technical novices. Data and assumptions are programs, clear and concise manuals, easily placed into Excel input file worksheets. documentation of underlying formulae, and Outputs are similarly presented as Excel troubleshooting, technical support. Model worksheets. Users require: assumptions are transparent and sensitivity  Microsoft Windows Operating Systems analysis is readily accessible.  Microsoft Office Excel 2003 or 2007.  The Bank distributes the model to country  Multi-lingual input/output is possible. counterparts engaged in a reform program and  Basic knowledge of Microsoft Excel. licenses its use to users attending a one week  User knowledge of basic economics, finance training program. The Bank carries out such and some actuarial concepts. training programs for client countries upon request.  The Bank also utilizes the model in the Data requirements formulation of policy options for The model is data intensive in order to support the consideration by country counterparts. robustness of its results. The key required data are:  The model’s quality improvement is a result  Population fertility and mortality rates by age of worldwide exposure and constant vetting by and gender. users. The World Bank constantly evaluates  Labor force participation rates, unemployment user feedback and regularly updates the rates by age and gender. program to uphold its state-of-the-art features,  Numbers of contributors and beneficiaries, functionality and adaptability. their contribution and retirement patterns by age and gender.  Wages and pensions by age and gender, income distribution for contributors and pensioners.