This Selected Issues paper discusses Romania’s modeling monetary policy. A simple Forecasting and Policy Analysis System (FPAS) for Romania has been designed to help in the preparation of the IMF staff’s forecasts and policy assessments. A major advantage of this approach is that it allows the systematic and rapid analysis of different policy options. The model embodies the key principle that, in an inflation-targeting framework, the role of monetary policy is to provide an anchor for inflation and inflation expectations. The development and calibration of this model is an ongoing process.


This Selected Issues paper discusses Romania’s modeling monetary policy. A simple Forecasting and Policy Analysis System (FPAS) for Romania has been designed to help in the preparation of the IMF staff’s forecasts and policy assessments. A major advantage of this approach is that it allows the systematic and rapid analysis of different policy options. The model embodies the key principle that, in an inflation-targeting framework, the role of monetary policy is to provide an anchor for inflation and inflation expectations. The development and calibration of this model is an ongoing process.

III. The Impact of Aging on the Public Sector in Romania28

A. Introduction

84. This paper addresses the expected impact of aging on Romania’s public sector and, especially, its pension system. Unlike some previous work on this topic in Romania, this paper takes a more macroeconomic perspective, considering not only the effects on the pension system, but also on some other expenditures. Among the possible reforms, the paper considers, in addition to direct pension reform, some more general economic reforms that could support the sustainability of the pension system.

85. The paper finds that the impact of aging on the public finances will be large, and that reforms already under way, while helpful, are insufficient to ensure sustainability. Combining them with further increases in the retirement age and improvements in the employment rate could, however, achieve sustainability, provided revenue gains are not immediately used for pension increases. More generally, the paper argues in favor of a more predictable pension increases, following an agreed formula rather than yearly ad hoc decisions. The paper begins, in section B, by providing an overview of the demographic developments and projections in Romania and compares them to other economies. Section C describes the existing pension system, including its main features and the challenges it faces. This is followed, in section D, by a discussion of possible reforms, including direct pension and more general economic reforms. Section E briefly concludes.

B. Demographic Developments

86. The Romanian population is aging and shrinking, as a result of rising life expectancy and low fertility. At a level of 1.3 births per woman, the total fertility rate is far below the rate necessary for replacement of about 2.1. With life expectancy at birth of just 71.3 years, there is much scope for further improvement.

article image
Source: UN (2007)

87. Compared to its immediate neighbors, the Romanian demographic position is not unusual (Table 1). Dependency ratios and life expectancies are very similar across all neighbors. Fertility rates are higher in Serbia and Montenegro, but similar elsewhere, while emigration is almost twice as high in Serbia and Montenegro and in Ukraine. Interestingly, Hungary was a net immigration country, showing that not all transition economies lose population due to emigration. Employment rates vary more across countries, with the Moldavian rate almost 10 percentage points lower and the Ukrainian rate about 8 percentage points higher than the Romanian one.

Table 1:

Demographic statistics for Romania and selected other countries 1/

article image

Fertility rate and life expectancy data: 2004, all other data: 2005.

No separate data yet available for Serbia.

Sources: World Development Indicators (WDI) and World Economic Outlook (WEO).

88. Compared to the largest Western European economies and the U.S., however, there are important differences. Romania’s dependency ratios are about 5 percentage points lower, as a result of lower life expectancy of on average 7.3 years and lower fertility. Net migration is negative, thus exacerbating rather than counteracting (even if to a limited extent) the low fertility rate. The most striking difference is however that employment rates are much lower, by on average 13.7 percentage points. These comparisons can be summarized by noting that stock variables appear to be of a lesser concern in Romania, while flow variables suggest that changes are occurring at a faster pace.

89. The exceptionally low fertility rate is assumed to increase gradually over the medium-term, but without reaching replacement rates.29 According to the UN central population projection, the total fertility rate is expected to be almost the same as the Eastern European average, but below the average of any other European region. Projections by the US Census Bureau equally assume a slow increase in fertility rates, reaching 1.7 by 2050. Clearly these projections are crucial for the predicted population size, but because of the uncertainties of future fertility, they are also prone to large forecast errors. It is thus inevitable that population projections and all simulations based on them will be subject to a large margin of error and therefore more than one scenario will be considered.


Total fertility rate

Citation: IMF Staff Country Reports 2007, 220; 10.5089/9781451832884.002.A003

90. While the share of old-aged people is still relatively low in Romania, it is expected to exceed Northern, Eastern and, marginally, Western European averages by 2050. Compared to Southern Europe, however, the share will still be much lower. The dependency rate will increase slightly less, because the share of young dependents will fall as a result of low fertility.


Share of people aged 65 and over

Citation: IMF Staff Country Reports 2007, 220; 10.5089/9781451832884.002.A003

91. While population projections by various institutions differ substantially, all predict a significant shrinking of the Romanian population over the medium to longterm (Table 2). This already started 1991, although initially at a very slow pace. The rate of decrease will reach ½ percent of the population per year by around 2011 (UN central projection) or 2038 (US Bureau of the Census projection).

Table 2:

Population data and projections

(in millions)

article image
Sources: Romanian Authorities, National Institute of Statistics (2005), United Nations (2007) and US Bureau of the Census (2006).

C. The Pension System


92. Romania has an unfunded (or “pay as you go”) pension system, in which the currently working generation finances the pensions of the retired generation. A voluntary and a compulsory funded system are soon to be phased in, but will remain small in comparison to the unfunded component. The main features of the current system and imminent reforms are summarized in Box 1.

93. Contributions rates and pensions have periodically been adjusted by ad hoc political decisions rather than a pre-determined formula. An advantage of this is that it allows the government to weigh pension contribution increases and their benefits (higher pensions) directly against other tax and spending decisions, thus potentially improving efficiency. There are, however, also political economy risks, given that the legislative period is very short compared to the horizon of the pension system, so that unsustainable choices may appear attractive to legislators (and their myopic voters). In an aging society this may be a particularly risky, as the vote of the elderly will be increasingly important.30

Main features of the Romanian Public Pensions System

Pillar I: unfunded system

Retirement age and required years of contributions:

article image


Earnings of at least RON 318 per month and up to 5 times the gross average wage (RON 6350) are liable for pension contributions. The employee contributes 9.5 percent and the employer pays the balance to achieve the following overall rates:

  • Ordinary working conditions: 29 percent

  • Difficult working conditions: 34 percent

  • Most difficult working conditions: 39 percent

The average pension is worth about 1/3 of the average wage.

Pillar II: mandatory funded system

Operational from January 1, 2008.

Compulsory for employees up to 35 years of age and optional for employees aged between 35 and 45.

Contributions: 2 percentage points out of the total contribution in 2008, rising by 0.5 percentage points per year until 6 percentage points are reached in 2016.

Pillar III: voluntary funded system

Expected to become operational during 2007. Contributions are made out of gross earnings.

94. The planned introduction of funded pillars in addition to the existing unfunded system does not directly address the problems created by aging, but may be beneficial for other reasons. The advantages include that contributions may be less distortionary, because they would be considered (forced) savings rather than taxes by individuals. Moreover, under most circumstances the rate of return on funded systems is higher than on unfunded ones. A related advantage is that national savings may increase, which would help strengthen the current account. There will however also be costs, notably the cost of transition, during which retiring individuals will still qualify for full state pensions, but contributions will partially go towards the funded system. As the reform has not yet been implemented, the calculations below will consider both the implications of aging on the current system if it were to remain static, and on a system that is reformed according to the planned schedule.

95. As long as an unfunded element remains, there will be fiscal pressures as a result of aging. A higher funded element would increase the fiscal costs in the medium term, because of the need to finance the transition, but would reduce them in the longer term. According to current reform plans the unfunded part of the pension system will remain the most important of the three pillars. However, even if the unfunded element were fully abolished, some fiscal pressures would remain, as some individuals would not be able to build up sufficiently large pension funds and hence require welfare payments after retirement.

Sustainability of the pension system

96. A crucial analytical choice in the assessment of the sustainability of a pension system is which feature to keep constant. On one hand, one could assume a fixed contribution rate, and forecast the implied replacement rate based on assumptions about the developments of the population and the economy. On the other hand, one could assume a fixed replacement rate, and calculate the necessary contribution rate. In both cases one would then look at the calculated rate to see whether it could be an acceptable outcome. If not, then clearly the system is not sustainable and deeper reforms than simple adjustments of one of the rates are necessary. The choice between both approaches is normally given by the pension system, which often implies one of the two approaches in the absence of a major reform. In Romania, however, as both rates are subject to ad hoc political decisions the choice is less obvious. Politicians facing a deteriorating dependency ratio are most likely to make adjustments on both sides, to dampen negative effects on both working and retired generations.

97. To illustrate the implications of aging, this paper assumes that the government will keep the relative living standard of pensioners constant. This assumption reflects the low level of current pensions compared to earnings, which means that it would be politically difficult to allow pensions to fall back even further in relative terms. Specifically, the government is assumed to keep the ratio of pensions per old-aged person constant as a share of per capita income rather than fixing the contribution or replacement rate.31 An implication of this assumption is that, if employment rates rise, the additional contributions do not increase pensions of the average old-aged person. Further assumptions are detailed in Box 2. These assumptions aim to focus on the effect of demographic change and abstract from other changes. Hence education expenditure is assumed to grow with GDP, but to fall with the number of young people. While there could well be a case for increasing per-student expenditure in excess of GDP growth, this is a separate question, not addressed here, and would need to be financed by other expenditure cuts or tax increases. Table 3 presents the results, both under the assumption of the current system remaining static, and under the assumption that reforms are implemented as announced.

Assumptions underlying the baseline scenario of the simulation


  • - Pension contributions: constant as a share of GDP in no reform scenario. In the reform scenario, it is assumed that half of all individuals for whom the second pillar is optional will decide to enroll.

  • - Other revenues: constant as a share of GDP.32


  • - Pensions: increasing at the rate of increase of the number of old-aged people and per capita GDP. In the reform scenario, it is assumed that the real interest rate exceeds the real rate of economic growth by 50 percent. Hence, first pillar expenditure can be cut by more than the reduction in contributions, without leaving pensioners worse off.

  • - Education: falling at the rate of decrease of young people and increasing with GDP.

  • - Other non-interest: constant as a share of GDP.34

Other assumptions:

  • - Growth of GDP per working adult: as in macroeconomic framework until 2012, 5.5 percent in 2013, falling to 3 percent by 2050.

  • - Population growth: UN WPP central forecast, unless otherwise noted.

  • - Employment rate: constant (relaxed later).

Table 3:

Simulation based on current pension system and planned reforms

(percent of GDP)

article image
Source: Staff calculations, using data from the Romanian authorities, United Nations (2007), and WEO.

98. The results of the simulation are dramatic and reveal the necessity of an enormous fiscal effort if relative living standards of pensioners are to be kept. Initially, the worsening of the primary balance is modest, but after 2030 it accelerates in line with the much faster increase in ratio of old-aged people per working adult (Figure 1). The planned reforms would at first increase the strain on the public pensions system, but by 2050 a saving of 0.3 percent of GDP could be achieved, which would rise further over time until the transition is complete.35 As shown in Table 4, the cost will also depend on fertility, with a difference between the low and high fertility scenarios of about 1½ percent of GDP. Despite the long horizon over which this deterioration is expected to occur, the size of the effect warrants early reforms.

Table 4:

Primary balances under different fertiliy scenarios

(percent of GDP)

article image
Source: Staff calculations, using data from the Romanian authorities, United Nations (2007), and WEO.

D. Options for Further Reform

99. Ultimately, within an unfunded pension system, no reform can prevent that aging will lead to a reduction in pension benefits relative to contributions paid. A reform therefore cannot solve the fundamental problem, it can merely determine who will bear most of the burden (pensioners or contributors, current of future generations) and how the burden will be borne (higher contributions or longer working lives). These choices are very important, as they may have very different effects on economic efficiency.

Reform of the current unfunded pension system

100. Increasing retirement ages has proved to be the most effective parametric reform in improving the sustainability of unfunded pension systems.36 Such a reform works by both reducing the time of benefit entitlement and increasing the time of compulsory contributions. However, for this to work, it must be the case that the longer working years do not lead to actuarially larger entitlements. The current retirement ages are already scheduled to increase to 60 for women and 65 for men by 2013. The retirement age for women then reached will still be low by international standards, and differentiation of retirement age by sex is becoming increasingly rare. The retirement age for men will be typical, although other countries strongly affected by aging have legislated increases in retirement ages to 67 (e.g., Denmark, Germany, Norway, United States). The estimated improvement in the primary balance as a result of increasing retirement ages by one year are shown in Table 5. The savings in 2050 range from ½ percent of GDP if fertility turns out to be high and the introduction of the second pillar is implemented as planned to 1 percent of GDP if fertility is low and no second pillar is introduced.

Table 5:

Savings from increasing the retirement age by 1 year

(percent of GDP)

article image
Source: Staff calculations, using data from the Romanian authorities, United Nations (2007), and WEO.

101. Another useful reform would be to introduce a more predictable mechanism for contribution and benefit rates. This would improve the transparency of the existing system, allowing individuals to better gauge the value of pension benefits they will receive relative to their contributions. To ensure a stable mechanism that would not be in need of adjustment, such a formula would have to take demographic statistics into account. In the extreme case, each individual could have a notional account, in which case contributions would be perceived as investments rather than tax payments.37 However, even if the system remained actuarially unfair, e.g., by maintaining elements of redistribution, the benefits of contributions would still be more obvious to individuals than currently.

Further move towards a fully funded pension system

102. Despite the predicted benefits of the introduction of funded pillars, a complete abolition of the unfunded pillar would be risky. The negative effects of aging on pensions would remain under a funded system, as increasing longevity will reduce annuity revenues on retirement. Moreover, the much smaller younger cohorts will need to supply the relatively large group of old-aged people with non-tradable goods and services. This can be expected to put upward pressure on their prices and thus reduce the real value of pensions. Another disadvantage of a fully funded system would be that individuals would bear all investment risks.

Other economic reforms which affect the pension system

103. Increasing Romania’s very low employment rate would be a powerful way of increasing pension contributions of the current system. The current low employment rate of just 56 percent means that each working adult is supporting 0.38 old-aged people.38 Increasing the employment rate can counteract the increase in old-aged people due to aging. Norway, Europe’s best performer has an employment rate of just over 75 percent. If the Romanian rate is steadily increased to such a level by 2050, then the ratio of old aged-persons per working adult will increase only modestly until around 2032. Thereafter, a steep increase is unavoidable, and every working adult would have to support 0.7 old-aged people by 2050. Without an increase in employment rates, however, there would be almost one (0.95) old-aged person per working adult by 2050.

Figure 1.
Figure 1.

Old-aged persons per working adult

Citation: IMF Staff Country Reports 2007, 220; 10.5089/9781451832884.002.A003

104. An increase in the employment rate would greatly reduce the fiscal cost of aging, provided pension entitlements are not increased following the increased participation. If the employment rate were increased to 75 percent, the primary balance would improve by between about 2½ and 4 percent (Table 6). The benefit would be particularly large in case of the low fertility scenario, in which case it would also be particularly necessary. Examples of such policies include: improved education, including retraining for employees with obsolete skills; encouragement of female labor participation by subsidized child care provision; welfare reform to ensure strong work incentives; and promotion of general labor market flexibility.

Table 6:

Savings from increasing the employment rate to EU best performance

(percent of GDP)

article image
Source: Staff calculations, using data from the Romanian authorities, United Nations (2007), and WEO.

105. Policies aimed at increasing fertility rates could also support the sustainability of the pension system, although the gains would be modest over the medium term. As shown in Table 4, the effects of increasing fertility rates to the UN’s optimistic scenario would be small, and in the case of the reformed scenario almost negligible. This is because expenditure on education would rise immediately, while stronger contributions can only be expected when individuals reach working age. Nevertheless, a number of policies aimed at increasing fertility rates were recently introduced, including generous maternity leave, increased child benefit, allowances for large and one-parent families, and even payments for first marriages and newborns. Empirical evidence from other countries (Gauthier and Hatzius, 1997) suggests that such policies can be expected to have a statistically significant—though modest—effect on fertility.

E. Conclusion

106. The aging of the Romanian population poses a challenge for the public pensions system and fiscal sustainability. This is not insurmountable, particularly if reforms are tackled soon.

107. Reforms under way are helpful, but insufficient for fiscal sustainability over the long term. In addition to direct changes to the pension system, such as further increase in retirement ages, especially for women, other structural reforms, particularly aimed at increasing the employment rate, appear most promising.

108. A requirement for successful fiscal consolidation is that improvements in the revenue side are not immediately spent. If for example the increased pension fund revenues that would result from higher employment rates were immediately spent on current pensioners, fiscal sustainability would not improve. And even if savings were achieved inside the pension system, they would not improve overall fiscal sustainability, if they led to higher current spending in other budgetary areas, as would occur if government tried to achieve a fixed overall fiscal balance.


  • Chand, Sheetal K. and Albert Jaeger (1996) “Aging Populations and Public Pension Schemes” Occasional Paper 147, International Monetary Fund.

    • Search Google Scholar
    • Export Citation
  • Gauthier, Anne H. and Jan Hatzius (1997), “Family Benefits and Fertility: An Econometric Analysis”, Population Studies 51(3), pp. 295-306.

    • Search Google Scholar
    • Export Citation
  • Goudswaard, Kees and Hans van de Kar (2001) “The Impact of Demographic Change on Tax Revenue”, Atlantic Economic Journal 22(3), pp. 52-60.

    • Search Google Scholar
    • Export Citation
  • Hagemann, Robert P. and Giuseppe Nicoletti (1989) “Population Ageing: Economic Effects and Some Policy Implications for Financing Public Pensions” OECD Economic Studies 12, pp. 51-96.

    • Search Google Scholar
    • Export Citation
  • Kohler, Hans-Peter, Francesco C. Billari, and Jose Antonio Ortega (2002) “The Emergence of Lowest-Low Fertility in Europe during the 1990s” Population and Development Review 28(4), pp. 641-680.

    • Search Google Scholar
    • Export Citation
  • Kohler, Hans-Peter, and Jose Antonio Ortega (2002) “Tempo-adjusted period parity progression measures, fertility postponement and completed cohort fertility”, Demographic Research 6(6), pp. 91-144.

    • Search Google Scholar
    • Export Citation
  • Luski, I. and J. Weinblatt (1998) “A Dynamic Analysis of Fiscal Pressure and Demographic Transition”, Applied Economics 30, pp. 1431-1442.

    • Search Google Scholar
    • Export Citation
  • National Institute of Statistics (2005), National Statistical Yearbook 2004, Bucharest.

  • Profeta, Paola (2002), “Aging and Retirement: Evidence Across Countries” International Tax and Public Finance 9, pp. 651-672.

  • United Nations, Department of Economic and Social Affairs, Population Division, (2007), World Population Prospects: The 2006 Revision.

  • US Bureau of the Census, International Programs Center (2006), International Database.


Prepared by Alexander Klemm.


The most common fertility measure is the total fertility rate. This is calculated by first obtaining the fertility rates (births per woman) for population groups defined by five-year intervals, which are then aggregated. Another measure is the cohort fertility rate, which is calculated as the total number of children born by women of the same cohort. As this measure can only be calculated for cohorts past child-bearing age, it is often approximated by the total fertility rate. This will, however, provide a downward-biased projection if the average mother’s age at birth is increasing and vice versa (for details see inter alia Kohler et al. (2002), for attempts to adjust for this see, e.g., Kohler and Ortega (2002)).


Cross-country empirical evidence suggests that countries with large shares of elderly in the population tend to have more generous pension system (Profeta, 2002).


In case of stable employment ratios and wage shares in GDP, this would be equivalent to fixing the replacement rate.


Goudswaard and van de Kar (2001) argue that aging may have a positive effect on tax revenues, as wages and hence tax payments of older employees are higher because of career progression. In Romania, however, such an effect should be small, because of the low progressivity of the flat income tax.


For empirical underpinnings of these assumptions, see Luski and Weinblatt (1998). Using cross-sectional data they find that expenditure shares in GNI of education and welfare spending do not change with per-capita GDP (except in poor countries). They also estimate a proportional relation of both with the share of young and old people in the population. They cannot confirm any scale effects in the provision of education.


Some expenditures, notably health, may increase with the share of old people, while others will decrease.


This will occur when the last pensioner who has not contributed to the second pillar is deceased.


See for example simulations in Chand and Jaeger (1996).


An incentive to underreport income may continue to exist, as the rate of return on this investment may not be very high and not reflect an individual’s preferred use of resources, but it would be reduced.


This is thus an adjusted dependency ratio, as only employed people are counted in the denominator. The logic could be taken further, by including only those old-aged persons who are entitled to a pension in the numerator, as for example done in Hagemann and Nicoletti (1989). Including all old-aged persons however, has the advantage that it shows the future burden in very general terms, as even those people without pension benefits will have to be supported somehow by public funds, except those who have sufficient private wealth, which are unlikely to form a significant share.

Romania: Selected Issues
Author: International Monetary Fund